diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/Makefile.am | 5 | ||||
| -rw-r--r-- | src/tree.h | 27 | ||||
| -rw-r--r-- | src/ucl_chartable.h | 3 | ||||
| -rw-r--r-- | src/ucl_emitter.c | 136 | ||||
| -rw-r--r-- | src/ucl_emitter_utils.c | 8 | ||||
| -rw-r--r-- | src/ucl_hash.c | 171 | ||||
| -rw-r--r-- | src/ucl_internal.h | 155 | ||||
| -rw-r--r-- | src/ucl_msgpack.c | 1600 | ||||
| -rw-r--r-- | src/ucl_parser.c | 476 | ||||
| -rw-r--r-- | src/ucl_sexp.c | 224 | ||||
| -rw-r--r-- | src/ucl_util.c | 901 | ||||
| -rw-r--r-- | src/xxhash.c | 666 | ||||
| -rw-r--r-- | src/xxhash.h | 139 |
13 files changed, 4089 insertions, 422 deletions
diff --git a/src/Makefile.am b/src/Makefile.am index c3f0c9fffb60..c2f3a4b42347 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -11,6 +11,8 @@ libucl_la_SOURCES= ucl_emitter.c \ ucl_parser.c \ ucl_schema.c \ ucl_util.c \ + ucl_msgpack.c \ + ucl_sexp.c \ xxhash.c libucl_la_CFLAGS= $(libucl_common_cflags) \ @CURL_CFLAGS@ @@ -20,7 +22,8 @@ libucl_la_LIBADD= @LIBFETCH_LIBS@ \ @LIBREGEX_LIB@ \ @CURL_LIBS@ -include_HEADERS= $(top_srcdir)/include/ucl.h +include_HEADERS= $(top_srcdir)/include/ucl.h \ + $(top_srcdir)/include/ucl++.h noinst_HEADERS= ucl_internal.h \ xxhash.h \ ucl_hash.h \ diff --git a/src/tree.h b/src/tree.h index cee9373698c2..404b4a867be9 100644 --- a/src/tree.h +++ b/src/tree.h @@ -44,6 +44,13 @@ #define TREE_DELTA_MAX 1 +#ifndef _HU_FUNCTION +# if defined(__GNUC__) || defined(__clang__) +# define _HU_FUNCTION(x) __attribute__((__unused__)) x +# else +# define _HU_FUNCTION(x) x +# endif +#endif #define TREE_ENTRY(type) \ struct { \ @@ -68,9 +75,9 @@ #define TREE_DEFINE(node, field) \ \ - struct node *TREE_BALANCE_##node##_##field(struct node *); \ + static struct node *_HU_FUNCTION(TREE_BALANCE_##node##_##field)(struct node *); \ \ - struct node *TREE_ROTL_##node##_##field(struct node *self) \ + static struct node *_HU_FUNCTION(TREE_ROTL_##node##_##field)(struct node *self) \ { \ struct node *r= self->field.avl_right; \ self->field.avl_right= r->field.avl_left; \ @@ -78,7 +85,7 @@ return TREE_BALANCE_##node##_##field(r); \ } \ \ - struct node *TREE_ROTR_##node##_##field(struct node *self) \ + static struct node *_HU_FUNCTION(TREE_ROTR_##node##_##field)(struct node *self) \ { \ struct node *l= self->field.avl_left; \ self->field.avl_left= l->field.avl_right; \ @@ -86,7 +93,7 @@ return TREE_BALANCE_##node##_##field(l); \ } \ \ - struct node *TREE_BALANCE_##node##_##field(struct node *self) \ + static struct node *_HU_FUNCTION(TREE_BALANCE_##node##_##field)(struct node *self) \ { \ int delta= TREE_DELTA(self, field); \ \ @@ -111,7 +118,7 @@ return self; \ } \ \ - struct node *TREE_INSERT_##node##_##field \ + static struct node *_HU_FUNCTION(TREE_INSERT_##node##_##field) \ (struct node *self, struct node *elm, int (*compare)(struct node *lhs, struct node *rhs)) \ { \ if (!self) \ @@ -123,7 +130,7 @@ return TREE_BALANCE_##node##_##field(self); \ } \ \ - struct node *TREE_FIND_##node##_##field \ + static struct node *_HU_FUNCTION(TREE_FIND_##node##_##field) \ (struct node *self, struct node *elm, int (*compare)(struct node *lhs, struct node *rhs)) \ { \ if (!self) \ @@ -136,7 +143,7 @@ return TREE_FIND_##node##_##field(self->field.avl_right, elm, compare); \ } \ \ - struct node *TREE_MOVE_RIGHT(struct node *self, struct node *rhs) \ + static struct node *_HU_FUNCTION(TREE_MOVE_RIGHT)(struct node *self, struct node *rhs) \ { \ if (!self) \ return rhs; \ @@ -144,7 +151,7 @@ return TREE_BALANCE_##node##_##field(self); \ } \ \ - struct node *TREE_REMOVE_##node##_##field \ + static struct node *_HU_FUNCTION(TREE_REMOVE_##node##_##field) \ (struct node *self, struct node *elm, int (*compare)(struct node *lhs, struct node *rhs)) \ { \ if (!self) return 0; \ @@ -163,7 +170,7 @@ return TREE_BALANCE_##node##_##field(self); \ } \ \ - void TREE_FORWARD_APPLY_ALL_##node##_##field \ + static void _HU_FUNCTION(TREE_FORWARD_APPLY_ALL_##node##_##field) \ (struct node *self, void (*function)(struct node *node, void *data), void *data) \ { \ if (self) \ @@ -174,7 +181,7 @@ } \ } \ \ - void TREE_REVERSE_APPLY_ALL_##node##_##field \ + static void _HU_FUNCTION(TREE_REVERSE_APPLY_ALL_##node##_##field) \ (struct node *self, void (*function)(struct node *node, void *data), void *data) \ { \ if (self) \ diff --git a/src/ucl_chartable.h b/src/ucl_chartable.h index 5248e117c63d..db9f02900c02 100644 --- a/src/ucl_chartable.h +++ b/src/ucl_chartable.h @@ -26,7 +26,7 @@ #include "ucl_internal.h" -static const unsigned int ucl_chartable[255] = { +static const unsigned int ucl_chartable[256] = { UCL_CHARACTER_VALUE_END, UCL_CHARACTER_DENIED, UCL_CHARACTER_DENIED, UCL_CHARACTER_DENIED, UCL_CHARACTER_DENIED, UCL_CHARACTER_DENIED, UCL_CHARACTER_DENIED, UCL_CHARACTER_DENIED, @@ -255,6 +255,7 @@ UCL_CHARACTER_KEY_START|UCL_CHARACTER_KEY|UCL_CHARACTER_VALUE_STR, UCL_CHARACTER_KEY_START|UCL_CHARACTER_KEY|UCL_CHARACTER_VALUE_STR, UCL_CHARACTER_KEY_START|UCL_CHARACTER_KEY|UCL_CHARACTER_VALUE_STR, UCL_CHARACTER_KEY_START|UCL_CHARACTER_KEY|UCL_CHARACTER_VALUE_STR, +UCL_CHARACTER_KEY_START|UCL_CHARACTER_KEY|UCL_CHARACTER_VALUE_STR, UCL_CHARACTER_KEY_START|UCL_CHARACTER_KEY|UCL_CHARACTER_VALUE_STR }; diff --git a/src/ucl_emitter.c b/src/ucl_emitter.c index 9ddf3584a25b..8bfbf09b8bd7 100644 --- a/src/ucl_emitter.c +++ b/src/ucl_emitter.c @@ -62,6 +62,7 @@ UCL_EMIT_TYPE_OPS(json); UCL_EMIT_TYPE_OPS(json_compact); UCL_EMIT_TYPE_OPS(config); UCL_EMIT_TYPE_OPS(yaml); +UCL_EMIT_TYPE_OPS(msgpack); #define UCL_EMIT_TYPE_CONTENT(type) { \ .ucl_emitter_write_elt = ucl_emit_ ## type ## _elt, \ @@ -71,12 +72,12 @@ UCL_EMIT_TYPE_OPS(yaml); .ucl_emitter_end_array = ucl_emit_ ## type ##_end_array \ } - const struct ucl_emitter_operations ucl_standartd_emitter_ops[] = { [UCL_EMIT_JSON] = UCL_EMIT_TYPE_CONTENT(json), [UCL_EMIT_JSON_COMPACT] = UCL_EMIT_TYPE_CONTENT(json_compact), [UCL_EMIT_CONFIG] = UCL_EMIT_TYPE_CONTENT(config), - [UCL_EMIT_YAML] = UCL_EMIT_TYPE_CONTENT(yaml) + [UCL_EMIT_YAML] = UCL_EMIT_TYPE_CONTENT(yaml), + [UCL_EMIT_MSGPACK] = UCL_EMIT_TYPE_CONTENT(msgpack) }; /* @@ -469,19 +470,150 @@ UCL_EMIT_TYPE_IMPL(json_compact, true) UCL_EMIT_TYPE_IMPL(config, false) UCL_EMIT_TYPE_IMPL(yaml, false) +static void +ucl_emit_msgpack_elt (struct ucl_emitter_context *ctx, + const ucl_object_t *obj, bool first, bool print_key) +{ + ucl_object_iter_t it; + struct ucl_object_userdata *ud; + const char *ud_out; + const ucl_object_t *cur, *celt; + + switch (obj->type) { + case UCL_INT: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + ucl_emitter_print_int_msgpack (ctx, ucl_object_toint (obj)); + break; + + case UCL_FLOAT: + case UCL_TIME: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + ucl_emitter_print_double_msgpack (ctx, ucl_object_todouble (obj)); + break; + + case UCL_BOOLEAN: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + ucl_emitter_print_bool_msgpack (ctx, ucl_object_toboolean (obj)); + break; + + case UCL_STRING: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + + if (obj->flags & UCL_OBJECT_BINARY) { + ucl_emitter_print_binary_string_msgpack (ctx, obj->value.sv, + obj->len); + } + else { + ucl_emitter_print_string_msgpack (ctx, obj->value.sv, obj->len); + } + break; + + case UCL_NULL: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + ucl_emitter_print_null_msgpack (ctx); + break; + + case UCL_OBJECT: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + ucl_emit_msgpack_start_obj (ctx, obj, print_key); + it = NULL; + + while ((cur = ucl_iterate_object (obj, &it, true)) != NULL) { + LL_FOREACH (cur, celt) { + ucl_emit_msgpack_elt (ctx, celt, false, true); + /* XXX: + * in msgpack the length of objects is encoded within a single elt + * so in case of multi-value keys we are using merely the first + * element ignoring others + */ + break; + } + } + + break; + + case UCL_ARRAY: + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + ucl_emit_msgpack_start_array (ctx, obj, print_key); + it = NULL; + + while ((cur = ucl_iterate_object (obj, &it, true)) != NULL) { + ucl_emit_msgpack_elt (ctx, cur, false, false); + } + + break; + + case UCL_USERDATA: + ud = (struct ucl_object_userdata *)obj; + ucl_emitter_print_key_msgpack (print_key, ctx, obj); + + if (ud->emitter) { + ud_out = ud->emitter (obj->value.ud); + if (ud_out == NULL) { + ud_out = "null"; + } + } + ucl_emitter_print_string_msgpack (ctx, obj->value.sv, obj->len); + break; + } +} + +static void +ucl_emit_msgpack_start_obj (struct ucl_emitter_context *ctx, + const ucl_object_t *obj, bool print_key) +{ + ucl_emitter_print_object_msgpack (ctx, obj->len); +} + +static void +ucl_emit_msgpack_start_array (struct ucl_emitter_context *ctx, + const ucl_object_t *obj, bool print_key) +{ + ucl_emitter_print_array_msgpack (ctx, obj->len); +} + +static void +ucl_emit_msgpack_end_object (struct ucl_emitter_context *ctx, + const ucl_object_t *obj) +{ + +} + +static void +ucl_emit_msgpack_end_array (struct ucl_emitter_context *ctx, + const ucl_object_t *obj) +{ + +} + unsigned char * ucl_object_emit (const ucl_object_t *obj, enum ucl_emitter emit_type) { + return ucl_object_emit_len (obj, emit_type, NULL); +} + +unsigned char * +ucl_object_emit_len (const ucl_object_t *obj, enum ucl_emitter emit_type, + size_t *outlen) +{ unsigned char *res = NULL; struct ucl_emitter_functions *func; + UT_string *s; + if (obj == NULL) { return NULL; } func = ucl_object_emit_memory_funcs ((void **)&res); + s = func->ud; if (func != NULL) { ucl_object_emit_full (obj, emit_type, func); + + if (outlen != NULL) { + *outlen = s->i; + } + ucl_object_emit_funcs_free (func); } diff --git a/src/ucl_emitter_utils.c b/src/ucl_emitter_utils.c index 91cad78bcbf6..95ac9a5d5776 100644 --- a/src/ucl_emitter_utils.c +++ b/src/ucl_emitter_utils.c @@ -62,6 +62,12 @@ static const struct ucl_emitter_context ucl_standard_emitters[] = { .id = UCL_EMIT_YAML, .func = NULL, .ops = &ucl_standartd_emitter_ops[UCL_EMIT_YAML] + }, + [UCL_EMIT_MSGPACK] = { + .name = "msgpack", + .id = UCL_EMIT_MSGPACK, + .func = NULL, + .ops = &ucl_standartd_emitter_ops[UCL_EMIT_MSGPACK] } }; @@ -73,7 +79,7 @@ static const struct ucl_emitter_context ucl_standard_emitters[] = { const struct ucl_emitter_context * ucl_emit_get_standard_context (enum ucl_emitter emit_type) { - if (emit_type >= UCL_EMIT_JSON && emit_type <= UCL_EMIT_YAML) { + if (emit_type >= UCL_EMIT_JSON && emit_type < UCL_EMIT_MAX) { return &ucl_standard_emitters[emit_type]; } diff --git a/src/ucl_hash.c b/src/ucl_hash.c index 275e84d478aa..7de8196fa014 100644 --- a/src/ucl_hash.c +++ b/src/ucl_hash.c @@ -26,6 +26,9 @@ #include "khash.h" #include "kvec.h" +#include <time.h> +#include <limits.h> + struct ucl_hash_elt { const ucl_object_t *obj; size_t ar_idx; @@ -37,11 +40,78 @@ struct ucl_hash_struct { bool caseless; }; +static uint64_t +ucl_hash_seed (void) +{ + static uint64_t seed; + + if (seed == 0) { +#ifdef UCL_RANDOM_FUNCTION + seed = UCL_RANDOM_FUNCTION; +#else + /* Not very random but can be useful for our purposes */ + seed = time (NULL); +#endif + } + + return seed; +} + +static const unsigned char lc_map[256] = { + 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, + 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, + 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, + 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, + 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, + 0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, + 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, + 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, + 0x40, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, + 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, + 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, + 0x78, 0x79, 0x7a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, + 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, + 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, + 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, + 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, + 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, + 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f, + 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, + 0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, + 0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, + 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, + 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, + 0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, + 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7, + 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, + 0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, + 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf, + 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, + 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, + 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, + 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff +}; + +#if (defined(WORD_BIT) && WORD_BIT == 64) || \ + (defined(__WORDSIZE) && __WORDSIZE == 64) || \ + defined(__x86_64__) || \ + defined(__amd64__) +#define UCL64_BIT_HASH 1 +#endif + +#ifdef UCL64_BIT_HASH +static inline uint32_t +ucl_hash_func (const ucl_object_t *o) +{ + return XXH64 (o->key, o->keylen, ucl_hash_seed ()); +} +#else static inline uint32_t ucl_hash_func (const ucl_object_t *o) { - return XXH32 (o->key, o->keylen, 0xdeadbeef); + return XXH32 (o->key, o->keylen, ucl_hash_seed ()); } +#endif static inline int ucl_hash_equal (const ucl_object_t *k1, const ucl_object_t *k2) @@ -56,33 +126,91 @@ ucl_hash_equal (const ucl_object_t *k1, const ucl_object_t *k2) KHASH_INIT (ucl_hash_node, const ucl_object_t *, struct ucl_hash_elt, 1, ucl_hash_func, ucl_hash_equal) +#ifdef UCL64_BIT_HASH static inline uint32_t ucl_hash_caseless_func (const ucl_object_t *o) { - void *xxh = XXH32_init (0xdeadbeef); - char hash_buf[64], *c; - const char *p; - ssize_t remain = o->keylen; + unsigned len = o->keylen; + unsigned leftover = o->keylen % 4; + unsigned fp, i; + const uint8_t* s = (const uint8_t*)o->key; + union { + struct { + unsigned char c1, c2, c3, c4; + } c; + uint32_t pp; + } u; + XXH64_state_t st; - p = o->key; - c = &hash_buf[0]; + fp = len - leftover; + XXH64_reset (&st, ucl_hash_seed ()); - while (remain > 0) { - *c++ = tolower (*p++); + for (i = 0; i != fp; i += 4) { + u.c.c1 = s[i], u.c.c2 = s[i + 1], u.c.c3 = s[i + 2], u.c.c4 = s[i + 3]; + u.c.c1 = lc_map[u.c.c1]; + u.c.c2 = lc_map[u.c.c2]; + u.c.c3 = lc_map[u.c.c3]; + u.c.c4 = lc_map[u.c.c4]; + XXH64_update (&st, &u.pp, sizeof (u)); + } - if (c - &hash_buf[0] == sizeof (hash_buf)) { - XXH32_update (xxh, hash_buf, sizeof (hash_buf)); - c = &hash_buf[0]; - } - remain --; + u.pp = 0; + switch (leftover) { + case 3: + u.c.c3 = lc_map[(unsigned char)s[i++]]; + case 2: + u.c.c2 = lc_map[(unsigned char)s[i++]]; + case 1: + u.c.c1 = lc_map[(unsigned char)s[i]]; + XXH64_update (&st, &u.pp, leftover); + break; } - if (c - &hash_buf[0] != 0) { - XXH32_update (xxh, hash_buf, c - &hash_buf[0]); + return XXH64_digest (&st); +} +#else +static inline uint32_t +ucl_hash_caseless_func (const ucl_object_t *o) +{ + unsigned len = o->keylen; + unsigned leftover = o->keylen % 4; + unsigned fp, i; + const uint8_t* s = (const uint8_t*)o->key; + union { + struct { + unsigned char c1, c2, c3, c4; + } c; + uint32_t pp; + } u; + XXH32_state_t st; + + fp = len - leftover; + XXH32_reset (&st, ucl_hash_seed ()); + + for (i = 0; i != fp; i += 4) { + u.c.c1 = s[i], u.c.c2 = s[i + 1], u.c.c3 = s[i + 2], u.c.c4 = s[i + 3]; + u.c.c1 = lc_map[u.c.c1]; + u.c.c2 = lc_map[u.c.c2]; + u.c.c3 = lc_map[u.c.c3]; + u.c.c4 = lc_map[u.c.c4]; + XXH32_update (&st, &u.pp, sizeof (u)); + } + + u.pp = 0; + switch (leftover) { + case 3: + u.c.c3 = lc_map[(unsigned char)s[i++]]; + case 2: + u.c.c2 = lc_map[(unsigned char)s[i++]]; + case 1: + u.c.c1 = lc_map[(unsigned char)s[i]]; + XXH32_update (&st, &u.pp, leftover); + break; } - return XXH32_digest (xxh); + return XXH32_digest (&st); } +#endif static inline int ucl_hash_caseless_equal (const ucl_object_t *k1, const ucl_object_t *k2) @@ -254,6 +382,11 @@ ucl_hash_iterate (ucl_hash_t *hashlin, ucl_hash_iter_t *iter) if (it == NULL) { it = UCL_ALLOC (sizeof (*it)); + + if (it == NULL) { + return NULL; + } + it->cur = &hashlin->ar.a[0]; it->end = it->cur + hashlin->ar.n; } @@ -336,7 +469,7 @@ ucl_hash_delete (ucl_hash_t* hashlin, const ucl_object_t *obj) k = kh_get (ucl_hash_caseless_node, h, obj); if (k != kh_end (h)) { elt = &kh_value (h, k); - kv_A (hashlin->ar, elt->ar_idx) = NULL; + kv_del (const ucl_object_t *, hashlin->ar, elt->ar_idx); kh_del (ucl_hash_caseless_node, h, k); } } @@ -346,7 +479,7 @@ ucl_hash_delete (ucl_hash_t* hashlin, const ucl_object_t *obj) k = kh_get (ucl_hash_node, h, obj); if (k != kh_end (h)) { elt = &kh_value (h, k); - kv_A (hashlin->ar, elt->ar_idx) = NULL; + kv_del (const ucl_object_t *, hashlin->ar, elt->ar_idx); kh_del (ucl_hash_node, h, k); } } diff --git a/src/ucl_internal.h b/src/ucl_internal.h index bdbe691d092b..31d6b1361383 100644 --- a/src/ucl_internal.h +++ b/src/ucl_internal.h @@ -145,15 +145,19 @@ enum ucl_character_type { struct ucl_macro { char *name; - ucl_macro_handler handler; + union { + ucl_macro_handler handler; + ucl_context_macro_handler context_handler; + } h; void* ud; + bool is_context; UT_hash_handle hh; }; struct ucl_stack { ucl_object_t *obj; struct ucl_stack *next; - int level; + uint64_t level; }; struct ucl_chunk { @@ -164,6 +168,8 @@ struct ucl_chunk { unsigned int line; unsigned int column; unsigned priority; + enum ucl_duplicate_strategy strategy; + enum ucl_parse_type parse_type; struct ucl_chunk *next; }; @@ -191,8 +197,12 @@ struct ucl_parser { enum ucl_parser_state prev_state; unsigned int recursion; int flags; + unsigned default_priority; + int err_code; ucl_object_t *top_obj; ucl_object_t *cur_obj; + ucl_object_t *trash_objs; + ucl_object_t *includepaths; char *cur_file; struct ucl_macro *macroes; struct ucl_stack *stack; @@ -220,13 +230,21 @@ size_t ucl_unescape_json_string (char *str, size_t len); * Handle include macro * @param data include data * @param len length of data + * @param args UCL object representing arguments to the macro * @param ud user data - * @param err error ptr * @return */ bool ucl_include_handler (const unsigned char *data, size_t len, const ucl_object_t *args, void* ud); +/** + * Handle tryinclude macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ud user data + * @return + */ bool ucl_try_include_handler (const unsigned char *data, size_t len, const ucl_object_t *args, void* ud); @@ -234,17 +252,52 @@ bool ucl_try_include_handler (const unsigned char *data, size_t len, * Handle includes macro * @param data include data * @param len length of data + * @param args UCL object representing arguments to the macro * @param ud user data - * @param err error ptr * @return */ bool ucl_includes_handler (const