From: Bruno Haible <bruno@clisp.org>
Date: Thu, 7 Feb 2002 13:14:55 +0000 (+0000)
Subject: New awk backend.
X-Git-Tag: v0.11.1~86
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=332d33b34c195c34cafa11594a013d97093f08f0;p=thirdparty%2Fgettext.git

New awk backend.
---

diff --git a/doc/ChangeLog b/doc/ChangeLog
index fc8c11bc2..b8f8079aa 100644
--- a/doc/ChangeLog
+++ b/doc/ChangeLog
@@ -1,3 +1,7 @@
+2002-01-27  Bruno Haible  <bruno@clisp.org>
+
+	* gettext.texi (gawk): Update.
+
 2002-02-02  Bruno Haible  <bruno@clisp.org>
 
 	* gettext.texi (Python): Update.
diff --git a/doc/gettext.texi b/doc/gettext.texi
index 600bfa822..6edc36dec 100644
--- a/doc/gettext.texi
+++ b/doc/gettext.texi
@@ -6943,7 +6943,7 @@ gawk 3.1 or newer
 @code{_"abc"}
 
 @item gettext/ngettext functions
-@code{dcgettext}
+@code{dcgettext}, missing @code{dcngettext} in gawk-3.1.0
 
 @item textdomain
 @code{TEXTDOMAIN} variable
@@ -6961,14 +6961,15 @@ automatic, but missing @code{setlocale (LC_MESSAGES, "")} in gawk-3.1.0
 use
 
 @item Extractor
-@code{gawk --gen-po}
+@code{xgettext}
 
 @item Formatting with positions
 @code{printf "%2$d %1$d"} (GNU awk only)
 
 @item Portability
 On platforms without gettext, no translation.  On non-GNU awks, you must
-define @code{dcgettext} and @code{bindtextdomain} yourself.
+define @code{dcgettext}, @code{dcngettext} and @code{bindtextdomain}
+yourself.
 
 @item po-mode marking
 ---
diff --git a/src/ChangeLog b/src/ChangeLog
index 4742b46cd..ef6eb4ce1 100644
--- a/src/ChangeLog
+++ b/src/ChangeLog
@@ -1,3 +1,22 @@
+2002-01-27  Bruno Haible  <bruno@clisp.org>
+
+	* message.h (format_type): New enum value 'format_awk'.
+	(NFORMATS): Increment.
+	* message.c (format_language): Add format_awk entry.
+	(format_language_pretty): Likewise.
+	* format.h (formatstring_awk): New declaration.
+	* format-awk.c: New file.
+	* format.c (formatstring_parsers): Add formatstring_awk.
+	* x-awk.h: New file.
+	* x-awk.c: New file.
+	* xgettext.c: Include x-awk.h.
+	(main): Call x_awk_extract_all, x_awk_keyword.
+	(language_to_scanner): Add awk rule.
+	(extension_to_language): Add awk rule.
+	* Makefile.am (noinst_HEADERS): Add x-awk.h.
+	(FORMAT_SOURCE): Add format-awk.c.
+	(xgettext_SOURCES): Add x-awk.c.
+
 2002-02-02  Bruno Haible  <bruno@clisp.org>
 
 	* x-python.h: New file.
diff --git a/src/Makefile.am b/src/Makefile.am
index cffbd9ee9..1b5ed1fb2 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -33,7 +33,7 @@ po-gram-gen.h po-hash-gen.h msgl-charset.h msgl-equal.h msgl-iconv.h \
 msgl-ascii.h msgl-cat.h msgl-english.h msgfmt.h msgunfmt.h read-mo.h \
 write-mo.h read-java.h write-java.h po-time.h plural-table.h format.h \
 xgettext.h x-c.h x-po.h x-python.h x-lisp.h x-elisp.h x-librep.h x-java.h \
-x-ycp.h x-rst.h
+x-awk.h x-ycp.h x-rst.h
 
 EXTRA_DIST = FILES project-id \
 gnu/gettext/DumpResource.java gnu/gettext/GetURL.java
@@ -67,8 +67,8 @@ open-po.c dir-list.c str-list.c
 
 # xgettext and msgfmt deal with format strings.
 FORMAT_SOURCE = format.c \
-format-c.c format-java.c format-lisp.c format-elisp.c format-librep.c \
-format-python.c format-pascal.c format-ycp.c
+format-c.c format-python.c format-lisp.c format-elisp.c format-librep.c \
+format-java.c format-awk.c format-pascal.c format-ycp.c
 
