Add some patches from Fedora and xmlgawk support #57857 by Alcino Dall Igna Junior.

(Portage version: 2.0.51-r15)

7 files changed, 418 insertions, 8 deletions

diff --git a/sys-apps/gawk/ChangeLog b/sys-apps/gawk/ChangeLog
index afb697932d81..eebf07733b72 100644
--- a/sys-apps/gawk/ChangeLog
+++ b/sys-apps/gawk/ChangeLog
@@ -1,6 +1,15 @@
 # ChangeLog for sys-apps/gawk
 # Copyright 2002-2005 Gentoo Foundation; Distributed under the GPL v2
-# $Header: /var/cvsroot/gentoo-x86/sys-apps/gawk/ChangeLog,v 1.50 2005/01/29 07:00:17 vapier Exp $
+# $Header: /var/cvsroot/gentoo-x86/sys-apps/gawk/ChangeLog,v 1.51 2005/02/12 04:16:58 vapier Exp $
+
+*gawk-3.1.4-r2 (11 Feb 2005)
+
+  11 Feb 2005; Mike Frysinger <vapier@gentoo.org>
+  +files/gawk-3.1.3-getpgrp_void.patch, +files/gawk-3.1.4-flonum.patch,
+  +files/gawk-3.1.4-nextc.patch, +files/gawk-3.1.4-uplow.patch,
+  -gawk-3.1.4-r1.ebuild, +gawk-3.1.4-r2.ebuild:
+  Add some patches from Fedora and xmlgawk support #57857 by Alcino Dall Igna
+  Junior.
 
 *gawk-3.1.4-r1 (29 Jan 2005)
 
