New port: kenlm.

KenLM Language Model Toolkit.

KenLM estimates, filters, and queries language models.

WWW: https://kheafield.com/code/kenlm/

7 files changed, 63 insertions, 0 deletions

diff --git a/textproc/Makefile b/textproc/Makefile
index 65789e378431..cd9feb227dfa 100644
--- a/textproc/Makefile
+++ b/textproc/Makefile
@@ -324,6 +324,7 @@
     SUBDIR += jrefentry
     SUBDIR += jshon
     SUBDIR += kdiff3
+    SUBDIR += kenlm
     SUBDIR += kibana
     SUBDIR += kmfl-european-latin
     SUBDIR += kmfl-khmer
diff --git a/textproc/kenlm/Makefile b/textproc/kenlm/Makefile
new file mode 100644
index 000000000000..e4dbd4aba08a
--- /dev/null
+++ b/textproc/kenlm/Makefile
@@ -0,0 +1,27 @@
+# Created by: Dmitry Sivachenko <demon@FreeBSD.org>
+# $FreeBSD$
+
+PORTNAME=	kenlm
+PORTVERSION=	20140830
+CATEGORIES=	textproc
+MASTER_SITES=	http://kheafield.com/code/
+DISTNAME=	${PORTNAME}
+
+MAINTAINER=	demon@FreeBSD.org
+COMMENT=	KenLM Language Model Toolkit
+
+BUILD_DEPENDS=	bjam:${PORTSDIR}/devel/boost-jam
+LIB_DEPENDS=	libboost_thread.so:${PORTSDIR}/devel/boost-libs
+
+USES+=		compiler:c++11-lang
+USE_LDCONFIG=	yes
+WRKSRC=		${WRKDIR}/${PORTNAME}
+MAKE_ENV+=	BOOST_BUILD_PATH=${WRKSRC}/jam-files/boost-build
+
+do-build:
+	cd ${WRKSRC} && ${SETENV} ${MAKE_ENV} ${LOCALBASE}/bin/bjam --toolset=${CHOSEN_COMPILER_TYPE} --with-boost=${LOCALBASE} ${_MAKE_JOBS}
+
+do-install:
+	cd ${WRKSRC}/bin && ${INSTALL_PROGRAM} lmplz fragment query filter build_binary dump_counts cat_compressed ${STAGEDIR}${PREFIX}/bin/
+
+.include <bsd.port.mk>
diff --git a/textproc/kenlm/distinfo b/textproc/kenlm/distinfo
new file mode 100644
index 000000000000..9ff4a22d6400
--- /dev/null
+++ b/textproc/kenlm/distinfo
@@ -0,0 +1,2 @@
+SHA256 (kenlm.tar.gz) = e9e5d985d563ade1f0ba89a8058e9bcc5c491f77efe8b48906e8de94e9c4f569
+SIZE (kenlm.tar.gz) = 849428
diff --git a/textproc/kenlm/files/patch-Jamroot b/textproc/kenlm/files/patch-Jamroot
new file mode 100644
index 000000000000..2da5ab30c245
--- /dev/null
+++ b/textproc/kenlm/files/patch-Jamroot
@@ -0,0 +1,10 @@
+--- Jamroot.orig	2014-04-02 19:33:24.000000000 +0400
++++ Jamroot	2014-08-31 00:34:40.000000000 +0400
+@@ -63,6 +63,6 @@
+ 
+ install-bin-libs lm//programs util//programs kenlm ;
+ install-headers headers : [ glob-tree *.hh : dist include ] : . ;
+-alias install : prefix-bin prefix-lib prefix-include ; 
++alias install : prefix-bin prefix-lib ; 
+ explicit headers ;
+ explicit install ;
diff --git a/textproc/kenlm/files/patch-jam-files-sanity.jam b/textproc/kenlm/files/patch-jam-files-sanity.jam
new file mode 100644
index 000000000000..344ddccb47bf
--- /dev/null
+++ b/textproc/kenlm/files/patch-jam-files-sanity.jam
@@ -0,0 +1,11 @@
+--- jam-files/sanity.jam.orig	2014-04-21 07:58:19.000000000 +0400
++++ jam-files/sanity.jam	2014-08-31 00:16:40.000000000 +0400
+@@ -148,7 +148,7 @@ rule boost-lib ( name macro : deps * ) {
+ 
+ #Argument is e.g. 103600
+ rule boost ( min-version ) {
+-  local cmd = "bash -c \"g++ "$(I-boost-include)" -dM -x c++ -E /dev/null -include boost/version.hpp 2>/dev/null |grep '#define BOOST_'\"" ;
++  local cmd = "bash -c \"CC "$(I-boost-include)" -dM -x c++ -E /dev/null -include boost/version.hpp 2>/dev/null |grep '#define BOOST_'\"" ;
+   local boost-shell = [ SHELL "$(cmd)" : exit-status ] ;
+   if $(boost-shell[2]) != 0 && $(CLEANING) = no {
+     echo Failed to run "$(cmd)" ;
diff --git a/textproc/kenlm/pkg-descr b/textproc/kenlm/pkg-descr
new file mode 100644
index 000000000000..5c0e6df3caa6
--- /dev/null
+++ b/textproc/kenlm/pkg-descr
@@ -0,0 +1,5 @@
+KenLM Language Model Toolkit.
+
+KenLM estimates, filters, and queries language models.
+
+WWW: https://kheafield.com/code/kenlm/
diff --git a/textproc/kenlm/pkg-plist b/textproc/kenlm/pkg-plist
new file mode 100644
index 000000000000..92ce43cf7df5
--- /dev/null
+++ b/textproc/kenlm/pkg-plist
@@ -0,0 +1,7 @@
+bin/build_binary
+bin/cat_compressed
+bin/dump_counts
+bin/filter
+bin/fragment
+bin/lmplz
+bin/query