From: Bruno Haible <bruno@clisp.org>
Date: Tue, 20 May 2025 17:35:20 +0000 (+0200)
Subject: xgettext: JavaScript: Parse RegExp literals according to ECMA-262 15th edition.
X-Git-Tag: v0.26~147
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=2fb3d868b278e4014cef10c1d97ec1f85806e530;p=thirdparty%2Fgettext.git

xgettext: JavaScript: Parse RegExp literals according to ECMA-262 15th edition.

Reported by Christian Wasserthal <aperion@fsfe.org> in
<https://lists.gnu.org/archive/html/bug-gettext/2025-05/msg00038.html>.

* gettext-tools/src/x-javascript.c (phase5_scan_regexp): Report an error when
end-of-line is reached. Parse character classes. Recognize more modifier flags.
* gettext-tools/tests/xgettext-javascript-2: Add two more test cases.
* NEWS: Mention the change.
---

diff --git a/NEWS b/NEWS
index d2a87986d..d44410482 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,10 @@
+Version 0.26 - July 2025
+
+# Programming languages support:
+  * JavaScript:
+    - xgettext now parses regular expressions with character classes
+      correctly.
+
 Version 0.25 - May 2025
 
 # Programming languages support:
diff --git a/gettext-tools/src/x-javascript.c b/gettext-tools/src/x-javascript.c
index d2b17335c..f77e01f01 100644
--- a/gettext-tools/src/x-javascript.c
+++ b/gettext-tools/src/x-javascript.c
@@ -1075,37 +1075,79 @@ is_after_expression (void)
     }
 }
 
+/* Scans a regular expression literal.
+   The leading '/' has already been consumed.
+   See ECMA-262 15th edition sections
+   - § 12.9.5 Regular Expression Literals
+   - § 22.2.3.3 RegExpInitialize  */
 static void
 phase5_scan_regexp (void)
 {
+  bool at_start;
   int c;
 
   /* Scan for end of RegExp literal ('/').  */
-  for (;;)
+  for (at_start = true; ; at_start = false)
     {
       /* Must use phase2 as there can't be comments.  */
       c = phase2_getc ();
+      if (c == UEOF || c == '\n' || c == '\r' || c == 0x2028 || c == 0x2029)
+        goto unterminated;
+      if (at_start && c == '*')
+        {
+          if_error (IF_SEVERITY_WARNING,
+                    logical_file_name, line_number, (size_t)(-1), false,
+                    _("invalid RegExp literal"));
+          return;
+        }
       if (c == '/')
         break;
       if (c == '\\')
         {
           c = phase2_getc ();
-          if (c != UEOF)
-            continue;
+          if (c == UEOF || c == '\n' || c == '\r' || c == 0x2028 || c == 0x2029)
+            goto unterminated;
         }
-      if (c == UEOF)
+      else if (c == '[')
         {
-          if_error (IF_SEVERITY_WARNING,
-                    logical_file_name, line_number, (size_t)(-1), false,
-                    _("RegExp literal terminated too early"));
-          return;
+          for (;;)
+            {
+              c = phase2_getc ();
+              if (c == UEOF
+                  || c == '\n' || c == '\r' || c == 0x2028 || c == 0x2029)
+                goto unterminated_in_class;
+              if (c == ']')
+                break;
+              if (c == '\\')
+                {
+                  c = phase2_getc ();
+                  if (c == UEOF
+                      || c == '\n' || c == '\r' || c == 0x2028 || c == 0x2029)
+                    goto unterminated_in_class;
+                }
+            }
         }
     }
 
-  /* Scan for modifier flags (ECMA-262 5th section 15.10.4.1).  */
+  /* Scan for modifier flags (ECMA-262 15th edition § 22.2.3.3).  */
   c = phase2_getc ();
-  if (!(c == 'g' || c == 'i' || c == 'm'))
+  if (!(c == 'd' || c == 'g' || c == 'i' || c == 'm' || c == 's'
+        || c == 'u' || c == 'v' || c == 'y'))
     phase2_ungetc (c);
+
+  return;
+
+ unterminated:
+  if_error (IF_SEVERITY_WARNING,
+            logical_file_name, line_number, (size_t)(-1), false,
+            _("RegExp literal terminated too early"));
+  return;
+
+ unterminated_in_class:
+  if_error (IF_SEVERITY_WARNING,
+            logical_file_name, line_number, (size_t)(-1), false,
+            _("RegExp literal lacks a ']' to match the '['"));
+  return;
 }
 
 /* Various syntactic constructs can be nested:
diff --git a/gettext-tools/tests/xgettext-javascript-2 b/gettext-tools/tests/xgettext-javascript-2
index d3d642814..806c74642 100755
--- a/gettext-tools/tests/xgettext-javascript-2
+++ b/gettext-tools/tests/xgettext-javascript-2
@@ -28,6 +28,8 @@ else
 var s7 = /a\/\f\r\n\t\v\0\b\s\S\w\W\d\D\b\Bb/.test(_("RegExp test string #14"));
 var s8 = /(?=(a+))a*b\1/.exec(_("RegExp test string #15"));
 var s9 = /_("a+")/.exec(_("RegExp test string #16"));
+var s10 = /[abc/def]/.exec(_("RegExp test string #17"));
+var s11 = /[abc/`def]/.exec(_("RegExp test string #18"));
 EOF
 
 : ${XGETTEXT=xgettext}
@@ -101,6 +103,12 @@ msgstr ""
 
 msgid "RegExp test string #16"
 msgstr ""
+
+msgid "RegExp test string #17"
+msgstr ""
+
+msgid "RegExp test string #18"
+msgstr ""
 EOF
 
 : ${DIFF=diff}