diff --git a/sys-apps/gawk/files/digest-gawk-3.1.4-r2 b/sys-apps/gawk/files/digest-gawk-3.1.4-r2
new file mode 100644
index 000000000000..96e092b1b324
--- /dev/null
+++ b/sys-apps/gawk/files/digest-gawk-3.1.4-r2
@@ -0,0 +1,2 @@
+MD5 3c8935efb9fdc7202720894279ad04a7 gawk-3.1.4.tar.gz 2260582
+MD5 7ff33a3753997f38a2a8343cc14796d2 patch_3.1.4__xml_20040813 50667
diff --git a/sys-apps/gawk/files/gawk-3.1.3-getpgrp_void.patch b/sys-apps/gawk/files/gawk-3.1.3-getpgrp_void.patch
new file mode 100644
index 000000000000..13d01310211b
--- /dev/null
+++ b/sys-apps/gawk/files/gawk-3.1.3-getpgrp_void.patch
@@ -0,0 +1,47 @@
+Ripped from Fedora.
+
+--- gawk-3.1.3/main.c.getpgrp_void	2004-01-26 12:08:26.000000000 +0100
++++ gawk-3.1.3/main.c	2004-01-26 12:15:22.000000000 +0100
+@@ -874,14 +874,11 @@
+ 
+ 	PROCINFO_node = install("PROCINFO",
+ 			node((NODE *) NULL, Node_var_array, (NODE *) NULL));
+-
+ #ifdef GETPGRP_VOID
+-#define getpgrp_arg() /* nothing */
++	value = getpgrp();
+ #else
+-#define getpgrp_arg() getpid()
++	value = getpgrp(getpid());
+ #endif
+-
+-	value = getpgrp(getpgrp_arg());
+ 	aptr = assoc_lookup(PROCINFO_node, tmp_string("pgrpid", 6), FALSE);
+ 	*aptr = make_number(value);
+ 
+--- gawk-3.1.3/io.c.getpgrp_void	2004-01-26 12:08:36.000000000 +0100
++++ gawk-3.1.3/io.c	2004-01-26 12:16:04.000000000 +0100
+@@ -1422,11 +1422,6 @@
+ 	return 0;
+ }
+ 
+-#ifdef GETPGRP_VOID
+-#define getpgrp_arg() /* nothing */
+-#else
+-#define getpgrp_arg() getpid()
+-#endif
+ 
+ /* pidopen --- "open" /dev/pid, /dev/ppid, and /dev/pgrpid */
+ 
+@@ -1440,7 +1435,11 @@
+ 	warning(_("use `PROCINFO[\"%s\"]' instead of `%s'"), cp, name);
+ 
+ 	if (name[6] == 'g')
++#ifdef GETPGRP_VOID
++		sprintf(tbuf, "%d\n", (int) getpgrp());
++#else
+ 		sprintf(tbuf, "%d\n", (int) getpgrp(getpgrp_arg()));
++#endif
+ 	else if (name[6] == 'i')
+ 		sprintf(tbuf, "%d\n", (int) getpid());
+ 	else
diff --git a/sys-apps/gawk/files/gawk-3.1.4-flonum.patch b/sys-apps/gawk/files/gawk-3.1.4-flonum.patch
new file mode 100644
index 000000000000..5ba45cdd05ad
--- /dev/null
+++ b/sys-apps/gawk/files/gawk-3.1.4-flonum.patch
@@ -0,0 +1,105 @@
+Ripped from Fedora.
+
+--- gawk-3.1.4/node.c.flonum	2004-07-28 15:45:04.000000000 +0200
++++ gawk-3.1.4/node.c	2004-11-09 15:53:03.353119096 +0100
+@@ -91,7 +91,7 @@
+ 
+ 	if (do_non_decimal_data) {
+ 		errno = 0;
+-		if (! do_traditional && isnondecimal(cp)) {
++		if (! do_traditional && isnondecimal(cp, TRUE)) {
+ 			n->numbr = nondec2awknum(cp, cpend - cp);
+ 			goto finish;
+ 		}
+@@ -584,3 +584,38 @@
+ 		return c;
+ 	}
+ }
++
++/* isnondecimal --- return true if number is not a decimal number */
++
++int
++isnondecimal(const char *str, int use_locale)
++{
++	int dec_point = '.';
++#if ENABLE_NLS && defined(HAVE_LOCALE_H)
++	if (use_locale)
++		dec_point = loc.decimal_point[0];	/* XXX --- assumes one char */
++#endif
++
++	if (str[0] != '0')
++		return FALSE;
++
++	/* leading 0x or 0X */
++	if (str[1] == 'x' || str[1] == 'X')
++		return TRUE;
++
++	/*
++	 * Numbers with '.', 'e', or 'E' are decimal.
++	 * Have to check so that things like 00.34 are handled right.
++	 *
++	 * These beasts can have trailing whitespace. Deal with that too.
++	 */
++	for (; *str != '\0'; str++) {
++		if (*str == 'e' || *str == 'E' || *str == dec_point)
++			return FALSE;
