5 files changed, 168 insertions, 0 deletions

diff --git a/textproc/Makefile b/textproc/Makefile
index ab0ddaea691f..c7ea1e749472 100644
--- a/textproc/Makefile
+++ b/textproc/Makefile
@@ -200,6 +200,7 @@
     SUBDIR += lacheck
     SUBDIR += latex2html
     SUBDIR += latte
+    SUBDIR += lemmatizer
     SUBDIR += libcroco
     SUBDIR += libebml
     SUBDIR += libextractor
diff --git a/textproc/lemmatizer/Makefile b/textproc/lemmatizer/Makefile
new file mode 100644
index 000000000000..ecdb7ab6ac70
--- /dev/null
+++ b/textproc/lemmatizer/Makefile
@@ -0,0 +1,55 @@
+# New ports collection makefile for:	lemmatizer
+# Date created:				26 May 2005
+# Whom:					Andrei V. Shetuhin
+#
+# $FreeBSD$
+#
+
+PORTNAME=	lemmatizer
+PORTVERSION=	1.0
+CATEGORIES=	textproc
+MASTER_SITES=	http://reki.ru/products/lemmatizer/
+DISTFILES=	lemmatizer-1.0.tar.gz \
+		rus-src-morph.tar.gz  \
+		eng-src-morph.tar.gz  \
+		ger-src-morph.tar.gz
+
+MAINTAINER=	shetuhin@corp.mail.ru
+COMMENT=	Russian / German / English lemmatizer library
+
+LIB_DEPENDS=	pcre:${PORTSDIR}/devel/pcre
+
+DB_DIR?=	/var/db/lemmatizer
+
+DIST_SUBDIR=	${PORTNAME}
+
+USE_GMAKE=	YES
+GNU_CONFIGURE=	YES
+
+CONFIGURE_ARGS=	--with-datadir=${DB_DIR}
+
+INSTALLS_SHLIB=	yes
+LDCONFIG_DIRS=	%%PREFIX%%/lib
+
+pre-fetch:
+	@${ECHO} ""
+	@${ECHO} "You may use the following build options:"
+	@${ECHO} ""
+	@${ECHO} "      DB_DIR=directory        Set alternate directory for database files"
+	@${ECHO} "                              (default is /var/db/lemmatizer)"
+	@${ECHO} ""
+
+post-extract:
+	@cd ${WRKDIR} && ${CP} -r Dicts ${WRKSRC}
+
+generatemorph:
+	@cd ${WRKSRC} \
+	&& ${ECHO_MSG} "===>  Creating dictionaries" \
+	&& ${GMAKE} generatemorph
+
+installmorph:
+	@cd ${WRKSRC} \
+	&& ${ECHO_MSG} "===>  Installing dictionaries" \
+	&& ${GMAKE} installmorph
+
+.include <bsd.port.mk>
diff --git a/textproc/lemmatizer/distinfo b/textproc/lemmatizer/distinfo
new file mode 100644
index 000000000000..36b7b0f336fa
--- /dev/null
+++ b/textproc/lemmatizer/distinfo
@@ -0,0 +1,8 @@
+MD5 (lemmatizer/lemmatizer-1.0.tar.gz) = 6c07df401c1f59cb0a90a519d5a553d4
+SIZE (lemmatizer/lemmatizer-1.0.tar.gz) = 529258
+MD5 (lemmatizer/rus-src-morph.tar.gz) = bac8e094c5ad640ec30daf262781c57b
+SIZE (lemmatizer/rus-src-morph.tar.gz) = 1107916
+MD5 (lemmatizer/eng-src-morph.tar.gz) = 89b2931339aa660cd38ca2b7e9f389e6
+SIZE (lemmatizer/eng-src-morph.tar.gz) = 411698
+MD5 (lemmatizer/ger-src-morph.tar.gz) = a0fccbc7ed1885b22b1c9a144c1ae061
+SIZE (lemmatizer/ger-src-morph.tar.gz) = 1243945
diff --git a/textproc/lemmatizer/pkg-descr b/textproc/lemmatizer/pkg-descr
new file mode 100644
index 000000000000..a547ee314428
--- /dev/null
+++ b/textproc/lemmatizer/pkg-descr
@@ -0,0 +1,8 @@
+Library for automatic morphological analysis of English,
+Russian and German Languages.
+Finds the lemmas (all forms) of a word.
+Written in C++.