unsigned char *data, size_t len, const ucl_object_t *args, void* ud); +/** + * Handle priority macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ud user data + * @return + */ +bool ucl_priority_handler (const unsigned char *data, size_t len, + const ucl_object_t *args, void* ud); + +/** + * Handle load macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ud user data + * @return + */ +bool ucl_load_handler (const unsigned char *data, size_t len, + const ucl_object_t *args, void* ud); +/** + * Handle inherit macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ctx the current context object + * @param ud user data + * @return + */ +bool ucl_inherit_handler (const unsigned char *data, size_t len, + const ucl_object_t *args, const ucl_object_t *ctx, void* ud); + size_t ucl_strlcpy (char *dst, const char *src, size_t siz); size_t ucl_strlcpy_unsafe (char *dst, const char *src, size_t siz); size_t ucl_strlcpy_tolower (char *dst, const char *src, size_t siz); +char *ucl_strnstr (const char *s, const char *find, int len); +char *ucl_strncasestr (const char *s, const char *find, int len); #ifdef __GNUC__ static inline void @@ -252,9 +305,10 @@ ucl_create_err (UT_string **err, const char *fmt, ...) __attribute__ (( format( printf, 2, 3) )); #endif +#undef UCL_FATAL_ERRORS + static inline void ucl_create_err (UT_string **err, const char *fmt, ...) - { if (*err == NULL) { utstring_new (*err); @@ -263,6 +317,10 @@ ucl_create_err (UT_string **err, const char *fmt, ...) utstring_printf_va (*err, fmt, ap); va_end (ap); } + +#ifdef UCL_FATAL_ERRORS + assert (0); +#endif } /** @@ -311,7 +369,7 @@ ucl_maybe_parse_boolean (ucl_object_t *obj, const unsigned char *start, size_t l } } - if (ret) { + if (ret && obj != NULL) { obj->type = UCL_BOOLEAN; obj->value.iv = val; } @@ -396,4 +454,89 @@ unsigned char * ucl_object_emit_single_json (const ucl_object_t *obj); */ bool ucl_maybe_long_string (const ucl_object_t *obj); +/** + * Print integer to the msgpack output + * @param ctx + * @param val + */ +void ucl_emitter_print_int_msgpack (struct ucl_emitter_context *ctx, + int64_t val); +/** + * Print integer to the msgpack output + * @param ctx + * @param val + */ +void ucl_emitter_print_double_msgpack (struct ucl_emitter_context *ctx, + double val); +/** + * Print double to the msgpack output + * @param ctx + * @param val + */ +void ucl_emitter_print_bool_msgpack (struct ucl_emitter_context *ctx, + bool val); +/** + * Print string to the msgpack output + * @param ctx + * @param s + * @param len + */ +void ucl_emitter_print_string_msgpack (struct ucl_emitter_context *ctx, + const char *s, size_t len); + +/** + * Print binary string to the msgpack output + * @param ctx + * @param s + * @param len + */ +void ucl_emitter_print_binary_string_msgpack (struct ucl_emitter_context *ctx, + const char *s, size_t len); + +/** + * Print array preamble for msgpack + * @param ctx + * @param len + */ +void ucl_emitter_print_array_msgpack (struct ucl_emitter_context *ctx, + size_t len); + +/** + * Print object preamble for msgpack + * @param ctx + * @param len + */ +void ucl_emitter_print_object_msgpack (struct ucl_emitter_context *ctx, + size_t len); +/** + * Print NULL to the msgpack output + * @param ctx + */ +void ucl_emitter_print_null_msgpack (struct ucl_emitter_context *ctx); +/** + * Print object's key if needed to the msgpack output + * @param print_key + * @param ctx + * @param obj + */ +void ucl_emitter_print_key_msgpack (bool print_key, + struct ucl_emitter_context *ctx, + const ucl_object_t *obj); + +/** + * Add new element to an object using the current merge strategy and priority + * @param parser + * @param nobj + * @return + */ +bool ucl_parser_process_object_element (struct ucl_parser *parser, + ucl_object_t *nobj); + +/** + * Parse msgpack chunk + * @param parser + * @return + */ +bool ucl_parse_msgpack (struct ucl_parser *parser); + #endif /* UCL_INTERNAL_H_ */ diff --git a/src/ucl_msgpack.c b/src/ucl_msgpack.c new file mode 100644 index 000000000000..e8ebfba8a35e --- /dev/null +++ b/src/ucl_msgpack.c @@ -0,0 +1,1600 @@ +/* + * Copyright (c) 2015, Vsevolod Stakhov + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are met: + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY AUTHOR ''AS IS'' AND ANY + * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + * DISCLAIMED. IN NO EVENT SHALL AUTHOR BE LIABLE FOR ANY + * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include "ucl.h" +#include "ucl_internal.h" + +#ifdef HAVE_ENDIAN_H +#include <endian.h> +#elif defined(HAVE_SYS_ENDIAN_H) +#include <sys/endian.h> +#elif defined(HAVE_MACHINE_ENDIAN_H) +#include <machine/endian.h> +#endif + +#if !defined(__LITTLE_ENDIAN__) && !defined(__BIG_ENDIAN__) + #if __BYTE_ORDER == __LITTLE_ENDIAN + #define __LITTLE_ENDIAN__ + #elif __BYTE_ORDER == __BIG_ENDIAN + #define __BIG_ENDIAN__ + #elif _WIN32 + #define __LITTLE_ENDIAN__ + #endif +#endif + +#define SWAP_LE_BE16(val) ((uint16_t) ( \ + (uint16_t) ((uint16_t) (val) >> 8) | \ + (uint16_t) ((uint16_t) (val) << 8))) + +#if defined(__clang__) || (defined(__GNUC__) && __GNUC__ >= 4 && defined (__GNUC_MINOR__) && __GNUC_MINOR__ >= 3) +# define SWAP_LE_BE32(val) ((uint32_t)__builtin_bswap32 ((uint32_t)(val))) +# define SWAP_LE_BE64(val) ((uint64_t)__builtin_bswap64 ((uint64_t)(val))) +#else + #define SWAP_LE_BE32(val) ((uint32_t)( \ + (((uint32_t)(val) & (uint32_t)0x000000ffU) << 24) | \ + (((uint32_t)(val) & (uint32_t)0x0000ff00U) << 8) | \ + (((uint32_t)(val) & (uint32_t)0x00ff0000U) >> 8) | \ + (((uint32_t)(val) & (uint32_t)0xff000000U) >> 24))) + + #define SWAP_LE_BE64(val) ((uint64_t)( \ + (((uint64_t)(val) & \ + (uint64_t)(0x00000000000000ffULL)) << 56) | \ + (((uint64_t)(val) & \ + (uint64_t)(0x000000000000ff00ULL)) << 40) | \ + (((uint64_t)(val) & \ + (uint64_t)(0x0000000000ff0000ULL)) << 24) | \ + (((uint64_t)(val) & \ + (uint64_t) (0x00000000ff000000ULL)) << 8) | \ + (((uint64_t)(val) & \ + (uint64_t)(0x000000ff00000000ULL)) >> 8) | \ + (((uint64_t)(val) & \ + (uint64_t)(0x0000ff0000000000ULL)) >> 24) | \ + (((uint64_t)(val) & \ + (uint64_t)(0x00ff000000000000ULL)) >> 40) | \ + (((uint64_t)(val) & \ + (uint64_t)(0xff00000000000000ULL)) >> 56))) +#endif + +#ifdef __LITTLE_ENDIAN__ +#define TO_BE16 SWAP_LE_BE16 +#define TO_BE32 SWAP_LE_BE32 +#define TO_BE64 SWAP_LE_BE64 +#define FROM_BE16 SWAP_LE_BE16 +#define FROM_BE32 SWAP_LE_BE32 +#define FROM_BE64 SWAP_LE_BE64 +#else +#define TO_BE16(val) (uint16_t)(val) +#define TO_BE32(val) (uint32_t)(val) +#define TO_BE64(val) (uint64_t)(val) +#define FROM_BE16(val) (uint16_t)(val) +#define FROM_BE32(val) (uint32_t)(val) +#define FROM_BE64(val) (uint64_t)(val) +#endif + +void +ucl_emitter_print_int_msgpack (struct ucl_emitter_context *ctx, int64_t val) +{ + const struct ucl_emitter_functions *func = ctx->func; + unsigned char buf[sizeof(uint64_t) + 1]; + const unsigned char mask_positive = 0x7f, mask_negative = 0xe0, + uint8_ch = 0xcc, uint16_ch = 0xcd, uint32_ch = 0xce, uint64_ch = 0xcf, + int8_ch = 0xd0, int16_ch = 0xd1, int32_ch = 0xd2, int64_ch = 0xd3; + unsigned len; + + if (val >= 0) { + if (val <= 0x7f) { + /* Fixed num 7 bits */ + len = 1; + buf[0] = mask_positive & val; + } + else if (val <= 0xff) { + len = 2; + buf[0] = uint8_ch; + buf[1] = val & 0xff; + } + else if (val <= 0xffff) { + uint16_t v = TO_BE16 (val); + + len = 3; + buf[0] = uint16_ch; + memcpy (&buf[1], &v, sizeof (v)); + } + else if (val <= 0xffffffff) { + uint32_t v = TO_BE32 (val); + + len = 5; + buf[0] = uint32_ch; + memcpy (&buf[1], &v, sizeof (v)); + } + else { + uint64_t v = TO_BE64 (val); + + len = 9; + buf[0] = uint64_ch; + memcpy (&buf[1], &v, sizeof (v)); + } + } + else { + uint64_t uval; + /* Bithack abs */ + uval = ((val ^ (val >> 63)) - (val >> 63)); + + if (val > -(1 << 5)) { + len = 1; + buf[0] = (mask_negative | uval) & 0xff; + } + else if (uval <= 0xff) { + len = 2; + buf[0] = int8_ch; + buf[1] = (unsigned char)val; + } + else if (uval <= 0xffff) { + uint16_t v = TO_BE16 (val); + + len = 3; + buf[0] = int16_ch; + memcpy (&buf[1], &v, sizeof (v)); + } + else if (uval <= 0xffffffff) { + uint32_t v = TO_BE32 (val); + + len = 5; + buf[0] = int32_ch; + memcpy (&buf[1], &v, sizeof (v)); + } + else { + uint64_t v = TO_BE64 (val); + + len = 9; + buf[0] = int64_ch; + memcpy (&buf[1], &v, sizeof (v)); + } + } + + func->ucl_emitter_append_len (buf, len, func->ud); +} + +void +ucl_emitter_print_double_msgpack (struct ucl_emitter_context *ctx, double val) +{ + const struct ucl_emitter_functions *func = ctx->func; + union { + double d; + uint64_t i; + } u; + const unsigned char dbl_ch = 0xcb; + unsigned char buf[sizeof(double) + 1]; + + /* Convert to big endian */ + u.d = val; + u.i = TO_BE64 (u.i); + + buf[0] = dbl_ch; + memcpy (&buf[1], &u.d, sizeof (double)); + func->ucl_emitter_append_len (buf, sizeof (buf), func->ud); +} + +void +ucl_emitter_print_bool_msgpack (struct ucl_emitter_context *ctx, bool val) +{ + const struct ucl_emitter_functions *func = ctx->func; + const unsigned char true_ch = 0xc3, false_ch = 0xc2; + + func->ucl_emitter_append_character (val ? true_ch : false_ch, 1, func->ud); +} + +void +ucl_emitter_print_string_msgpack (struct ucl_emitter_context *ctx, + const char *s, size_t len) +{ + const struct ucl_emitter_functions *func = ctx->func; + const unsigned char fix_mask = 0xA0, l8_ch = 0xd9, l16_ch = 0xda, l32_ch = 0xdb; + unsigned char buf[5]; + unsigned blen; + + if (len <= 0x1F) { + blen = 1; + buf[0] = (len | fix_mask) & 0xff; + } + else if (len <= 0xff) { + blen = 2; + buf[0] = l8_ch; + buf[1] = len & 0xff; + } + else if (len <= 0xffff) { + uint16_t bl = TO_BE16 (len); + + blen = 3; + buf[0] = l16_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + else { + uint32_t bl = TO_BE32 (len); + + blen = 5; + buf[0] = l32_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + + func->ucl_emitter_append_len (buf, blen, func->ud); + func->ucl_emitter_append_len (s, len, func->ud); +} + +void +ucl_emitter_print_binary_string_msgpack (struct ucl_emitter_context *ctx, + const char *s, size_t len) +{ + const struct ucl_emitter_functions *func = ctx->func; + const unsigned char l8_ch = 0xc4, l16_ch = 0xc5, l32_ch = 0xc6; + unsigned char buf[5]; + unsigned blen; + + if (len <= 0xff) { + blen = 2; + buf[0] = l8_ch; + buf[1] = len & 0xff; + } + else if (len <= 0xffff) { + uint16_t bl = TO_BE16 (len); + + blen = 3; + buf[0] = l16_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + else { + uint32_t bl = TO_BE32 (len); + + blen = 5; + buf[0] = l32_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + + func->ucl_emitter_append_len (buf, blen, func->ud); + func->ucl_emitter_append_len (s, len, func->ud); +} + +void +ucl_emitter_print_null_msgpack (struct ucl_emitter_context *ctx) +{ + const struct ucl_emitter_functions *func = ctx->func; + const unsigned char nil = 0xc0; + + func->ucl_emitter_append_character (nil, 1, func->ud); +} + +void +ucl_emitter_print_key_msgpack (bool print_key, struct ucl_emitter_context *ctx, + const ucl_object_t *obj) +{ + if (print_key) { + ucl_emitter_print_string_msgpack (ctx, obj->key, obj->keylen); + } +} + +void +ucl_emitter_print_array_msgpack (struct ucl_emitter_context *ctx, size_t len) +{ + const struct ucl_emitter_functions *func = ctx->func; + const unsigned char fix_mask = 0x90, l16_ch = 0xdc, l32_ch = 0xdd; + unsigned char buf[5]; + unsigned blen; + + if (len <= 0xF) { + blen = 1; + buf[0] = (len | fix_mask) & 0xff; + } + else if (len <= 0xffff) { + uint16_t bl = TO_BE16 (len); + + blen = 3; + buf[0] = l16_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + else { + uint32_t bl = TO_BE32 (len); + + blen = 5; + buf[0] = l32_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + + func->ucl_emitter_append_len (buf, blen, func->ud); +} + +void +ucl_emitter_print_object_msgpack (struct ucl_emitter_context *ctx, size_t len) +{ + const struct ucl_emitter_functions *func = ctx->func; + const unsigned char fix_mask = 0x80, l16_ch = 0xde, l32_ch = 0xdf; + unsigned char buf[5]; + unsigned blen; + + if (len <= 0xF) { + blen = 1; + buf[0] = (len | fix_mask) & 0xff; + } + else if (len <= 0xffff) { + uint16_t bl = TO_BE16 (len); + + blen = 3; + buf[0] = l16_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + else { + uint32_t bl = TO_BE32 (len); + + blen = 5; + buf[0] = l32_ch; + memcpy (&buf[1], &bl, sizeof (bl)); + } + + func->ucl_emitter_append_len (buf, blen, func->ud); +} + + +enum ucl_msgpack_format { + msgpack_positive_fixint = 0, + msgpack_fixmap, + msgpack_fixarray, + msgpack_fixstr, + msgpack_nil, + msgpack_false, + msgpack_true, + msgpack_bin8, + msgpack_bin16, + msgpack_bin32, + msgpack_ext8, + msgpack_ext16, + msgpack_ext32, + msgpack_float32, + msgpack_float64, + msgpack_uint8, + msgpack_uint16, + msgpack_uint32, + msgpack_uint64, + msgpack_int8, + msgpack_int16, + msgpack_int32, + msgpack_int64, + msgpack_fixext1, + msgpack_fixext2, + msgpack_fixext4, + msgpack_fixext8, + msgpack_fixext16, + msgpack_str8, + msgpack_str16, + msgpack_str32, + msgpack_array16, + msgpack_array32, + msgpack_map16, + msgpack_map32, + msgpack_negative_fixint, + msgpack_invalid +}; + +typedef ssize_t (*ucl_msgpack_parse_function)(struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); + +static ssize_t ucl_msgpack_parse_map (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_array (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_string (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_int (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_float (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_bool (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_null (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); +static ssize_t ucl_msgpack_parse_ignore (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain); + +#define MSGPACK_FLAG_FIXED (1 << 0) +#define MSGPACK_FLAG_CONTAINER (1 << 1) +#define MSGPACK_FLAG_TYPEVALUE (1 << 2) +#define MSGPACK_FLAG_EXT (1 << 3) +#define MSGPACK_FLAG_ASSOC (1 << 4) +#define MSGPACK_FLAG_KEY (1 << 5) +#define MSGPACK_CONTAINER_BIT (1ULL << 62) + +/* + * Search tree packed in array + */ +struct ucl_msgpack_parser { + uint8_t prefix; /* Prefix byte */ + uint8_t prefixlen; /* Length of prefix in bits */ + uint8_t fmt; /* The desired format */ + uint8_t len; /* Length of the object + (either length bytes + or length of value in case + of fixed objects */ + uint8_t flags; /* Flags of the specified type */ + ucl_msgpack_parse_function func; /* Parser function */ +} parsers[] = { + { + 0xa0, + 3, + msgpack_fixstr, + 0, + MSGPACK_FLAG_FIXED|MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0x0, + 1, + msgpack_positive_fixint, + 0, + MSGPACK_FLAG_FIXED|MSGPACK_FLAG_TYPEVALUE, + ucl_msgpack_parse_int + }, + { + 0xe0, + 3, + msgpack_negative_fixint, + 0, + MSGPACK_FLAG_FIXED|MSGPACK_FLAG_TYPEVALUE, + ucl_msgpack_parse_int + }, + { + 0x80, + 4, + msgpack_fixmap, + 0, + MSGPACK_FLAG_FIXED|MSGPACK_FLAG_CONTAINER|MSGPACK_FLAG_ASSOC, + ucl_msgpack_parse_map + }, + { + 0x90, + 4, + msgpack_fixarray, + 0, + MSGPACK_FLAG_FIXED|MSGPACK_FLAG_CONTAINER, + ucl_msgpack_parse_array + }, + { + 0xd9, + 8, + msgpack_str8, + 1, + MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0xc4, + 8, + msgpack_bin8, + 1, + MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0xcf, + 8, + msgpack_uint64, + 8, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xd3, + 8, + msgpack_int64, + 8, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xce, + 8, + msgpack_uint32, + 4, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xd2, + 8, + msgpack_int32, + 4, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xcb, + 8, + msgpack_float64, + 8, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_float + }, + { + 0xca, + 8, + msgpack_float32, + 4, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_float + }, + { + 0xc2, + 8, + msgpack_false, + 1, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_TYPEVALUE, + ucl_msgpack_parse_bool + }, + { + 0xc3, + 8, + msgpack_true, + 1, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_TYPEVALUE, + ucl_msgpack_parse_bool + }, + { + 0xcc, + 8, + msgpack_uint8, + 1, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xcd, + 8, + msgpack_uint16, + 2, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xd0, + 8, + msgpack_int8, + 1, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xd1, + 8, + msgpack_int16, + 2, + MSGPACK_FLAG_FIXED, + ucl_msgpack_parse_int + }, + { + 0xc0, + 8, + msgpack_nil, + 0, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_TYPEVALUE, + ucl_msgpack_parse_null + }, + { + 0xda, + 8, + msgpack_str16, + 2, + MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0xdb, + 8, + msgpack_str32, + 4, + MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0xc5, + 8, + msgpack_bin16, + 2, + MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0xc6, + 8, + msgpack_bin32, + 4, + MSGPACK_FLAG_KEY, + ucl_msgpack_parse_string + }, + { + 0xdc, + 8, + msgpack_array16, + 2, + MSGPACK_FLAG_CONTAINER, + ucl_msgpack_parse_array + }, + { + 0xdd, + 8, + msgpack_array32, + 4, + MSGPACK_FLAG_CONTAINER, + ucl_msgpack_parse_array + }, + { + 0xde, + 8, + msgpack_map16, + 2, + MSGPACK_FLAG_CONTAINER|MSGPACK_FLAG_ASSOC, + ucl_msgpack_parse_map + }, + { + 0xdf, + 8, + msgpack_map32, + 4, + MSGPACK_FLAG_CONTAINER|MSGPACK_FLAG_ASSOC, + ucl_msgpack_parse_map + }, + { + 0xc7, + 8, + msgpack_ext8, + 1, + MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xc8, + 8, + msgpack_ext16, + 2, + MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xc9, + 8, + msgpack_ext32, + 4, + MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xd4, + 8, + msgpack_fixext1, + 1, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xd5, + 8, + msgpack_fixext2, + 2, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xd6, + 8, + msgpack_fixext4, + 4, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xd7, + 8, + msgpack_fixext8, + 8, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + }, + { + 0xd8, + 8, + msgpack_fixext16, + 16, + MSGPACK_FLAG_FIXED | MSGPACK_FLAG_EXT, + ucl_msgpack_parse_ignore + } +}; + +#undef MSGPACK_DEBUG_PARSER + +static inline struct ucl_msgpack_parser * +ucl_msgpack_get_parser_from_type (unsigned char t) +{ + unsigned int i, shift, mask; + + for (i = 0; i < sizeof (parsers) / sizeof (parsers[0]); i ++) { + shift = CHAR_BIT - parsers[i].prefixlen; + mask = parsers[i].prefix >> shift; + + if (mask == (t >> shift)) { + return &parsers[i]; + } + } + + return NULL; +} + +static inline struct ucl_stack * +ucl_msgpack_get_container (struct ucl_parser *parser, + struct ucl_msgpack_parser *obj_parser, uint64_t len) +{ + struct ucl_stack *stack; + + assert (obj_parser != NULL); + + if (obj_parser->flags & MSGPACK_FLAG_CONTAINER) { + assert ((len & MSGPACK_CONTAINER_BIT) == 0); + /* + * Insert new container to the stack + */ + if (parser->stack == NULL) { + parser->stack = calloc (1, sizeof (struct ucl_stack)); + + if (parser->stack == NULL) { + ucl_create_err (&parser->err, "no memory"); + return NULL; + } + } + else { + stack = calloc (1, sizeof (struct ucl_stack)); + + if (stack == NULL) { + ucl_create_err (&parser->err, "no memory"); + return NULL; + } + + stack->next = parser->stack; + parser->stack = stack; + } + + parser->stack->level = len | MSGPACK_CONTAINER_BIT; + +#ifdef MSGPACK_DEBUG_PARSER + stack = parser->stack; + while (stack) { + fprintf(stderr, "+"); + stack = stack->next; + } + + fprintf(stderr, "%s -> %d\n", obj_parser->flags & MSGPACK_FLAG_ASSOC ? "object" : "array", (int)len); +#endif + } + else { + /* + * Get the current stack top + */ + if (parser->stack) { + return parser->stack; + } + else { + ucl_create_err (&parser->err, "bad top level object for msgpack"); + return NULL; + } + } + + return parser->stack; +} + +static bool +ucl_msgpack_is_container_finished (struct ucl_stack *container) +{ + uint64_t level; + + assert (container != NULL); + + if (container->level & MSGPACK_CONTAINER_BIT) { + level = container->level & ~MSGPACK_CONTAINER_BIT; + + if (level == 0) { + return true; + } + } + + return false; +} + +static bool +ucl_msgpack_insert_object (struct ucl_parser *parser, + const unsigned char *key, + size_t keylen, ucl_object_t *obj) +{ + uint64_t level; + struct ucl_stack *container; + + container = parser->stack; + assert (container != NULL); + assert (container->level > 0); + assert (obj != NULL); + assert (container->obj != NULL); + + if (container->obj->type == UCL_ARRAY) { + ucl_array_append (container->obj, obj); + } + else if (container->obj->type == UCL_OBJECT) { + if (key == NULL || keylen == 0) { + ucl_create_err (&parser->err, "cannot insert object with no key"); + return false; + } + + obj->key = key; + obj->keylen = keylen; + + if (!