++		else if (! ISDIGIT(*str))
++			break;
++	}
++
++	return TRUE;
++}
++
+--- gawk-3.1.4/awkgram.y.flonum	2004-07-26 16:11:12.000000000 +0200
++++ gawk-3.1.4/awkgram.y	2004-11-09 15:53:03.355118792 +0100
+@@ -2069,12 +2069,14 @@
+ 			eof_warned = TRUE;
+ 		}
+ 		tokadd('\0');
+-		if (! do_traditional && isnondecimal(tokstart)) {
+-			static short warned = FALSE;
+-			if (do_lint && ! warned) {
+-				warned = TRUE;
+-				lintwarn("numeric constant `%.*s' treated as octal or hexadecimal",
+-					strlen(tokstart)-1, tokstart);
++		if (! do_traditional && isnondecimal(tokstart, FALSE)) {
++			if (do_lint) {
++				if (ISDIGIT(tokstart[1]))       /* not an 'x' or 'X' */
++					lintwarn("numeric constant `%.*s' treated as octal",
++						strlen(tokstart)-1, tokstart);
++				else if (tokstart[1] == 'x' || tokstart[1] == 'X')
++					lintwarn("numeric constant `%.*s' treated as hexadecimal",
++						strlen(tokstart)-1, tokstart);
+ 			}
+ 			yylval.nodeval = make_number(nondec2awknum(tokstart, strlen(tokstart)));
+ 		} else
+--- gawk-3.1.4/builtin.c.flonum	2004-07-13 09:55:28.000000000 +0200
++++ gawk-3.1.4/builtin.c	2004-11-09 15:53:03.357118488 +0100
+@@ -2779,7 +2779,7 @@
+ 
+ 	if ((tmp->flags & (NUMBER|NUMCUR)) != 0)
+ 		d = (double) force_number(tmp);
+-	else if (isnondecimal(tmp->stptr))
++	else if (isnondecimal(tmp->stptr, TRUE))
+ 		d = nondec2awknum(tmp->stptr, tmp->stlen);
+ 	else
+ 		d = (double) force_number(tmp);
+--- gawk-3.1.4/awk.h.flonum	2004-07-26 16:11:05.000000000 +0200
++++ gawk-3.1.4/awk.h	2004-11-09 15:53:03.358118336 +0100
+@@ -742,8 +742,6 @@
+ /* ------------------------- Pseudo-functions ------------------------- */
+ 
+ #define is_identchar(c)		(isalnum(c) || (c) == '_')
+-#define isnondecimal(str)	(((str)[0]) == '0' && (ISDIGIT((str)[1]) \
+-					|| (str)[1] == 'x' || (str)[1] == 'X'))
+ 
+ #define var_uninitialized(n)	((n)->var_value == Nnull_string)
+ 
+@@ -1138,6 +1136,7 @@
+ extern int avoid_dfa P((NODE *re, char *str, size_t len));	/* temporary */
+ extern int reisstring P((const char *text, size_t len, Regexp *re, const char *buf));
+ extern int remaybelong P((const char *text, size_t len));
++extern int isnondecimal P((const char *str, int use_locale));
+ 
+ /* strncasecmp.c */
+ #ifndef BROKEN_STRNCASECMP
diff --git a/sys-apps/gawk/files/gawk-3.1.4-nextc.patch b/sys-apps/gawk/files/gawk-3.1.4-nextc.patch
new file mode 100644
index 000000000000..bbed6591e24a
--- /dev/null
+++ b/sys-apps/gawk/files/gawk-3.1.4-nextc.patch
@@ -0,0 +1,28 @@
+Ripped from Fedora.
+
+--- gawk-3.1.4/awkgram.y.nextc	2004-11-09 15:53:03.355118792 +0100
++++ gawk-3.1.4/awkgram.y	2004-11-09 16:05:46.091165288 +0100
+@@ -1433,6 +1433,11 @@
+ nextc(void)
+ {
+ 	if (gawk_mb_cur_max > 1)	{
++		if (!lexptr || lexptr >= lexend)
++			if (!get_src_buf()) {
++				return EOF;
++			}
++
+ 		/* Update the buffer index.  */
+ 		cur_ring_idx = (cur_ring_idx == RING_BUFFER_SIZE - 1)? 0 :
+ 			cur_ring_idx + 1;
+@@ -1444,11 +1449,6 @@
+ 			mbstate_t tmp_state;
+ 			size_t mbclen;
+ 
+-			if (!lexptr || lexptr >= lexend)
+-				if (!get_src_buf()) {
+-					return EOF;
+-				}
+-
+ 			for (idx = 0 ; lexptr + idx < lexend ; idx++) {
+ 				tmp_state = cur_mbstate;
+ 				mbclen = mbrlen(lexptr, idx + 1, &tmp_state);