 # libgettextsrc contains all code that is needed by at least two programs.
 libgettextsrc_la_SOURCES = \
@@ -87,8 +87,8 @@ msgfmt_SOURCES = msgfmt.c write-mo.c write-java.c plural-eval.c
 msgmerge_SOURCES = msgmerge.c
 msgunfmt_SOURCES = msgunfmt.c read-mo.c read-java.c
 xgettext_SOURCES = xgettext.c \
-  x-c.c x-po.c x-python.c x-lisp.c x-elisp.c x-librep.c x-java.l x-ycp.c \
-  x-rst.c
+  x-c.c x-po.c x-python.c x-lisp.c x-elisp.c x-librep.c x-java.l x-awk.c \
+  x-ycp.c x-rst.c
 msgattrib_SOURCES = msgattrib.c
 msgcat_SOURCES = msgcat.c
 msgcomm_SOURCES = msgcomm.c
diff --git a/src/format-awk.c b/src/format-awk.c
new file mode 100644
index 000000000..add968fd8
--- /dev/null
+++ b/src/format-awk.c
@@ -0,0 +1,629 @@
+/* awk format strings.
+   Copyright (C) 2001-2002 Free Software Foundation, Inc.
+   Written by Bruno Haible <haible@clisp.cons.org>, 2002.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2, or (at your option)
+   any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software Foundation,
+   Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
+
+#ifdef HAVE_CONFIG_H
+# include <config.h>
+#endif
+
+#include <stdbool.h>
+#include <stdlib.h>
+
+#include "format.h"
+#include "xmalloc.h"
+#include "error.h"
+#include "progname.h"
+#include "gettext.h"
+
+#define _(str) gettext (str)
+
+/* awk format strings are described in the gawk-3.1 documentation and
+   implemented in gawk-3.1.0/builtin.c: format_tree().
+   A directive
+   - starts with '%' or '%m$' where m is a positive integer,
+   - is optionally followed by any of the characters '#', '0', '-', ' ', '+',
+     each of which acts as a flag,
+   - is optionally followed by a width specification: '*' (reads an argument)
+     or '*m$' or a nonempty digit sequence,
+   - is optionally followed by '.' and a precision specification: '*' (reads
+     an argument) or '*m$' or a nonempty digit sequence,
+   - is finished by a specifier
+       - '%', that needs no argument,
+       - 'c', that need a character argument,
+       - 's', that need a string argument,
+       - 'i', 'd', that need a signed integer argument,
+       - 'o', 'u', 'x', 'X', that need an unsigned integer argument,
+       - 'e', 'E', 'f', 'g', 'G', that need a floating-point argument.
+   Numbered ('%m$' or '*m$') and unnumbered argument specifications cannot
+   be used in the same string.
+ */
+
+enum format_arg_type
+{
+  FAT_NONE,
+  FAT_CHARACTER,
+  FAT_STRING,
+  FAT_INTEGER,
+  FAT_UNSIGNED_INTEGER,
+  FAT_FLOAT
+};
+
+struct numbered_arg
+{
+  unsigned int number;
+  enum format_arg_type type;
+};
+
+struct spec
+{
+  unsigned int directives;
+  unsigned int numbered_arg_count;
+  unsigned int allocated;
+  struct numbered_arg *numbered;
+};
+
+/* Locale independent test for a decimal digit.
+   Argument can be  'char' or 'unsigned char'.  (Whereas the argument of
+   <ctype.h> isdigit must be an 'unsigned char'.)  */
+#undef isdigit
+#define isdigit(c) ((unsigned int) ((c) - '0') < 10)
+
+
+/* Prototypes for local functions.  Needed to ensure compiler checking of
+   function argument counts despite of K&R C function definition syntax.  */
+static int numbered_arg_compare PARAMS ((const void *p1, const void *p2));
+static void *format_parse PARAMS ((const char *format));
+static void format_free PARAMS ((void *descr));
+static int format_get_number_of_directives PARAMS ((void *descr));
+static bool format_check PARAMS ((const lex_pos_ty *pos,
+				  void *msgid_descr, void *msgstr_descr,
+				  bool equality,
+				  bool noisy, const char *pretty_msgstr));
+
+
+static int
+numbered_arg_compare (p1, p2)
+     const void *p1;
+     const void *p2;
+{
+  unsigned int n1 = ((const struct numbered_arg *) p1)->number;
+  unsigned int n2 = ((const struct numbered_arg *) p2)->number;
+
+  return (n1 > n2 ? 1 : n1 < n2 ? -1 : 0);
+}
+
+static void *
+format_parse (format)
+     const char *format;
+{
+  struct spec spec;
+  unsigned int unnumbered_arg_count;
+  struct spec *result;
+
+  spec.directives = 0;
+  spec.numbered_arg_count = 0;
+  spec.allocated = 0;
+  spec.numbered = NULL;
+  unnumbered_arg_count = 0;
+
+  for (; *format != '\0';)
+    if (*format++ == '%')
+      {
+	/* A directive.  */
+	unsigned int number = 0;
+	enum format_arg_type type;
+
+	spec.directives++;
+
+	if (isdigit (*format))
+	  {
+	    const char *f = format;
+	    unsigned int m = 0;
+
+	    do
+	      {
+		m = 10 * m + (*f - '0');
+		f++;
+	      }
+	    while (isdigit (*f));
+
+	    if (*f == '$')
+	      {
+		if (m == 0)
+		  goto bad_format;
+		number = m;
+		format = ++f;
+	      }
+	  }
+
+	/* Parse flags.  */
+	while (*format == ' ' || *format == '+' || *format == '-'
+	       || *format == '#' || *format == '0')
+	  format++;
+
+	/* Parse width.  */
+	if (*format == '*')
+	  {
+	    unsigned int width_number = 0;
+
+	    format++;
+
+	    if (isdigit (*format))
+	      {
+		const char *f = format;
+		unsigned int m = 0;
+
+		do
+		  {
+		    m = 10 * m + (*f - '0');
+		    f++;
+		  }
+		while (isdigit (*f));
+
+		if (*f == '$')
+		  {
+		    if (m == 0)
+		      goto bad_format;
+		    width_number = m;
+		    format = ++f;
+		  }
+	      }
+
+	    if (width_number)
+	      {
+		/* Numbered argument.  */
+
+		/* Numbered and unnumbered specifications are exclusive.  */
+		if (unnumbered_arg_count > 0)
+		  goto bad_format;
+
+		if (spec.allocated == spec.numbered_arg_count)
+		  {
+		    spec.allocated = 2 * spec.allocated + 1;
+		    spec.numbered = (struct numbered_arg *) xrealloc (spec.numbered, spec.allocated * sizeof (struct numbered_arg));
+		  }
+		spec.numbered[spec.numbered_arg_count].number = width_number;
+		spec.numbered[spec.numbered_arg_count].type = FAT_INTEGER;
+		spec.numbered_arg_count++;
+	      }
+	    else
+	      {
+		/* Unnumbered argument.  */
+
+		/* Numbered and unnumbered specifications are exclusive.  */
+		if (spec.numbered_arg_count > 0)
+		  goto bad_format;
+
+		if (spec.allocated == unnumbered_arg_count)
+		  {
+		    spec.allocated = 2 * spec.allocated + 1;
+		    spec.numbered = (struct numbered_arg *) xrealloc (spec.numbered, spec.allocated * sizeof (struct numbered_arg));
+		  }
+		spec.numbered[unnumbered_arg_count].number = unnumbered_arg_count + 1;
+		spec.numbered[unnumbered_arg_count].type = FAT_INTEGER;
+		unnumbered_arg_count++;
+	      }
+	  }
+	else if (isdigit (*format))
+	  {
+	    do format++; while (isdigit (*format));
+	  }
+
+	/* Parse precision.  */
+	if (*format == '.')
+	  {
+	    format++;
+
+	    if (*format == '*')
+	      {
+		unsigned int precision_number = 0;
+
+		format++;
+
+		if (isdigit (*format))
+		  {
+		    const char *f = format;
+		    unsigned int m = 0;
+
+		    do
+		      {
+			m = 10 * m + (*f - '0');
+			f++;
+		      }
+		    while (isdigit (*f));
+
+		    if (*f == '$')
+		      {
+			if (m == 0)
+			  goto bad_format;
+			precision_number = m;
+			format = ++f;
+		      }
+		  }
+
+		if (precision_number)
+		  {
+		    /* Numbered argument.  */
+
+		    /* Numbered and unnumbered specifications are exclusive.  */
+		    if (unnumbered_arg_count > 0)
+		      goto bad_format;
+
+		    if (spec.allocated == spec.numbered_arg_count)
+		      {
+			spec.allocated = 2 * spec.allocated + 1;
+			spec.numbered = (struct numbered_arg *) xrealloc (spec.numbered, spec.allocated * sizeof (struct numbered_arg));
+		      }
+		    spec.numbered[spec.numbered_arg_count].number = precision_number;
+		    spec.numbered[spec.numbered_arg_count].type = FAT_INTEGER;
+		    spec.numbered_arg_count++;
+		  }
+		else
+		  {
+		    /* Unnumbered argument.  */
+
+		    /* Numbered and unnumbered specifications are exclusive.  */
+		    if (spec.numbered_arg_count > 0)
+		      goto bad_format;
+
+		    if (spec.allocated == unnumbered_arg_count)
+		      {
+			spec.allocated = 2 * spec.allocated + 1;
+			spec.numbered = (struct numbered_arg *) xrealloc (spec.numbered, spec.allocated * sizeof (struct numbered_arg));
+		      }
+		    spec.numbered[unnumbered_arg_count].type = unnumbered_arg_count + 1;
+		    spec.numbered[unnumbered_arg_count].type = FAT_INTEGER;
+		    unnumbered_arg_count++;
+		  }
+	      }
+	    else if (isdigit (*format))
+	      {
+		do format++; while (isdigit (*format));
+	      }
+	  }
+
+	switch (*format)
+	  {
+	  case '%':
+	    type = FAT_NONE;
+	    break;
+	  case 'c':
+	    type = FAT_CHARACTER;
+	    break;
+	  case 's':
+	    type = FAT_STRING;
+	    break;
+	  case 'i': case 'd':
