From: Bruno Haible <bruno@clisp.org>
Date: Sun, 1 Jun 2025 23:23:13 +0000 (+0200)
Subject: xgettext: PHP: Improve heuristic for format strings.
X-Git-Tag: v0.26~129
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=dacdbfec87a01589480fe120db9395b528f2c922;p=thirdparty%2Fgettext.git

xgettext: PHP: Improve heuristic for format strings.

* gettext-tools/src/format-php.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_php): Use it.
* gettext-tools/tests/format-php-3: New file.
* gettext-tools/tests/Makefile.am (TESTS): Add it.
---

diff --git a/gettext-tools/src/format-php.c b/gettext-tools/src/format-php.c
index b43fd5a99..11b55a0d7 100644
--- a/gettext-tools/src/format-php.c
+++ b/gettext-tools/src/format-php.c
@@ -72,6 +72,11 @@ 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 'php-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;
 };
@@ -92,6 +97,7 @@ format_parse (const char *format, bool translated, char *fdi,
 {
   const char *const format_start = format;
   unsigned int directives;
+  unsigned int likely_intentional_directives;
   unsigned int numbered_arg_count;
   struct numbered_arg *numbered;
   unsigned int numbered_allocated;
@@ -99,6 +105,7 @@ format_parse (const char *format, bool translated, char *fdi,
   struct spec *result;
 
   directives = 0;
+  likely_intentional_directives = 0;
   numbered_arg_count = 0;
   numbered = NULL;
   numbered_allocated = 0;
@@ -108,6 +115,8 @@ format_parse (const char *format, bool translated, char *fdi,
     if (*format++ == '%')
       {
         /* A directive.  */
+        bool likely_intentional = true;
+
         FDI_SET (format - 1, FMTDIR_START);
         directives++;
 
@@ -148,7 +157,11 @@ format_parse (const char *format, bool translated, char *fdi,
             for (;;)
               {
                 if (*format == '0' || *format == '-' || *format == ' ')
-                  format++;
+                  {
+                    if (*format == ' ')
+                      likely_intentional = false;
+                    format++;
+                  }
                 else if (*format == '\'')
                   {
                     format++;
@@ -231,6 +244,8 @@ format_parse (const char *format, bool translated, char *fdi,
             numbered_arg_count++;
           }
 
+        if (likely_intentional)
+          likely_intentional_directives++;
         FDI_SET (format, FMTDIR_END);
 
         format++;
@@ -285,6 +300,7 @@ format_parse (const char *format, bool translated, char *fdi,
 
   result = XMALLOC (struct spec);
   result->directives = directives;
+  result->likely_intentional_directives = likely_intentional_directives;
   result->numbered_arg_count = numbered_arg_count;
   result->numbered = numbered;
   return result;
@@ -313,6 +329,14 @@ 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,
@@ -397,7 +421,7 @@ struct formatstring_parser formatstring_php =
   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 355e47e62..edf08b489 100644
--- a/gettext-tools/tests/Makefile.am
+++ b/gettext-tools/tests/Makefile.am
@@ -214,7 +214,7 @@ TESTS = gettext-1 gettext-2 \
 	format-lisp-1 format-lisp-2 \
 	format-lua-1 format-lua-2 \
 	format-modula2-1 format-modula2-2 \
-	format-php-1 format-php-2 \
+	format-php-1 format-php-2 format-php-3 \
 	format-python-1 format-python-2 \
 	format-python-brace-1 format-python-brace-2 \
 	format-pascal-1 format-pascal-2 \
diff --git a/gettext-tools/tests/format-php-3 b/gettext-tools/tests/format-php-3
new file mode 100755
index 000000000..969665e17
--- /dev/null
+++ b/gettext-tools/tests/format-php-3
@@ -0,0 +1,49 @@
+#! /bin/sh
+. "${srcdir=.}/init.sh"; path_prepend_ . ../src
+
+# Test heuristic recognition of PHP format strings like "100% complete".
+
+cat <<\EOF > f-ph-3.php
+<?php
+gettext("Test 1a is 100% complete")
+printf(gettext("Test 1b is 100% complete"), 120)
+gettext("Test 2a from 0% complete to 100% complete")
+printf(gettext("Test 2b from 0% complete to 100% complete"), 120, 121)
+gettext("Test 3a of %s is 100% complete")
+printf(gettext("Test 3b of %s is 100% complete"), "foo", 120)
+?>
+EOF
+
+: ${XGETTEXT=xgettext}
+${XGETTEXT} --omit-header --no-location -d f-ph-3.tmp f-ph-3.php || Exit 1
+LC_ALL=C tr -d '\r' < f-ph-3.tmp.po > f-ph-3.po || Exit 1
+
+cat <<\EOF > f-ph-3.ok
+msgid "Test 1a is 100% complete"
+msgstr ""
+
+#, php-format
+msgid "Test 1b is 100% complete"
+msgstr ""
+
+msgid "Test 2a from 0% complete to 100% complete"
+msgstr ""
+
+#, php-format
+msgid "Test 2b from 0% complete to 100% complete"
+msgstr ""
+
+#, php-format
+msgid "Test 3a of %s is 100% complete"
+msgstr ""
+
+#, php-format
+msgid "Test 3b of %s is 100% complete"
+msgstr ""
+EOF
+
+: ${DIFF=diff}
+${DIFF} f-ph-3.ok f-ph-3.po
+result=$?
+
+exit $result