Library for automatic morphological analysis of English,

Russian and German Languages. Version 2.
Finds the lemmas (all forms) of a word.
Written in C++.

WWW: http://www.aot.ru/
- Andrei V. Shetuhin
slonik-v-domene@mail.ru
reki@reki.ru

PR:		ports/110137
Submitted by:	Andrei V. Shetuhin

5 files changed, 178 insertions, 0 deletions

diff --git a/textproc/Makefile b/textproc/Makefile
index bb2781ff1b3d..886e10a6ec69 100644
--- a/textproc/Makefile
+++ b/textproc/Makefile
@@ -243,6 +243,7 @@
     SUBDIR += latex2html
     SUBDIR += latte
     SUBDIR += lemmatizer
+    SUBDIR += lemmatizer2
     SUBDIR += libcroco
     SUBDIR += libebml
     SUBDIR += libextractor
diff --git a/textproc/lemmatizer2/Makefile b/textproc/lemmatizer2/Makefile
new file mode 100644
index 000000000000..8a7c5b2ba003
--- /dev/null
+++ b/textproc/lemmatizer2/Makefile
@@ -0,0 +1,54 @@
+# New ports collection makefile for:	lemmatizer2
+# Date created:				9 Mar 2007
+# Whom:					Andrei V. Shetuhin
+#
+# $FreeBSD$
+#
+
+PORTNAME=	lemmatizer2
+PORTVERSION=	2.0.1
+CATEGORIES=	textproc
+MASTER_SITES=	http://reki.ru/products/lemmatizer/ \
+		http://havoc.ru/products/lemmatizer/
+DISTFILES=	lemmatizer2-2.0.1.tar.gz \
+		rus-src-morph.tar.gz  \
+		eng-src-morph.tar.gz  \
+		ger-src-morph.tar.gz
+
+MAINTAINER=	reki@reki.ru
+COMMENT=	Russian / German / English lemmatizer library
+
+BUILD_DEPENDS=	cmake:${PORTSDIR}/devel/cmake
+LIB_DEPENDS=	pcre:${PORTSDIR}/devel/pcre
+
+DB_DIR?=	${DESTDIR}/var/db/lemmatizer
+
+DIST_SUBDIR=	${PORTNAME}
+
+USE_LDCONFIG=	${PREFIX}/lib/lemmatizer2
+
+pre-fetch:
+	@${ECHO} ""
+	@${ECHO} "You may use the following build options:"
+	@${ECHO} ""
+	@${ECHO} "      DB_DIR=directory        Set alternate directory for database files"
+	@${ECHO} "                              (default is ${DESTDIR}/var/db/lemmatizer)"
+	@${ECHO} ""
+
+do-configure:
+	@cd ${WRKSRC} && ${LOCALBASE}/bin/cmake ${WRKSRC}
+
+post-extract:
+	@cd ${WRKDIR} && ${CP} -r Dicts ${WRKSRC}
+
+generatemorph:
+	@cd ${WRKSRC} && RML=${WRKSRC};export RML && ./GenerateMorph.sh Russian
+	@cd ${WRKSRC} && RML=${WRKSRC};export RML && ./GenerateMorph.sh English
+	@cd ${WRKSRC} && RML=${WRKSRC};export RML && ./GenerateMorph.sh German
+
+installmorph:
+	@${MKDIR} ${DB_DIR} && ${MKDIR} ${DB_DIR}/Bin \
+	&& cd ${WRKSRC} && ${CP} ./Bin/rml.ini ${DB_DIR}/Bin \
+	&& ${CP} -r Dicts ${DB_DIR}
+
+.include <bsd.port.mk>
diff --git a/textproc/lemmatizer2/distinfo b/textproc/lemmatizer2/distinfo
new file mode 100644
index 000000000000..9cc4f1201c8f
--- /dev/null
+++ b/textproc/lemmatizer2/distinfo
@@ -0,0 +1,12 @@
+MD5 (lemmatizer2/lemmatizer2-2.0.1.tar.gz) = 718ca416af9fa3aed6f032f6f48073c1
+SHA256 (lemmatizer2/lemmatizer2-2.0.1.tar.gz) = c5254b288e6bbe447639985e893ab6a5d00be5993575c64352b305df088404d3