+	    type = FAT_INTEGER;
+	    break;
+	  case 'u': case 'o': case 'x': case 'X':
+	    type = FAT_UNSIGNED_INTEGER;
+	    break;
+	  case 'e': case 'E': case 'f': case 'g': case 'G':
+	    type = FAT_FLOAT;
+	    break;
+	  default:
+	    goto bad_format;
+	  }
+
+	if (type != FAT_NONE)
+	  {
+	    if (number)
+	      {
+		/* Numbered argument.  */
+
+		/* Numbered and unnumbered specifications are exclusive.  */
+		if (unnumbered_arg_count > 0)
+		  goto bad_format;
+
+		if (spec.allocated == spec.numbered_arg_count)
+		  {
+		    spec.allocated = 2 * spec.allocated + 1;
+		    spec.numbered = (struct numbered_arg *) xrealloc (spec.numbered, spec.allocated * sizeof (struct numbered_arg));
+		  }
+		spec.numbered[spec.numbered_arg_count].number = number;
+		spec.numbered[spec.numbered_arg_count].type = type;
+		spec.numbered_arg_count++;
+	      }
+	    else
+	      {
+		/* Unnumbered argument.  */
+
+		/* Numbered and unnumbered specifications are exclusive.  */
+		if (spec.numbered_arg_count > 0)
+		  goto bad_format;
+
+		if (spec.allocated == unnumbered_arg_count)
+		  {
+		    spec.allocated = 2 * spec.allocated + 1;
+		    spec.numbered = (struct numbered_arg *) xrealloc (spec.numbered, spec.allocated * sizeof (struct numbered_arg));
+		  }
+		spec.numbered[unnumbered_arg_count].number = unnumbered_arg_count + 1;
+		spec.numbered[unnumbered_arg_count].type = type;
+		unnumbered_arg_count++;
+	      }
+	  }
+
+	format++;
+      }
+
+  /* Convert the unnumbered argument array to numbered arguments.  */
+  if (unnumbered_arg_count > 0)
+    spec.numbered_arg_count = unnumbered_arg_count;
+  /* Sort the numbered argument array, and eliminate duplicates.  */
+  else if (spec.numbered_arg_count > 1)
+    {
+      unsigned int i, j;
+      bool err;
+
+      qsort (spec.numbered, spec.numbered_arg_count,
+	     sizeof (struct numbered_arg), numbered_arg_compare);
+
+      /* Remove duplicates: Copy from i to j, keeping 0 <= j <= i.  */
+      err = false;
+      for (i = j = 0; i < spec.numbered_arg_count; i++)
+	if (j > 0 && spec.numbered[i].number == spec.numbered[j-1].number)
+	  {
+	    enum format_arg_type type1 = spec.numbered[i].type;
+	    enum format_arg_type type2 = spec.numbered[j-1].type;
+	    enum format_arg_type type_both;
+
+	    if (type1 == type2)
+	      type_both = type1;
+	    else
+	      /* Incompatible types.  */
+	      type_both = FAT_NONE, err = true;
+
+	    spec.numbered[j-1].type = type_both;
+	  }
+	else
+	  {
+	    if (j < i)
+	      {
+		spec.numbered[j].number = spec.numbered[i].number;
+		spec.numbered[j].type = spec.numbered[i].type;
+	      }
+	    j++;
+	  }
+      spec.numbered_arg_count = j;
+      if (err)
+	goto bad_format;
+    }
+
+  result = (struct spec *) xmalloc (sizeof (struct spec));
+  *result = spec;
+  return result;
+
+ bad_format:
+  if (spec.numbered != NULL)
+    free (spec.numbered);
+  return NULL;
+}
+
+static void
+format_free (descr)
+     void *descr;
+{
+  struct spec *spec = (struct spec *) descr;
+
+  if (spec->numbered != NULL)
+    free (spec->numbered);
+  free (spec);
+}
+
+static int
+format_get_number_of_directives (descr)
+     void *descr;
+{
+  struct spec *spec = (struct spec *) descr;
+
+  return spec->directives;
+}
+
+static bool
+format_check (pos, msgid_descr, msgstr_descr, equality, noisy, pretty_msgstr)
+     const lex_pos_ty *pos;
+     void *msgid_descr;
+     void *msgstr_descr;
+     bool equality;
+     bool noisy;
+     const char *pretty_msgstr;
+{
+  struct spec *spec1 = (struct spec *) msgid_descr;
+  struct spec *spec2 = (struct spec *) msgstr_descr;
+  bool err = false;
+
+  if (spec1->numbered_arg_count + spec2->numbered_arg_count > 0)
+    {
+      unsigned int i, j;
+      unsigned int n1 = spec1->numbered_arg_count;
+      unsigned int n2 = spec2->numbered_arg_count;
+
+      /* Check the argument names are the same.
+	 Both arrays are sorted.  We search for the first difference.  */
+      for (i = 0, j = 0; i < n1 || j < n2; )
+	{
+	  int cmp = (i >= n1 ? 1 :
+		     j >= n2 ? -1 :
+		     spec1->numbered[i].number > spec2->numbered[j].number ? 1 :
+		     spec1->numbered[i].number < spec2->numbered[j].number ? -1 :
+		     0);
+
+	  if (cmp > 0)
+	    {
+	      if (noisy)
+		{
+		  error_with_progname = false;
+		  error_at_line (0, 0, pos->file_name, pos->line_number,
+				 _("a format specification for argument %u, as in '%s', doesn't exist in 'msgid'"),
+				 spec2->numbered[j].number, pretty_msgstr);
+		  error_with_progname = true;
+		}
+	      err = true;
+	      break;
+	    }
+	  else if (cmp < 0)
+	    {
+	      if (equality)
+		{
+		  if (noisy)
+		    {
+		      error_with_progname = false;
+		      error_at_line (0, 0, pos->file_name, pos->line_number,
+				     _("a format specification for argument %u doesn't exist in '%s'"),
+				     spec1->numbered[i].number, pretty_msgstr);
+		      error_with_progname = true;
+		    }
+		  err = true;
+		  break;
+		}
+	      else
+		i++;
+	    }
+	  else
+	    j++, i++;
+	}
+      /* Check the argument types are the same.  */
+      if (!err)
+	for (i = 0, j = 0; j < n2; )
+	  {
+	    if (spec1->numbered[i].number == spec2->numbered[j].number)
+	      {
+		if (spec1->numbered[i].type != spec2->numbered[j].type)
+		  {
+		    if (noisy)
+		      {
+			error_with_progname = false;
+			error_at_line (0, 0, pos->file_name, pos->line_number,
+				       _("format specifications in 'msgid' and '%s' for argument %u are not the same"),
+				       pretty_msgstr,
+				       spec2->numbered[j].number);
+			error_with_progname = true;
+		      }
+		    err = true;
+		    break;
+		  }
+		j++, i++;
+	      }
+	    else
+	      i++;
+	  }
+    }
+
+  return err;
+}
+
+
+struct formatstring_parser formatstring_awk =
+{
+  format_parse,
+  format_free,
+  format_get_number_of_directives,
+  format_check
+};
+
+
+#ifdef TEST
+
+/* Test program: Print the argument list specification returned by
+   format_parse for strings read from standard input.  */
+
+#include <stdio.h>
+#include "getline.h"
+
+static void
+format_print (descr)
+     void *descr;
+{
+  struct spec *spec = (struct spec *) descr;
+  unsigned int last;
+  unsigned int i;
+
+  if (spec == NULL)
+    {
+      printf ("INVALID");
+      return;
+    }
+
+  printf ("(");
+  last = 1;
+  for (i = 0; i < spec->numbered_arg_count; i++)
+    {
+      unsigned int number = spec->numbered[i].number;
+
+      if (i > 0)
+	printf (" ");
+      if (number < last)
+	abort ();
+      for (; last < number; last++)
+	printf ("_ ");
+      switch (spec->numbered[i].type)
+	{
+	case FAT_CHARACTER:
+	  printf ("c");
+	  break;
+	case FAT_STRING:
+	  printf ("s");
+	  break;
+	case FAT_INTEGER:
+	  printf ("i");
+	  break;
+	case FAT_UNSIGNED_INTEGER:
+	  printf ("[unsigned]i");
+	  break;
+	case FAT_FLOAT:
+	  printf ("f");
+	  break;
+	default:
+	  abort ();
+	}
+      last = number + 1;
+    }
+  printf (")");
+}
+
+int
+main ()
+{
+  for (;;)
+    {
+      char *line = NULL;
+      size_t line_len = 0;
+      void *descr;
+
+      if (getline (&line, &line_len, stdin) < 0)
+	break;
+
+      descr = format_parse (line);
+
+      format_print (descr);
+      printf ("\n");
+
+      free (line);
+    }
+
+  return 0;
+}
+
+/*
+ * For Emacs M-x compile
+ * Local Variables:
+ * compile-command: "/bin/sh ../libtool --mode=link gcc -o a.out -static -O -g -Wall -I.. -I../lib -I../intl -DHAVE_CONFIG_H -DTEST format-awk.c ../lib/libgettextlib.la"
+ * End:
+ */
+
+#endif /* TEST */
diff --git a/src/format.c b/src/format.c
index e2c2fea68..6b4346059 100644
--- a/src/format.c
+++ b/src/format.c
@@ -33,6 +33,7 @@ struct formatstring_parser *formatstring_parsers[NFORMATS] =
   /* format_librep */		&formatstring_librep,
   /* format_smalltalk */	&formatstring_smalltalk,
   /* format_java */		&formatstring_java,
+  /* format_awk */		&formatstring_awk,
   /* format_pascal */		&formatstring_pascal,
   /* format_ycp */		&formatstring_ycp
 };
diff --git a/src/format.h b/src/format.h
index b50d052ae..1df1c14a6 100644
--- a/src/format.h
+++ b/src/format.h
@@ -61,6 +61,7 @@ extern struct formatstring_parser formatstring_elisp;
 extern struct formatstring_parser formatstring_librep;
 extern struct formatstring_parser formatstring_smalltalk;
 extern struct formatstring_parser formatstring_java;
+extern struct formatstring_parser formatstring_awk;
 extern struct formatstring_parser formatstring_pascal;
 extern struct formatstring_parser formatstring_ycp;
 
