[3.11] pythongh-95731: Fix module docstring extraction in pygettext (p…

…ythonGH-95732) (python#98281)

pythongh-95731: Fix module docstring extraction in pygettext (pythonGH-95732)
(cherry picked from commit 120b4ab)

Co-authored-by: Jakub Kuczys <me@jacken.men>

Lib/test/test_tools/test_i18n.py

@@ -155,6 +155,26 @@ class C:
         '''))
         self.assertFalse([msgid for msgid in msgids if 'doc' in msgid])
 
+    def test_moduledocstring(self):
+        for doc in ('"""doc"""', "r'''doc'''", "R'doc'", 'u"doc"'):
+            with self.subTest(doc):
+                msgids = self.extract_docstrings_from_str(dedent('''\
+                %s
+                ''' % doc))
+                self.assertIn('doc', msgids)
+
+    def test_moduledocstring_bytes(self):
+        msgids = self.extract_docstrings_from_str(dedent('''\
+        b"""doc"""
+        '''))
+        self.assertFalse([msgid for msgid in msgids if 'doc' in msgid])
+
+    def test_moduledocstring_fstring(self):
+        msgids = self.extract_docstrings_from_str(dedent('''\
+        f"""doc"""
+        '''))
+        self.assertFalse([msgid for msgid in msgids if 'doc' in msgid])
+
     def test_msgid(self):
         msgids = self.extract_docstrings_from_str(
                 '''_("""doc""" r'str' u"ing")''')

Misc/NEWS.d/next/Tools-Demos/2022-08-05-23-25-59.gh-issue-95731.N2KohU.rst

@@ -0,0 +1 @@
+Fix handling of module docstrings in :file:`Tools/i18n/pygettext.py`.

Tools/i18n/pygettext.py

@@ -335,9 +335,10 @@ def __waiting(self, ttype, tstring, lineno):
                 if ttype == tokenize.STRING and is_literal_string(tstring):
                     self.__addentry(safe_eval(tstring), lineno, isdocstring=1)
                     self.__freshmodule = 0
-                elif ttype not in (tokenize.COMMENT, tokenize.NL):
-                    self.__freshmodule = 0
-                return
+                    return
+                if ttype in (tokenize.COMMENT, tokenize.NL, tokenize.ENCODING):
+                    return
+                self.__freshmodule = 0
             # class or func/method docstring?
             if ttype == tokenize.NAME and tstring in ('class', 'def'):
                 self.__state = self.__suiteseen