(parser->flags & UCL_PARSER_ZEROCOPY)) { + ucl_copy_key_trash (obj); + } + + ucl_parser_process_object_element (parser, obj); + } + else { + ucl_create_err (&parser->err, "bad container type"); + return false; + } + + if (container->level & MSGPACK_CONTAINER_BIT) { + level = container->level & ~MSGPACK_CONTAINER_BIT; + container->level = (level - 1) | MSGPACK_CONTAINER_BIT; + } + + return true; +} + +static struct ucl_stack * +ucl_msgpack_get_next_container (struct ucl_parser *parser) +{ + struct ucl_stack *cur = NULL; + uint64_t level; + + cur = parser->stack; + + if (cur == NULL) { + return NULL; + } + + if (cur->level & MSGPACK_CONTAINER_BIT) { + level = cur->level & ~MSGPACK_CONTAINER_BIT; + + if (level == 0) { + /* We need to switch to the previous container */ + parser->stack = cur->next; + parser->cur_obj = cur->obj; + free (cur); + +#ifdef MSGPACK_DEBUG_PARSER + cur = parser->stack; + while (cur) { + fprintf(stderr, "-"); + cur = cur->next; + } + fprintf(stderr, "-%s -> %d\n", parser->cur_obj->type == UCL_OBJECT ? "object" : "array", (int)parser->cur_obj->len); +#endif + + return ucl_msgpack_get_next_container (parser); + } + } + + /* + * For UCL containers we don't know length, so we just insert the whole + * message pack blob into the top level container + */ + + assert (cur->obj != NULL); + + return cur; +} + +#define CONSUME_RET do { \ + if (ret != -1) { \ + p += ret; \ + remain -= ret; \ + obj_parser = NULL; \ + assert (remain >= 0); \ + } \ + else { \ + ucl_create_err (&parser->err, \ + "cannot parse type %d of len %u", \ + (int)obj_parser->fmt, \ + (unsigned)len); \ + return false; \ + } \ +} while(0) + +#define GET_NEXT_STATE do { \ + container = ucl_msgpack_get_next_container (parser); \ + if (container == NULL) { \ + ucl_create_err (&parser->err, \ + "empty container"); \ + return false; \ + } \ + next_state = container->obj->type == UCL_OBJECT ? \ + read_assoc_key : read_array_value; \ +} while(0) + +static bool +ucl_msgpack_consume (struct ucl_parser *parser) +{ + const unsigned char *p, *end, *key = NULL; + struct ucl_stack *container; + enum e_msgpack_parser_state { + read_type, + start_assoc, + start_array, + read_assoc_key, + read_assoc_value, + finish_assoc_value, + read_array_value, + finish_array_value, + error_state + } state = read_type, next_state = error_state; + struct ucl_msgpack_parser *obj_parser; + uint64_t len; + ssize_t ret, remain, keylen = 0; +#ifdef MSGPACK_DEBUG_PARSER + uint64_t i; + enum e_msgpack_parser_state hist[256]; +#endif + + p = parser->chunks->begin; + remain = parser->chunks->remain; + end = p + remain; + + + while (p < end) { +#ifdef MSGPACK_DEBUG_PARSER + hist[i++ % 256] = state; +#endif + switch (state) { + case read_type: + obj_parser = ucl_msgpack_get_parser_from_type (*p); + + if (obj_parser == NULL) { + ucl_create_err (&parser->err, "unknown msgpack format: %x", + (unsigned int)*p); + + return false; + } + /* Now check length sanity */ + if (obj_parser->flags & MSGPACK_FLAG_FIXED) { + if (obj_parser->len == 0) { + /* We have an embedded size */ + len = *p & ~obj_parser->prefix; + } + else { + if (remain < obj_parser->len) { + ucl_create_err (&parser->err, "not enough data remain to " + "read object's length: %u remain, %u needed", + (unsigned)remain, obj_parser->len); + + return false; + } + + len = obj_parser->len; + } + + if (!(obj_parser->flags & MSGPACK_FLAG_TYPEVALUE)) { + /* We must pass value as the second byte */ + if (remain > 0) { + p ++; + remain --; + } + } + else { + /* Len is irrelevant now */ + len = 0; + } + } + else { + /* Length is not embedded */ + if (remain < obj_parser->len) { + ucl_create_err (&parser->err, "not enough data remain to " + "read object's length: %u remain, %u needed", + (unsigned)remain, obj_parser->len); + + return false; + } + + p ++; + remain --; + + switch (obj_parser->len) { + case 1: + len = *p; + break; + case 2: + len = FROM_BE16 (*(uint16_t *)p); + break; + case 4: + len = FROM_BE32 (*(uint32_t *)p); + break; + case 8: + len = FROM_BE64 (*(uint64_t *)p); + break; + default: + assert (0); + break; + } + + p += obj_parser->len; + remain -= obj_parser->len; + } + + if (obj_parser->flags & MSGPACK_FLAG_ASSOC) { + /* We have just read the new associative map */ + state = start_assoc; + } + else if (obj_parser->flags & MSGPACK_FLAG_CONTAINER){ + state = start_array; + } + else { + state = next_state; + } + + break; + case start_assoc: + parser->cur_obj = ucl_object_new_full (UCL_OBJECT, + parser->chunks->priority); + /* Insert to the previous level container */ + if (parser->stack && !ucl_msgpack_insert_object (parser, + key, keylen, parser->cur_obj)) { + return false; + } + /* Get new container */ + container = ucl_msgpack_get_container (parser, obj_parser, len); + + if (container == NULL) { + return false; + } + + ret = obj_parser->func (parser, container, len, obj_parser->fmt, + p, remain); + CONSUME_RET; + key = NULL; + keylen = 0; + + if (len > 0) { + state = read_type; + next_state = read_assoc_key; + } + else { + /* Empty object */ + state = finish_assoc_value; + } + break; + + case start_array: + parser->cur_obj = ucl_object_new_full (UCL_ARRAY, + parser->chunks->priority); + /* Insert to the previous level container */ + if (parser->stack && !ucl_msgpack_insert_object (parser, + key, keylen, parser->cur_obj)) { + return false; + } + /* Get new container */ + container = ucl_msgpack_get_container (parser, obj_parser, len); + + if (container == NULL) { + return false; + } + + ret = obj_parser->func (parser, container, len, obj_parser->fmt, + p, remain); + CONSUME_RET; + + if (len > 0) { + state = read_type; + next_state = read_array_value; + } + else { + /* Empty array */ + state = finish_array_value; + } + break; + + case read_array_value: + /* + * p is now at the value start, len now contains length read and + * obj_parser contains the corresponding specific parser + */ + container = parser->stack; + + if (container == NULL) { + return false; + } + + ret = obj_parser->func (parser, container, len, obj_parser->fmt, + p, remain); + CONSUME_RET; + + + /* Insert value to the container and check if we have finished array */ + if (!ucl_msgpack_insert_object (parser, NULL, 0, + parser->cur_obj)) { + return false; + } + + if (ucl_msgpack_is_container_finished (container)) { + state = finish_array_value; + } + else { + /* Read more elements */ + state = read_type; + next_state = read_array_value; + } + + break; + + case read_assoc_key: + /* + * Keys must have string type for ucl msgpack + */ + if (!(obj_parser->flags & MSGPACK_FLAG_KEY)) { + ucl_create_err (&parser->err, "bad type for key: %u, expected " + "string", (unsigned)obj_parser->fmt); + + return false; + } + + key = p; + keylen = len; + + if (keylen > remain || keylen == 0) { + ucl_create_err (&parser->err, "too long or empty key"); + return false; + } + + p += len; + remain -= len; + + state = read_type; + next_state = read_assoc_value; + break; + + case read_assoc_value: + /* + * p is now at the value start, len now contains length read and + * obj_parser contains the corresponding specific parser + */ + container = parser->stack; + + if (container == NULL) { + return false; + } + + ret = obj_parser->func (parser, container, len, obj_parser->fmt, + p, remain); + CONSUME_RET; + + assert (key != NULL && keylen > 0); + + if (!ucl_msgpack_insert_object (parser, key, keylen, + parser->cur_obj)) { + return false; + } + + key = NULL; + keylen = 0; + + if (ucl_msgpack_is_container_finished (container)) { + state = finish_assoc_value; + } + else { + /* Read more elements */ + state = read_type; + next_state = read_assoc_key; + } + break; + + case finish_array_value: + case finish_assoc_value: + GET_NEXT_STATE; + state = read_type; + break; + + case error_state: + ucl_create_err (&parser->err, "invalid state machine state"); + + return false; + } + } + + /* Check the finishing state */ + switch (state) { + case start_array: + case start_assoc: + /* Empty container at the end */ + if (len != 0) { + ucl_create_err (&parser->err, "invalid non-empty container at the end"); + + return false; + } + + parser->cur_obj = ucl_object_new_full ( + state == start_array ? UCL_ARRAY : UCL_OBJECT, + parser->chunks->priority); + /* Insert to the previous level container */ + if (!ucl_msgpack_insert_object (parser, + key, keylen, parser->cur_obj)) { + return false; + } + /* Get new container */ + container = ucl_msgpack_get_container (parser, obj_parser, len); + + if (container == NULL) { + return false; + } + + ret = obj_parser->func (parser, container, len, obj_parser->fmt, + p, remain); + break; + + case read_array_value: + case read_assoc_value: + if (len != 0) { + ucl_create_err (&parser->err, "unfinished value at the end"); + + return false; + } + + container = parser->stack; + + if (container == NULL) { + return false; + } + + ret = obj_parser->func (parser, container, len, obj_parser->fmt, + p, remain); + CONSUME_RET; + + + /* Insert value to the container and check if we have finished array */ + if (!ucl_msgpack_insert_object (parser, NULL, 0, + parser->cur_obj)) { + return false; + } + break; + case finish_array_value: + case finish_assoc_value: + case read_type: + /* Valid finishing state */ + break; + default: + /* Invalid finishing state */ + ucl_create_err (&parser->err, "invalid state machine finishing state: %d", + state); + + return false; + } + + /* Rewind to the top level container */ + ucl_msgpack_get_next_container (parser); + assert (parser->stack == NULL || + (parser->stack->level & MSGPACK_CONTAINER_BIT) == 0); + + return true; +} + +bool +ucl_parse_msgpack (struct ucl_parser *parser) +{ + ucl_object_t *container = NULL; + const unsigned char *p; + bool ret; + + assert (parser != NULL); + assert (parser->chunks != NULL); + assert (parser->chunks->begin != NULL); + assert (parser->chunks->remain != 0); + + p = parser->chunks->begin; + + if (parser->stack) { + container = parser->stack->obj; + } + + /* + * When we start parsing message pack chunk, we must ensure that we + * have either a valid container or the top object inside message pack is + * of container type + */ + if (container == NULL) { + if ((*p & 0x80) != 0x80 && !(*p >= 0xdc && *p <= 0xdf)) { + ucl_create_err (&parser->err, "bad top level object for msgpack"); + return false; + } + } + + ret = ucl_msgpack_consume (parser); + + if (ret && parser->top_obj == NULL) { + parser->top_obj = parser->cur_obj; + } + + return ret; +} + +static ssize_t +ucl_msgpack_parse_map (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + container->obj = parser->cur_obj; + + return 0; +} + +static ssize_t +ucl_msgpack_parse_array (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + container->obj = parser->cur_obj; + + return 0; +} + +static ssize_t +ucl_msgpack_parse_string (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + ucl_object_t *obj; + + if (len > remain) { + return -1; + } + + obj = ucl_object_new_full (UCL_STRING, parser->chunks->priority); + obj->value.sv = pos; + obj->len = len; + + if (fmt >= msgpack_bin8 && fmt <= msgpack_bin32) { + obj->flags |= UCL_OBJECT_BINARY; + } + + if (!(parser->flags & UCL_PARSER_ZEROCOPY)) { + if (obj->flags & UCL_OBJECT_BINARY) { + obj->trash_stack[UCL_TRASH_VALUE] = malloc (len); + + if (obj->trash_stack[UCL_TRASH_VALUE] != NULL) { + memcpy (obj->trash_stack[UCL_TRASH_VALUE], pos, len); + } + } + else { + ucl_copy_value_trash (obj); + } + } + + parser->cur_obj = obj; + + return len; +} + +static ssize_t +ucl_msgpack_parse_int (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + ucl_object_t *obj; + + if (len > remain) { + return -1; + } + + obj = ucl_object_new_full (UCL_INT, parser->chunks->priority); + + switch (fmt) { + case msgpack_positive_fixint: + obj->value.iv = (*pos & 0x7f); + len = 1; + break; + case msgpack_negative_fixint: + obj->value.iv = - (*pos & 0x1f); + len = 1; + break; + case msgpack_uint8: + obj->value.iv = (unsigned char)*pos; + len = 1; + break; + case msgpack_int8: + obj->value.iv = (signed char)*pos; + len = 1; + break; + case msgpack_int16: + obj->value.iv = FROM_BE16 (*(int16_t *)pos); + len = 2; + break; + case msgpack_uint16: + obj->value.iv = FROM_BE16 (*(uint16_t *)pos); + len = 2; + break; + case msgpack_int32: + obj->value.iv = FROM_BE32 (*(int32_t *)pos); + len = 4; + break; + case msgpack_uint32: + obj->value.iv = FROM_BE32 (*(uint32_t *)pos); + len = 4; + break; + case msgpack_int64: + obj->value.iv = FROM_BE64 (*(int64_t *)pos); + len = 8; + break; + case msgpack_uint64: + obj->value.iv = FROM_BE64 (*(uint64_t *)pos); + len = 8; + break; + default: + assert (0); + break; + } + + parser->cur_obj = obj; + + return len; +} + +static ssize_t +ucl_msgpack_parse_float (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + ucl_object_t *obj; + union { + uint32_t i; + float f; + } d; + + if (len > remain) { + return -1; + } + + obj = ucl_object_new_full (UCL_FLOAT, parser->chunks->priority); + + switch (fmt) { + case msgpack_float32: + d.i = FROM_BE32 (*(uint32_t *)pos); + /* XXX: can be slow */ + obj->value.dv = d.f; + len = 4; + break; + case msgpack_float64: + obj->value.iv = FROM_BE64 (*(uint64_t *)pos); + len = 8; + break; + default: + assert (0); + break; + } + + parser->cur_obj = obj; + + return len; +} + +static ssize_t +ucl_msgpack_parse_bool (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + ucl_object_t *obj; + + if (len > remain) { + return -1; + } + + obj = ucl_object_new_full (UCL_BOOLEAN, parser->chunks->priority); + + switch (fmt) { + case msgpack_true: + obj->value.iv = true; + break; + case msgpack_false: + obj->value.iv = false; + break; + default: + assert (0); + break; + } + + parser->cur_obj = obj; + + return 1; +} + +static ssize_t +ucl_msgpack_parse_null (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + ucl_object_t *obj; + + if (len > remain) { + return -1; + } + + obj = ucl_object_new_full (UCL_NULL, parser->chunks->priority); + parser->cur_obj = obj; + + return 1; +} + +static ssize_t +ucl_msgpack_parse_ignore (struct ucl_parser *parser, + struct ucl_stack *container, size_t len, enum ucl_msgpack_format fmt, + const unsigned char *pos, size_t remain) +{ + if (len > remain) { + return -1; + } + + switch (fmt) { + case msgpack_fixext1: + len = 2; + break; + case msgpack_fixext2: + len = 3; + break; + case msgpack_fixext4: + len = 5; + break; + case msgpack_fixext8: + len = 9; + break; + case msgpack_fixext16: + len = 17; + break; + case msgpack_ext8: + case msgpack_ext16: + case msgpack_ext32: + len = len + 1; + break; + default: + ucl_create_err (&parser->err, "bad type: %x", (unsigned)fmt); + return -1; + } + + return len; +} diff --git a/src/ucl_parser.c b/src/ucl_parser.c index 75acba8ecbd3..9bd41391ba36 100644 --- a/src/ucl_parser.c +++ b/src/ucl_parser.c @@ -67,6 +67,7 @@ ucl_set_err (struct ucl_parser *parser, int code, const char *str, UT_string **e else { filename = "<unknown>"; } + if (chunk->pos < chunk->end) { if (isgraph (*chunk->pos)) { fmt_string = "error while parsing %s: " @@ -84,6 +85,8 @@ ucl_set_err (struct ucl_parser *parser, int code, const char *str, UT_string **e ucl_create_err (err, "error while parsing %s: at the end of chunk: %s", filename, str); } + + parser->err_code = code; } /** @@ -513,7 +516,7 @@ ucl_copy_or_store_ptr (struct ucl_parser *parser, /* Copy string */ *dst = UCL_ALLOC (in_len + 1); if (*dst == NULL) { - ucl_set_err (parser, 0, "cannot allocate memory for a string", + ucl_set_err (parser, UCL_EINTERNAL, "cannot allocate memory for a string", &parser->err); return false; } @@ -559,7 +562,8 @@ ucl_copy_or_store_ptr (struct ucl_parser *parser, * @return */ static inline ucl_object_t * -ucl_add_parser_stack (ucl_object_t *obj, struct ucl_parser *parser, bool is_array, int level) +ucl_parser_add_container (ucl_object_t *obj, struct ucl_parser *parser, + bool is_array, int level) { struct ucl_stack *st; @@ -570,7 +574,9 @@ ucl_add_parser_stack (ucl_object_t *obj, struct ucl_parser *parser, bool is_arra else { obj->type = UCL_OBJECT; } - obj->value.ov = ucl_hash_create (parser->flags & UCL_PARSER_KEY_LOWERCASE); + if (obj->value.ov == NULL) { + obj->value.ov = ucl_hash_create (parser->flags & UCL_PARSER_KEY_LOWERCASE); + } parser->state = UCL_STATE_KEY; } else { @@ -585,7 +591,7 @@ ucl_add_parser_stack (ucl_object_t *obj, struct ucl_parser *parser, bool is_arra st = UCL_ALLOC (sizeof (struct ucl_stack)); if (st == NULL) { - ucl_set_err (parser, 0, "cannot allocate memory for an object", + ucl_set_err (parser, UCL_EINTERNAL, "cannot allocate memory for an object", &parser->err); ucl_object_unref (obj); return NULL; @@ -834,19 +840,21 @@ ucl_maybe_parse_number (ucl_object_t *obj, *pos = c; return EINVAL; - set_obj: - if (allow_double && (need_double || is_time)) { - if (!is_time) { - obj->type = UCL_FLOAT; +set_obj: + if (obj != NULL) { + if (allow_double && (need_double || is_time)) { + if (!is_time) { + obj->type = UCL_FLOAT; + } + else { + obj->type = UCL_TIME; + } + obj->value.dv = is_neg ? (-dv) : dv; } else { - obj->type = UCL_TIME; + obj->type = UCL_INT; + obj->value.iv = is_neg ? (-lv) : lv; } - obj->value.dv = is_neg ? (-dv) : dv; - } - else { - obj->type = UCL_INT; - obj->value.iv = is_neg ? (-lv) : lv; } *pos = p; return 0; @@ -856,6 +864,7 @@ ucl_maybe_parse_number (ucl_object_t *obj, * Parse possible number * @param parser * @param chunk + * @param obj * @return true if a number has been parsed */ static bool @@ -875,7 +884,8 @@ ucl_lex_number (struct ucl_parser *parser, return true; } else if (ret == ERANGE) { - ucl_set_err (parser, ERANGE, "numeric value out of range", &parser->err); + ucl_set_err (parser, UCL_ESYNTAX, "numeric value out of range", + &parser->err); } return false; @@ -885,6 +895,9 @@ ucl_lex_number (struct ucl_parser *parser, * Parse quoted string with possible escapes * @param parser * @param chunk + * @param need_unescape + * @param ucl_escape + * @param var_expand * @return true if a string has been parsed */ static bool @@ -971,6 +984,7 @@ ucl_parser_append_elt (struct ucl_parser *parser, ucl_hash_t *cont, /* Implicit array */ top->flags |= UCL_OBJECT_MULTIVALUE; DL_APPEND (top, elt); + parser->stack->obj->len ++; } else { if ((top->flags & UCL_OBJECT_MULTIVALUE) != 0) { @@ -979,34 +993,127 @@ ucl_parser_append_elt (struct ucl_parser *parser, ucl_hash_t *cont, } else { /* Convert to an array */ - ucl_hash_delete (cont, top); nobj = ucl_object_typed_new (UCL_ARRAY); nobj->key = top->key; nobj->keylen = top->keylen; nobj->flags |= UCL_OBJECT_MULTIVALUE; ucl_array_append (nobj, top); ucl_array_append (nobj, elt); - ucl_hash_insert (cont, nobj, nobj->key, nobj->keylen); + ucl_hash_replace (cont, top, nobj); + } + } +} + +bool +ucl_parser_process_object_element (struct ucl_parser *parser, ucl_object_t *nobj) +{ + ucl_hash_t *container; + ucl_object_t *tobj; + + container = parser->stack->obj->value.ov; + + tobj = __DECONST (ucl_object_t *, ucl_hash_search_obj (container, nobj)); + if (tobj == NULL) { + container = ucl_hash_insert_object (container, nobj, + parser->flags & UCL_PARSER_KEY_LOWERCASE); + nobj->prev = nobj; + nobj->next = NULL; + parser->stack->obj->len ++; + } + else { + unsigned priold = ucl_object_get_priority (tobj), + prinew = ucl_object_get_priority (nobj); + switch (parser->chunks->strategy) { + + case UCL_DUPLICATE_APPEND: + /* + * The logic here is the following: + * + * - if we have two objects with the same priority, then we form an + * implicit or explicit array + * - if a new object has bigger priority, then we overwrite an old one + * - if a new object has lower priority, then we ignore it + */ + + + /* Special case for inherited objects */ + if (tobj->flags & UCL_OBJECT_INHERITED) { + prinew = priold + 1; + } + + if (priold == prinew) { + ucl_parser_append_elt (parser, container, tobj, nobj); + } + else if (priold > prinew) { + /* + * We add this new object to a list of trash objects just to ensure + * that it won't come to any real object + * XXX: rather inefficient approach + */ + DL_APPEND (parser->trash_objs, nobj); + } + else { + ucl_hash_replace (container, tobj, nobj); + ucl_object_unref (tobj); + } + + break; + + case UCL_DUPLICATE_REWRITE: + /* We just rewrite old values regardless of priority */ + ucl_hash_replace (container, tobj, nobj); + ucl_object_unref (tobj); + + break; + + case UCL_DUPLICATE_ERROR: + ucl_create_err (&parser->err, "error while parsing %s: " + "line: %d, column: %d: duplicate element for key '%s' " + "has been found", + parser->cur_file ? parser->cur_file : "<unknown>", + parser->chunks->line, parser->chunks->column, nobj->key); + return false; + + case UCL_DUPLICATE_MERGE: + /* + * Here we do have some old object so we just push it on top of objects stack + */ + if (tobj->type == UCL_OBJECT || tobj->type == UCL_ARRAY) { + ucl_object_unref (nobj); + nobj = tobj; + } + else { + /* For other types we create implicit array as usual */ + ucl_parser_append_elt (parser, container, tobj, nobj); + } + break; } } + + parser->stack->obj->value.ov = container; + parser->cur_obj = nobj; + + return true; } /** * Parse a key in an object * @param parser * @param chunk + * @param next_key + * @param end_of_object * @return true if a key has been parsed */ static bool -ucl_parse_key (struct ucl_parser *parser, struct ucl_chunk *chunk, bool *next_key, bool *end_of_object) +ucl_parse_key (struct ucl_parser *parser, struct ucl_chunk *chunk, + bool *next_key, bool *end_of_object) { const unsigned char *p, *c = NULL, *end, *t; const char *key = NULL; bool got_quote = false, got_eq = false, got_semicolon = false, need_unescape = false, ucl_escape = false, var_expand = false, got_content = false, got_sep = false; - ucl_object_t *nobj, *tobj; - ucl_hash_t *container; + ucl_object_t *nobj; ssize_t keylen; p = chunk->pos; @@ -1191,47 +1298,17 @@ ucl_parse_key (struct ucl_parser *parser, struct ucl_chunk *chunk, bool *next_ke return false; } - container = parser->stack->obj->value.ov; nobj->key = key; nobj->keylen = keylen; - tobj = __DECONST (ucl_object_t *, ucl_hash_search_obj (container, nobj)); - if (tobj == NULL) { - container = ucl_hash_insert_object (container, nobj, - parser->flags & UCL_PARSER_KEY_LOWERCASE); - nobj->prev = nobj; - nobj->next = NULL; - parser->stack->obj->len ++; - } - else { - /* - * The logic here is the following: - * - * - if we have two objects with the same priority, then we form an - * implicit or explicit array - * - if a new object has bigger priority, then we overwrite an old one - * - if a new object has lower priority, then we ignore it - */ - unsigned priold = ucl_object_get_priority (tobj), - prinew = ucl_object_get_priority (nobj); - if (priold == prinew) { - ucl_parser_append_elt (parser, container, tobj, nobj); - } - else if (priold > prinew) { - ucl_object_unref (nobj); - return true; - } - else { - ucl_hash_replace (container, tobj, nobj); - ucl_object_unref (tobj); - } + + if (!ucl_parser_process_object_element (parser, nobj)) { + return false; } if (ucl_escape) { nobj->flags |= UCL_OBJECT_NEED_KEY_ESCAPE; } - parser->stack->obj->value.ov = container; - parser->cur_obj = nobj; return true; } @@ -1240,6 +1317,8 @@ ucl_parse_key (struct ucl_parser *parser, struct ucl_chunk *chunk, bool *next_ke * Parse a cl string * @param parser * @param chunk + * @param var_expand + * @param need_unescape * @return true if a key has been parsed */ static bool @@ -1309,6 +1388,8 @@ ucl_parse_string_value (struct ucl_parser *parser, * @param chunk * @param term * @param term_len + * @param beg + * @param var_expand * @return size of multiline string or 0 in case of error */ static int @@ -1360,8 +1441,8 @@ ucl_parse_multiline_string (struct ucl_parser *parser, return len; } -static ucl_object_t* -ucl_get_value_object (struct ucl_parser *parser) +static inline ucl_object_t* +ucl_parser_get_container (struct ucl_parser *parser) { ucl_object_t *t, *obj = NULL; @@ -1373,7 +1454,12 @@ ucl_get_value_object (struct ucl_parser *parser) /* Object must be allocated */ obj = ucl_object_new_full (UCL_NULL, parser->chunks->priority); t = parser->stack->obj; - ucl_array_append (t, obj); + + if (!ucl_array_append (t, obj)) { + ucl_object_unref (obj); + return NULL; + } + parser->cur_obj = obj; } else { @@ -1417,42 +1503,51 @@ ucl_parse_value (struct ucl_parser *parser, struct ucl_chunk *chunk) c = p; switch (*p) { case '"': - obj = ucl_get_value_object (parser); ucl_chunk_skipc (chunk, p); - if (!ucl_lex_json_string (parser, chunk, &need_unescape, &ucl_escape, &var_expand)) { + + if (!ucl_lex_json_string (parser, chunk, &need_unescape, &ucl_escape, + &var_expand)) { return false; } + + obj = ucl_parser_get_container (parser); str_len = chunk->pos - c - 2; obj->type = UCL_STRING; - if ((str_len = ucl_copy_or_store_ptr (parser, c + 1, &obj->trash_stack[UCL_TRASH_VALUE], - &obj->value.sv, str_len, need_unescape, false, var_expand)) == -1) { + if ((str_len = ucl_copy_or_store_ptr (parser, c + 1, + &obj->trash_stack[UCL_TRASH_VALUE], + &obj->value.sv, str_len, need_unescape, false, + var_expand)) == -1) { return false; } obj->len = str_len; + parser->state = UCL_STATE_AFTER_VALUE; p = chunk->pos; + return true; break; case '{': - obj = ucl_get_value_object (parser); + obj = ucl_parser_get_container (parser); /* We have a new object */ - obj = ucl_add_parser_stack (obj, parser, false, parser->stack->level); + obj = ucl_parser_add_container (obj, parser, false, parser->stack->level); if (obj == NULL) { return false; } ucl_chunk_skipc (chunk, p); + return true; break; case '[': - obj = ucl_get_value_object (parser); + obj = ucl_parser_get_container (parser); /* We have a new array */ - obj = ucl_add_parser_stack (obj, parser, true, parser->stack->level); + obj = ucl_parser_add_container (obj, parser, true, parser->stack->level); if (obj == NULL) { return false; } ucl_chunk_skipc (chunk, p); + return true; break; case ']': @@ -1466,7 +1561,7 @@ ucl_parse_value (struct ucl_parser *parser, struct ucl_chunk *chunk) } break; case '<': - obj = ucl_get_value_object (parser); + obj = ucl_parser_get_container (parser); /* We have something like multiline value, which must be <<[A-Z]+\n */ if (chunk->end - p > 3) { if (memcmp (p, "<<", 2) == 0) { @@ -1488,13 +1583,19 @@ ucl_parse_value (struct ucl_parser *parser, struct ucl_chunk *chunk) "unterminated multiline value", &parser->err); return false; } + obj->type = UCL_STRING; - if ((str_len = ucl_copy_or_store_ptr (parser, c, &obj->trash_stack[UCL_TRASH_VALUE], - &obj->value.sv, str_len - 1, false, false, var_expand)) == -1) { + obj->flags |= UCL_OBJECT_MULTILINE; + if ((str_len = ucl_copy_or_store_ptr (parser, c, + &obj->trash_stack[UCL_TRASH_VALUE], + &obj->value.sv, str_len - 1, false, + false, var_expand)) == -1) { return false; } obj->len = str_len; + parser->state = UCL_STATE_AFTER_VALUE; + return true; } } @@ -1503,8 +1604,9 @@ ucl_parse_value (struct ucl_parser *parser, struct ucl_chunk *chunk) default: parse_string: if (obj == NULL) { - obj = ucl_get_value_object (parser); + obj = ucl_parser_get_container (parser); } + /* Parse atom */ if (ucl_test_character (*p, UCL_CHARACTER_VALUE_DIGIT_START)) { if (!ucl_lex_number (parser, chunk, obj)) { @@ -1519,7 +1621,8 @@ parse_string: /* Fallback to normal string */ } - if (!ucl_parse_string_value (parser, chunk, &var_expand, &need_unescape)) { + if (!ucl_parse_string_value (parser, chunk, &var_expand, + &need_unescape)) { return false; } /* Cut trailing spaces */ @@ -1530,7 +1633,7 @@ parse_string: } str_len = chunk->pos - c - stripped_spaces; if (str_len <= 0) { - ucl_set_err (parser, 0, "string value must not be empty", + ucl_set_err (parser, UCL_ESYNTAX, "string value must not be empty", &parser->err); return false; } @@ -1540,7 +1643,8 @@ parse_string: } else if (!ucl_maybe_parse_boolean (obj, c, str_len)) { obj->type = UCL_STRING; - if ((str_len = ucl_copy_or_store_ptr (parser, c, &obj->trash_stack[UCL_TRASH_VALUE], + if ((str_len = ucl_copy_or_store_ptr (parser, c, + &obj->trash_stack[UCL_TRASH_VALUE], &obj->value.sv, str_len, need_unescape, false, var_expand)) == -1) { return false; @@ -1652,6 +1756,9 @@ ucl_parse_after_value (struct ucl_parser *parser, struct ucl_chunk *chunk) * Handle macro data * @param parser * @param chunk + * @param marco + * @param macro_start + * @param macro_len * @return */ static bool @@ -1790,6 +1897,7 @@ ucl_parse_macro_arguments (struct ucl_parser *parser, if (chunk->remain == 0) { goto restore_chunk; } + args_len ++; ucl_chunk_skipc (chunk, p); break; case 99: @@ -1842,8 +1950,6 @@ restore_chunk: /** * Handle the main states of rcl parser * @param parser parser structure - * @param data the pointer to the beginning of a chunk - * @param len the length of a chunk * @return true if chunk has been parsed and false in case of error */ static bool @@ -1858,17 +1964,6 @@ ucl_state_machine (struct ucl_parser *parser) bool next_key = false, end_of_object = false, ret; if (parser->top_obj == NULL) { - if (*chunk->pos == '[') { - obj = ucl_add_parser_stack (NULL, parser, true, 0); - } - else { - obj = ucl_add_parser_stack (NULL, parser, false, 0); - } - if (obj == NULL) { - return false; - } - parser->top_obj = obj; - parser->cur_obj = obj; parser->state = UCL_STATE_INIT; } @@ -1892,7 +1987,9 @@ ucl_state_machine (struct ucl_parser *parser) UCL_CHARACTER_WHITESPACE_UNSAFE)) { ucl_chunk_skipc (chunk, p); } + p = chunk->pos; + if (*p == '[') { parser->state = UCL_STATE_VALUE; ucl_chunk_skipc (chunk, p); @@ -1903,6 +2000,23 @@ ucl_state_machine (struct ucl_parser *parser) ucl_chunk_skipc (chunk, p); } } + + if (parser->top_obj == NULL) { + if (parser->state == UCL_STATE_VALUE) { + obj = ucl_parser_add_container (NULL, parser, true, 0); + } + else { + obj = ucl_parser_add_container (NULL, parser, false, 0); + } + + if (obj == NULL) { + return false; + } + + parser->top_obj = obj; + parser->cur_obj = obj; + } + } break; case UCL_STATE_KEY: @@ -1936,7 +2050,7 @@ ucl_state_machine (struct ucl_parser *parser) else if (parser->state != UCL_STATE_MACRO_NAME) { if (next_key && parser->stack->obj->type == UCL_OBJECT) { /* Parse more keys and nest objects accordingly */ - obj = ucl_add_parser_stack (parser->cur_obj, parser, false, + obj = ucl_parser_add_container (parser->cur_obj, parser, false, parser->stack->level + 1); if (obj == NULL) { return false; @@ -1967,6 +2081,7 @@ ucl_state_machine (struct ucl_parser *parser) parser->state = UCL_STATE_ERROR; return false; } + if (parser->stack != NULL) { if (parser->stack->obj->type == UCL_OBJECT) { parser->state = UCL_STATE_KEY; @@ -1987,20 +2102,36 @@ ucl_state_machine (struct ucl_parser *parser) *p != '(') { ucl_chunk_skipc (chunk, p); } - else if (p - c > 0) { - /* We got macro name */ - macro_len = (size_t)(p - c); - HASH_FIND (hh, parser->macroes, c, macro_len, macro); - if (macro == NULL) { - ucl_create_err (&parser->err, "error on line %d at column %d: " - "unknown macro: '%.*s', character: '%c'", - chunk->line, chunk->column, (int)(p - c), c, *chunk->pos); + else { + if (p - c > 0) { + /* We got macro name */ + macro_len = (size_t) (p - c); + HASH_FIND (hh, parser->macroes, c, macro_len, macro); + if (macro == NULL) { + ucl_create_err (&parser->err, + "error on line %d at column %d: " + "unknown macro: '%.*s', character: '%c'", + chunk->line, + chunk->column, + (int) (p - c), + c, + *chunk->pos); + parser->state = UCL_STATE_ERROR; + return false; + } + /* Now we need to skip all spaces */ + SKIP_SPACES_COMMENTS(parser, chunk, p); + parser->state = UCL_STATE_MACRO; + } + else { + /* We have invalid macro name */ + ucl_create_err (&parser->err, + "error on line %d at column %d: invalid macro name", + chunk->line, + chunk->column); parser->state = UCL_STATE_ERROR; return false; } - /* Now we need to skip all spaces */ - SKIP_SPACES_COMMENTS(parser, chunk, p); - parser->state = UCL_STATE_MACRO; } break; case UCL_STATE_MACRO: @@ -2024,14 +2155,36 @@ ucl_state_machine (struct ucl_parser *parser) macro_start, macro_len); parser->state = parser->prev_state; if (macro_escaped == NULL) { - ret = macro->handler (macro_start, macro_len, macro_args, - macro->ud); + if (macro->is_context) { + ret = macro->h.context_handler (macro_start, macro_len, + macro_args, + parser->top_obj, + macro->ud); + } + else { + ret = macro->h.handler (macro_start, macro_len, macro_args, + macro->ud); + } } else { - ret = macro->handler (macro_escaped, macro_len, macro_args, + if (macro->is_context) { + ret = macro->h.context_handler (macro_escaped, macro_len, + macro_args, + parser->top_obj, + macro->ud); + } + else { + ret = macro->h.handler (macro_escaped, macro_len, macro_args, macro->ud); + } + UCL_FREE (macro_len + 1, macro_escaped); } + + /* + * Chunk can be modified within macro handler + */ + chunk = parser->chunks; p = chunk->pos; if (macro_args) { ucl_object_unref (macro_args); @@ -2061,13 +2214,18 @@ ucl_parser_new (int flags) if (new == NULL) { return NULL; } + memset (new, 0, sizeof (struct ucl_parser)); ucl_parser_register_macro (new, "include", ucl_include_handler, new); ucl_parser_register_macro (new, "try_include", ucl_try_include_handler, new); ucl_parser_register_macro (new, "includes", ucl_includes_handler, new); + ucl_parser_register_macro (new, "priority", ucl_priority_handler, new); + ucl_parser_register_macro (new, "load", ucl_load_handler, new); + ucl_parser_register_context_macro (new, "inherit", ucl_inherit_handler, new); new->flags = flags; + new->includepaths = NULL; /* Initial assumption about filevars */ ucl_parser_set_filevars (new, NULL, false); @@ -2075,6 +2233,17 @@ ucl_parser_new (int flags) return new; } +bool +ucl_parser_set_default_priority (struct ucl_parser *parser, unsigned prio) +{ + if (parser == NULL) { + return false; + } + + parser->default_priority = prio; + + return true; +} void ucl_parser_register_macro (struct ucl_parser *parser, const char *macro, @@ -2085,14 +2254,39 @@ ucl_parser_register_macro (struct ucl_parser *parser, const char *macro, if (macro == NULL || handler == NULL) { return; } + + new = UCL_ALLOC (sizeof (struct ucl_macro)); + if (new == NULL) { + return; + } + + memset (new, 0, sizeof (struct ucl_macro)); + new->h.handler = handler; + new->name = strdup (macro); + new->ud = ud; + HASH_ADD_KEYPTR (hh, parser->macroes, new->name, strlen (new->name), new); +} + +void +ucl_parser_register_context_macro (struct ucl_parser *parser, const char *macro, + ucl_context_macro_handler handler, void* ud) +{ + struct ucl_macro *new; + + if (macro == NULL || handler == NULL) { + return; + } + new = UCL_ALLOC (sizeof (struct ucl_macro)); if (new == NULL) { return; } + memset (new, 0, sizeof (struct ucl_macro)); - new->handler = handler; + new->h.context_handler = handler; new->name = strdup (macro); new->ud = ud; + new->is_context = true; HASH_ADD_KEYPTR (hh, parser->macroes, new->name, strlen (new->name), new); } @@ -2159,11 +2353,16 @@ ucl_parser_set_variables_handler (struct ucl_parser *parser, } bool -ucl_parser_add_chunk_priority (struct ucl_parser *parser, const unsigned char *data, - size_t len, unsigned priority) +ucl_parser_add_chunk_full (struct ucl_parser *parser, const unsigned char *data, + size_t len, unsigned priority, enum ucl_duplicate_strategy strat, + enum ucl_parse_type parse_type) { struct ucl_chunk *chunk; + if (parser == NULL) { + return false; + } + if (data == NULL) { ucl_create_err (&parser->err, "invalid chunk added"); return false; @@ -2185,14 +2384,24 @@ ucl_parser_add_chunk_priority (struct ucl_parser *parser, const unsigned char *d chunk->line = 1; chunk->column = 0; chunk->priority = priority; + chunk->strategy = strat; + chunk->parse_type = parse_type; LL_PREPEND (parser->chunks, chunk); parser->recursion ++; + if (parser->recursion > UCL_MAX_RECURSION) { ucl_create_err (&parser->err, "maximum include nesting limit is reached: %d", parser->recursion); return false; } - return ucl_state_machine (parser); + + switch (parse_type) { + default: + case UCL_PARSE_UCL: + return ucl_state_machine (parser); + case UCL_PARSE_MSGPACK: + return ucl_parse_msgpack (parser); + } } ucl_create_err (&parser->err, "a parser is in an invalid state"); @@ -2201,15 +2410,33 @@ ucl_parser_add_chunk_priority (struct ucl_parser *parser, const unsigned char *d } bool +ucl_parser_add_chunk_priority (struct ucl_parser *parser, + const unsigned char *data, size_t len, unsigned priority) +{ + /* We dereference parser, so this check is essential */ + if (parser == NULL) { + return false; + } + + return ucl_parser_add_chunk_full (parser, data, len, + priority, UCL_DUPLICATE_APPEND, UCL_PARSE_UCL); +} + +bool ucl_parser_add_chunk (struct ucl_parser *parser, const unsigned char *data, size_t len) { - return ucl_parser_add_chunk_priority (parser, data, len, 0); + if (parser == NULL) { + return false; + } + + return ucl_parser_add_chunk_full (parser, data, len, + parser->default_priority, UCL_DUPLICATE_APPEND, UCL_PARSE_UCL); } bool -ucl_parser_add_string (struct ucl_parser *parser, const char *data, - size_t len) +ucl_parser_add_string_priority (struct ucl_parser *parser, const char *data, + size_t len, unsigned priority) { if (data == NULL) { ucl_create_err (&parser->err, "invalid string added"); @@ -2219,5 +2446,40 @@ ucl_parser_add_string (struct ucl_parser *parser, const char *data, len = strlen (data); } - return ucl_parser_add_chunk (parser, (const unsigned char *)data, len); + return ucl_parser_add_chunk_priority (parser, + (const unsigned char *)data, len, priority); +} + +bool +ucl_parser_add_string (struct ucl_parser *parser, const char *data, + size_t len) +{ + if (parser == NULL) { + return false; + } + + return ucl_parser_add_string_priority (parser, + (const unsigned char *)data, len, parser->default_priority); +} + +bool +ucl_set_include_path (struct ucl_parser *parser, ucl_object_t *paths) +{ + if (parser == NULL || paths == NULL) { + return false; + } + + if (parser->includepaths == NULL) { + parser->includepaths = ucl_object_copy (paths); + } + else { + ucl_object_unref (parser->includepaths); + parser->includepaths = ucl_object_copy (paths); + } + + if (parser->includepaths == NULL) { + return false; + } + + return true; } diff --git a/src/ucl_sexp.c b/src/ucl_sexp.c new file mode 100644 index 000000000000..3ca7eb102925 --- /dev/null +++ b/src/ucl_sexp.c @@ -0,0 +1,224 @@ +/* + * Copyright (c) 2015, Vsevolod Stakhov + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are met: + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY AUTHOR ''AS IS'' AND ANY + * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE + * DISCLAIMED. IN NO EVENT SHALL AUTHOR BE LIABLE FOR ANY + * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND + * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include <ucl.h> +#include "ucl.h" +#include "ucl_internal.h" +#include "utlist.h" + +#define NEXT_STATE do { \ +if (p >= end) { \ + if (state != read_ebrace) { \ + ucl_create_err (&parser->err,\ + "extra data");\ + state = parse_err; \ + } \ +} \ +else { \ +switch (*p) { \ + case '(': \ + state = read_obrace; \ + break; \ + case ')': \ + state = read_ebrace; \ + break; \ + default: \ + len = 0; \ + mult = 1; \ + state = read_length; \ + break; \ + } \ +} \ +} while(0) + +bool +ucl_parse_csexp (struct ucl_parser *parser) +{ + const unsigned char *p, *end; + ucl_object_t *obj; + struct ucl_stack *st; + uint64_t len = 0, mult = 1; + enum { + start_parse, + read_obrace, + read_length, + read_value, + read_ebrace, + parse_err + } state = start_parse; + + assert (parser != NULL); + assert (parser->chunks != NULL); + assert (parser->chunks->begin != NULL); + assert (parser->chunks->remain != 0); + + p = parser->chunks->begin; + end = p + parser->chunks->remain; + + while (p < end) { + switch (state) { + case start_parse: + /* At this point we expect open brace */ + if (*p == '(') { + state = read_obrace; + } + else { + ucl_create_err (&parser->err, "bad starting character for " + "sexp block: %x", (int)*p); + state = parse_err; + } + break; + + case read_obrace: + st = calloc (1, sizeof (*st)); + + if (st == NULL) { + ucl_create_err (&parser->err, "no memory"); + state = parse_err; + continue; + } + + st->obj = ucl_object_typed_new (UCL_ARRAY); + + if (st->obj == NULL) { + ucl_create_err (&parser->err, "no memory"); + state = parse_err; + continue; + } + + if (parser->stack == NULL) { + /* We have no stack */ + parser->stack = st; + + if (parser->top_obj == NULL) { + parser->top_obj = st->obj; + } + } + else { + /* Prepend new element to the stack */ + LL_PREPEND (parser->stack, st); + } + + p ++; + NEXT_STATE; + + break; + + case read_length: + if (*p == ':') { + if (len == 0) { + ucl_create_err (&parser->err, "zero length element"); + state = parse_err; + continue; + } + + state = read_value; + } + else if (*p >= '0' && *p <= '9') { + len += (*p - '0') * mult; + mult *= 10; + + if (len > UINT32_MAX) { + ucl_create_err (&parser->err, "too big length of an " + "element"); + state = parse_err; + continue; + } + } + else { + ucl_create_err (&parser->err, "bad length character: %x", + (int)*p); + state = parse_err; + continue; + } + + p ++; + break; + + case read_value: + if ((uint64_t)(end - p) > len || len == 0) { + ucl_create_err (&parser->err, "invalid length: %llu, %ld " + "remain", (long long unsigned)len, (long)(end - p)); + state = parse_err; + continue; + } + obj = ucl_object_typed_new (UCL_STRING); + + obj->value.sv = (const char*)p; + obj->len = len; + obj->flags |= UCL_OBJECT_BINARY; + + if (!(parser->flags & UCL_PARSER_ZEROCOPY)) { + ucl_copy_value_trash (obj); + } + + ucl_array_append (parser->stack->obj, obj); + p += len; + NEXT_STATE; + break; + + case read_ebrace: + if (parser->stack == NULL) { + /* We have an extra end brace */ + ucl_create_err (&parser->err, "invalid length: %llu, %ld " + "remain", (long long unsigned)len, (long)(end - p)); + state = parse_err; + continue; + } + /* Pop the container */ + st = parser->stack; + parser->stack = st->next; + + if (parser->stack->obj->type == UCL_ARRAY) { + ucl_array_append (parser->stack->obj, st->obj); + } + else { + ucl_create_err (&parser->err, "bad container object, array " + "expected"); + state = parse_err; + continue; + } + + free (st); + p++; + NEXT_STATE; + break; + + case parse_err: + default: + return false; + } + } + + if (state != read_ebrace) { + ucl_create_err (&parser->err, "invalid finishing state: %d", state); + return false; + } + + return true; +}
\ No newline at end of file diff --git a/src/ucl_util.c b/src/ucl_util.c index 41e012bf15bb..730a5c4afd57 100644 --- a/src/ucl_util.c +++ b/src/ucl_util.c @@ -1,4 +1,5 @@ /* Copyright (c) 2013, Vsevolod Stakhov + * Copyright (c) 2015 Allan Jude <allanjude@freebsd.org> * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -25,6 +26,7 @@ #include "ucl_internal.h" #include "ucl_chartable.h" #include "kvec.h" +#include <stdarg.h> #ifndef _WIN32 #include <glob.h> @@ -230,11 +232,13 @@ ucl_object_free_internal (ucl_object_t *obj, bool allow_rec, ucl_object_dtor dto kv_destroy (*vec); UCL_FREE (sizeof (*vec), vec); } + obj->value.av = NULL; } else if (obj->type == UCL_OBJECT) { if (obj->value.ov != NULL) { ucl_hash_destroy (obj->value.ov, (ucl_hash_free_func *)dtor); } + obj->value.ov = NULL; } tmp = obj->next; dtor (obj); @@ -266,6 +270,17 @@ ucl_unescape_json_string (char *str, size_t len) while (len) { if (*h == '\\') { h ++; + + if (len == 1) { + /* + * If \ is last, then do not try to go further + * Issue: #74 + */ + len --; + *t++ = '\\'; + continue; + } + switch (*h) { case 'n': *t++ = '\n'; @@ -350,7 +365,10 @@ ucl_unescape_json_string (char *str, size_t len) else { *t++ = *h++; } - len --; + + if (len > 0) { + len --; + } } *t = '\0'; @@ -392,11 +410,24 @@ ucl_copy_value_trash (const ucl_object_t *obj) if (obj->type == UCL_STRING) { /* Special case for strings */ - deconst->trash_stack[UCL_TRASH_VALUE] = malloc (obj->len + 1); - if (deconst->trash_stack[UCL_TRASH_VALUE] != NULL) { - memcpy (deconst->trash_stack[UCL_TRASH_VALUE], obj->value.sv, obj->len); - deconst->trash_stack[UCL_TRASH_VALUE][obj->len] = '\0'; - deconst->value.sv = obj->trash_stack[UCL_TRASH_VALUE]; + if (obj->flags & UCL_OBJECT_BINARY) { + deconst->trash_stack[UCL_TRASH_VALUE] = malloc (obj->len); + if (deconst->trash_stack[UCL_TRASH_VALUE] != NULL) { + memcpy (deconst->trash_stack[UCL_TRASH_VALUE], + obj->value.sv, + obj->len); + deconst->value.sv = obj->trash_stack[UCL_TRASH_VALUE]; + } + } + else { + deconst->trash_stack[UCL_TRASH_VALUE] = malloc (obj->len + 1); + if (deconst->trash_stack[UCL_TRASH_VALUE] != NULL) { + memcpy (deconst->trash_stack[UCL_TRASH_VALUE], + obj->value.sv, + obj->len); + deconst->trash_stack[UCL_TRASH_VALUE][obj->len] = '\0'; + deconst->value.sv = obj->trash_stack[UCL_TRASH_VALUE]; + } } } else { @@ -406,10 +437,11 @@ ucl_copy_value_trash (const ucl_object_t *obj) } deconst->flags |= UCL_OBJECT_ALLOCATED_VALUE; } + return obj->trash_stack[UCL_TRASH_VALUE]; } -UCL_EXTERN ucl_object_t* +ucl_object_t* ucl_parser_get_object (struct ucl_parser *parser) { if (parser->state != UCL_STATE_ERROR && parser->top_obj != NULL) { @@ -419,7 +451,7 @@ ucl_parser_get_object (struct ucl_parser *parser) return NULL; } -UCL_EXTERN void +void ucl_parser_free (struct ucl_parser *parser) { struct ucl_stack *stack, *stmp; @@ -427,6 +459,7 @@ ucl_parser_free (struct ucl_parser *parser) struct ucl_chunk *chunk, *ctmp; struct ucl_pubkey *key, *ktmp; struct ucl_variable *var, *vtmp; + ucl_object_t *tr, *trtmp; if (parser == NULL) { return; @@ -436,6 +469,10 @@ ucl_parser_free (struct ucl_parser *parser) ucl_object_unref (parser->top_obj); } + if (parser->includepaths != NULL) { + ucl_object_unref (parser->includepaths); + } + LL_FOREACH_SAFE (parser->stack, stack, stmp) { free (stack); } @@ -455,6 +492,9 @@ ucl_parser_free (struct ucl_parser *parser) free (var->var); UCL_FREE (sizeof (struct ucl_variable), var); } + LL_FOREACH_SAFE (parser->trash_objs, tr, trtmp) { + ucl_object_free_internal (tr, false, ucl_object_dtor_free); + } if (parser->err != NULL) { utstring_free (parser->err); @@ -467,29 +507,61 @@ ucl_parser_free (struct ucl_parser *parser) UCL_FREE (sizeof (struct ucl_parser), parser); } -UCL_EXTERN const char * +const char * ucl_parser_get_error(struct ucl_parser *parser) { if (parser == NULL) { return NULL; } - if (parser->err == NULL) + if (parser->err == NULL) { return NULL; + } - return utstring_body(parser->err); + return utstring_body (parser->err); } -UCL_EXTERN void +int +ucl_parser_get_error_code(struct ucl_parser *parser) +{ + if (parser == NULL) { + return 0; + } + + return parser->err_code; +} + +unsigned +ucl_parser_get_column(struct ucl_parser *parser) +{ + if (parser == NULL || parser->chunks == NULL) { + return 0; + } + + return parser->chunks->column; +} + +unsigned +ucl_parser_get_linenum(struct ucl_parser *parser) +{ + if (parser == NULL || parser->chunks == NULL) { + return 0; + } + + return parser->chunks->line; +} + +void ucl_parser_clear_error(struct ucl_parser *parser) { if (parser != NULL && parser->err != NULL) { utstring_free(parser->err); parser->err = NULL; + parser->err_code = 0; } } -UCL_EXTERN bool +bool ucl_pubkey_add (struct ucl_parser *parser, const unsigned char *key, size_t len) { #ifndef HAVE_OPENSSL @@ -738,6 +810,20 @@ ucl_sig_check (const unsigned char *data, size_t datalen, } #endif +struct ucl_include_params { + bool check_signature; + bool must_exist; + bool use_glob; + bool use_prefix; + bool soft_fail; + bool allow_glob; + unsigned priority; + enum ucl_duplicate_strategy strat; + enum ucl_parse_type parse_type; + const char *prefix; + const char *target; +}; + /** * Include an url to configuration * @param data @@ -748,8 +834,8 @@ ucl_sig_check (const unsigned char *data, size_t datalen, */ static bool ucl_include_url (const unsigned char *data, size_t len, - struct ucl_parser *parser, bool check_signature, bool must_exist, - unsigned priority) + struct ucl_parser *parser, + struct ucl_include_params *params) { bool res; @@ -761,11 +847,11 @@ ucl_include_url (const unsigned char *data, size_t len, snprintf (urlbuf, sizeof (urlbuf), "%.*s", (int)len, data); - if (!ucl_fetch_url (urlbuf, &buf, &buflen, &parser->err, must_exist)) { - return (!must_exist || false); + if (!ucl_fetch_url (urlbuf, &buf, &buflen, &parser->err, params->must_exist)) { + return (!params->must_exist || false); } - if (check_signature) { + if (params->check_signature) { #if (defined(HAVE_OPENSSL) && OPENSSL_VERSION_NUMBER >= 0x10000000L) unsigned char *sigbuf = NULL; size_t siglen = 0; @@ -792,7 +878,8 @@ ucl_include_url (const unsigned char *data, size_t len, prev_state = parser->state; parser->state = UCL_STATE_INIT; - res = ucl_parser_add_chunk_priority (parser, buf, buflen, priority); + res = ucl_parser_add_chunk_full (parser, buf, buflen, params->priority, + params->strat, params->parse_type); if (res == true) { /* Remove chunk from the stack */ chunk = parser->chunks; @@ -821,22 +908,27 @@ ucl_include_url (const unsigned char *data, size_t len, */ static bool ucl_include_file_single (const unsigned char *data, size_t len, - struct ucl_parser *parser, bool check_signature, bool must_exist, - unsigned priority) + struct ucl_parser *parser, struct ucl_include_params *params) { bool res; struct ucl_chunk *chunk; unsigned char *buf = NULL; - char *old_curfile; - size_t buflen; + char *old_curfile, *ext; + size_t buflen = 0; char filebuf[PATH_MAX], realbuf[PATH_MAX]; int prev_state; struct ucl_variable *cur_var, *tmp_var, *old_curdir = NULL, *old_filename = NULL; + ucl_object_t *nest_obj = NULL, *old_obj = NULL, *new_obj = NULL; + ucl_hash_t *container = NULL; + struct ucl_stack *st = NULL; snprintf (filebuf, sizeof (filebuf), "%.*s", (int)len, data); if (ucl_realpath (filebuf, realbuf) == NULL) { - if (!must_exist) { + if (params->soft_fail) { + return false; + } + if (!params->must_exist) { return true; } ucl_create_err (&parser->err, "cannot open file %s: %s", @@ -847,16 +939,23 @@ ucl_include_file_single (const unsigned char *data, size_t len, if (parser->cur_file && strcmp (realbuf, parser->cur_file) == 0) { /* We are likely including the file itself */ + if (params->soft_fail) { + return false; + } + ucl_create_err (&parser->err, "trying to include the file %s from itself", realbuf); return false; } - if (!ucl_fetch_file (realbuf, &buf, &buflen, &parser->err, must_exist)) { - return (!must_exist || false); + if (!ucl_fetch_file (realbuf, &buf, &buflen, &parser->err, params->must_exist)) { + if (params->soft_fail) { + return false; + } + return (!params->must_exist || false); } - if (check_signature) { + if (params->check_signature) { #if (defined(HAVE_OPENSSL) && OPENSSL_VERSION_NUMBER >= 0x10000000L) unsigned char *sigbuf = NULL; size_t siglen = 0; @@ -900,14 +999,128 @@ ucl_include_file_single (const unsigned char *data, size_t len, prev_state = parser->state; parser->state = UCL_STATE_INIT; - res = ucl_parser_add_chunk_priority (parser, buf, buflen, priority); - if (!res && !must_exist) { + if (params->use_prefix && params->prefix == NULL) { + /* Auto generate a key name based on the included filename */ + params->prefix = basename (realbuf); + ext = strrchr (params->prefix, '.'); + if (ext != NULL && (strcmp (ext, ".conf") == 0 || strcmp (ext, ".ucl") == 0)) { + /* Strip off .conf or .ucl */ + *ext = '\0'; + } + } + if (params->prefix != NULL) { + /* This is a prefixed include */ + container = parser->stack->obj->value.ov; + + old_obj = __DECONST (ucl_object_t *, ucl_hash_search (container, + params->prefix, strlen (params->prefix))); + + if (strcasecmp (params->target, "array") == 0 && old_obj == NULL) { + /* Create an array with key: prefix */ + old_obj = ucl_object_new_full (UCL_ARRAY, params->priority); + old_obj->key = params->prefix; + old_obj->keylen = strlen (params->prefix); + ucl_copy_key_trash(old_obj); + old_obj->prev = old_obj; + old_obj->next = NULL; + + container = ucl_hash_insert_object (container, old_obj, + parser->flags & UCL_PARSER_KEY_LOWERCASE); + parser->stack->obj->len ++; + + nest_obj = ucl_object_new_full (UCL_OBJECT, params->priority); + nest_obj->prev = nest_obj; + nest_obj->next = NULL; + + ucl_array_append (old_obj, nest_obj); + } + else if (old_obj == NULL) { + /* Create an object with key: prefix */ + nest_obj = ucl_object_new_full (UCL_OBJECT, params->priority); + nest_obj->key = params->prefix; + nest_obj->keylen = strlen (params->prefix); + ucl_copy_key_trash(nest_obj); + nest_obj->prev = nest_obj; + nest_obj->next = NULL; + + container = ucl_hash_insert_object (container, nest_obj, + parser->flags & UCL_PARSER_KEY_LOWERCASE); + parser->stack->obj->len ++; + } + else if (strcasecmp (params->target, "array") == 0 || + ucl_object_type(old_obj) == UCL_ARRAY) { + if (ucl_object_type(old_obj) == UCL_ARRAY) { + /* Append to the existing array */ + nest_obj = ucl_object_new_full (UCL_OBJECT, params->priority); + nest_obj->prev = nest_obj; + nest_obj->next = NULL; + + ucl_array_append (old_obj, nest_obj); + } + else { + /* Convert the object to an array */ + new_obj = ucl_object_typed_new (UCL_ARRAY); + new_obj->key = old_obj->key; + new_obj->keylen = old_obj->keylen; + new_obj->flags |= UCL_OBJECT_MULTIVALUE; + new_obj->prev = new_obj; + new_obj->next = NULL; + + nest_obj = ucl_object_new_full (UCL_OBJECT, params->priority); + nest_obj->prev = nest_obj; + nest_obj->next = NULL; + + ucl_array_append (new_obj, old_obj); + ucl_array_append (new_obj, nest_obj); + ucl_hash_replace (container, old_obj, new_obj); + } + } + else { + if (ucl_object_type (old_obj) == UCL_OBJECT) { + /* Append to existing Object*/ + nest_obj = old_obj; + } + else { + /* The key is not an object */ + ucl_create_err (&parser->err, + "Conflicting type for key: %s", + params->prefix); + return false; + } + } + + /* Put all of the content of the include inside that object */ + parser->stack->obj->value.ov = container; + + if (nest_obj != NULL) { + st = UCL_ALLOC (sizeof (struct ucl_stack)); + if (st == NULL) { + ucl_create_err (&parser->err, "cannot allocate memory for an object"); + ucl_object_unref (nest_obj); + return NULL; + } + st->obj = nest_obj; + st->level = parser->stack->level; + LL_PREPEND (parser->stack, st); + parser->cur_obj = nest_obj; + } + } + + res = ucl_parser_add_chunk_full (parser, buf, buflen, params->priority, + params->strat, params->parse_type); + if (!res && !params->must_exist) { /* Free error */ utstring_free (parser->err); parser->err = NULL; parser->state = UCL_STATE_AFTER_VALUE; } + /* Stop nesting the include, take 1 level off the stack */ + if (params->prefix != NULL && nest_obj != NULL) { + parser->stack = st->next; + UCL_FREE (sizeof (struct ucl_stack), st); + } + /* Remove chunk from the stack */ chunk = parser->chunks; if (chunk != NULL) { @@ -917,6 +1130,10 @@ ucl_include_file_single (const unsigned char *data, size_t len, } /* Restore old file vars */ + if (parser->cur_file) { + free (parser->cur_file); + } + parser->cur_file = old_curfile; DL_FOREACH_SAFE (parser->variables, cur_var, tmp_var) { if (strcmp (cur_var->var, "CURDIR") == 0 && old_curdir) { @@ -938,9 +1155,6 @@ ucl_include_file_single (const unsigned char *data, size_t len, if (old_curdir) { DL_APPEND (parser->variables, old_curdir); } - if (old_curfile) { - free (old_curfile); - } parser->state = prev_state; @@ -961,8 +1175,7 @@ ucl_include_file_single (const unsigned char *data, size_t len, */ static bool ucl_include_file (const unsigned char *data, size_t len, - struct ucl_parser *parser, bool check_signature, bool must_exist, - bool allow_glob, unsigned priority) + struct ucl_parser *parser, struct ucl_include_params *params) { const unsigned char *p = data, *end = data + len; bool need_glob = false; @@ -971,9 +1184,8 @@ ucl_include_file (const unsigned char *data, size_t len, size_t i; #ifndef _WIN32 - if (!allow_glob) { - return ucl_include_file_single (data, len, parser, check_signature, - must_exist, priority); + if (!params->allow_glob) { + return ucl_include_file_single (data, len, parser, params); } else { /* Check for special symbols in a filename */ @@ -987,14 +1199,17 @@ ucl_include_file (const unsigned char *data, size_t len, if (need_glob) { glob_t globbuf; memset (&globbuf, 0, sizeof (globbuf)); - ucl_strlcpy (glob_pattern, (const char *)data, sizeof (glob_pattern)); + ucl_strlcpy (glob_pattern, (const char *)data, + (len + 1 < sizeof (glob_pattern) ? len + 1 : sizeof (glob_pattern))); if (glob (glob_pattern, 0, NULL, &globbuf) != 0) { - return (!must_exist || false); + return (!params->must_exist || false); } for (i = 0; i < globbuf.gl_pathc; i ++) { if (!ucl_include_file_single ((unsigned char *)globbuf.gl_pathv[i], - strlen (globbuf.gl_pathv[i]), parser, check_signature, - must_exist, priority)) { + strlen (globbuf.gl_pathv[i]), parser, params)) { + if (params->soft_fail) { + continue; + } globfree (&globbuf); return false; } @@ -1002,22 +1217,20 @@ ucl_include_file (const unsigned char *data, size_t len, } globfree (&globbuf); - if (cnt == 0 && must_exist) { + if (cnt == 0 && params->must_exist) { ucl_create_err (&parser->err, "cannot match any files for pattern %s", glob_pattern); return false; } } else { - return ucl_include_file_single (data, len, parser, check_signature, - must_exist, priority); + return ucl_include_file_single (data, len, parser, params); } } #else /* Win32 compilers do not support globbing. Therefore, for Win32, treat allow_glob/need_glob as a NOOP and just return */ - return ucl_include_file_single (data, len, parser, check_signature, - must_exist, priority); + return ucl_include_file_single (data, len, parser, params); #endif return true; @@ -1039,52 +1252,123 @@ ucl_include_common (const unsigned char *data, size_t len, bool default_try, bool default_sign) { - bool try_load, allow_glob, allow_url, need_sign; - unsigned priority; + bool allow_url, search; + const char *duplicate; const ucl_object_t *param; - ucl_object_iter_t it = NULL; + ucl_object_iter_t it = NULL, ip = NULL; + char ipath[PATH_MAX]; + struct ucl_include_params params; /* Default values */ - try_load = default_try; - allow_glob = false; - allow_url = true; - need_sign = default_sign; - priority = 0; + params.soft_fail = default_try; + params.allow_glob = false; + params.check_signature = default_sign; + params.use_prefix = false; + params.target = "object"; + params.prefix = NULL; + params.priority = 0; + params.parse_type = UCL_PARSE_UCL; + params.strat = UCL_DUPLICATE_APPEND; + params.must_exist = !default_try; + + search = false; /* Process arguments */ if (args != NULL && args->type == UCL_OBJECT) { while ((param = ucl_iterate_object (args, &it, true)) != NULL) { if (param->type == UCL_BOOLEAN) { - if (strcmp (param->key, "try") == 0) { - try_load = ucl_object_toboolean (param); + if (strncmp (param->key, "try", param->keylen) == 0) { + params.must_exist = !ucl_object_toboolean (param); + } + else if (strncmp (param->key, "sign", param->keylen) == 0) { + params.check_signature = ucl_object_toboolean (param); } - else if (strcmp (param->key, "sign") == 0) { - need_sign = ucl_object_toboolean (param); + else if (strncmp (param->key, "glob", param->keylen) == 0) { + params.allow_glob = ucl_object_toboolean (param); } - else if (strcmp (param->key, "glob") == 0) { - allow_glob = ucl_object_toboolean (param); + else if (strncmp (param->key, "url", param->keylen) == 0) { + allow_url = ucl_object_toboolean (param); + } + else if (strncmp (param->key, "prefix", param->keylen) == 0) { + params.use_prefix = ucl_object_toboolean (param); + } + } + else if (param->type == UCL_STRING) { + if (strncmp (param->key, "key", param->keylen) == 0) { + params.prefix = ucl_object_tostring (param); + } + else if (strncmp (param->key, "target", param->keylen) == 0) { + params.target = ucl_object_tostring (param); + } + else if (strncmp (param->key, "duplicate", param->keylen) == 0) { + duplicate = ucl_object_tostring (param); + + if (strcmp (duplicate, "append") == 0) { + params.strat = UCL_DUPLICATE_APPEND; + } + else if (strcmp (duplicate, "merge") == 0) { + params.strat = UCL_DUPLICATE_MERGE; + } + else if (strcmp (duplicate, "rewrite") == 0) { + params.strat = UCL_DUPLICATE_REWRITE; + } + else if (strcmp (duplicate, "error") == 0) { + params.strat = UCL_DUPLICATE_ERROR; + } } - else if (strcmp (param->key, "url") == 0) { - allow_url = ucl_object_toboolean (param); + } + else if (param->type == UCL_ARRAY) { + if (strncmp (param->key, "path", param->keylen) == 0) { + ucl_set_include_path (parser, __DECONST(ucl_object_t *, param)); } } else if (param->type == UCL_INT) { - if (strcmp (param->key, "priority") == 0) { - priority = ucl_object_toint (param); + if (strncmp (param->key, "priority", param->keylen) == 0) { + params.priority = ucl_object_toint (param); } } } } - if (*data == '/' || *data == '.') { - /* Try to load a file */ - return ucl_include_file (data, len, parser, need_sign, !try_load, - allow_glob, priority); + if (parser->includepaths == NULL) { + if (allow_url && ucl_strnstr (data, "://", len) != NULL) { + /* Globbing is not used for URL's */ + return ucl_include_url (data, len, parser, ¶ms); + } + else if (data != NULL) { + /* Try to load a file */ + return ucl_include_file (data, len, parser, ¶ms); + } } - else if (allow_url) { - /* Globbing is not used for URL's */ - return ucl_include_url (data, len, parser, need_sign, !try_load, - priority); + else { + if (allow_url && ucl_strnstr (data, "://", len) != NULL) { + /* Globbing is not used for URL's */ + return ucl_include_url (data, len, parser, ¶ms); + } + + ip = ucl_object_iterate_new (parser->includepaths); + while ((param = ucl_object_iterate_safe (ip, true)) != NULL) { + if (ucl_object_type(param) == UCL_STRING) { + snprintf (ipath, sizeof (ipath), "%s/%.*s", ucl_object_tostring(param), + (int)len, data); + if ((search = ucl_include_file (ipath, strlen (ipath), + parser, ¶ms))) { + if (!params.allow_glob) { + break; + } + } + } + } + ucl_object_iterate_free (ip); + if (search == true) { + return true; + } + else { + ucl_create_err (&parser->err, + "cannot find file: %.*s in search path", + (int)len, data); + return false; + } } return false; @@ -1094,11 +1378,11 @@ ucl_include_common (const unsigned char *data, size_t len, * Handle include macro * @param data include data * @param len length of data + * @param args UCL object representing arguments to the macro * @param ud user data - * @param err error ptr * @return */ -UCL_EXTERN bool +bool ucl_include_handler (const unsigned char *data, size_t len, const ucl_object_t *args, void* ud) { @@ -1111,11 +1395,11 @@ ucl_include_handler (const unsigned char *data, size_t len, * Handle includes macro * @param data include data * @param len length of data + * @param args UCL object representing arguments to the macro * @param ud user data - * @param err error ptr * @return */ -UCL_EXTERN bool +bool ucl_includes_handler (const unsigned char *data, size_t len, const ucl_object_t *args, void* ud) { @@ -1124,8 +1408,15 @@ ucl_includes_handler (const unsigned char *data, size_t len, return ucl_include_common (data, len, args, parser, false, true); } - -UCL_EXTERN bool +/** + * Handle tryinclude macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ud user data + * @return + */ +bool ucl_try_include_handler (const unsigned char *data, size_t len, const ucl_object_t *args, void* ud) { @@ -1134,7 +1425,251 @@ ucl_try_include_handler (const unsigned char *data, size_t len, return ucl_include_common (data, len, args, parser, true, false); } -UCL_EXTERN bool +/** + * Handle priority macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ud user data + * @return + */ +bool +ucl_priority_handler (const unsigned char *data, size_t len, + const ucl_object_t *args, void* ud) +{ + struct ucl_parser *parser = ud; + unsigned priority = 255; + const ucl_object_t *param; + bool found = false; + char *value = NULL, *leftover = NULL; + ucl_object_iter_t it = NULL; + + if (parser == NULL) { + return false; + } + + /* Process arguments */ + if (args != NULL && args->type == UCL_OBJECT) { + while ((param = ucl_iterate_object (args, &it, true)) != NULL) { + if (param->type == UCL_INT) { + if (strncmp (param->key, "priority", param->keylen) == 0) { + priority = ucl_object_toint (param); + found = true; + } + } + } + } + + if (len > 0) { + value = malloc(len + 1); + ucl_strlcpy(value, (const char *)data, len + 1); + priority = strtol(value, &leftover, 10); + if (*leftover != '\0') { + ucl_create_err (&parser->err, "Invalid priority value in macro: %s", + value); + free(value); + return false; + } + free(value); + found = true; + } + + if (found == true) { + parser->chunks->priority = priority; + return true; + } + + ucl_create_err (&parser->err, "Unable to parse priority macro"); + return false; +} + +/** + * Handle load macro + * @param data include data + * @param len length of data + * @param args UCL object representing arguments to the macro + * @param ud user data + * @return + */ +bool +ucl_load_handler (const unsigned char *data, size_t len, + const ucl_object_t *args, void* ud) +{ + struct ucl_parser *parser = ud; + const ucl_object_t *param; + ucl_object_t *obj, *old_obj; + ucl_object_iter_t it = NULL; + bool try_load, multiline, test; + const char *target, *prefix; + char *load_file, *tmp; + unsigned char *buf; + size_t buflen; + unsigned priority; + int64_t iv; + ucl_hash_t *container = NULL; + enum ucl_string_flags flags; + + /* Default values */ + try_load = false; + multiline = false; + test = false; + target = "string"; + prefix = NULL; + load_file = NULL; + buf = NULL; + buflen = 0; + priority = 0; + obj = NULL; + old_obj = NULL; + flags = 0; + + if (parser == NULL) { + return false; + } + + /* Process arguments */ + if (args != NULL && args->type == UCL_OBJECT) { + while ((param = ucl_iterate_object (args, &it, true)) != NULL) { + if (param->type == UCL_BOOLEAN) { + if (strncmp (param->key, "try", param->keylen) == 0) { + try_load = ucl_object_toboolean (param); + } + else if (strncmp (param->key, "multiline", param->keylen) == 0) { + multiline = ucl_object_toboolean (param); + } + else if (strncmp (param->key, "escape", param->keylen) == 0) { + test = ucl_object_toboolean (param); + if (test) { + flags |= UCL_STRING_ESCAPE; + } + } + else if (strncmp (param->key, "trim", param->keylen) == 0) { + test = ucl_object_toboolean (param); + if (test) { + flags |= UCL_STRING_TRIM; + } + } + } + else if (param->type == UCL_STRING) { + if (strncmp (param->key, "key", param->keylen) == 0) { + prefix = ucl_object_tostring (param); + } + else if (strncmp (param->key, "target", param->keylen) == 0) { + target = ucl_object_tostring (param); + } + } + else if (param->type == UCL_INT) { + if (strncmp (param->key, "priority", param->keylen) == 0) { + priority = ucl_object_toint (param); + } + } + } + } + + if (prefix == NULL || strlen(prefix) == 0) { + ucl_create_err (&parser->err, "No Key specified in load macro"); + return false; + } + + if (len > 0) { + asprintf (&load_file, "%.*s", (int)len, data); + if (!ucl_fetch_file (load_file, &buf, &buflen, &parser->err, !try_load)) { + return (try_load || false); + } + + container = parser->stack->obj->value.ov; + old_obj = __DECONST (ucl_object_t *, ucl_hash_search (container, prefix, strlen (prefix))); + if (old_obj != NULL) { + ucl_create_err (&parser->err, "Key %s already exists", prefix); + return false; + } + + if (strcasecmp (target, "string") == 0) { + obj = ucl_object_fromstring_common (buf, buflen, flags); + ucl_copy_value_trash (obj); + if (multiline) { + obj->flags |= UCL_OBJECT_MULTILINE; + } + } + else if (strcasecmp (target, "int") == 0) { + asprintf(&tmp, "%.*s", (int)buflen, buf); + iv = strtoll(tmp, NULL, 10); + obj = ucl_object_fromint(iv); + } + + if (buflen > 0) { + ucl_munmap (buf, buflen); + } + + if (obj != NULL) { + obj->key = prefix; + obj->keylen = strlen (prefix); + ucl_copy_key_trash(obj); + obj->prev = obj; + obj->next = NULL; + ucl_object_set_priority (obj, priority); + container = ucl_hash_insert_object (container, obj, + parser->flags & UCL_PARSER_KEY_LOWERCASE); + parser->stack->obj->value.ov = container; + } + return true; + } + + ucl_create_err (&parser->err, "Unable to parse load macro"); + return false; +} + +bool +ucl_inherit_handler (const unsigned char *data, size_t len, + const ucl_object_t *args, const ucl_object_t *ctx, void* ud) +{ + const ucl_object_t *parent, *cur; + ucl_object_t *target, *copy; + ucl_object_iter_t it = NULL; + bool replace = false; + struct ucl_parser *parser = ud; + + parent = ucl_object_find_keyl (ctx, data, len); + + /* Some sanity checks */ + if (parent == NULL || ucl_object_type (parent) != UCL_OBJECT) { + ucl_create_err (&parser->err, "Unable to find inherited object %*.s", + (int)len, data); + return false; + } + + if (parser->stack == NULL || parser->stack->obj == NULL || + ucl_object_type (parser->stack->obj) != UCL_OBJECT) { + ucl_create_err (&parser->err, "Invalid inherit context"); + return false; + } + + target = parser->stack->obj; + + if (args && (cur = ucl_object_find_key (args, "replace")) != NULL) { + replace = ucl_object_toboolean (cur); + } + + while ((cur = ucl_iterate_object (parent, &it, true))) { + /* We do not replace existing keys */ + if (!replace && ucl_object_find_keyl (target, cur->key, cur->keylen)) { + continue; + } + + copy = ucl_object_copy (cur); + + if (!replace) { + copy->flags |= UCL_OBJECT_INHERITED; + } + + ucl_object_insert_key (target, copy, copy->key, + copy->keylen, false); + } + + return true; +} + +bool ucl_parser_set_filevars (struct ucl_parser *parser, const char *filename, bool need_expand) { char realbuf[PATH_MAX], *curdir; @@ -1164,8 +1699,9 @@ ucl_parser_set_filevars (struct ucl_parser *parser, const char *filename, bool n return true; } -UCL_EXTERN bool -ucl_parser_add_file (struct ucl_parser *parser, const char *filename) +bool +ucl_parser_add_file_priority (struct ucl_parser *parser, const char *filename, + unsigned priority) { unsigned char *buf; size_t len; @@ -1188,7 +1724,7 @@ ucl_parser_add_file (struct ucl_parser *parser, const char *filename) } parser->cur_file = strdup (realbuf); ucl_parser_set_filevars (parser, realbuf, false); - ret = ucl_parser_add_chunk (parser, buf, len); + ret = ucl_parser_add_chunk_priority (parser, buf, len, priority); if (len > 0) { ucl_munmap (buf, len); @@ -1197,8 +1733,20 @@ ucl_parser_add_file (struct ucl_parser *parser, const char *filename) return ret; } -UCL_EXTERN bool -ucl_parser_add_fd (struct ucl_parser *parser, int fd) +bool +ucl_parser_add_file (struct ucl_parser *parser, const char *filename) +{ + if (parser == NULL) { + return false; + } + + return ucl_parser_add_file_priority(parser, filename, + parser->default_priority); +} + +bool +ucl_parser_add_fd_priority (struct ucl_parser *parser, int fd, + unsigned priority) { unsigned char *buf; size_t len; @@ -1221,7 +1769,7 @@ ucl_parser_add_fd (struct ucl_parser *parser, int fd) } parser->cur_file = NULL; len = st.st_size; - ret = ucl_parser_add_chunk (parser, buf, len); + ret = ucl_parser_add_chunk_priority (parser, buf, len, priority); if (len > 0) { ucl_munmap (buf, len); @@ -1230,6 +1778,16 @@ ucl_parser_add_fd (struct ucl_parser *parser, int fd) return ret; } +bool +ucl_parser_add_fd (struct ucl_parser *parser, int fd) +{ + if (parser == NULL) { + return false; + } + + return ucl_parser_add_fd_priority(parser, fd, parser->default_priority); +} + size_t ucl_strlcpy (char *dst, const char *src, size_t siz) { @@ -1285,6 +1843,51 @@ ucl_strlcpy_tolower (char *dst, const char *src, size_t siz) return (s - src); /* count does not include NUL */ } +/* + * Find the first occurrence of find in s + */ +char * +ucl_strnstr (const char *s, const char *find, int len) +{ + char c, sc; + int mlen; + + if ((c = *find++) != 0) { + mlen = strlen (find); + do { + do { + if ((sc = *s++) == 0 || len-- == 0) + return (NULL); + } while (sc != c); + } while (strncmp (s, find, mlen) != 0); + s--; + } + return ((char *)s); +} + +/* + * Find the first occurrence of find in s, ignore case. + */ +char * +ucl_strncasestr (const char *s, const char *find, int len) +{ + char c, sc; + int mlen; + + if ((c = *find++) != 0) { + c = tolower (c); + mlen = strlen (find); + do { + do { + if ((sc = *s++) == 0 || len-- == 0) + return (NULL); + } while (tolower (sc) != c); + } while (strncasecmp (s, find, mlen) != 0); + s--; + } + return ((char *)s); +} + ucl_object_t * ucl_object_fromstring_common (const char *str, size_t len, enum ucl_string_flags flags) { @@ -1539,7 +2142,7 @@ ucl_object_delete_keyl (ucl_object_t *top, const char *key, size_t keylen) bool ucl_object_delete_key (ucl_object_t *top, const char *key) { - return ucl_object_delete_keyl (top, key, strlen(key)); + return ucl_object_delete_keyl (top, key, strlen (key)); } ucl_object_t* @@ -1564,7 +2167,7 @@ ucl_object_pop_keyl (ucl_object_t *top, const char *key, size_t keylen) ucl_object_t* ucl_object_pop_key (ucl_object_t *top, const char *key) { - return ucl_object_pop_keyl (top, key, strlen(key)); + return ucl_object_pop_keyl (top, key, strlen (key)); } bool @@ -1642,10 +2245,45 @@ ucl_object_find_keyl (const ucl_object_t *obj, const char *key, size_t klen) const ucl_object_t * ucl_object_find_key (const ucl_object_t *obj, const char *key) { - if (key == NULL) + if (key == NULL) { return NULL; + } - return ucl_object_find_keyl (obj, key, strlen(key)); + return ucl_object_find_keyl (obj, key, strlen (key)); +} + +const ucl_object_t* +ucl_object_find_any_key (const ucl_object_t *obj, + const char *key, ...) +{ + va_list ap; + const ucl_object_t *ret = NULL; + const char *nk = NULL; + + if (obj == NULL || key == NULL) { + return NULL; + } + + ret = ucl_object_find_keyl (obj, key, strlen (key)); + + if (ret == NULL) { + va_start (ap, key); + + while (ret == NULL) { + nk = va_arg (ap, const char *); + + if (nk == NULL) { + break; + } + else { + ret = ucl_object_find_keyl (obj, nk, strlen (nk)); + } + } + + va_end (ap); + } + + return