diff --git a/src/message.c b/src/message.c
index 130821fd0..84ac2d005 100644
--- a/src/message.c
+++ b/src/message.c
@@ -47,6 +47,7 @@ const char *const format_language[NFORMATS] =
   /* format_librep */		"librep",
   /* format_smalltalk */	"smalltalk",
   /* format_java */		"java",
+  /* format_awk */		"awk",
   /* format_pascal */		"object-pascal",
   /* format_ycp */		"ycp"
 };
@@ -60,6 +61,7 @@ const char *const format_language_pretty[NFORMATS] =
   /* format_librep */		"librep",
   /* format_smalltalk */	"Smalltalk",
   /* format_java */		"Java",
+  /* format_awk */		"awk",
   /* format_pascal */		"Object Pascal",
   /* format_ycp */		"YCP"
 };
diff --git a/src/message.h b/src/message.h
index 1cd8d0085..aad981b7e 100644
--- a/src/message.h
+++ b/src/message.h
@@ -41,10 +41,11 @@ enum format_type
   format_librep,
   format_smalltalk,
   format_java,
+  format_awk,
   format_pascal,
   format_ycp
 };
-#define NFORMATS 9	/* Number of format_type enum values.  */
+#define NFORMATS 10	/* Number of format_type enum values.  */
 extern const char *const format_language[NFORMATS];
 extern const char *const format_language_pretty[NFORMATS];
 