diff --git a/sys-apps/gawk/files/gawk-3.1.4-uplow.patch b/sys-apps/gawk/files/gawk-3.1.4-uplow.patch
new file mode 100644
index 000000000000..2db5185595df
--- /dev/null
+++ b/sys-apps/gawk/files/gawk-3.1.4-uplow.patch
@@ -0,0 +1,212 @@
+Ripped from Fedora.
+
+--- gawk-3.1.4/builtin.c.uplow	2004-11-09 15:53:03.357118488 +0100
++++ gawk-3.1.4/builtin.c	2004-11-09 15:55:25.455516248 +0100
+@@ -1652,51 +1652,116 @@
+ 		fflush(rp->fp);
+ }
+ 
++/* wide_tolower_toupper --- lower- or uppercase a multibute string */
++
++typedef int (*isw_func)(wint_t);
++typedef wint_t (*tow_func)(wint_t);
++
++static NODE *
++wide_tolower_toupper(NODE *t1, isw_func iswu, tow_func towl)
++{
++#ifdef MBS_SUPPORT
++	register unsigned char *cp, *cpe;
++	register unsigned char *cp2;
++	size_t mbclen;
++	mbstate_t mbs, prev_mbs;
++	wchar_t wc;
++	NODE *t2;
++	/*
++	 * Since the lowercase char and its uppercase equivalent may occupy
++	 * different number of bytes (Turkish `i'), we cannot say the length
++	 * of the output string.
++	 * This approach is adapted from format_tree().
++	 */
++	unsigned char *obuf;
++	size_t osiz, ofre;
++
++	/*
++	 * Better 2 spare bytes than 1, consistently with make_str_node().
++	 * And we need gawk_mb_cur_max free bytes before we convert the last
++	 * char, so we add (gawk_mb_cur_max - 1).
++	 */
++	osiz = t1->stlen + 2 + (gawk_mb_cur_max - 1);
++	ofre = osiz - 2;
++	emalloc(obuf, char *, osiz, "wide_tolower_toupper");
++
++	memset(&mbs, 0, sizeof(mbstate_t));
++	cp = (unsigned char *)t1->stptr;
++	cpe = (unsigned char *)(t1->stptr + t1->stlen);
++	cp2 = obuf;
++	while (cp < cpe) {
++		if (ofre < gawk_mb_cur_max) {
++			size_t olen = cp2 - obuf;
++			ofre += osiz;
++			osiz *= 2;
++			erealloc(obuf, char *, osiz, "wide_tolower_toupper");
++			cp2 = obuf + olen;
++		}
++		prev_mbs = mbs;
++		mbclen = (size_t) mbrtowc(&wc, (char *) cp, cpe - cp,
++					  &mbs);
++		if (mbclen == 0 || mbclen == (size_t) -1 || mbclen == (size_t) -2) {
++			/* Null wide char, or a problem appeared. */
++			*cp2++ = *cp++;
++			ofre--;
++			continue;
++		}
++			 
++		/* If the character doesn't need change, copy it. */
++		if (!(*iswu)(wc)) {
++			ofre -= mbclen;
++			while (mbclen--)
++				*cp2++ = *cp++;
++			continue;
++		}
++
++		/* Increment the input pointer. */
++		cp += mbclen;
++
++		/* Write the modified wide character. */
++		mbclen = wcrtomb((char *) cp2, (*towl)(wc), &prev_mbs);
++
++		if (mbclen > 0 && mbclen < (size_t) -2) {
++			/* Increment the output pointer. */
++			cp2 += mbclen;
++			ofre -= mbclen;
++		} else {
++			/* A problem appeared. */
++			cp2++;
++			ofre--;
++		}
++	}
++	t2 = make_str_node(obuf, cp2 - obuf, ALREADY_MALLOCED);
++	t2->flags |= TEMP;
++	return t2;
++#else
++	cant_happen();
++	return NULL;
++#endif
++}
++
+ /* do_tolower --- lower case a string */
+ 
+ NODE *
+ do_tolower(NODE *tree)
+ {
+ 	NODE *t1, *t2;
+-	register unsigned char *cp, *cp2;
+-#ifdef MBS_SUPPORT
+-	size_t mbclen = 0;
+-	mbstate_t mbs, prev_mbs;
+-	if (gawk_mb_cur_max > 1)
+-		memset(&mbs, 0, sizeof(mbstate_t));
+-#endif
+ 
+ 	t1 = tree_eval(tree->lnode);
+ 	if (do_lint && (t1->flags & (STRING|STRCUR)) == 0)
+ 		lintwarn(_("tolower: received non-string argument"));
+ 	t1 = force_string(t1);
+-	t2 = tmp_string(t1->stptr, t1->stlen);
+-	for (cp = (unsigned char *)t2->stptr,