+SIZE (lemmatizer2/lemmatizer2-2.0.1.tar.gz) = 440639
+MD5 (lemmatizer2/rus-src-morph.tar.gz) = c7508fd6964e66b7248df0bdfa1671fe
+SHA256 (lemmatizer2/rus-src-morph.tar.gz) = 6ff6caa4771a89c69ce1cd582b7994b87f321bb457b64de8c431d8a434f95992
+SIZE (lemmatizer2/rus-src-morph.tar.gz) = 1162534
+MD5 (lemmatizer2/eng-src-morph.tar.gz) = de917c1f85b3fc108c3d405681370378
+SHA256 (lemmatizer2/eng-src-morph.tar.gz) = 654d4ffe70c036d2e1b7449cac7ac7f0e2df502de51a4b88d90e7bf7fa3396b2
+SIZE (lemmatizer2/eng-src-morph.tar.gz) = 415748
+MD5 (lemmatizer2/ger-src-morph.tar.gz) = 4acd751c3727df23957af8d5c997d752
+SHA256 (lemmatizer2/ger-src-morph.tar.gz) = 145a592327d8c07dd4b4b5b4effe8de1a637a3c35126a372e904c4884b354495
+SIZE (lemmatizer2/ger-src-morph.tar.gz) = 1247810
diff --git a/textproc/lemmatizer2/pkg-descr b/textproc/lemmatizer2/pkg-descr
new file mode 100644
index 000000000000..2efbb5e67dac
--- /dev/null
+++ b/textproc/lemmatizer2/pkg-descr
@@ -0,0 +1,9 @@
+Library for automatic morphological analysis of English,
+Russian and German Languages. Version 2.
+Finds the lemmas (all forms) of a word.
+Written in C++.
+
+WWW: http://www.aot.ru/
+- Andrei V. Shetuhin
+slonik-v-domene@mail.ru
+reki@reki.ru
diff --git a/textproc/lemmatizer2/pkg-plist b/textproc/lemmatizer2/pkg-plist
new file mode 100644
index 000000000000..96867478f890
--- /dev/null
+++ b/textproc/lemmatizer2/pkg-plist
@@ -0,0 +1,102 @@
+bin/FileLem
+bin/LemClientTest
+bin/MorphGen
+bin/StructDictLoader
+bin/TestLem
+include/lemmatizer2/AgramtabLib/EngGramTab.h
+include/lemmatizer2/AgramtabLib/GerGramTab.h
+include/lemmatizer2/AgramtabLib/RusGramTab.h
+include/lemmatizer2/AgramtabLib/StdGramtab.h
+include/lemmatizer2/AgramtabLib/agramtab_.h
+include/lemmatizer2/AgramtabLib/eng_consts.h
+include/lemmatizer2/AgramtabLib/ger_consts.h
+include/lemmatizer2/AgramtabLib/morph_const.h
+include/lemmatizer2/AgramtabLib/rus_consts.h
+include/lemmatizer2/GraphanLib/Consent.h
+include/lemmatizer2/GraphanLib/Descriptors.h
+include/lemmatizer2/GraphanLib/GraphanDicts.h
+include/lemmatizer2/GraphanLib/GraphmatFile.h
+include/lemmatizer2/GraphanLib/HTMLConv.h
+include/lemmatizer2/GraphanLib/StdGraph.h
+include/lemmatizer2/GraphanLib/UnitHolder.h
+include/lemmatizer2/GraphanLib/abbrev.h
+include/lemmatizer2/GraphanLib/graline.h
+include/lemmatizer2/LemClient.hpp
+include/lemmatizer2/LemmatizerLib/Lemmatizers.h
+include/lemmatizer2/LemmatizerLib/MorphAutomBuilder.h
+include/lemmatizer2/LemmatizerLib/MorphAutomat.h
+include/lemmatizer2/LemmatizerLib/MorphDict.h
+include/lemmatizer2/LemmatizerLib/MorphDictBuilder.h
+include/lemmatizer2/LemmatizerLib/PLMLineCollection.h
+include/lemmatizer2/LemmatizerLib/Paradigm.h
+include/lemmatizer2/LemmatizerLib/Predict.h
+include/lemmatizer2/LemmatizerLib/StaticVectorMap.h