diff --git a/src/x-awk.c b/src/x-awk.c
new file mode 100644
index 000000000..8915e5310
--- /dev/null
+++ b/src/x-awk.c
@@ -0,0 +1,866 @@
+/* xgettext awk backend.
+   Copyright (C) 2002 Free Software Foundation, Inc.
+
+   This file was written by Bruno Haible <haible@clisp.cons.org>, 2002.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2, or (at your option)
+   any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software Foundation,
+   Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
+
+#ifdef HAVE_CONFIG_H
+# include "config.h"
+#endif
+
+#include <errno.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "message.h"
+#include "x-awk.h"
+#include "xgettext.h"
+#include "error.h"
+#include "progname.h"
+#include "xmalloc.h"
+#include "exit.h"
+#include "gettext.h"
+
+#define _(s) gettext(s)
+
+#if HAVE_C_BACKSLASH_A
+# define ALERT_CHAR '\a'
+#else
+# define ALERT_CHAR '\7'
+#endif
+
+
+/* The awk syntax is defined in the gawk manual page and documentation.
+   See also gawk/awkgram.y.  */
+
+enum token_type_ty
+{
+  token_type_eof,
+  token_type_lparen,		/* ( */
+  token_type_rparen,		/* ) */
+  token_type_comma,		/* , */
+  token_type_string,		/* "abc" */
+  token_type_i18nstring,	/* _"abc" */
+  token_type_symbol,		/* symbol, number */
+  token_type_other		/* regexp, misc. operator */
+};
+typedef enum token_type_ty token_type_ty;
+
+typedef struct token_ty token_ty;
+struct token_ty
+{
+  token_type_ty type;
+  char *string;		/* for token_type_{symbol,string,i18nstring} */
+  int line_number;
+};
+
+
+/* Prototypes for local functions.  Needed to ensure compiler checking of
+   function argument counts despite of K&R C function definition syntax.  */
+static void init_keywords PARAMS ((void));
+static int phase1_getc PARAMS ((void));
+static void phase1_ungetc PARAMS ((int c));
+static int phase2_getc PARAMS ((void));
+static void phase2_ungetc PARAMS ((int c));
+static int phase7_getc PARAMS ((void));
+static inline void free_token PARAMS ((token_ty *tp));
+static void x_awk_lex PARAMS ((token_ty *tp));
+static bool extract_parenthesized PARAMS ((message_list_ty *mlp,
+					   int commas_to_skip,
+					   int plural_commas));
+
+
+/* ====================== Keyword set customization.  ====================== */
+
+/* If true extract all strings.  */
+static bool extract_all = false;
+
+static hash_table keywords;
+static bool default_keywords = true;
+
+
+void
+x_awk_extract_all ()
+{
+  extract_all = true;
+}
+
+
+void
+x_awk_keyword (name)
+     const char *name;
+{
+  if (name == NULL)
+    default_keywords = false;
+  else
+    {
+      const char *end;
+      int argnum1;
+      int argnum2;
+      const char *colon;
+
+      if (keywords.table == NULL)
+	init_hash (&keywords, 100);
+
+      split_keywordspec (name, &end, &argnum1, &argnum2);
+
+      /* The characters between name and end should form a valid C identifier.
+	 A colon means an invalid parse in split_keywordspec().  */
+      colon = strchr (name, ':');
+      if (colon == NULL || colon >= end)
+	{
+	  if (argnum1 == 0)
+	    argnum1 = 1;
+	  insert_entry (&keywords, name, end - name,
+			(void *) (long) (argnum1 + (argnum2 << 10)));
+	}
+    }
+}
+
+/* Finish initializing the keywords hash table.
+   Called after argument processing, before each file is processed.  */
+static void
+init_keywords ()
+{
+  if (default_keywords)
+    {
+      x_awk_keyword ("dcgettext");
+      x_awk_keyword ("dcngettext:1,2");
+      default_keywords = false;
+    }
+}
+
+
+/* ================== Reading of characters and tokens.  =================== */
+
+/* Real filename, used in error messages about the input file.  */
+static const char *real_file_name;
+
+/* Logical filename and line number, used to label the extracted messages.  */
+static char *logical_file_name;
+static int line_number;
+
+/* The input file stream.  */
+static FILE *fp;
+
+/* These are for tracking whether comments count as immediately before
+   keyword.  */
+static int last_comment_line;
+static int last_non_comment_line;
+
+
+/* 1. line_number handling.  */
+
+static int
+phase1_getc ()
+{
+  int c = getc (fp);
+
+  if (c == EOF)
+    {
+      if (ferror (fp))
+	error (EXIT_FAILURE, errno, _("error while reading \"%s\""),
+	       real_file_name);
+      return EOF;
+    }
+
+  if (c == '\n')
+    line_number++;
+
+  return c;
+}
+
+static void
+phase1_ungetc (c)
+     int c;
+{
+  if (c != EOF)
+    {
+      if (c == '\n')
+	--line_number;
+
+      ungetc (c, fp);
+    }
+}
+
+
+/* 2. Replace each comment that is not inside a string literal or regular
+   expression with a newline character.  We need to remember the comment
+   for later, because it may be attached to a keyword string.  */
+
+static int
+phase2_getc ()
+{
+  static char *buffer;
+  static size_t bufmax;
+  size_t buflen;
+  int lineno;
+  int c;
+
+  c = phase1_getc ();
+  if (c == '#')
+    {
+      buflen = 0;
+      lineno = line_number;
+      for (;;)
+	{
+	  c = phase1_getc ();
+	  if (c == '\n' || c == EOF)
+	    break;
+	  if (buflen >= bufmax)
+	    {
+	      bufmax += 100;
+	      buffer = xrealloc (buffer, bufmax);
+	    }
+	  buffer[buflen++] = c;
+	}
+      if (buflen >= bufmax)
+	{
+	  bufmax += 100;
+	  buffer = xrealloc (buffer, bufmax);
+	}
+      buffer[buflen] = '\0';
+      xgettext_comment_add (buffer);
+      last_comment_line = lineno;
+    }
+  return c;
+}
+
+static void
+phase2_ungetc (c)
+     int c;
+{
+  if (c != EOF)
+    phase1_ungetc (c);
+}
+
+
+/* 7. Replace escape sequences within character strings with their
+   single character equivalents.  */
+
+#define P7_QUOTES (1000 + '"')
+
+static int
+phase7_getc ()
+{
+  int c;
+
+  for (;;)
+    {
+      /* Use phase 1, because phase 2 elides comments.  */
+      c = phase1_getc ();
+
+      if (c == EOF || c == '\n')
+	break;
+      if (c == '"')
+	return P7_QUOTES;
+      if (c != '\\')
+	return c;
+      c = phase1_getc ();
+      if (c == EOF)
+	break;
+      if (c != '\n')
+	switch (c)
+	  {
+	  case 'a':
+	    return ALERT_CHAR;
+	  case 'b':
+	    return '\b';
+	  case 'f':
+	    return '\f';
+	  case 'n':
+	    return '\n';
+	  case 'r':
+	    return '\r';
+	  case 't':
+	    return '\t';
+	  case 'v':
+	    return '\v';
+	  case '0': case '1': case '2': case '3': case '4':
+	  case '5': case '6': case '7':
+	    {
+	      int n = c - '0';
+
+	      c = phase1_getc ();
+	      if (c != EOF)
+		{
+		  if (c >= '0' && c <= '7')
+		    {
+		      n = (n << 3) + (c - '0');
+		      c = phase1_getc ();
+		      if (c != EOF)
+			{
+			  if (c >= '0' && c <= '7')
+			    n = (n << 3) + (c - '0');
+			  else
+			    phase1_ungetc (c);
+			}
+		    }
+		  else
+		    phase1_ungetc (c);
+		}
+	      return (unsigned char) n;
+	    }
+	  case 'x':
+	    {
+	      int n = 0;
+
+	      for (;;)
+		{
+		  c = phase1_getc ();
+		  if (c == EOF)
+		    break;
+		  else if (c >= '0' && c <= '9')
+		    n = (n << 4) + (c - '0');
+		  else if (c >= 'A' && c <= 'F')
+		    n = (n << 4) + (c - 'A' + 10);
+		  else if (c >= 'a' && c <= 'f')
+		    n = (n << 4) + (c - 'a' + 10);
+		  else
+		    {
+		      phase1_ungetc (c);
+		      break;
+		    }
+		}
+	      return (unsigned char) n;
+	    }
+	  default:
+	    return c;
+	  }
+    }
+
+  phase1_ungetc (c);
+  error_with_progname = false;
+  error (0, 0, _("%s:%d: warning: unterminated string"), logical_file_name,
+	 line_number);
+  error_with_progname = true;
+  return P7_QUOTES;
+}
+
+
+/* Free the memory pointed to by a 'struct token_ty'.  */
+static inline void
+free_token (tp)
+     token_ty *tp;
+{
+  switch (tp->type)
+    {
+    case token_type_string:
+    case token_type_i18nstring:
+    case token_type_symbol:
+      free (tp->string);
+      break;
+    default:
+      break;
+    }
+}
+
+
+/* Combine characters into tokens.  Discard whitespace.  */
+
+/* There is an ambiguity about '/': It can start a division operator ('/' or
+   '/=') or it can start a regular expression.  The distinction is important
+   because inside regular expressions, '#' and '"' lose its special meanings.
+   If you look at the awk grammar, you see that the operator is only allowed
+   right after a 'variable' or 'simp_exp' nonterminal, and these nonterminals
+   can only end in the NAME, LENGTH, YSTRING, YNUMBER, ')', ']' terminals.
+   So we prefer the division operator interpretation only right after
+   symbol, string, number, ')', ']', with whitespace but no newline allowed
+   in between.  */
+static bool prefer_division_over_regexp;
+
+static void
+x_awk_lex (tp)
+     token_ty *tp;
+{
+  static char *buffer;
+  static int bufmax;
+  int bufpos;
+  int c;
+
+  for (;;)
+    {
+      tp->line_number = line_number;
+      c = phase2_getc ();
+
+      switch (c)
+	{
+	case EOF:
+	  tp->type = token_type_eof;
+	  return;
+
+	case '\n':
+	  if (last_non_comment_line > last_comment_line)
+	    xgettext_comment_reset ();
+	  /* Newline is not allowed inside expressions.  It usually
+	     introduces a fresh statement.  */
+	  prefer_division_over_regexp = false;
+	  /* FALLTHROUGH */
+	case '\t':
+	case ' ':
+	  /* Ignore whitespace and comments.  */
+	  continue;
+
+	case '\\':
+	  /* Backslash ought to be immediately followed by a newline.  */
+	  continue;
+	}
+
+      last_non_comment_line = tp->line_number;
+
+      switch (c)
+	{
+	case '.':
+	  {
+	    int c2 = phase2_getc ();
+	    phase2_ungetc (c2);
+	    if (!(c2 >= '0' && c2 <= '9'))
+	      {
+
+		tp->type = token_type_other;
+		prefer_division_over_regexp = false;
+		return;
+	      }
+	  }
+	  /* FALLTHROUGH */
+	case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
+	case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
+	case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
+	case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
+	case 'Y': case 'Z':
+	case '_':
+	case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
+	case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
+	case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
+	case 's': case 't': case 'u': case 'v': case 'w': case 'x':
+	case 'y': case 'z':
+	case '0': case '1': case '2': case '3': case '4':
+	case '5': case '6': case '7': case '8': case '9':
+	  /* Symbol, or part of a number.  */
+	  bufpos = 0;
+	  for (;;)
+	    {
+	      if (bufpos >= bufmax)
+		{
+		  bufmax += 100;
+		  buffer = xrealloc (buffer, bufmax);
+		}
+	      buffer[bufpos++] = c;
+	      c = phase2_getc ();
+	      switch (c)
+		{
+		case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
+		case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
+		case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
+		case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
+		case 'Y': case 'Z':
+		case '_':
+		case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
+		case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
+		case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
+		case 's': case 't': case 'u': case 'v': case 'w': case 'x':
+		case 'y': case 'z':
+		case '0': case '1': case '2': case '3': case '4':