+
+WWW: http://www.aot.ru/
+- Andrei V. Shetuhin
+shetuhin@corp.mail.ru
diff --git a/textproc/lemmatizer/pkg-plist b/textproc/lemmatizer/pkg-plist
new file mode 100644
index 000000000000..d532b4cf33d2
--- /dev/null
+++ b/textproc/lemmatizer/pkg-plist
@@ -0,0 +1,96 @@
+bin/MorphGen
+bin/TestLem
+bin/TestLemClient
+include/lemmatizer/AgramtabLib/EngGramTab.h
+include/lemmatizer/AgramtabLib/GerGramTab.h
+include/lemmatizer/AgramtabLib/RusGramTab.h
+include/lemmatizer/AgramtabLib/StdGramtab.h
+include/lemmatizer/AgramtabLib/agramtab_.h
+include/lemmatizer/AgramtabLib/eng_consts.h
+include/lemmatizer/AgramtabLib/ger_consts.h
+include/lemmatizer/AgramtabLib/morph_const.h
+include/lemmatizer/AgramtabLib/rus_consts.h
+include/lemmatizer/GraphanLib/Consent.h
+include/lemmatizer/GraphanLib/Descriptors.h
+include/lemmatizer/GraphanLib/GraphanDicts.h
+include/lemmatizer/GraphanLib/GraphmatFile.h
+include/lemmatizer/GraphanLib/HTMLConv.h
+include/lemmatizer/GraphanLib/StdGraph.h
+include/lemmatizer/GraphanLib/UnitHolder.h
+include/lemmatizer/GraphanLib/abbrev.h
+include/lemmatizer/GraphanLib/graline.h
+include/lemmatizer/LemClient/LemClient.hpp
+include/lemmatizer/LemClient/LemClientC.h
+include/lemmatizer/LemmatizerLib/Lemmatizers.h
+include/lemmatizer/LemmatizerLib/MorphAutomBuilder.h
+include/lemmatizer/LemmatizerLib/MorphAutomat.h
+include/lemmatizer/LemmatizerLib/MorphDict.h
+include/lemmatizer/LemmatizerLib/MorphDictBuilder.h
+include/lemmatizer/LemmatizerLib/PLMLineCollection.h
+include/lemmatizer/LemmatizerLib/Paradigm.h
+include/lemmatizer/LemmatizerLib/Predict.h
+include/lemmatizer/LemmatizerLib/StaticVectorMap.h
+include/lemmatizer/LemmatizerLib/Statistic.h
+include/lemmatizer/LemmatizerLib/StdMorph.h
+include/lemmatizer/MorphWizardLib/FormInfo.h
+include/lemmatizer/MorphWizardLib/OperationMeter.h
+include/lemmatizer/MorphWizardLib/wizard.h
+include/lemmatizer/PCRE/config.h
+include/lemmatizer/PCRE/internal.h
+include/lemmatizer/PCRE/pcre.h
+include/lemmatizer/PCRE/pcre_rml.h
+include/lemmatizer/StructDictLib/Domen.h
+include/lemmatizer/StructDictLib/Field.h
+include/lemmatizer/StructDictLib/ItemsContainer.h
+include/lemmatizer/StructDictLib/LessDomItem.h
+include/lemmatizer/StructDictLib/Ross.h
+include/lemmatizer/StructDictLib/Signat.h
+include/lemmatizer/StructDictLib/StdRoss.h
+include/lemmatizer/StructDictLib/StructDictConsts.h
+include/lemmatizer/StructDictLib/TempArticle.h
+include/lemmatizer/StructDictLib/TextField.h
+include/lemmatizer/common/COMSyntaxHolder.h
+include/lemmatizer/common/Chunk.h
+include/lemmatizer/common/DDCInternalError.h
+include/lemmatizer/common/DDC_common.h
+include/lemmatizer/common/DwdsThesaurus.h
+include/lemmatizer/common/GramInfo.h
+include/lemmatizer/common/Graspace.h
+include/lemmatizer/common/MorphologyHolder.h
+include/lemmatizer/common/PlmLine.h
+include/lemmatizer/common/SyntaxHolder.h
+include/lemmatizer/common/cgic.h
+include/lemmatizer/common/cortege.h
+include/lemmatizer/common/imports.h
+include/lemmatizer/common/rus_numerals.h
+include/lemmatizer/common/string_socket.h
+include/lemmatizer/common/string_tokenizer.h
+include/lemmatizer/common/util_classes.h
+include/lemmatizer/common/utilit.h
+lib/libAgramtab.a
+lib/libAgramtab.so
+lib/libGraphan.a
+lib/libGraphan.so
+lib/libLemClient.a
+lib/libLemClient.so
+lib/libLemCommon.a
+lib/libLemCommon.so
+lib/libLemPCRE.a
+lib/libLemPCRE.so
+lib/libLemmatizer.a
+lib/libLemmatizer.so
+lib/libMorphWizard.a
+lib/libMorphWizard.so
+lib/libStructDict.a
+lib/libStructDict.so
+@dirrm include/lemmatizer/common
+@dirrm include/lemmatizer/StructDictLib
+@dirrm include/lemmatizer/PCRE
+@dirrm include/lemmatizer/MorphWizardLib
+@dirrm include/lemmatizer/LemmatizerLib
+@dirrm include/lemmatizer/LemClient
+@dirrm include/lemmatizer/GraphanLib
+@dirrm include/lemmatizer/AgramtabLib
+@dirrm include/lemmatizer
+
+@unexec echo "If you going to remove lemmatizer permanently, you should also remove dictionaries directory. E.g.: rm -rf /var/db/lemmatize" | /usr/bin/fmt