ret; } const ucl_object_t* @@ -1792,6 +2430,12 @@ ucl_object_iterate_free (ucl_object_iter_t it) const ucl_object_t * ucl_lookup_path (const ucl_object_t *top, const char *path_in) { + return ucl_lookup_path_char (top, path_in, '.'); +} + + +const ucl_object_t * +ucl_lookup_path_char (const ucl_object_t *top, const char *path_in, const char sep) { const ucl_object_t *o = NULL, *found; const char *p, *c; char *err_str; @@ -1805,20 +2449,20 @@ ucl_lookup_path (const ucl_object_t *top, const char *path_in) { p = path_in; /* Skip leading dots */ - while (*p == '.') { + while (*p == sep) { p ++; } c = p; while (*p != '\0') { p ++; - if (*p == '.' || *p == '\0') { + if (*p == sep || *p == '\0') { if (p > c) { switch (top->type) { case UCL_ARRAY: /* Key should be an int */ index = strtoul (c, &err_str, 10); - if (err_str != NULL && (*err_str != '.' && *err_str != '\0')) { + if (err_str != NULL && (*err_str != sep && *err_str != '\0')) { return NULL; } o = ucl_array_find_index (top, index); @@ -1913,6 +2557,10 @@ ucl_object_new_userdata (ucl_userdata_dtor dtor, ucl_userdata_emitter emitter) ucl_type_t ucl_object_type (const ucl_object_t *obj) { + if (obj == NULL) { + return UCL_NULL; + } + return obj->type; } @@ -1981,6 +2629,11 @@ ucl_array_append (ucl_object_t *top, ucl_object_t *elt) if (vec == NULL) { vec = UCL_ALLOC (sizeof (*vec)); + + if (vec == NULL) { + return false; + } + kv_init (*vec); top->value.av = (void *)vec; } @@ -2021,14 +2674,23 @@ bool ucl_array_merge (ucl_object_t *top, ucl_object_t *elt, bool copy) { unsigned i; + ucl_object_t *cp = NULL; ucl_object_t **obj; - UCL_ARRAY_GET (v1, top); - UCL_ARRAY_GET (v2, elt); if (elt == NULL || top == NULL || top->type != UCL_ARRAY || elt->type != UCL_ARRAY) { return false; } + if (copy) { + cp = ucl_object_copy (elt); + } + else { + cp = ucl_object_ref (elt); + } + + UCL_ARRAY_GET (v1, top); + UCL_ARRAY_GET (v2, cp); + kv_concat (ucl_object_t *, *v1, *v2); for (i = v2->n; i < v1->n; i ++) { @@ -2036,14 +2698,7 @@ ucl_array_merge (ucl_object_t *top, ucl_object_t *elt, bool copy) if (*obj == NULL) { continue; } - top->len ++; - if (copy) { - *obj = ucl_object_copy (*obj); - } - else { - ucl_object_ref (*obj); - } } return true; @@ -2056,6 +2711,10 @@ ucl_array_delete (ucl_object_t *top, ucl_object_t *elt) ucl_object_t *ret = NULL; unsigned i; + if (vec == NULL) { + return NULL; + } + for (i = 0; i < vec->n; i ++) { if (kv_A (*vec, i) == elt) { kv_del (ucl_object_t *, *vec, i); @@ -2073,7 +2732,8 @@ ucl_array_head (const ucl_object_t *top) { UCL_ARRAY_GET (vec, top); - if (top == NULL || top->type != UCL_ARRAY || top->value.av == NULL) { + if (vec == NULL || top == NULL || top->type != UCL_ARRAY || + top->value.av == NULL) { return NULL; } @@ -2136,6 +2796,25 @@ ucl_array_find_index (const ucl_object_t *top, unsigned int index) return NULL; } +unsigned int +ucl_array_index_of (ucl_object_t *top, ucl_object_t *elt) +{ + UCL_ARRAY_GET (vec, top); + unsigned i; + + if (vec == NULL) { + return (unsigned int)(-1); + } + + for (i = 0; i < vec->n; i ++) { + if (kv_A (*vec, i) == elt) { + return i; + } + } + + return (unsigned int)(-1); +} + ucl_object_t * ucl_array_replace_index (ucl_object_t *top, ucl_object_t *elt, unsigned int index) @@ -2265,7 +2944,9 @@ ucl_object_tostring_safe (const ucl_object_t *obj, const char **target) switch (obj->type) { case UCL_STRING: - *target = ucl_copy_value_trash (obj); + if (!(obj->flags & UCL_OBJECT_BINARY)) { + *target = ucl_copy_value_trash (obj); + } break; default: return false; @@ -2286,7 +2967,12 @@ ucl_object_tostring (const ucl_object_t *obj) const char * ucl_object_tostring_forced (const ucl_object_t *obj) { - return ucl_copy_value_trash (obj); + /* TODO: For binary strings we might encode string here */ + if (!(obj->flags & UCL_OBJECT_BINARY)) { + return ucl_copy_value_trash (obj); + } + + return NULL; } bool @@ -2533,7 +3219,7 @@ ucl_object_compare (const ucl_object_t *o1, const ucl_object_t *o2) void ucl_object_array_sort (ucl_object_t *ar, - int (*cmp)(const ucl_object_t *o1, const ucl_object_t *o2)) + int (*cmp)(const ucl_object_t **o1, const ucl_object_t **o2)) { UCL_ARRAY_GET (vec, ar); @@ -2563,6 +3249,9 @@ ucl_object_set_priority (ucl_object_t *obj, { if (obj != NULL) { priority &= (0x1 << PRIOBITS) - 1; - obj->flags |= priority << ((sizeof (obj->flags) * NBBY) - PRIOBITS); + priority <<= ((sizeof (obj->flags) * NBBY) - PRIOBITS); + priority |= obj->flags & ((1 << ((sizeof (obj->flags) * NBBY) - + PRIOBITS)) - 1); + obj->flags = priority; } } diff --git a/src/xxhash.c b/src/xxhash.c index fc4463395c5e..3473eb46a3b0 100644 --- a/src/xxhash.c +++ b/src/xxhash.c @@ -1,6 +1,6 @@ /* xxHash - Fast Hash algorithm -Copyright (C) 2012-2013, Yann Collet. +Copyright (C) 2012-2014, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) Redistribution and use in source and binary forms, with or without @@ -28,6 +28,7 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. You can contact the author at : - xxHash source repository : http://code.google.com/p/xxhash/ +- public discussion board : https://groups.google.com/forum/#!forum/lz4c */ @@ -47,7 +48,7 @@ You can contact the author at : // When this option is enabled, xxHash output for null input pointers will be the same as a null-length input. // This option has a very small performance cost (only measurable on small inputs). // By default, this option is disabled. To enable it, uncomment below define : -//#define XXH_ACCEPT_NULL_INPUT_POINTER 1 +// #define XXH_ACCEPT_NULL_INPUT_POINTER 1 // XXH_FORCE_NATIVE_FORMAT : // By default, xxHash library provides endian-independant Hash values, based on little-endian convention. @@ -58,7 +59,6 @@ You can contact the author at : // This option has no impact on Little_Endian CPU. #define XXH_FORCE_NATIVE_FORMAT 0 - //************************************** // Compiler Specific Options //************************************** @@ -68,28 +68,30 @@ You can contact the author at : #endif #ifdef _MSC_VER // Visual Studio -# define forceinline static __forceinline -#else +# define FORCE_INLINE static __forceinline +#else # ifdef __GNUC__ -# define forceinline static inline __attribute__((always_inline)) +# define FORCE_INLINE static inline __attribute__((always_inline)) # else -# define forceinline static inline +# define FORCE_INLINE static inline # endif #endif - //************************************** // Includes & Memory related functions //************************************** #include "xxhash.h" -// Modify the local functions below should you wish to use some other memory related routines +// Modify the local functions below should you wish to use some other memory routines // for malloc(), free() #include <stdlib.h> -forceinline void* XXH_malloc(size_t s) { return malloc(s); } -forceinline void XXH_free (void* p) { free(p); } +static void* XXH_malloc(size_t s) { return malloc(s); } +static void XXH_free (void* p) { free(p); } // for memcpy() #include <string.h> -forceinline void* XXH_memcpy(void* dest, const void* src, size_t size) { return memcpy(dest,src,size); } +static void* XXH_memcpy(void* dest, const void* src, size_t size) +{ + return memcpy(dest,src,size); +} //************************************** @@ -97,17 +99,17 @@ forceinline void* XXH_memcpy(void* dest, const void* src, size_t size) { return //************************************** #if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L // C99 # include <stdint.h> - typedef uint8_t BYTE; - typedef uint16_t U16; - typedef uint32_t U32; - typedef int32_t S32; - typedef uint64_t U64; +typedef uint8_t BYTE; +typedef uint16_t U16; +typedef uint32_t U32; +typedef int32_t S32; +typedef uint64_t U64; #else - typedef unsigned char BYTE; - typedef unsigned short U16; - typedef unsigned int U32; - typedef signed int S32; - typedef unsigned long long U64; +typedef unsigned char BYTE; +typedef unsigned short U16; +typedef unsigned int U32; +typedef signed int S32; +typedef uint64_t U64; #endif #if defined(__GNUC__) && !defined(XXH_USE_UNALIGNED_ACCESS) @@ -124,13 +126,21 @@ forceinline void* XXH_memcpy(void* dest, const void* src, size_t size) { return # endif #endif -typedef struct _U32_S { U32 v; } _PACKED U32_S; +typedef struct _U32_S +{ + U32 v; +} _PACKED U32_S; +typedef struct _U64_S +{ + U64 v; +} _PACKED U64_S; #if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) # pragma pack(pop) #endif #define A32(x) (((U32_S *)(x))->v) +#define A64(x) (((U64_S *)(x))->v) //*************************************** @@ -141,20 +151,37 @@ typedef struct _U32_S { U32 v; } _PACKED U32_S; // Note : although _rotl exists for minGW (GCC under windows), performance seems poor #if defined(_MSC_VER) # define XXH_rotl32(x,r) _rotl(x,r) +# define XXH_rotl64(x,r) _rotl64(x,r) #else # define XXH_rotl32(x,r) ((x << r) | (x >> (32 - r))) +# define XXH_rotl64(x,r) ((x << r) | (x >> (64 - r))) #endif #if defined(_MSC_VER) // Visual Studio # define XXH_swap32 _byteswap_ulong -#elif GCC_VERSION >= 403 +# define XXH_swap64 _byteswap_uint64 +#elif GCC_VERSION >= 403 || defined(__clang__) # define XXH_swap32 __builtin_bswap32 +# define XXH_swap64 __builtin_bswap64 #else -static inline U32 XXH_swap32 (U32 x) { +static inline U32 XXH_swap32 (U32 x) +{ return ((x << 24) & 0xff000000 ) | - ((x << 8) & 0x00ff0000 ) | - ((x >> 8) & 0x0000ff00 ) | - ((x >> 24) & 0x000000ff );} + ((x << 8) & 0x00ff0000 ) | + ((x >> 8) & 0x0000ff00 ) | + ((x >> 24) & 0x000000ff ); +} +static inline U64 XXH_swap64 (U64 x) +{ + return ((x << 56) & 0xff00000000000000ULL) | + ((x << 40) & 0x00ff000000000000ULL) | + ((x << 24) & 0x0000ff0000000000ULL) | + ((x << 8) & 0x000000ff00000000ULL) | + ((x >> 8) & 0x00000000ff000000ULL) | + ((x >> 24) & 0x0000000000ff0000ULL) | + ((x >> 40) & 0x000000000000ff00ULL) | + ((x >> 56) & 0x00000000000000ffULL); +} #endif @@ -167,13 +194,18 @@ static inline U32 XXH_swap32 (U32 x) { #define PRIME32_4 668265263U #define PRIME32_5 374761393U +#define PRIME64_1 11400714785074694791ULL +#define PRIME64_2 14029467366897019727ULL +#define PRIME64_3 1609587929392839161ULL +#define PRIME64_4 9650029242287828579ULL +#define PRIME64_5 2870177450012600261ULL //************************************** // Architecture Macros //************************************** typedef enum { XXH_bigEndian=0, XXH_littleEndian=1 } XXH_endianess; #ifndef XXH_CPU_LITTLE_ENDIAN // It is possible to define XXH_CPU_LITTLE_ENDIAN externally, for example using a compiler switch - static const int one = 1; +static const int one = 1; # define XXH_CPU_LITTLE_ENDIAN (*(char*)(&one)) #endif @@ -189,28 +221,49 @@ typedef enum { XXH_bigEndian=0, XXH_littleEndian=1 } XXH_endianess; //**************************** typedef enum { XXH_aligned, XXH_unaligned } XXH_alignment; -forceinline U32 XXH_readLE32_align(const U32* ptr, XXH_endianess endian, XXH_alignment align) -{ +FORCE_INLINE U32 XXH_readLE32_align(const void* ptr, XXH_endianess endian, XXH_alignment align) +{ + if (align==XXH_unaligned) + return endian==XXH_littleEndian ? A32(ptr) : XXH_swap32(A32(ptr)); + else + return endian==XXH_littleEndian ? *(U32*)ptr : XXH_swap32(*(U32*)ptr); +} + +FORCE_INLINE U32 XXH_readLE32(const void* ptr, XXH_endianess endian) +{ + return XXH_readLE32_align(ptr, endian, XXH_unaligned); +} + +FORCE_INLINE U64 XXH_readLE64_align(const void* ptr, XXH_endianess endian, XXH_alignment align) +{ if (align==XXH_unaligned) - return endian==XXH_littleEndian ? A32(ptr) : XXH_swap32(A32(ptr)); + return endian==XXH_littleEndian ? A64(ptr) : XXH_swap64(A64(ptr)); else - return endian==XXH_littleEndian ? *ptr : XXH_swap32(*ptr); + return endian==XXH_littleEndian ? *(U64*)ptr : XXH_swap64(*(U64*)ptr); } -forceinline U32 XXH_readLE32(const U32* ptr, XXH_endianess endian) { return XXH_readLE32_align(ptr, endian, XXH_unaligned); } +FORCE_INLINE U64 XXH_readLE64(const void* ptr, XXH_endianess endian) +{ + return XXH_readLE64_align(ptr, endian, XXH_unaligned); +} //**************************** // Simple Hash Functions //**************************** -forceinline U32 XXH32_endian_align(const void* input, int len, U32 seed, XXH_endianess endian, XXH_alignment align) +FORCE_INLINE U32 XXH32_endian_align(const void* input, size_t len, U32 seed, XXH_endianess endian, XXH_alignment align) { const BYTE* p = (const BYTE*)input; - const BYTE* const bEnd = p + len; + const BYTE* bEnd = p + len; U32 h32; +#define XXH_get32bits(p) XXH_readLE32_align(p, endian, align) #ifdef XXH_ACCEPT_NULL_INPUT_POINTER - if (p==NULL) { len=0; p=(const BYTE*)(size_t)16; } + if (p==NULL) + { + len=0; + bEnd=p=(const BYTE*)(size_t)16; + } #endif if (len>=16) @@ -223,11 +276,24 @@ forceinline U32 XXH32_endian_align(const void* input, int len, U32 seed, XXH_end do { - v1 += XXH_readLE32_align((const U32*)p, endian, align) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4; - v2 += XXH_readLE32_align((const U32*)p, endian, align) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4; - v3 += XXH_readLE32_align((const U32*)p, endian, align) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4; - v4 += XXH_readLE32_align((const U32*)p, endian, align) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4; - } while (p<=limit); + v1 += XXH_get32bits(p) * PRIME32_2; + v1 = XXH_rotl32(v1, 13); + v1 *= PRIME32_1; + p+=4; + v2 += XXH_get32bits(p) * PRIME32_2; + v2 = XXH_rotl32(v2, 13); + v2 *= PRIME32_1; + p+=4; + v3 += XXH_get32bits(p) * PRIME32_2; + v3 = XXH_rotl32(v3, 13); + v3 *= PRIME32_1; + p+=4; + v4 += XXH_get32bits(p) * PRIME32_2; + v4 = XXH_rotl32(v4, 13); + v4 *= PRIME32_1; + p+=4; + } + while (p<=limit); h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18); } @@ -238,9 +304,9 @@ forceinline U32 XXH32_endian_align(const void* input, int len, U32 seed, XXH_end h32 += (U32) len; - while (p<=bEnd-4) + while (p+4<=bEnd) { - h32 += XXH_readLE32_align((const U32*)p, endian, align) * PRIME32_3; + h32 += XXH_get32bits(p) * PRIME32_3; h32 = XXH_rotl32(h32, 17) * PRIME32_4 ; p+=4; } @@ -262,18 +328,19 @@ forceinline U32 XXH32_endian_align(const void* input, int len, U32 seed, XXH_end } -U32 XXH32(const void* input, int len, U32 seed) +unsigned int XXH32 (const void* input, size_t len, unsigned seed) { #if 0 // Simple version, good for code maintenance, but unfortunately slow for small inputs - void* state = XXH32_init(seed); - XXH32_update(state, input, len); - return XXH32_digest(state); + XXH32_state_t state; + XXH32_reset(&state, seed); + XXH32_update(&state, input, len); + return XXH32_digest(&state); #else XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; # if !defined(XXH_USE_UNALIGNED_ACCESS) - if (!(((size_t)input) & 3)) // Input is aligned, let's leverage the speed advantage + if ((((size_t)input) & 3) == 0) // Input is aligned, let's leverage the speed advantage { if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned); @@ -289,12 +356,152 @@ U32 XXH32(const void* input, int len, U32 seed) #endif } +FORCE_INLINE U64 XXH64_endian_align(const void* input, size_t len, U64 seed, XXH_endianess endian, XXH_alignment align) +{ + const BYTE* p = (const BYTE*)input; + const BYTE* bEnd = p + len; + U64 h64; +#define XXH_get64bits(p) XXH_readLE64_align(p, endian, align) + +#ifdef XXH_ACCEPT_NULL_INPUT_POINTER + if (p==NULL) + { + len=0; + bEnd=p=(const BYTE*)(size_t)32; + } +#endif + + if (len>=32) + { + const BYTE* const limit = bEnd - 32; + U64 v1 = seed + PRIME64_1 + PRIME64_2; + U64 v2 = seed + PRIME64_2; + U64 v3 = seed + 0; + U64 v4 = seed - PRIME64_1; + + do + { + v1 += XXH_get64bits(p) * PRIME64_2; + p+=8; + v1 = XXH_rotl64(v1, 31); + v1 *= PRIME64_1; + v2 += XXH_get64bits(p) * PRIME64_2; + p+=8; + v2 = XXH_rotl64(v2, 31); + v2 *= PRIME64_1; + v3 += XXH_get64bits(p) * PRIME64_2; + p+=8; + v3 = XXH_rotl64(v3, 31); + v3 *= PRIME64_1; + v4 += XXH_get64bits(p) * PRIME64_2; + p+=8; + v4 = XXH_rotl64(v4, 31); + v4 *= PRIME64_1; + } + while (p<=limit); + + h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 12) + XXH_rotl64(v4, 18); + + v1 *= PRIME64_2; + v1 = XXH_rotl64(v1, 31); + v1 *= PRIME64_1; + h64 ^= v1; + h64 = h64 * PRIME64_1 + PRIME64_4; + + v2 *= PRIME64_2; + v2 = XXH_rotl64(v2, 31); + v2 *= PRIME64_1; + h64 ^= v2; + h64 = h64 * PRIME64_1 + PRIME64_4; + + v3 *= PRIME64_2; + v3 = XXH_rotl64(v3, 31); + v3 *= PRIME64_1; + h64 ^= v3; + h64 = h64 * PRIME64_1 + PRIME64_4; + + v4 *= PRIME64_2; + v4 = XXH_rotl64(v4, 31); + v4 *= PRIME64_1; + h64 ^= v4; + h64 = h64 * PRIME64_1 + PRIME64_4; + } + else + { + h64 = seed + PRIME64_5; + } + + h64 += (U64) len; + + while (p+8<=bEnd) + { + U64 k1 = XXH_get64bits(p); + k1 *= PRIME64_2; + k1 = XXH_rotl64(k1,31); + k1 *= PRIME64_1; + h64 ^= k1; + h64 = XXH_rotl64(h64,27) * PRIME64_1 + PRIME64_4; + p+=8; + } + + if (p+4<=bEnd) + { + h64 ^= (U64)(XXH_get32bits(p)) * PRIME64_1; + h64 = XXH_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; + p+=4; + } + + while (p<bEnd) + { + h64 ^= (*p) * PRIME64_5; + h64 = XXH_rotl64(h64, 11) * PRIME64_1; + p++; + } + + h64 ^= h64 >> 33; + h64 *= PRIME64_2; + h64 ^= h64 >> 29; + h64 *= PRIME64_3; + h64 ^= h64 >> 32; + + return h64; +} -//**************************** -// Advanced Hash Functions -//**************************** -struct XXH_state32_t +uint64_t XXH64 (const void* input, size_t len, uint64_t seed) +{ +#if 0 + // Simple version, good for code maintenance, but unfortunately slow for small inputs + XXH64_state_t state; + XXH64_reset(&state, seed); + XXH64_update(&state, input, len); + return XXH64_digest(&state); +#else + XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; + +# if !defined(XXH_USE_UNALIGNED_ACCESS) + if ((((size_t)input) & 7)==0) // Input is aligned, let's leverage the speed advantage + { + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH64_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned); + else + return XXH64_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned); + } +# endif + + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH64_endian_align(input, len, seed, XXH_littleEndian, XXH_unaligned); + else + return XXH64_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned); +#endif +} + +/**************************************************** + * Advanced Hash Functions +****************************************************/ + +/*** Allocation ***/ +typedef struct { U64 total_len; U32 seed; @@ -302,21 +509,60 @@ struct XXH_state32_t U32 v2; U32 v3; U32 v4; - int memsize; - char memory[16]; -}; + U32 mem32[4]; /* defined as U32 for alignment */ + U32 memsize; +} XXH_istate32_t; + +typedef struct +{ + U64 total_len; + U64 seed; + U64 v1; + U64 v2; + U64 v3; + U64 v4; + U64 mem64[4]; /* defined as U64 for alignment */ + U32 memsize; +} XXH_istate64_t; -int XXH32_sizeofState(void) +XXH32_state_t* XXH32_createState(void) { - XXH_STATIC_ASSERT(XXH32_SIZEOFSTATE >= sizeof(struct XXH_state32_t)); // A compilation error here means XXH32_SIZEOFSTATE is not large enough - return sizeof(struct