+		case '5': case '6': case '7': case '8': case '9':
+		  continue;
+		default:
+		  if (bufpos == 1 && buffer[0] == '_' && c == '"')
+		    {
+		      tp->type = token_type_i18nstring;
+		      goto case_string;
+		    }
+		  phase2_ungetc (c);
+		  break;
+		}
+	      break;
+	    }
+	  if (bufpos >= bufmax)
+	    {
+	      bufmax += 100;
+	      buffer = xrealloc (buffer, bufmax);
+	    }
+	  buffer[bufpos] = '\0';
+	  tp->string = xstrdup (buffer);
+	  tp->type = token_type_symbol;
+	  /* Most identifiers can be variable names; after them we must
+	     interpret '/' as division operator.  But for awk's builtin
+	     keywords we have three cases:
+	     (a) Must interpret '/' as division operator. "length".
+	     (b) Must interpret '/' as start of a regular expression.
+		 "do", "exit", "print", "printf", "return".
+	     (c) '/' after this keyword in invalid anyway. All others.
+	     I used the following script for the distinction.
+		for k in $awk_keywords; do
+		  echo; echo $k; awk "function foo () { $k / 10 }" < /dev/null
+		done
+	   */
+	  if (strcmp (buffer, "do") == 0
+	      || strcmp (buffer, "exit") == 0
+	      || strcmp (buffer, "print") == 0
+	      || strcmp (buffer, "printf") == 0
+	      || strcmp (buffer, "return") == 0)
+	    prefer_division_over_regexp = false;
+	  else
+	    prefer_division_over_regexp = true;
+	  return;
+
+	case '"':
+	  tp->type = token_type_string;
+	case_string:
+	  bufpos = 0;
+	  for (;;)
+	    {
+	      c = phase7_getc ();
+	      if (c == EOF || c == P7_QUOTES)
+		break;
+	      if (bufpos >= bufmax)
+		{
+		  bufmax += 100;
+		  buffer = xrealloc (buffer, bufmax);
+		}
+	      buffer[bufpos++] = c;
+	    }
+	  if (bufpos >= bufmax)
+	    {
+	      bufmax += 100;
+	      buffer = xrealloc (buffer, bufmax);
+	    }
+	  buffer[bufpos] = '\0';
+	  tp->string = xstrdup (buffer);
+	  prefer_division_over_regexp = true;
+	  return;
+
+	case '(':
+	  tp->type = token_type_lparen;
+	  prefer_division_over_regexp = false;
+	  return;
+
+	case ')':
+	  tp->type = token_type_rparen;
+	  prefer_division_over_regexp = true;
+	  return;
+
+	case ',':
+	  tp->type = token_type_comma;
+	  prefer_division_over_regexp = false;
+	  return;
+
+	case ']':
+	  tp->type = token_type_other;
+	  prefer_division_over_regexp = true;
+	  return;
+
+	case '/':
+	  if (!prefer_division_over_regexp)
+	    {
+	      /* Regular expression.
+	         Counting brackets is non-trivial. [[] is balanced, and so is
+	         [\]]. Also, /[/]/ is balanced and ends at the third slash.
+	         Do not count [ or ] if either one is preceded by a \.
+	         A '[' should be counted if
+	          a) it is the first one so far (brackets == 0), or
+	          b) it is the '[' in '[:'.
+	         A ']' should be counted if not preceded by a \.
+	         According to POSIX, []] is how you put a ] into a set.
+	         Try to handle that too.
+	       */
+	      int brackets = 0;
+	      bool pos0 = true;		/* true at start of regexp */
+	      bool pos1_open = false;	/* true after [ at start of regexp */
+	      bool pos2_open_not = false; /* true after [^ at start of regexp */
+
+	      for (;;)
+		{
+		  c = phase1_getc ();
+
+		  if (c == EOF || c == '\n')
+		    {
+		      phase1_ungetc (c);
+		      error_with_progname = false;
+		      error (0, 0, _("%s:%d: warning: unterminated regular expression"),
+			     logical_file_name, line_number);
+		      error_with_progname = true;
+		      break;
+		    }
+		  else if (c == '[')
+		    {
+		      if (brackets == 0)
+			brackets++;
+		      else
+			{
+			  c = phase1_getc ();
+			  if (c == ':')
+			    brackets++;
+			  phase1_ungetc (c);
+			}
+		      if (pos0)
+			{
+			  pos0 = false;
+			  pos1_open = true;
+			  continue;
+			}
+		    }
+		  else if (c == ']')
+		    {
+		      if (!(pos1_open || pos2_open_not))
+			brackets--;
+		    }
+		  else if (c == '^')
+		    {
+		      if (pos1_open)
+			{
+			  pos1_open = false;
+			  pos2_open_not = true;
+			  continue;
+			}
+		    }
+		  else if (c == '\\')
+		    {
+		      c = phase1_getc ();
+		      /* Backslash-newline is valid and ignored.  */
+		    }
+		  else if (c == '/')
+		    {
+		      if (brackets <= 0)
+			break;
+		    }
+
+		  pos0 = false;
+		  pos1_open = false;
+		  pos2_open_not = false;
+		}
+
+	      tp->type = token_type_other;
+	      prefer_division_over_regexp = false;
+	      return;
+	    }
+	  /* FALLTHROUGH */
+
+	default:
+	  /* We could carefully recognize each of the 2 and 3 character
+	     operators, but it is not necessary, as we only need to recognize
+	     gettext invocations.  Don't bother.  */
+	  tp->type = token_type_other;
+	  prefer_division_over_regexp = false;
+	  return;
+	}
+    }
+}
+
+
+/* ========================= Extracting strings.  ========================== */
+
+/* The file is broken into tokens.  Scan the token stream, looking for
+   a keyword, followed by a left paren, followed by a string.  When we
+   see this sequence, we have something to remember.  We assume we are
+   looking at a valid C or C++ program, and leave the complaints about
+   the grammar to the compiler.
+
+     Normal handling: Look for
+       keyword ( ... msgid ... )
+     Plural handling: Look for
+       keyword ( ... msgid ... msgid_plural ... )
+
+   We use recursion because the arguments before msgid or between msgid
+   and msgid_plural can contain subexpressions of the same form.  */
+
+
+/* Extract messages until the next balanced closing parenthesis.
+   Extracted messages are added to MLP.
+   When a specific argument shall be extracted, COMMAS_TO_SKIP >= 0 and,
+   if also a plural argument shall be extracted, PLURAL_COMMAS > 0,
+   otherwise PLURAL_COMMAS = 0.
+   When no specific argument shall be extracted, COMMAS_TO_SKIP < 0.
+   Return true upon eof, false upon closing parenthesis.  */
+static bool
+extract_parenthesized (mlp, commas_to_skip, plural_commas)
+     message_list_ty *mlp;
+     int commas_to_skip;
+     int plural_commas;
+{
+  /* Remember the message containing the msgid, for msgid_plural.  */
+  message_ty *plural_mp = NULL;
+
+  /* 0 when no keyword has been seen.  1 right after a keyword is seen.  */
+  int state;
+  /* Parameters of the keyword just seen.  Defined only in state 1.  */
+  int next_commas_to_skip = -1;
+  int next_plural_commas = 0;
+
+  /* Start state is 0.  */
+  state = 0;
+
+  while (1)
+    {
+      token_ty token;
+
+      x_awk_lex (&token);
+      switch (token.type)
+	{
+	case token_type_symbol:
+	  /* No need to bother if we extract all strings anyway.  */
+	  if (!extract_all)
+	    {
+	      void *keyword_value;
+
+	      if (find_entry (&keywords, token.string, strlen (token.string),
+			      &keyword_value)
+		  == 0)
+		{
+		  int argnum1 = (int) (long) keyword_value & ((1 << 10) - 1);
+		  int argnum2 = (int) (long) keyword_value >> 10;
+
+		  next_commas_to_skip = argnum1 - 1;
+		  next_plural_commas = (argnum2 > argnum1 ? argnum2 - argnum1 : 0);
+		  state = 1;
+		}
+	      else
+		state = 0;
+	    }
+	  free (token.string);
+	  continue;
+
+	case token_type_lparen:
+	  /* No need to recurse if we extract all strings anyway.  */
+	  if (extract_all)
+	    continue;
+	  if (state
+	      ?  extract_parenthesized (mlp, next_commas_to_skip,
+					next_plural_commas)
+	      : extract_parenthesized (mlp, -1, 0))
+	    return true;
+	  state = 0;
+	  continue;
+
+	case token_type_rparen:
+	  /* No need to return if we extract all strings anyway.  */
+	  if (extract_all)
+	    continue;
+	  return false;
+
+	case token_type_comma:
+	  /* No need to bother if we extract all strings anyway.  */
+	  if (extract_all)
+	    continue;
+	  if (commas_to_skip >= 0)
+	    {
+	      if (commas_to_skip > 0)
+		commas_to_skip--;
+	      else
+		if (plural_mp != NULL && plural_commas > 0)
+		  {
+		    commas_to_skip = plural_commas - 1;
+		    plural_commas = 0;
+		  }
+		else
+		  commas_to_skip = -1;
+	    }
+	  state = 0;
+	  continue;
+
+	case token_type_string:
+	  {
+	    lex_pos_ty pos;
+	    pos.file_name = logical_file_name;
+	    pos.line_number = token.line_number;
+
+	    if (extract_all)
+	      remember_a_message (mlp, token.string, &pos);
+	    else
+	      {
+		if (commas_to_skip == 0)
+		  {
+		    if (plural_mp == NULL)
+		      {
+			/* Seen an msgid.  */
+			message_ty *mp = remember_a_message (mlp, token.string,
+							     &pos);
+			if (plural_commas > 0)
+			  plural_mp = mp;
+		      }
+		    else
+		      {
+			/* Seen an msgid_plural.  */
+			remember_a_message_plural (plural_mp, token.string,
+						   &pos);
+			plural_mp = NULL;
+		      }
+		  }
+		else
+		  free (token.string);
+		state = 0;
+	      }
+	    continue;
+	  }
+
+	case token_type_i18nstring:
+	  {
+	    lex_pos_ty pos;
+	    pos.file_name = logical_file_name;
+	    pos.line_number = token.line_number;
+
+	    remember_a_message (mlp, token.string, &pos);
+	  }
+	  state = 0;
+	  continue;
+
+	case token_type_eof:
+	  return true;
+
+	case token_type_other:
+	  state = 0;
+	  continue;
+
+	default:
+	  abort ();
+	}
+    }
+}
+
+
+void
+extract_awk (f, real_filename, logical_filename, mdlp)
+     FILE *f;
+     const char *real_filename;
+     const char *logical_filename;
+     msgdomain_list_ty *mdlp;
+{
+  message_list_ty *mlp = mdlp->item[0]->messages;
+
+  fp = f;
+  real_file_name = real_filename;
+  logical_file_name = xstrdup (logical_filename);
+  line_number = 1;
+
+  last_comment_line = -1;
+  last_non_comment_line = -1;
+
+  prefer_division_over_regexp = false;
+
+  init_keywords ();
+
+  /* Eat tokens until eof is seen.  When extract_parenthesized returns
+     due to an unbalanced closing parenthesis, just restart it.  */
+  while (!extract_parenthesized (mlp, -1, 0))
+    ;
+
+  fp = NULL;
+  real_file_name = NULL;
+  logical_file_name = NULL;
+  line_number = 0;
+}
diff --git a/src/x-awk.h b/src/x-awk.h
new file mode 100644
index 000000000..0b10b0fb5
--- /dev/null
+++ b/src/x-awk.h
@@ -0,0 +1,32 @@
+/* xgettext awk backend.
+   Copyright (C) 2002 Free Software Foundation, Inc.
+   Written by Bruno Haible <haible@clisp.cons.org>, 2002.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2, or (at your option)
+   any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software Foundation,
+   Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
+
+
+#define EXTENSIONS_AWK \
+  { "awk",    "awk"   },						\
+
+#define SCANNERS_AWK \
+  { "awk",        extract_awk, &formatstring_awk },			\
+
+/* Scan an awk file and add its translatable strings to mdlp.  */
+extern void extract_awk PARAMS ((FILE *fp, const char *real_filename,
+				 const char *logical_filename,
+				 msgdomain_list_ty *mdlp));
+
+extern void x_awk_keyword PARAMS ((const char *keyword));
+extern void x_awk_extract_all PARAMS ((void));
diff --git a/src/xgettext.c b/src/xgettext.c
index d288a30c8..11f174ddd 100644
--- a/src/xgettext.c
+++ b/src/xgettext.c
@@ -64,6 +64,7 @@
 #include "x-elisp.h"
 #include "x-librep.h"
 #include "x-java.h"
+#include "x-awk.h"
 #include "x-ycp.h"
 #include "x-rst.h"
 
