From: Bruno Haible <bruno@clisp.org>
Date: Sun, 1 Jun 2025 17:53:09 +0000 (+0200)
Subject: xgettext: awk: Improve heuristic for format strings.
X-Git-Tag: v0.26~138
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=b330ce78045430003b353b0d4f71327c251c6b75;p=thirdparty%2Fgettext.git

xgettext: awk: Improve heuristic for format strings.

* gettext-tools/src/format-awk.c (struct spec): Add field
'likely_intentional_directives'.
(format_parse): Set it to the number of directives that don't contain a space.
(format_is_unlikely_intentional): New function.
(formatstring_awk): Use it.
* gettext-tools/tests/format-awk-3: New file.
* gettext-tools/tests/Makefile.am (TESTS): Add it.
---

diff --git a/gettext-tools/src/format-awk.c b/gettext-tools/src/format-awk.c
index bdacf3eea..ef2036bc2 100644
--- a/gettext-tools/src/format-awk.c
+++ b/gettext-tools/src/format-awk.c
@@ -71,6 +71,11 @@ struct numbered_arg
 struct spec
 {
   unsigned int directives;
+  /* We consider a directive as "likely intentional" if it does not contain a
+     space.  This prevents xgettext from flagging strings like "100% complete"
+     as 'awk-format' if they don't occur in a context that requires a format
+     string.  */
+  unsigned int likely_intentional_directives;
   unsigned int numbered_arg_count;
   struct numbered_arg *numbered;
 };
@@ -96,6 +101,7 @@ format_parse (const char *format, bool translated, char *fdi,
   struct spec *result;
 
   spec.directives = 0;
+  spec.likely_intentional_directives = 0;
   spec.numbered_arg_count = 0;
   spec.numbered = NULL;
   numbered_allocated = 0;
@@ -108,6 +114,7 @@ format_parse (const char *format, bool translated, char *fdi,
         /* A directive.  */
         unsigned int number = 0;
         enum format_arg_type type;
+        bool likely_intentional = true;
 
         FDI_SET (format - 1, FMTDIR_START);
         spec.directives++;
@@ -140,7 +147,11 @@ format_parse (const char *format, bool translated, char *fdi,
         /* Parse flags.  */
         while (*format == ' ' || *format == '+' || *format == '-'
                || *format == '#' || *format == '0')
-          format++;
+          {
+            if (*format == ' ')
+              likely_intentional = false;
+            format++;
+          }
 
         /* Parse width.  */
         if (*format == '*')
@@ -390,6 +401,8 @@ format_parse (const char *format, bool translated, char *fdi,
               }
           }
 
+        if (likely_intentional)
+          spec.likely_intentional_directives++;
         FDI_SET (format, FMTDIR_END);
 
         format++;
@@ -473,6 +486,14 @@ format_get_number_of_directives (void *descr)
   return spec->directives;
 }
 
+static bool
+format_is_unlikely_intentional (void *descr)
+{
+  struct spec *spec = (struct spec *) descr;
+
+  return spec->likely_intentional_directives == 0;
+}
+
 static bool
 format_check (void *msgid_descr, void *msgstr_descr, bool equality,
               formatstring_error_logger_t error_logger, void *error_logger_data,
@@ -557,7 +578,7 @@ struct formatstring_parser formatstring_awk =
   format_parse,
   format_free,
   format_get_number_of_directives,
-  NULL,
+  format_is_unlikely_intentional,
   format_check
 };
 
diff --git a/gettext-tools/tests/Makefile.am b/gettext-tools/tests/Makefile.am
index 833c504c1..7801c9631 100644
--- a/gettext-tools/tests/Makefile.am
+++ b/gettext-tools/tests/Makefile.am
@@ -195,7 +195,7 @@ TESTS = gettext-1 gettext-2 \
 	xgettext-ycp-1 xgettext-ycp-2 xgettext-ycp-3 xgettext-ycp-4 \
 	xgettext-ycp-stackovfl-1 xgettext-ycp-stackovfl-2 \
 	xgettext-ycp-stackovfl-3 xgettext-ycp-stackovfl-4 \
-	format-awk-1 format-awk-2 \
+	format-awk-1 format-awk-2 format-awk-3 \
 	format-boost-1 format-boost-2 \
 	format-c-1 format-c-2 format-c-3 format-c-4 format-c-5 \
 	format-c++-brace-1 format-c++-brace-2 \
diff --git a/gettext-tools/tests/format-awk-3 b/gettext-tools/tests/format-awk-3
new file mode 100755
index 000000000..daa8246bc
--- /dev/null
+++ b/gettext-tools/tests/format-awk-3
@@ -0,0 +1,47 @@
+#! /bin/sh
+. "${srcdir=.}/init.sh"; path_prepend_ . ../src
+
+# Test heuristic recognition of awk format strings like "100% complete".
+
+cat <<\EOF > f-a-3.awk
+dcgettext("Test 1a is 100% complete");
+printf dcgettext("Test 1b is 100% complete"), 120;
+dcgettext("Test 2a from 0% complete to 100% complete");
+printf dcgettext("Test 2b from 0% complete to 100% complete"), 120, 121;
+dcgettext("Test 3a of %s is 100% complete");
+printf dcgettext("Test 3b of %s is 100% complete"), "foo", 120;
+EOF
+
+: ${XGETTEXT=xgettext}
+${XGETTEXT} --omit-header --no-location -d f-a-3.tmp f-a-3.awk || Exit 1
+LC_ALL=C tr -d '\r' < f-a-3.tmp.po > f-a-3.po || Exit 1
+
+cat <<\EOF > f-a-3.ok
+msgid "Test 1a is 100% complete"
+msgstr ""
+
+#, awk-format
+msgid "Test 1b is 100% complete"
+msgstr ""
+
+msgid "Test 2a from 0% complete to 100% complete"
+msgstr ""
+
+#, awk-format
+msgid "Test 2b from 0% complete to 100% complete"
+msgstr ""
+
+#, awk-format
+msgid "Test 3a of %s is 100% complete"
+msgstr ""
+
+#, awk-format
+msgid "Test 3b of %s is 100% complete"
+msgstr ""
+EOF
+
+: ${DIFF=diff}
+${DIFF} f-a-3.ok f-a-3.po
+result=$?
+
+exit $result