Add script to compile/install/distribute antiword

This days Git on Windows is used not only for code, but more and more
for managing documentation workflow. And on Windows a lot of people use
*.doc as document format.

So the idea is to provide textconv diff & blame support for *.doc out of
the box, for which we'll need default textconv driver.

Personally I've been using antiword on Linux for such porpose for quite
some time already, and while not perfect, antiword works ok -- I can see
what's changed in git, and `git gui blame` docs, which is great to me.

NOTE: antiword needs to access files with encoding mapping tables, and since we
are going to add antiword to distribution (next patch) for this I had to patch
it to determine installation prefix at runtime, just like git does.

Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>

src/antiword/.gitignore

@@ -0,0 +1 @@
+/antiword-git/

src/antiword/patches/0001-Add-.gititgnore.patch

@@ -0,0 +1,23 @@
+From 6e97f006d17627866155bbe27344b18df11a3c96 Mon Sep 17 00:00:00 2001
+From: Kirill Smelkov <kirr@mns.spb.ru>
+Date: Wed, 24 Nov 2010 21:24:41 +0300
+Subject: [PATCH 1/4] Add .gititgnore
+
+Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
+---
+ .gitignore |    3 +++
+ 1 files changed, 3 insertions(+), 0 deletions(-)
+ create mode 100644 .gitignore
+
+diff --git a/.gitignore b/.gitignore
+new file mode 100644
+index 0000000..28febf5
+--- /dev/null
++++ b/.gitignore
+@@ -0,0 +1,3 @@
++*.o
++*.exe
++tags
+-- 
+1.7.3.2.245.g03276
+

src/antiword/patches/0002-antiword-Tweak-Makefile.Linux-for-msysGit.patch

@@ -0,0 +1,50 @@
+From c1b802eeb0307db3e90bd2fca702256652331758 Mon Sep 17 00:00:00 2001
+From: Kirill Smelkov <kirr@mns.spb.ru>
+Date: Wed, 10 Nov 2010 20:02:19 +0300
+Subject: [PATCH 2/4] antiword: Tweak Makefile.Linux for msysGit
+
+It is easier to start from Makefile.Linux because even if there is
+Makefile.cygming as we'll see in later patches, it does not serve our purposes
+well - we'll still need to change installed location and runtime path for
+resources...
+
+Also we don't need kantiword, and /usr/local/bin/ -> /usr/bin/
+
+Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
+---
+ Makefile.Linux |    8 ++++----
+ 1 files changed, 4 insertions(+), 4 deletions(-)
+
+diff --git a/Makefile.Linux b/Makefile.Linux
+index eeb2d15..65af1df 100644
+--- a/Makefile.Linux
++++ b/Makefile.Linux
+@@ -30,13 +30,13 @@ OBJS	=\
+ 	wordlib.o wordmac.o wordole.o wordwin.o xmalloc.o xml.o
+
+ PROGS =\
+-	antiword\
+-	kantiword
++	antiword.exe
++#	kantiword
+
+ LOCAL_INSTALL_DIR = $(HOME)/bin
+ LOCAL_RESOURCES_DIR = $(HOME)/.antiword
+
+-GLOBAL_INSTALL_DIR = /usr/local/bin
++GLOBAL_INSTALL_DIR = /usr/bin
+ GLOBAL_RESOURCES_DIR = /usr/share/antiword
+
+ all:		$(PROGS)
+@@ -71,7 +71,7 @@ clean:
+ 	rm -f $(OBJS)
+ 	rm -f $(PROGS)
+
+-antiword:	$(OBJS)
++antiword.exe:	$(OBJS)
+ 	@rm -f $@
+ 	$(LD) $(LDFLAGS) $(OBJS) $(LDLIBS) -o $@
+ 	@chmod 750 $@
+-- 
+1.7.3.2.245.g03276
+

src/antiword/patches/0003-Tech-szBasename-about-and-path-separators-win32-only.patch

