Issue #10182: The re module doesn't truncate indices to 32 bits anymore.

pitrou · pitrou · commit 9a2b26748b73 · 2012-12-02T12:54:28.000+01:00
Patch by Serhiy Storchaka.
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
@@ -1,4 +1,4 @@
-from test.support import verbose, run_unittest, gc_collect
+from test.support import verbose, run_unittest, gc_collect, bigmemtest, _2G
 import io
 import re
 from re import Scanner
@@ -949,6 +949,21 @@ def test_compile(self):
         # Test behaviour when not given a string or pattern as parameter
         self.assertRaises(TypeError, re.compile, 0)
 
+    # The huge memuse is because of re.sub() using a list and a join()
+    # to create the replacement result.
+    @bigmemtest(size=_2G, memuse=20)
+    def test_large(self, size):
+        # Issue #10182: indices were 32-bit-truncated.
+        s = 'a' * size
+        m = re.search('$', s)
+        self.assertIsNotNone(m)
+        self.assertEqual(m.start(), size)
+        self.assertEqual(m.end(), size)
+        r, n = re.subn('', '', s)
+        self.assertEqual(r, s)
+        self.assertEqual(n, size + 1)
+
+
 def run_re_tests():
     from test.re_tests import tests, SUCCEED, FAIL, SYNTAX_ERROR
     if verbose:
diff --git a/Misc/NEWS b/Misc/NEWS
@@ -98,6 +98,9 @@ Core and Builtins
 Library
 -------
 
+- Issue #10182: The re module doesn't truncate indices to 32 bits anymore.
+  Patch by Serhiy Storchaka.
+
 - Issue #16573: In 2to3, treat enumerate() like a consuming call, so superfluous
   list() calls aren't added to filter(), map(), and zip() which are directly
   passed enumerate().
diff --git a/Modules/_sre.c b/Modules/_sre.c
@@ -1618,7 +1618,7 @@ sre_literal_template(int charsize, char* ptr, Py_ssize_t len)
 static PyObject *
 sre_codesize(PyObject* self, PyObject *unused)
 {
-    return Py_BuildValue("l", sizeof(SRE_CODE));
+    return PyLong_FromSize_t(sizeof(SRE_CODE));
 }
 
 static PyObject *
@@ -2435,7 +2435,7 @@ pattern_subx(PatternObject* self, PyObject* ptemplate, PyObject* string,
         return NULL;
 
     if (subn)
-        return Py_BuildValue("Ni", item, n);
+        return Py_BuildValue("Nn", item, n);
 
     return item;
 
@@ -3387,7 +3387,7 @@ match_start(MatchObject* self, PyObject* args)
     }
 
     /* mark is -1 if group is undefined */
-    return Py_BuildValue("i", self->mark[index*2]);
+    return PyLong_FromSsize_t(self->mark[index*2]);
 }
 
 static PyObject*
@@ -3410,7 +3410,7 @@ match_end(MatchObject* self, PyObject* args)
     }
 
     /* mark is -1 if group is undefined */
-    return Py_BuildValue("i", self->mark[index*2+1]);
+    return PyLong_FromSsize_t(self->mark[index*2+1]);
 }
 
 LOCAL(PyObject*)
@@ -3560,7 +3560,7 @@ static PyObject *
 match_lastindex_get(MatchObject *self)
 {
     if (self->lastindex >= 0)
-        return Py_BuildValue("i", self->lastindex);
+        return PyLong_FromSsize_t(self->lastindex);
     Py_INCREF(Py_None);
     return Py_None;
 }

Original file line number	Diff line number	Diff line change
`@@ -1618,7 +1618,7 @@ sre_literal_template(int charsize, char* ptr, Py_ssize_t len)`
`1618`	`1618`	`static PyObject *`
`1619`	`1619`	`sre_codesize(PyObject* self, PyObject *unused)`
`1620`	`1620`	`{`
`1621`		`- return Py_BuildValue("l", sizeof(SRE_CODE));`
	`1621`	`+ return PyLong_FromSize_t(sizeof(SRE_CODE));`
`1622`	`1622`	`}`
`1623`	`1623`
`1624`	`1624`	`static PyObject *`
`@@ -2435,7 +2435,7 @@ pattern_subx(PatternObject* self, PyObject* ptemplate, PyObject* string,`
`2435`	`2435`	`return NULL;`
`2436`	`2436`
`2437`	`2437`	`if (subn)`
`2438`		`- return Py_BuildValue("Ni", item, n);`
	`2438`	`+ return Py_BuildValue("Nn", item, n);`
`2439`	`2439`
`2440`	`2440`	`return item;`
`2441`	`2441`
`@@ -3387,7 +3387,7 @@ match_start(MatchObject* self, PyObject* args)`
`3387`	`3387`	`}`
`3388`	`3388`
`3389`	`3389`	`/* mark is -1 if group is undefined */`
`3390`		`- return Py_BuildValue("i", self->mark[index*2]);`
	`3390`	`+ return PyLong_FromSsize_t(self->mark[index*2]);`
`3391`	`3391`	`}`
`3392`	`3392`
`3393`	`3393`	`static PyObject*`
`@@ -3410,7 +3410,7 @@ match_end(MatchObject* self, PyObject* args)`
`3410`	`3410`	`}`
`3411`	`3411`
`3412`	`3412`	`/* mark is -1 if group is undefined */`
`3413`		`- return Py_BuildValue("i", self->mark[index*2+1]);`
	`3413`	`+ return PyLong_FromSsize_t(self->mark[index*2+1]);`
`3414`	`3414`	`}`
`3415`	`3415`
`3416`	`3416`	`LOCAL(PyObject*)`
`@@ -3560,7 +3560,7 @@ static PyObject *`
`3560`	`3560`	`match_lastindex_get(MatchObject *self)`
`3561`	`3561`	`{`
`3562`	`3562`	`if (self->lastindex >= 0)`
`3563`		`- return Py_BuildValue("i", self->lastindex);`
	`3563`	`+ return PyLong_FromSsize_t(self->lastindex);`
`3564`	`3564`	`Py_INCREF(Py_None);`
`3565`	`3565`	`return Py_None;`
`3566`	`3566`	`}`