XXH_state32_t); + XXH_STATIC_ASSERT(sizeof(XXH32_state_t) >= sizeof(XXH_istate32_t)); // A compilation error here means XXH32_state_t is not large enough + return (XXH32_state_t*)XXH_malloc(sizeof(XXH32_state_t)); } +void* XXH32_init (unsigned seed) +{ + XXH32_state_t *st = XXH32_createState(); + XXH32_reset(st, seed); + + return st; +} + +XXH_errorcode XXH32_freeState(XXH32_state_t* statePtr) +{ + XXH_free(statePtr); + return XXH_OK; +}; + +XXH64_state_t* XXH64_createState(void) +{ + XXH_STATIC_ASSERT(sizeof(XXH64_state_t) >= sizeof(XXH_istate64_t)); // A compilation error here means XXH64_state_t is not large enough + return (XXH64_state_t*)XXH_malloc(sizeof(XXH64_state_t)); +} +XXH_errorcode XXH64_freeState(XXH64_state_t* statePtr) +{ + XXH_free(statePtr); + return XXH_OK; +}; -XXH_errorcode XXH32_resetState(void* state_in, U32 seed) -{ - struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; + +/*** Hash feed ***/ + +XXH_errorcode XXH32_reset(XXH32_state_t* state_in, U32 seed) +{ + XXH_istate32_t* state = (XXH_istate32_t*) state_in; state->seed = seed; state->v1 = seed + PRIME32_1 + PRIME32_2; state->v2 = seed + PRIME32_2; @@ -327,18 +573,23 @@ XXH_errorcode XXH32_resetState(void* state_in, U32 seed) return XXH_OK; } - -void* XXH32_init (U32 seed) +XXH_errorcode XXH64_reset(XXH64_state_t* state_in, uint64_t seed) { - void* state = XXH_malloc (sizeof(struct XXH_state32_t)); - XXH32_resetState(state, seed); - return state; + XXH_istate64_t* state = (XXH_istate64_t*) state_in; + state->seed = seed; + state->v1 = seed + PRIME64_1 + PRIME64_2; + state->v2 = seed + PRIME64_2; + state->v3 = seed + 0; + state->v4 = seed - PRIME64_1; + state->total_len = 0; + state->memsize = 0; + return XXH_OK; } -forceinline XXH_errorcode XXH32_update_endian (void* state_in, const void* input, int len, XXH_endianess endian) +FORCE_INLINE XXH_errorcode XXH32_update_endian (XXH32_state_t* state_in, const void* input, size_t len, XXH_endianess endian) { - struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; + XXH_istate32_t* state = (XXH_istate32_t *) state_in; const BYTE* p = (const BYTE*)input; const BYTE* const bEnd = p + len; @@ -350,20 +601,32 @@ forceinline XXH_errorcode XXH32_update_endian (void* state_in, const void* input if (state->memsize + len < 16) // fill in tmp buffer { - XXH_memcpy(state->memory + state->memsize, input, len); - state->memsize += len; + XXH_memcpy((BYTE*)(state->mem32) + state->memsize, input, len); + state->memsize += (U32)len; return XXH_OK; } if (state->memsize) // some data left from previous update { - XXH_memcpy(state->memory + state->memsize, input, 16-state->memsize); + XXH_memcpy((BYTE*)(state->mem32) + state->memsize, input, 16-state->memsize); { - const U32* p32 = (const U32*)state->memory; - state->v1 += XXH_readLE32(p32, endian) * PRIME32_2; state->v1 = XXH_rotl32(state->v1, 13); state->v1 *= PRIME32_1; p32++; - state->v2 += XXH_readLE32(p32, endian) * PRIME32_2; state->v2 = XXH_rotl32(state->v2, 13); state->v2 *= PRIME32_1; p32++; - state->v3 += XXH_readLE32(p32, endian) * PRIME32_2; state->v3 = XXH_rotl32(state->v3, 13); state->v3 *= PRIME32_1; p32++; - state->v4 += XXH_readLE32(p32, endian) * PRIME32_2; state->v4 = XXH_rotl32(state->v4, 13); state->v4 *= PRIME32_1; p32++; + const U32* p32 = state->mem32; + state->v1 += XXH_readLE32(p32, endian) * PRIME32_2; + state->v1 = XXH_rotl32(state->v1, 13); + state->v1 *= PRIME32_1; + p32++; + state->v2 += XXH_readLE32(p32, endian) * PRIME32_2; + state->v2 = XXH_rotl32(state->v2, 13); + state->v2 *= PRIME32_1; + p32++; + state->v3 += XXH_readLE32(p32, endian) * PRIME32_2; + state->v3 = XXH_rotl32(state->v3, 13); + state->v3 *= PRIME32_1; + p32++; + state->v4 += XXH_readLE32(p32, endian) * PRIME32_2; + state->v4 = XXH_rotl32(state->v4, 13); + state->v4 *= PRIME32_1; + p32++; } p += 16-state->memsize; state->memsize = 0; @@ -379,11 +642,24 @@ forceinline XXH_errorcode XXH32_update_endian (void* state_in, const void* input do { - v1 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4; - v2 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4; - v3 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4; - v4 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4; - } while (p<=limit); + v1 += XXH_readLE32(p, endian) * PRIME32_2; + v1 = XXH_rotl32(v1, 13); + v1 *= PRIME32_1; + p+=4; + v2 += XXH_readLE32(p, endian) * PRIME32_2; + v2 = XXH_rotl32(v2, 13); + v2 *= PRIME32_1; + p+=4; + v3 += XXH_readLE32(p, endian) * PRIME32_2; + v3 = XXH_rotl32(v3, 13); + v3 *= PRIME32_1; + p+=4; + v4 += XXH_readLE32(p, endian) * PRIME32_2; + v4 = XXH_rotl32(v4, 13); + v4 *= PRIME32_1; + p+=4; + } + while (p<=limit); state->v1 = v1; state->v2 = v2; @@ -393,17 +669,17 @@ forceinline XXH_errorcode XXH32_update_endian (void* state_in, const void* input if (p < bEnd) { - XXH_memcpy(state->memory, p, bEnd-p); + XXH_memcpy(state->mem32, p, bEnd-p); state->memsize = (int)(bEnd-p); } return XXH_OK; } -XXH_errorcode XXH32_update (void* state_in, const void* input, int len) +XXH_errorcode XXH32_update (XXH32_state_t* state_in, const void* input, size_t len) { XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; - + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) return XXH32_update_endian(state_in, input, len, XXH_littleEndian); else @@ -412,11 +688,11 @@ XXH_errorcode XXH32_update (void* state_in, const void* input, int len) -forceinline U32 XXH32_intermediateDigest_endian (void* state_in, XXH_endianess endian) +FORCE_INLINE U32 XXH32_digest_endian (const XXH32_state_t* state_in, XXH_endianess endian) { - struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; - const BYTE * p = (const BYTE*)state->memory; - BYTE* bEnd = (BYTE*)state->memory + state->memsize; + XXH_istate32_t* state = (XXH_istate32_t*) state_in; + const BYTE * p = (const BYTE*)state->mem32; + BYTE* bEnd = (BYTE*)(state->mem32) + state->memsize; U32 h32; if (state->total_len >= 16) @@ -430,9 +706,9 @@ forceinline U32 XXH32_intermediateDigest_endian (void* state_in, XXH_endianess e h32 += (U32) state->total_len; - while (p<=bEnd-4) + while (p+4<=bEnd) { - h32 += XXH_readLE32((const U32*)p, endian) * PRIME32_3; + h32 += XXH_readLE32(p, endian) * PRIME32_3; h32 = XXH_rotl32(h32, 17) * PRIME32_4; p+=4; } @@ -449,27 +725,217 @@ forceinline U32 XXH32_intermediateDigest_endian (void* state_in, XXH_endianess e h32 ^= h32 >> 13; h32 *= PRIME32_3; h32 ^= h32 >> 16; - +#if 0 + XXH32_freeState((XXH32_state_t *)state_in); +#endif return h32; } -U32 XXH32_intermediateDigest (void* state_in) +U32 XXH32_digest (const XXH32_state_t* state_in) { XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; - + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) - return XXH32_intermediateDigest_endian(state_in, XXH_littleEndian); + return XXH32_digest_endian(state_in, XXH_littleEndian); else - return XXH32_intermediateDigest_endian(state_in, XXH_bigEndian); + return XXH32_digest_endian(state_in, XXH_bigEndian); } -U32 XXH32_digest (void* state_in) +FORCE_INLINE XXH_errorcode XXH64_update_endian (XXH64_state_t* state_in, const void* input, size_t len, XXH_endianess endian) { - U32 h32 = XXH32_intermediateDigest(state_in); + XXH_istate64_t * state = (XXH_istate64_t *) state_in; + const BYTE* p = (const BYTE*)input; + const BYTE* const bEnd = p + len; - XXH_free(state_in); +#ifdef XXH_ACCEPT_NULL_INPUT_POINTER + if (input==NULL) return XXH_ERROR; +#endif - return h32; + state->total_len += len; + + if (state->memsize + len < 32) // fill in tmp buffer + { + XXH_memcpy(((BYTE*)state->mem64) + state->memsize, input, len); + state->memsize += (U32)len; + return XXH_OK; + } + + if (state->memsize) // some data left from previous update + { + XXH_memcpy(((BYTE*)state->mem64) + state->memsize, input, 32-state->memsize); + { + const U64* p64 = state->mem64; + state->v1 += XXH_readLE64(p64, endian) * PRIME64_2; + state->v1 = XXH_rotl64(state->v1, 31); + state->v1 *= PRIME64_1; + p64++; + state->v2 += XXH_readLE64(p64, endian) * PRIME64_2; + state->v2 = XXH_rotl64(state->v2, 31); + state->v2 *= PRIME64_1; + p64++; + state->v3 += XXH_readLE64(p64, endian) * PRIME64_2; + state->v3 = XXH_rotl64(state->v3, 31); + state->v3 *= PRIME64_1; + p64++; + state->v4 += XXH_readLE64(p64, endian) * PRIME64_2; + state->v4 = XXH_rotl64(state->v4, 31); + state->v4 *= PRIME64_1; + p64++; + } + p += 32-state->memsize; + state->memsize = 0; + } + + if (p+32 <= bEnd) + { + const BYTE* const limit = bEnd - 32; + U64 v1 = state->v1; + U64 v2 = state->v2; + U64 v3 = state->v3; + U64 v4 = state->v4; + + do + { + v1 += XXH_readLE64(p, endian) * PRIME64_2; + v1 = XXH_rotl64(v1, 31); + v1 *= PRIME64_1; + p+=8; + v2 += XXH_readLE64(p, endian) * PRIME64_2; + v2 = XXH_rotl64(v2, 31); + v2 *= PRIME64_1; + p+=8; + v3 += XXH_readLE64(p, endian) * PRIME64_2; + v3 = XXH_rotl64(v3, 31); + v3 *= PRIME64_1; + p+=8; + v4 += XXH_readLE64(p, endian) * PRIME64_2; + v4 = XXH_rotl64(v4, 31); + v4 *= PRIME64_1; + p+=8; + } + while (p<=limit); + + state->v1 = v1; + state->v2 = v2; + state->v3 = v3; + state->v4 = v4; + } + + if (p < bEnd) + { + XXH_memcpy(state->mem64, p, bEnd-p); + state->memsize = (int)(bEnd-p); + } + + return XXH_OK; +} + +XXH_errorcode XXH64_update (XXH64_state_t* state_in, const void* input, size_t len) +{ + XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; + + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH64_update_endian(state_in, input, len, XXH_littleEndian); + else + return XXH64_update_endian(state_in, input, len, XXH_bigEndian); +} + + + +FORCE_INLINE U64 XXH64_digest_endian (const XXH64_state_t* state_in, XXH_endianess endian) +{ + XXH_istate64_t * state = (XXH_istate64_t *) state_in; + const BYTE * p = (const BYTE*)state->mem64; + BYTE* bEnd = (BYTE*)state->mem64 + state->memsize; + U64 h64; + + if (state->total_len >= 32) + { + U64 v1 = state->v1; + U64 v2 = state->v2; + U64 v3 = state->v3; + U64 v4 = state->v4; + + h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 12) + XXH_rotl64(v4, 18); + + v1 *= PRIME64_2; + v1 = XXH_rotl64(v1, 31); + v1 *= PRIME64_1; + h64 ^= v1; + h64 = h64*PRIME64_1 + PRIME64_4; + + v2 *= PRIME64_2; + v2 = XXH_rotl64(v2, 31); + v2 *= PRIME64_1; + h64 ^= v2; + h64 = h64*PRIME64_1 + PRIME64_4; + + v3 *= PRIME64_2; + v3 = XXH_rotl64(v3, 31); + v3 *= PRIME64_1; + h64 ^= v3; + h64 = h64*PRIME64_1 + PRIME64_4; + + v4 *= PRIME64_2; + v4 = XXH_rotl64(v4, 31); + v4 *= PRIME64_1; + h64 ^= v4; + h64 = h64*PRIME64_1 + PRIME64_4; + } + else + { + h64 = state->seed + PRIME64_5; + } + + h64 += (U64) state->total_len; + + while (p+8<=bEnd) + { + U64 k1 = XXH_readLE64(p, endian); + k1 *= PRIME64_2; + k1 = XXH_rotl64(k1,31); + k1 *= PRIME64_1; + h64 ^= k1; + h64 = XXH_rotl64(h64,27) * PRIME64_1 + PRIME64_4; + p+=8; + } + + if (p+4<=bEnd) + { + h64 ^= (U64)(XXH_readLE32(p, endian)) * PRIME64_1; + h64 = XXH_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; + p+=4; + } + + while (p<bEnd) + { + h64 ^= (*p) * PRIME64_5; + h64 = XXH_rotl64(h64, 11) * PRIME64_1; + p++; + } + + h64 ^= h64 >> 33; + h64 *= PRIME64_2; + h64 ^= h64 >> 29; + h64 *= PRIME64_3; + h64 ^= h64 >> 32; +#if 0 + XXH64_freeState((XXH64_state_t *)state_in); +#endif + return h64; +} + + +uint64_t XXH64_digest (const XXH64_state_t* state_in) +{ + XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; + + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH64_digest_endian(state_in, XXH_littleEndian); + else + return XXH64_digest_endian(state_in, XXH_bigEndian); } + + diff --git a/src/xxhash.h b/src/xxhash.h index b89238883e23..49b0239071e3 100644 --- a/src/xxhash.h +++ b/src/xxhash.h @@ -1,20 +1,20 @@ /* - xxHash - Fast Hash algorithm + xxHash - Extremely Fast Hash algorithm Header File - Copyright (C) 2012-2013, Yann Collet. + Copyright (C) 2012-2014, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: - + * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. - + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR @@ -52,113 +52,114 @@ CRC32 0.43 GB/s 9 MD5-32 0.33 GB/s 10 Ronald L. Rivest SHA1-32 0.28 GB/s 10 -Q.Score is a measure of quality of the hash function. -It depends on successfully passing SMHasher test set. +Q.Score is a measure of quality of the hash function. +It depends on successfully passing SMHasher test set. 10 is a perfect score. */ - -#pragma once +#ifndef LIBUCL_XXHASH_H +#define LIBUCL_XXHASH_H #if defined (__cplusplus) extern "C" { #endif -//**************************** -// Type -//**************************** +/***************************** + Includes +*****************************/ +#include <stddef.h> /* size_t */ +#include <stdint.h> + + +/***************************** + Type +*****************************/ typedef enum { XXH_OK=0, XXH_ERROR } XXH_errorcode; -//**************************** -// Simple Hash Functions -//**************************** +/***************************** + Simple Hash Functions +*****************************/ -unsigned int XXH32 (const void* input, int len, unsigned int seed); +unsigned int XXH32 (const void* input, size_t length, unsigned seed); +uint64_t XXH64 (const void* input, size_t length, uint64_t seed); /* XXH32() : - Calculate the 32-bits hash of sequence of length "len" stored at memory address "input". - The memory between input & input+len must be valid (allocated and read-accessible). + Calculate the 32-bits hash of sequence "length" bytes stored at memory address "input". + The memory between input & input+length must be valid (allocated and read-accessible). "seed" can be used to alter the result predictably. This function successfully passes all SMHasher tests. Speed on Core 2 Duo @ 3 GHz (single thread, SMHasher benchmark) : 5.4 GB/s - Note that "len" is type "int", which means it is limited to 2^31-1. - If your data is larger, use the advanced functions below. +XXH64() : + Calculate the 64-bits hash of sequence of length "len" stored at memory address "input". */ -//**************************** -// Advanced Hash Functions -//**************************** - -void* XXH32_init (unsigned int seed); -XXH_errorcode XXH32_update (void* state, const void* input, int len); -unsigned int XXH32_digest (void* state); +/***************************** + Advanced Hash Functions +*****************************/ +typedef struct { int64_t ll[ 6]; } XXH32_state_t; +typedef struct { int64_t ll[11]; } XXH64_state_t; /* -These functions calculate the xxhash of an input provided in several small packets, -as opposed to an input provided as a single block. - -It must be started with : -void* XXH32_init() -The function returns a pointer which holds the state of calculation. - -This pointer must be provided as "void* state" parameter for XXH32_update(). -XXH32_update() can be called as many times as necessary. -The user must provide a valid (allocated) input. -The function returns an error code, with 0 meaning OK, and any other value meaning there is an error. -Note that "len" is type "int", which means it is limited to 2^31-1. -If your data is larger, it is recommended to chunk your data into blocks -of size for example 2^30 (1GB) to avoid any "int" overflow issue. +These structures allow static allocation of XXH states. +States must then be initialized using XXHnn_reset() before first use. -Finally, you can end the calculation anytime, by using XXH32_digest(). -This function returns the final 32-bits hash. -You must provide the same "void* state" parameter created by XXH32_init(). -Memory will be freed by XXH32_digest(). +If you prefer dynamic allocation, please refer to functions below. */ +/* + * !!! + * Rspamd specific: we use the legacy method to free state when digest is obtained + * !!! + */ +void * XXH32_init (unsigned seed); +XXH32_state_t* XXH32_createState(void); +XXH_errorcode XXH32_freeState(XXH32_state_t* statePtr); -int XXH32_sizeofState(void); -XXH_errorcode XXH32_resetState(void* state, unsigned int seed); +XXH64_state_t* XXH64_createState(void); +XXH_errorcode XXH64_freeState(XXH64_state_t* statePtr); -#define XXH32_SIZEOFSTATE 48 -typedef struct { long long ll[(XXH32_SIZEOFSTATE+(sizeof(long long)-1))/sizeof(long long)]; } XXH32_stateSpace_t; /* -These functions allow user application to make its own allocation for state. +These functions create and release memory for XXH state. +States must then be initialized using XXHnn_reset() before first use. +*/ -XXH32_sizeofState() is used to know how much space must be allocated for the xxHash 32-bits state. -Note that the state must be aligned to access 'long long' fields. Memory must be allocated and referenced by a pointer. -This pointer must then be provided as 'state' into XXH32_resetState(), which initializes the state. -For static allocation purposes (such as allocation on stack, or freestanding systems without malloc()), -use the structure XXH32_stateSpace_t, which will ensure that memory space is large enough and correctly aligned to access 'long long' fields. -*/ +XXH_errorcode XXH32_reset (XXH32_state_t* statePtr, unsigned seed); +XXH_errorcode XXH32_update (XXH32_state_t* statePtr, const void* input, size_t length); +unsigned int XXH32_digest (const XXH32_state_t* statePtr); +XXH_errorcode XXH64_reset (XXH64_state_t* statePtr, uint64_t seed); +XXH_errorcode XXH64_update (XXH64_state_t* statePtr, const void* input, size_t length); +uint64_t XXH64_digest (const XXH64_state_t* statePtr); -unsigned int XXH32_intermediateDigest (void* state); /* -This function does the same as XXH32_digest(), generating a 32-bit hash, -but preserve memory context. -This way, it becomes possible to generate intermediate hashes, and then continue feeding data with XXH32_update(). -To free memory context, use XXH32_digest(), or free(). -*/ +These functions calculate the xxHash of an input provided in multiple smaller packets, +as opposed to an input provided as a single block. +XXH state space must first be allocated, using either static or dynamic method provided above. +Start a new hash by initializing state with a seed, using XXHnn_reset(). -//**************************** -// Deprecated function names -//**************************** -// The following translations are provided to ease code transition -// You are encouraged to no longer this function names -#define XXH32_feed XXH32_update -#define XXH32_result XXH32_digest -#define XXH32_getIntermediateResult XXH32_intermediateDigest +Then, feed the hash state by calling XXHnn_update() as many times as necessary. +Obviously, input must be valid, meaning allocated and read accessible. +The function returns an error code, with 0 meaning OK, and any other value meaning there is an error. + +Finally, you can produce a hash anytime, by using XXHnn_digest(). +This function returns the final nn-bits hash. +You can nonetheless continue feeding the hash state with more input, +and therefore get some new hashes, by calling again XXHnn_digest(). +When you are done, don't forget to free XXH state space, using typically XXHnn_freeState(). +*/ #if defined (__cplusplus) } #endif + +#endif |