@@ -0,0 +1,46 @@
+From 5985c61e19256d51f281d87e6610760663854b30 Mon Sep 17 00:00:00 2001
+From: Kirill Smelkov <kirr@mns.spb.ru>
+Date: Wed, 24 Nov 2010 21:27:58 +0300
+Subject: [PATCH 3/4] Tech szBasename about '/' and '\\' path separators (win32 only)
+
+Previously it was using '/' on linux and '\' on mingw. Since on windows
+path separator can be on of those two, teach szBasename about it.
+
+Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
+---
+ misc.c |   11 ++++++-----
+ 1 files changed, 6 insertions(+), 5 deletions(-)
+
+diff --git a/misc.c b/misc.c
+index 609a2f0..63d1c91 100644
+--- a/misc.c
++++ b/misc.c
+@@ -519,7 +519,7 @@ unilen(const UCHAR *s)
+ const char *
+ szBasename(const char *szFilename)
+ {
+-	const char	*szTmp;
++	const char	*tail;
+
+ 	fail(szFilename == NULL);
+
+@@ -527,11 +527,12 @@ szBasename(const char *szFilename)
+ 		return "null";
+ 	}
+
+-	szTmp = strrchr(szFilename, FILE_SEPARATOR[0]);
+-	if (szTmp == NULL) {
+-		return szFilename;
++	/* check for both '/' and '\\' */
++	for (tail = szFilename+strlen(szFilename)-1; tail >= szFilename; --tail) {
++		if (*tail == '/' || *tail == '\\')
++			return tail+1;
+ 	}
+-	return ++szTmp;
++	return szFilename;
+ } /* end of szBasename */
+
+ /*
+-- 
+1.7.3.2.245.g03276
+