+-	     cp2 = (unsigned char *)(t2->stptr + t2->stlen); cp < cp2; cp++)
+-#ifdef MBS_SUPPORT
+-		if (gawk_mb_cur_max > 1) {
+-			wchar_t wc;
+-			prev_mbs = mbs;
+-			mbclen = (size_t) mbrtowc(&wc, (char *) cp, cp2 - cp,
+-						  &mbs);
+-			if ((mbclen != 1) && (mbclen != (size_t) -1) &&
+-				(mbclen != (size_t) -2) && (mbclen != 0)) {
+-				/* a multibyte character.  */
+-				if (iswupper(wc)) {
+-					wc = towlower(wc);
+-					wcrtomb((char *) cp, wc, &prev_mbs);
+-				}
+-				/* Adjust the pointer.  */
+-				cp += mbclen - 1;
+-			} else {
+-				/* Otherwise we treat it as a singlebyte character.  */
+-				if (ISUPPER(*cp))
+-					*cp = tolower(*cp);
+-			}
+-		} else
+-#endif
+-		if (ISUPPER(*cp))
+-			*cp = TOLOWER(*cp);
++
++	if (gawk_mb_cur_max > 1)
++		t2 = wide_tolower_toupper(t1, &iswupper, &towlower);
++	else {
++		register unsigned char *cp, *cpe;
++		t2 = tmp_string(t1->stptr, t1->stlen);
++		for (cp = (unsigned char *)t2->stptr,
++		     cpe = (unsigned char *)(t2->stptr + t2->stlen); cp < cpe; cp++)
++			if (ISUPPER(*cp))
++				*cp = TOLOWER(*cp);
++	}
+ 	free_temp(t1);
+ 	return t2;
+ }
+@@ -1707,45 +1772,22 @@
+ do_toupper(NODE *tree)
+ {
+ 	NODE *t1, *t2;
+-	register unsigned char *cp, *cp2;
+-#ifdef MBS_SUPPORT
+-	size_t mbclen = 0;
+-	mbstate_t mbs, prev_mbs;
+-	if (gawk_mb_cur_max > 1)
+-		memset(&mbs, 0, sizeof(mbstate_t));
+-#endif
+ 
+ 	t1 = tree_eval(tree->lnode);
+ 	if (do_lint && (t1->flags & (STRING|STRCUR)) == 0)
+ 		lintwarn(_("toupper: received non-string argument"));
+ 	t1 = force_string(t1);
+-	t2 = tmp_string(t1->stptr, t1->stlen);
+-	for (cp = (unsigned char *)t2->stptr,
+-	     cp2 = (unsigned char *)(t2->stptr + t2->stlen); cp < cp2; cp++)
+-#ifdef MBS_SUPPORT
+-		if (gawk_mb_cur_max > 1) {
+-			wchar_t wc;
+-			prev_mbs = mbs;
+-			mbclen = (size_t) mbrtowc(&wc, (char *) cp, cp2 - cp,
+-						  &mbs);
+-			if ((mbclen != 1) && (mbclen != (size_t) -1) &&
+-				(mbclen != (size_t) -2) && (mbclen != 0)) {
+-				/* a multibyte character.  */
+-				if (iswlower(wc)) {
+-					wc = towupper(wc);
+-					wcrtomb((char *) cp, wc, &prev_mbs);
+-				}
+-				/* Adjust the pointer.  */
+-				cp += mbclen - 1;
+-			} else {
+-				/* Otherwise we treat it as a singlebyte character.  */
+-				if (ISLOWER(*cp))
+-					*cp = toupper(*cp);
+-			}
+-		} else
+-#endif
+-		if (ISLOWER(*cp))
+-			*cp = TOUPPER(*cp);
++
++	if (gawk_mb_cur_max > 1)
++		t2 = wide_tolower_toupper(t1, &iswlower, &towupper);
++	else {
++		register unsigned char *cp, *cpe;
++		t2 = tmp_string(t1->stptr, t1->stlen);
++		for (cp = (unsigned char *)t2->stptr,
++		     cpe = (unsigned char *)(t2->stptr + t2->stlen); cp < cpe; cp++)
++			if (ISLOWER(*cp))
++				*cp = TOUPPER(*cp);
++	}
+ 	free_temp(t1);
+ 	return t2;
+ }
diff --git a/sys-apps/gawk/gawk-3.1.4-r1.ebuild b/sys-apps/gawk/gawk-3.1.4-r2.ebuild
index b215d8b33af6..c071c31df383 100644
--- a/sys-apps/gawk/gawk-3.1.4-r1.ebuild
+++ b/sys-apps/gawk/gawk-3.1.4-r2.ebuild
@@ -1,30 +1,37 @@
 # Copyright 1999-2005 Gentoo Foundation
 # Distributed under the terms of the GNU General Public License v2
-# $Header: /var/cvsroot/gentoo-x86/sys-apps/gawk/gawk-3.1.4-r1.ebuild,v 1.1 2005/01/29 07:00:17 vapier Exp $
+# $Header: /var/cvsroot/gentoo-x86/sys-apps/gawk/gawk-3.1.4-r2.ebuild,v 1.1 2005/02/12 04:16:58 vapier Exp $
 