@@ -229,6 +230,7 @@ main (argc, argv)
 	x_elisp_extract_all ();
 	x_librep_extract_all ();
 	x_java_extract_all ();
+	x_awk_extract_all ();
 	break;
       case 'c':
 	if (optarg == NULL)
@@ -284,6 +286,7 @@ main (argc, argv)
 	    x_elisp_keyword (optarg);
 	    x_librep_keyword (optarg);
 	    x_java_keyword (optarg);
+	    x_awk_keyword (optarg);
 	  }
 	break;
       case 'K':
@@ -1252,10 +1255,11 @@ language_to_extractor (name)
     SCANNERS_ELISP
     SCANNERS_LIBREP
     SCANNERS_JAVA
+    SCANNERS_AWK
     SCANNERS_YCP
     SCANNERS_RST
-    /* Here will follow more languages and their scanners: awk, perl,
-       etc...  Make sure new scanners honor the --exclude-file option.  */
+    /* Here will follow more languages and their scanners: perl, etc...
+       Make sure new scanners honor the --exclude-file option.  */
   };
 
   table_ty *tp;
@@ -1295,6 +1299,7 @@ extension_to_language (extension)
     EXTENSIONS_ELISP
     EXTENSIONS_LIBREP
     EXTENSIONS_JAVA
+    EXTENSIONS_AWK
     EXTENSIONS_YCP
     EXTENSIONS_RST
     /* Here will follow more file extensions: sh, pl, tcl ... */
diff --git a/tests/ChangeLog b/tests/ChangeLog
index d58af35fe..762397c17 100644
--- a/tests/ChangeLog
+++ b/tests/ChangeLog
@@ -1,3 +1,10 @@
+2002-01-27  Bruno Haible  <bruno@clisp.org>
+
+	* format-awk-1: New file.
+	* format-awk-2: New file.
+	* lang-gawk: New file.
+	* Makefile.am (TESTS): Add format-awk-1, format-awk-2, lang-gawk.
+
 2002-02-02  Bruno Haible  <bruno@clisp.org>
 