src/antiword/patches/0004-Implement-runtime-prefix-detection.patch

@@ -0,0 +1,140 @@
+From 9860d61871f53e94d0780061fcab4e5951e10215 Mon Sep 17 00:00:00 2001
+From: Kirill Smelkov <kirr@mns.spb.ru>
+Date: Wed, 24 Nov 2010 21:29:41 +0300
+Subject: [PATCH 4/4] Implement runtime prefix detection
+
+Antiword needs prefix to know where to look for encoding map tables.
+Don't hardcode this prefix, and instead do, what Git does - determine it
+at runtime.
+
+Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
+---
+ antiword.h |    2 ++
+ fonts_u.c  |    6 ++++--
+ misc.c     |   44 ++++++++++++++++++++++++++++++++++++++++++++
+ options.c  |    8 ++++----
+ 4 files changed, 54 insertions(+), 6 deletions(-)
+
+diff --git a/antiword.h b/antiword.h
+index 3f4aad5..1e4a384 100644
+--- a/antiword.h
++++ b/antiword.h
+@@ -430,6 +430,8 @@ extern USHORT	usGetListValue(int, int, const style_block_type *);
+ #if !defined(__riscos)
+ extern const char	*szGetHomeDirectory(void);
+ extern const char	*szGetAntiwordDirectory(void);
++extern const char	*szGetAntiwordRuntimePrefix(void);
++extern const char	*szGetAntiwordRuntimeDir(void);
+ #endif /* !__riscos */
+ extern long	lGetFilesize(const char *);
+ #if defined(DEBUG)
+diff --git a/fonts_u.c b/fonts_u.c
+index a99f7d2..812830e 100644
+--- a/fonts_u.c
++++ b/fonts_u.c
+@@ -27,7 +27,8 @@ FILE *
+ pOpenFontTableFile(void)
+ {
+ 	FILE		*pFile;
+-	const char	*szHome, *szAntiword, *szGlobalFile;
++	const char	*szHome, *szAntiword;
++	char		szGlobalFile[PATH_MAX+1];
+ 	char		szEnvironmentFile[PATH_MAX+1];
+ 	char		szLocalFile[PATH_MAX+1];
+
+@@ -75,7 +76,8 @@ pOpenFontTableFile(void)
+ 	}
+
+ 	/* Try the global version of the fontnames file */
+-	szGlobalFile = GLOBAL_ANTIWORD_DIR FILE_SEPARATOR FONTNAMES_FILE;
++	snprintf(szGlobalFile, sizeof(szGlobalFile),
++		"%s" FILE_SEPARATOR FONTNAMES_FILE, szGetAntiwordRuntimeDir());
+ 	DBG_MSG(szGlobalFile);
+
+ 	pFile = fopen(szGlobalFile, "r");
+diff --git a/misc.c b/misc.c
+index 63d1c91..893faa9 100644
+--- a/misc.c
++++ b/misc.c
+@@ -71,6 +71,50 @@ szGetAntiwordDirectory(void)
+ #endif /* !__riscos */
+
+ /*
++ * szGetAntiwordRuntimePrefix
++ * (win32 only)
++ */
++const char *
++szGetAntiwordRuntimePrefix(void)
++{
++	static char prefix[PATH_MAX+1];
++	char *tail;
++	int ncut;
++
++	strncpy(prefix, __argv[0], sizeof(prefix));
++
++	/* do dirname() twice  prefix/bin/antiword.exe -> prefix */
++	for (ncut=0, tail = prefix+strlen(prefix)-1; ncut < 2 && tail >= prefix; --tail) {
++		if (*tail == '/' || *tail == '\\') {
++			*tail = '\0';
++			++ncut;
++		}
++	}
++
++	if (ncut != 2)
++		werr(1, "E: Cannot determine runtime prefix from \'%s\'", __argv[0]);
++
++	return prefix;
++}
++
++/*
++ * szGetAntiwordRuntimeDir
++ * (win32 only)
++ */
++const char *
++szGetAntiwordRuntimeDir(void)
++{
++	static char dir[PATH_MAX+1];
++	const char *prefix = szGetAntiwordRuntimePrefix();
++
++	strncpy(dir, prefix, sizeof(dir));
++	strncat(dir, FILE_SEPARATOR "share" FILE_SEPARATOR "antiword", sizeof(dir)-strlen(dir));
++
++	return dir;
++}
++
++
++/*
+  * Get the size of the specified file.
+  * Returns -1 if the file does not exist or is not a proper file.
+  */
+diff --git a/options.c b/options.c
+index 8379fef..9bcc88a 100644
+--- a/options.c
++++ b/options.c
+@@ -238,11 +238,11 @@ pOpenCharacterMappingFile(const char *szLeafname)
+ 	/* Try the global version of the mapping file */
+ 	if (tFilenameLen <
+ 	    sizeof(szMappingFile) -
+-	    sizeof(GLOBAL_ANTIWORD_DIR) -
++	    strlen(szGetAntiwordRuntimeDir()) -
+ 	    sizeof(FILE_SEPARATOR)) {
+ 		sprintf(szMappingFile,
+-			GLOBAL_ANTIWORD_DIR FILE_SEPARATOR "%s%s",
+-			szLeafname, szSuffix);
++			"%s" FILE_SEPARATOR "%s%s",
++			szGetAntiwordRuntimeDir(), szLeafname, szSuffix);
+ 		DBG_MSG(szMappingFile);
+ 		pFile = fopen(szMappingFile, "r");
+ 		if (pFile != NULL) {
+@@ -253,7 +253,7 @@ pOpenCharacterMappingFile(const char *szLeafname)
+ 	}
+ 	werr(0, "I can't open your mapping file (%s%s)\n"
+ 		"It is not in '%s" FILE_SEPARATOR ANTIWORD_DIR "' nor in '"
+-		GLOBAL_ANTIWORD_DIR "'.", szLeafname, szSuffix, szHome);
++		"%s" "'.", szLeafname, szSuffix, szHome, szGetAntiwordRuntimeDir());
+ 	return NULL;
+ #endif /* __riscos */
+ } /* end of pOpenCharacterMappingFile */
+-- 
+1.7.3.2.245.g03276
+

src/antiword/release.sh

@@ -0,0 +1,26 @@
+#!/bin/sh
+
+cd "$(dirname "$0")"
+
+REPO=git://repo.or.cz/antiword.git
+DIR=antiword-git
+HEAD=0680e7ee62e430e0905085e5b4cfb01d31db5936   # antiword 0.37
+
+die () {
+	echo "$*" >&2
+	exit 1
+}
+
+test -d $DIR ||
+( git clone -n $REPO $DIR && cd $DIR && git checkout $HEAD ) ||
+die "Could not clone $REPO"
+
+(cd $DIR && git am ../patches/*) ||
+die "Could not apply patches"
+
+(cd $DIR &&
+make -f Makefile.Linux antiword.exe &&
+index=$(/share/msysGit/pre-install.sh) &&
+make -f Makefile.Linux global_install &&
+/share/msysGit/post-install.sh $index "Install antiword (Git $HEAD)"
+) || die "Could not install antiword"