 inherit eutils toolchain-funcs
 
+XML_PATCH=patch_3.1.4__xml_20040813
 DESCRIPTION="GNU awk pattern-matching language"
 HOMEPAGE="http://www.gnu.org/software/gawk/gawk.html"
-SRC_URI="mirror://gnu/gawk/${P}.tar.gz"
+SRC_URI="mirror://gnu/gawk/${P}.tar.gz
+	xml? ( http://home1.vr-web.de/~Juergen.Kahrs/${XML_PATCH} )"
 
 LICENSE="GPL-2"
 SLOT="0"
 KEYWORDS="~alpha ~amd64 ~arm ~hppa ~ia64 ~mips ~ppc ~ppc64 ~s390 ~sh ~sparc ~x86"
-IUSE="nls build uclibc"
+IUSE="nls build xml"
 
 RDEPEND="virtual/libc"
 DEPEND="${RDEPEND}
 	nls? ( sys-devel/gettext )"
 
 src_unpack() {
-	unpack ${A}
+	unpack ${P}.tar.gz
 
 	# Copy filefuncs module's source over ...
 	cp -PR "${FILESDIR}"/filefuncs "${WORKDIR}"/ || die "cp failed"
 
-	cd ${S}
-	epatch ${FILESDIR}/${P}-disable-DFA.patch #78227
+	cd "${S}"
+	epatch "${FILESDIR}"/${P}-disable-DFA.patch #78227
+	epatch "${FILESDIR}"/${PN}-3.1.3-getpgrp_void.patch #fedora
+	epatch "${FILESDIR}"/${P}-flonum.patch #fedora
+	epatch "${FILESDIR}"/${P}-nextc.patch #fedora
+	epatch "${FILESDIR}"/${P}-uplow.patch #fedora
+	use xml && epatch "${DISTDIR}"/${XML_PATCH} #57857
 	# support for dec compiler.
 	[[ $(tc-getCC) == "ccc" ]] && epatch ${FILESDIR}/${PN}-3.1.2-dec-alpha-compiler.diff
 }
@@ -92,7 +99,7 @@ src_install() {
 	done
 
 	if ! use build ; then
-		cd ${S}
+		cd "${S}"
 		dosym gawk.1.gz /usr/share/man/man1/awk.1.gz
 		dodoc AUTHORS ChangeLog FUTURES LIMITATIONS NEWS PROBLEMS POSIX.STD README
 		docinto README_d