 	* xgettext-18: New file.
diff --git a/tests/Makefile.am b/tests/Makefile.am
index e7e7e335c..64a40284d 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -46,6 +46,7 @@ TESTS = gettext-1 gettext-2 \
 	xgettext-7 xgettext-8 xgettext-9 xgettext-10 xgettext-11 xgettext-12 \
 	xgettext-13 xgettext-14 xgettext-15 xgettext-16 xgettext-17 \
 	xgettext-18 \
+	format-awk-1 format-awk-2 \
 	format-c-1 format-c-2 \
 	format-elisp-1 format-elisp-2 \
 	format-java-1 format-java-2 \
@@ -55,7 +56,7 @@ TESTS = gettext-1 gettext-2 \
 	format-pascal-1 format-pascal-2 \
 	format-ycp-1 format-ycp-2 \
 	plural-1 plural-2 \
-	lang-c lang-c++ lang-objc lang-python lang-clisp lang-elisp lang-librep lang-java lang-pascal lang-ycp lang-po lang-rst \
+	lang-c lang-c++ lang-objc lang-python lang-clisp lang-elisp lang-librep lang-java lang-gawk lang-pascal lang-ycp lang-po lang-rst \
 	rpath-1a rpath-1b \
 	rpath-2aaa rpath-2aab rpath-2aac rpath-2aad \
 	rpath-2aba rpath-2abb rpath-2abc rpath-2abd \
diff --git a/tests/format-awk-1 b/tests/format-awk-1
new file mode 100755
index 000000000..e5a2b3a01
--- /dev/null
+++ b/tests/format-awk-1
@@ -0,0 +1,140 @@
+#! /bin/sh
+
+# Test recognition of awk format strings.
+
+tmpfiles=""
+trap 'rm -fr $tmpfiles' 1 2 3 15
+
+tmpfiles="$tmpfiles f-a-1.data"
+cat <<\EOF > f-a-1.data
+# Valid: no argument
+"abc%%"
+# Valid: one character argument
+"abc%c"
+# Valid: one string argument
+"abc%s"
+# Valid: one integer argument
+"abc%i"
+# Valid: one integer argument
+"abc%d"
+# Valid: one integer argument
+"abc%o"
+# Valid: one integer argument
+"abc%u"
+# Valid: one integer argument
+"abc%x"
+# Valid: one integer argument
+"abc%X"
+# Valid: one floating-point argument
+"abc%e"
+# Valid: one floating-point argument
+"abc%E"
+# Valid: one floating-point argument
+"abc%f"
+# Valid: one floating-point argument
+"abc%g"
+# Valid: one floating-point argument
+"abc%G"
+# Valid: one argument with flags
+"abc%0#g"
+# Valid: one argument with width
+"abc%2g"
+# Valid: one argument with width
+"abc%*g"
+# Valid: one argument with precision
+"abc%.4g"
+# Valid: one argument with precision
+"abc%.*g"
+# Valid: one argument with width and precision
+"abc%14.4g"
+# Valid: one argument with width and precision
+"abc%14.*g"
+# Valid: one argument with width and precision
+"abc%*.4g"
+# Valid: one argument with width and precision
+"abc%*.*g"
+# Invalid: unterminated
+"abc%"
+# Invalid: unknown format specifier
+"abc%y"
+# Invalid: unknown format specifier
+"abc%F"
+# Invalid: flags after width
+"abc%*0g"
+# Invalid: twice precision
+"abc%.4.2g"
+# Valid: three arguments
+"abc%d%u%u"
+# Valid: a numbered argument
+"abc%1$d"
+# Invalid: zero
+"abc%0$d"
+# Valid: two-digit numbered arguments
+"abc%11$def%10$dgh%9$dij%8$dkl%7$dmn%6$dop%5$dqr%4$dst%3$duv%2$dwx%1$dyz"
+# Invalid: unterminated number
+"abc%1"
+# Invalid: flags before number
+"abc%+1$d"
+# Valid: three arguments, two with same number
+"abc%1$4x,%2$c,%1$u"
+# Invalid: argument with conflicting types
+"abc%1$4x,%2$c,%1$s"
+# Valid: no conflict
+"abc%1$4x,%2$c,%1$u"
+# Invalid: mixing of numbered and unnumbered arguments
+"abc%d%2$x"
+# Valid: numbered argument with constant precision
+"abc%1$.9x"
+# Invalid: mixing of numbered and unnumbered arguments
+"abc%1$.*x"
+# Valid: missing non-final argument
+"abc%2$x%3$s"
+# Valid: permutation
+"abc%2$ddef%1$d"
+# Valid: multiple uses of same argument
+"abc%2$xdef%1$sghi%2$x"
+# Valid: one argument with width
+"abc%2$#*1$g"
+# Valid: one argument with width and precision
+"abc%3$*2$.*1$g"
+# Invalid: zero
+"abc%2$*0$.*1$g"
+EOF
+
+: ${XGETTEXT=xgettext}
+n=0
+while read comment; do
+  read string
+  n=`expr $n + 1`
+  tmpfiles="$tmpfiles f-a-1-$n.in f-a-1-$n.po"
+  cat <<EOF > f-a-1-$n.in
+dcgettext(${string});
+EOF
+  ${XGETTEXT} -L awk -o f-a-1-$n.po f-a-1-$n.in || exit 1
+  test -f f-a-1-$n.po || exit 1
+  fail=
+  if echo "$comment" | grep 'Valid:' > /dev/null; then
+    if grep awk-format f-a-1-$n.po > /dev/null; then
+      :
+    else
+      fail=yes
+    fi
+  else
+    if grep awk-format f-a-1-$n.po > /dev/null; then
+      fail=yes
+    else
+      :
+    fi
+  fi
+  if test -n "$fail"; then
+    echo "Format string recognition error:" 1>&2
+    cat f-a-1-$n.in 1>&2
+    echo "Got:" 1>&2
+    cat f-a-1-$n.po 1>&2
+    exit 1
+  fi
+done < f-a-1.data
+
+rm -fr $tmpfiles
+
+exit 0
diff --git a/tests/format-awk-2 b/tests/format-awk-2
new file mode 100755
index 000000000..3636cd060
--- /dev/null
+++ b/tests/format-awk-2
@@ -0,0 +1,144 @@
+#! /bin/sh
+
+# Test checking of awk format strings.
+
+tmpfiles=""
+trap 'rm -fr $tmpfiles' 1 2 3 15
+
+tmpfiles="$tmpfiles f-a-2.data"
+cat <<\EOF > f-a-2.data
+# Valid: %% doesn't count
+msgid  "abc%%def"
+msgstr "xyz"
+# Invalid: invalid msgstr
+msgid  "abc%%def"
+msgstr "xyz%"
+# Valid: same arguments
+msgid  "abc%s%gdef"
+msgstr "xyz%s%g"
+# Valid: same arguments, with different widths
+msgid  "abc%2sdef"
+msgstr "xyz%3s"
+# Valid: same arguments but in numbered syntax
+msgid  "abc%s%gdef"
+msgstr "xyz%1$s%2$g"
+# Valid: permutation
+msgid  "abc%s%g%cdef"
+msgstr "xyz%3$c%2$g%1$s"
+# Invalid: too few arguments
+msgid  "abc%2$udef%1$s"
+msgstr "xyz%1$s"
+# Invalid: too few arguments
+msgid  "abc%sdef%u"
+msgstr "xyz%s"
+# Invalid: too many arguments
+msgid  "abc%udef"
+msgstr "xyz%uvw%c"
+# Valid: same numbered arguments, with different widths
+msgid  "abc%2$5s%1$4s"
+msgstr "xyz%2$4s%1$5s"
+# Invalid: missing argument
+msgid  "abc%2$sdef%1$u"
+msgstr "xyz%1$u"
+# Invalid: missing argument
+msgid  "abc%1$sdef%2$u"
+msgstr "xyz%2$u"
+# Invalid: added argument
+msgid  "abc%1$udef"
+msgstr "xyz%1$uvw%2$c"
+# Valid: type compatibility
+msgid  "abc%i"
+msgstr "xyz%d"
+# Valid: type compatibility
+msgid  "abc%o"
+msgstr "xyz%u"
+# Valid: type compatibility
+msgid  "abc%u"
+msgstr "xyz%x"
+# Valid: type compatibility
+msgid  "abc%u"
+msgstr "xyz%X"
+# Valid: type compatibility
+msgid  "abc%e"
+msgstr "xyz%E"
+# Valid: type compatibility
+msgid  "abc%e"
+msgstr "xyz%f"
+# Valid: type compatibility
+msgid  "abc%e"
+msgstr "xyz%g"
+# Valid: type compatibility
+msgid  "abc%e"
+msgstr "xyz%G"
+# Invalid: type incompatibility
+msgid  "abc%c"
+msgstr "xyz%s"
+# Invalid: type incompatibility
+msgid  "abc%c"
+msgstr "xyz%i"
+# Invalid: type incompatibility
+msgid  "abc%c"
+msgstr "xyz%o"
+# Invalid: type incompatibility
+msgid  "abc%c"
+msgstr "xyz%e"
+# Invalid: type incompatibility
+msgid  "abc%s"
+msgstr "xyz%i"
+# Invalid: type incompatibility
+msgid  "abc%s"
+msgstr "xyz%o"
+# Invalid: type incompatibility
+msgid  "abc%s"
+msgstr "xyz%e"
+# Invalid: type incompatibility
+msgid  "abc%i"
+msgstr "xyz%o"
+# Invalid: type incompatibility
+msgid  "abc%i"
+msgstr "xyz%e"
+# Invalid: type incompatibility
+msgid  "abc%u"
+msgstr "xyz%e"
+# Invalid: type incompatibility for width
+msgid  "abc%g%*g"
+msgstr "xyz%*g%g"
+EOF
+
+: ${MSGFMT=msgfmt}
+n=0
+while read comment; do
+  read msgid_line
+  read msgstr_line
+  n=`expr $n + 1`
+  tmpfiles="$tmpfiles f-a-2-$n.po f-a-2-$n.mo"
+  cat <<EOF > f-a-2-$n.po
+#, awk-format
+${msgid_line}
+${msgstr_line}
+EOF
+  fail=
+  if echo "$comment" | grep 'Valid:' > /dev/null; then
+    if ${MSGFMT} --check-format -o f-a-2-$n.mo f-a-2-$n.po; then
+      :
+    else
+      fail=yes
+    fi
+  else
+    ${MSGFMT} --check-format -o f-a-2-$n.mo f-a-2-$n.po 2> /dev/null
+    if test $? = 1; then
+      :
+    else
+      fail=yes
+    fi
+  fi
+  if test -n "$fail"; then
+    echo "Format string checking error:" 1>&2
+    cat f-a-2-$n.po 1>&2
+    exit 1
+  fi
+done < f-a-2.data
+
+rm -fr $tmpfiles
+
+exit 0
diff --git a/tests/lang-gawk b/tests/lang-gawk
new file mode 100755
index 000000000..f83357b3c
--- /dev/null
+++ b/tests/lang-gawk
@@ -0,0 +1,105 @@
+#! /bin/sh
+
+# Test of gettext facilities in the GNU awk language.
+# Assumes an fr_FR locale is installed.
+# Assumes the following packages are installed: gawk.
+
+tmpfiles=""
+trap 'rm -fr $tmpfiles' 1 2 3 15
+
+tmpfiles="$tmpfiles prog.awk"
+cat <<\EOF > prog.awk
+BEGIN {
+  TEXTDOMAIN = "prog"
+  bindtextdomain ("./")
+
+  print _"'Your command, please?', asked the waiter."
+
+  printf dcngettext ("a piece of cake", "%d pieces of cake", n) "\n", n
+
+  printf _"%s is replaced by %s." "\n", "FF", "EUR"
+}
+EOF
+
+tmpfiles="$tmpfiles prog.pot"
+: ${XGETTEXT=xgettext}
+${XGETTEXT} -o prog.pot --omit-header --no-location prog.awk
+
+tmpfiles="$tmpfiles prog.ok"
+cat <<EOF > prog.ok
+msgid "'Your command, please?', asked the waiter."
+msgstr ""
+
+#, awk-format
+msgid "a piece of cake"
+msgid_plural "%d pieces of cake"
+msgstr[0] ""
+msgstr[1] ""
+
+#, awk-format
+msgid "%s is replaced by %s."
+msgstr ""
+EOF
+
+: ${DIFF=diff}
+${DIFF} prog.ok prog.pot || exit 1
+
+tmpfiles="$tmpfiles fr.po"
+cat <<\EOF > fr.po
+msgid ""
+msgstr ""
+"Content-Type: text/plain; charset=ISO-8859-1\n"
+"Plural-Forms: nplurals=2; plural=(n > 1);\n"
+
+msgid "'Your command, please?', asked the waiter."
+msgstr "«Votre commande, s'il vous plait», dit le garçon."
+
+# Les gateaux allemands sont les meilleurs du monde.
+#, awk-format
+msgid "a piece of cake"
+msgid_plural "%d pieces of cake"
+msgstr[0] "un morceau de gateau"
+msgstr[1] "%d morceaux de gateau"
+
+# Reverse the arguments.
+#, awk-format
+msgid "%s is replaced by %s."
+msgstr "%2$s remplace %1$s."
+EOF
+
+tmpfiles="$tmpfiles fr.po.new"
+: ${MSGMERGE=msgmerge}
+${MSGMERGE} -q -o fr.po.new fr.po prog.pot
+
+: ${DIFF=diff}
+${DIFF} fr.po fr.po.new || exit 1
+
+tmpfiles="$tmpfiles fr"
+test -d fr || mkdir fr
+test -d fr/LC_MESSAGES || mkdir fr/LC_MESSAGES
+
+: ${MSGFMT=msgfmt}
+${MSGFMT} -o fr/LC_MESSAGES/prog.mo fr.po
+
+tmpfiles="$tmpfiles prog.ok prog.out"
+: ${DIFF=diff}
+cat <<\EOF > prog.ok
+«Votre commande, s'il vous plait», dit le garçon.
+2 morceaux de gateau
+EUR remplace FF.
+EOF
+
+# Test for presence of gawk version 3.1.1 or newer.
+(gawk --version) >/dev/null 2>/dev/null \
+  || { echo "SKIP: lang-awk"; rm -fr $tmpfiles; exit 77; }
+case `gawk --version | sed -e 's/^[^0-9]*//'` in
+  0.* | 1.* | 2.* | 3.0* | 3.1.0*)
+    echo "SKIP: lang-awk"; rm -fr $tmpfiles; exit 77;;
+esac
+
+LANGUAGE= LC_ALL=fr_FR gawk -v n=2 -f prog.awk > prog.out || exit 1
+${DIFF} prog.ok prog.out || exit 1
+
+rm -fr $tmpfiles
+
+exit 0