+include/lemmatizer2/LemmatizerLib/Statistic.h
+include/lemmatizer2/LemmatizerLib/StdMorph.h
+include/lemmatizer2/MorphWizardLib/FormInfo.h
+include/lemmatizer2/MorphWizardLib/OperationMeter.h
+include/lemmatizer2/MorphWizardLib/wizard.h
+include/lemmatizer2/PCRE/pcre_rml.h
+include/lemmatizer2/StructDictLib/Domen.h
+include/lemmatizer2/StructDictLib/Field.h
+include/lemmatizer2/StructDictLib/ItemsContainer.h
+include/lemmatizer2/StructDictLib/LessDomItem.h
+include/lemmatizer2/StructDictLib/Ross.h
+include/lemmatizer2/StructDictLib/Signat.h
+include/lemmatizer2/StructDictLib/StdRoss.h
+include/lemmatizer2/StructDictLib/StructDictConsts.h
+include/lemmatizer2/StructDictLib/TempArticle.h
+include/lemmatizer2/StructDictLib/TextField.h
+include/lemmatizer2/common/CExpc.h
+include/lemmatizer2/common/COMSyntaxHolder.h
+include/lemmatizer2/common/Chunk.h
+include/lemmatizer2/common/DDCInternalError.h
+include/lemmatizer2/common/DDC_common.h
+include/lemmatizer2/common/DwdsThesaurus.h
+include/lemmatizer2/common/GramInfo.h
+include/lemmatizer2/common/Graspace.h
+include/lemmatizer2/common/MorphologyHolder.h
+include/lemmatizer2/common/PlmLine.h
+include/lemmatizer2/common/SyntaxHolder.h
+include/lemmatizer2/common/cgic.h
+include/lemmatizer2/common/cortege.h
+include/lemmatizer2/common/imports.h
+include/lemmatizer2/common/rus_numerals.h
+include/lemmatizer2/common/string_socket.h
+include/lemmatizer2/common/string_tokenizer.h
+include/lemmatizer2/common/util_classes.h
+include/lemmatizer2/common/utilit.h
+lib/lemmatizer2/libAgramtab-st.a
+lib/lemmatizer2/libAgramtab.so
+lib/lemmatizer2/libAgramtab.so.2
+lib/lemmatizer2/libAgramtab.so.2.0.1
+lib/lemmatizer2/libGraphan-st.a
+lib/lemmatizer2/libGraphan.so
+lib/lemmatizer2/libGraphan.so.2
+lib/lemmatizer2/libGraphan.so.2.0.1
+lib/lemmatizer2/libLemClient-st.a
+lib/lemmatizer2/libLemClient.so
+lib/lemmatizer2/libLemClient.so.2
+lib/lemmatizer2/libLemClient.so.2.0.1
+lib/lemmatizer2/libLemmatizer-st.a
+lib/lemmatizer2/libLemmatizer.so
+lib/lemmatizer2/libLemmatizer.so.2
+lib/lemmatizer2/libLemmatizer.so.2.0.1
+lib/lemmatizer2/libMorphWizard-st.a
+lib/lemmatizer2/libMorphWizard.so
+lib/lemmatizer2/libMorphWizard.so.2
+lib/lemmatizer2/libMorphWizard.so.2.0.1
+lib/lemmatizer2/libStructDict-st.a
+lib/lemmatizer2/libStructDict.so
+lib/lemmatizer2/libStructDict.so.2
+lib/lemmatizer2/libStructDict.so.2.0.1
+@dirrm include/lemmatizer2/AgramtabLib
+@dirrm include/lemmatizer2/GraphanLib
+@dirrm include/lemmatizer2/LemmatizerLib
+@dirrm include/lemmatizer2/MorphWizardLib
+@dirrm include/lemmatizer2/StructDictLib
+@dirrm include/lemmatizer2/PCRE
+@dirrm include/lemmatizer2/common
+@dirrm include/lemmatizer2
+@dirrm lib/lemmatizer2
+@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