[3.5] bpo-30375: Correct the stacklevel of regex compiling warnings. (GH-1595) (#1605)

Warnings emitted when compile a regular expression now always point
to the line in the user code.  Previously they could point into inners
of the re module if emitted from inside of groups or conditionals..
(cherry picked from commit c7ac728)

Author: serhiy-storchaka

Lib/sre_parse.py

@@ -309,7 +309,7 @@ def isname(name):
             return False
     return True
 
-def _class_escape(source, escape):
+def _class_escape(source, escape, nested):
     # handle escape code inside character class
     code = ESCAPES.get(escape)
     if code:
@@ -353,13 +353,13 @@ def _class_escape(source, escape):
             if c in ASCIILETTERS:
                 import warnings
                 warnings.warn('bad escape %s' % escape,
-                              DeprecationWarning, stacklevel=8)
+                              DeprecationWarning, stacklevel=nested + 6)
             return LITERAL, ord(escape[1])
     except ValueError:
         pass
     raise source.error("bad escape %s" % escape, len(escape))
 
-def _escape(source, escape, state):
+def _escape(source, escape, state, nested):
     # handle escape code in expression
     code = CATEGORIES.get(escape)
     if code:
@@ -420,21 +420,21 @@ def _escape(source, escape, state):
             if c in ASCIILETTERS:
                 import warnings
                 warnings.warn('bad escape %s' % escape,
-                              DeprecationWarning, stacklevel=8)
+                              DeprecationWarning, stacklevel=nested + 6)
             return LITERAL, ord(escape[1])
     except ValueError:
         pass
     raise source.error("bad escape %s" % escape, len(escape))
 
-def _parse_sub(source, state, nested=True):
+def _parse_sub(source, state, nested):
     # parse an alternation: a|b|c
 
     items = []
     itemsappend = items.append
     sourcematch = source.match
     start = source.tell()
     while True:
-        itemsappend(_parse(source, state))
+        itemsappend(_parse(source, state, nested + 1))
         if not sourcematch("|"):
             break
 
@@ -476,10 +476,10 @@ def _parse_sub(source, state, nested=True):
     subpattern.append((BRANCH, (None, items)))
     return subpattern
 
-def _parse_sub_cond(source, state, condgroup):
-    item_yes = _parse(source, state)
+def _parse_sub_cond(source, state, condgroup, nested):
+    item_yes = _parse(source, state, nested + 1)
     if source.match("|"):
-        item_no = _parse(source, state)
+        item_no = _parse(source, state, nested + 1)
         if source.next == "|":
             raise source.error("conditional backref with more than two branches")
     else:
@@ -488,7 +488,7 @@ def _parse_sub_cond(source, state, condgroup):
     subpattern.append((GROUPREF_EXISTS, (condgroup, item_yes, item_no)))
     return subpattern
 
-def _parse(source, state):
+def _parse(source, state, nested):
     # parse a simple pattern
     subpattern = SubPattern(state)
 
@@ -521,7 +521,7 @@ def _parse(source, state):
                 continue
 
         if this[0] == "\\":
-            code = _escape(source, this, state)
+            code = _escape(source, this, state, nested + 1)
             subpatternappend(code)
 
         elif this not in SPECIAL_CHARS:
@@ -546,7 +546,7 @@ def _parse(source, state):
                 if this == "]" and set != start:
                     break
                 elif this[0] == "\\":
-                    code1 = _class_escape(source, this)
+                    code1 = _class_escape(source, this, nested + 1)
                 else:
                     code1 = LITERAL, _ord(this)
                 if sourcematch("-"):
@@ -562,7 +562,7 @@ def _parse(source, state):
                         setappend((LITERAL, _ord("-")))
                         break
                     if that[0] == "\\":
-                        code2 = _class_escape(source, that)
+                        code2 = _class_escape(source, that, nested + 1)
                     else:
                         code2 = LITERAL, _ord(that)
                     if code1[0] != LITERAL or code2[0] != LITERAL:
@@ -713,7 +713,7 @@ def _parse(source, state):
                         lookbehindgroups = state.lookbehindgroups
                         if lookbehindgroups is None:
                             state.lookbehindgroups = state.groups
-                    p = _parse_sub(source, state)
+                    p = _parse_sub(source, state, nested + 1)
                     if dir < 0:
                         if lookbehindgroups is None:
                             state.lookbehindgroups = None
@@ -773,9 +773,9 @@ def _parse(source, state):
                 except error as err:
                     raise source.error(err.msg, len(name) + 1) from None
             if condgroup:
-                p = _parse_sub_cond(source, state, condgroup)
+                p = _parse_sub_cond(source, state, condgroup, nested + 1)
             else:
-                p = _parse_sub(source, state)
+                p = _parse_sub(source, state, nested + 1)
             if not source.match(")"):
                 raise source.error("missing ), unterminated subpattern",
                                    source.tell() - start)

Lib/test/test_re.py

@@ -638,14 +638,18 @@ def test_other_escapes(self):
         re.purge()  # for warnings
         for c in 'ceghijklmopqyzCEFGHIJKLMNOPQRTVXY':
             with self.subTest(c):
-                with self.assertWarns(DeprecationWarning):
+                with self.assertWarns(DeprecationWarning) as warns:
                     self.assertEqual(re.fullmatch('\\%c' % c, c).group(), c)
                     self.assertIsNone(re.match('\\%c' % c, 'a'))
+                self.assertRegex(str(warns.warnings[0].message), 'bad escape')
+                self.assertEqual(warns.warnings[0].filename, __file__)
         for c in 'ceghijklmopqyzABCEFGHIJKLMNOPQRTVXYZ':
             with self.subTest(c):
-                with self.assertWarns(DeprecationWarning):
+                with self.assertWarns(DeprecationWarning) as warns:
                     self.assertEqual(re.fullmatch('[\\%c]' % c, c).group(), c)
                     self.assertIsNone(re.match('[\\%c]' % c, 'a'))
+                self.assertRegex(str(warns.warnings[0].message), 'bad escape')
+                self.assertEqual(warns.warnings[0].filename, __file__)
 
     def test_string_boundaries(self):
         # See http://bugs.python.org/issue10713

Misc/NEWS

@@ -49,6 +49,10 @@ Extension Modules
 Library
 -------
 
+- bpo-30375: Warnings emitted when compile a regular expression now always
+  point to the line in the user code.  Previously they could point into inners
+  of the re module if emitted from inside of groups or conditionals.
+
 - bpo-30048: Fixed ``Task.cancel()`` can be ignored when the task is
   running coroutine and the coroutine returned without any more ``await``.