bpo-32827: Fix usage of _PyUnicodeWriter_Prepare() in decoding errors handler. (GH-5636)

miss-islington · serhiy-storchaka · web-flow · commit 310b424c476e · 2018-02-13T01:32:46.000-08:00
(cherry picked from commit b7e2d67) Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
@@ -4500,7 +4500,7 @@ unicode_decode_call_errorhandler_writer(
     }
     if (need_to_grow) {
         writer->overallocate = 1;
-        if (_PyUnicodeWriter_Prepare(writer, writer->min_length,
+        if (_PyUnicodeWriter_Prepare(writer, writer->min_length - writer->pos,
                             PyUnicode_MAX_CHAR_VALUE(repunicode)) == -1)
             goto onError;
     }
@@ -6176,9 +6176,7 @@ _PyUnicode_DecodeUnicodeEscape(const char *s,
                 &writer)) {
             goto onError;
         }
-        if (_PyUnicodeWriter_Prepare(&writer, writer.min_length, 127) < 0) {
-            goto onError;
-        }
+        assert(end - s <= writer.size - writer.pos);
 
 #undef WRITE_ASCII_CHAR
 #undef WRITE_CHAR
@@ -6455,9 +6453,7 @@ PyUnicode_DecodeRawUnicodeEscape(const char *s,
                 &writer)) {
             goto onError;
         }
-        if (_PyUnicodeWriter_Prepare(&writer, writer.min_length, 127) < 0) {
-            goto onError;
-        }
+        assert(end - s <= writer.size - writer.pos);
 
 #undef WRITE_CHAR
     }

Original file line number	Diff line number	Diff line change
`@@ -4500,7 +4500,7 @@ unicode_decode_call_errorhandler_writer(`
`4500`	`4500`	`}`
`4501`	`4501`	`if (need_to_grow) {`
`4502`	`4502`	`writer->overallocate = 1;`
`4503`		`- if (_PyUnicodeWriter_Prepare(writer, writer->min_length,`
	`4503`	`+ if (_PyUnicodeWriter_Prepare(writer, writer->min_length - writer->pos,`
`4504`	`4504`	`PyUnicode_MAX_CHAR_VALUE(repunicode)) == -1)`
`4505`	`4505`	`goto onError;`
`4506`	`4506`	`}`
`@@ -6176,9 +6176,7 @@ _PyUnicode_DecodeUnicodeEscape(const char *s,`
`6176`	`6176`	`&writer)) {`
`6177`	`6177`	`goto onError;`
`6178`	`6178`	`}`
`6179`		`- if (_PyUnicodeWriter_Prepare(&writer, writer.min_length, 127) < 0) {`
`6180`		`- goto onError;`
`6181`		`- }`
	`6179`	`+ assert(end - s <= writer.size - writer.pos);`
`6182`	`6180`
`6183`	`6181`	`#undef WRITE_ASCII_CHAR`
`6184`	`6182`	`#undef WRITE_CHAR`
`@@ -6455,9 +6453,7 @@ PyUnicode_DecodeRawUnicodeEscape(const char *s,`
`6455`	`6453`	`&writer)) {`
`6456`	`6454`	`goto onError;`
`6457`	`6455`	`}`
`6458`		`- if (_PyUnicodeWriter_Prepare(&writer, writer.min_length, 127) < 0) {`
`6459`		`- goto onError;`
`6460`		`- }`
	`6456`	`+ assert(end - s <= writer.size - writer.pos);`
`6461`	`6457`
`6462`	`6458`	`#undef WRITE_CHAR`
`6463`	`6459`	`}`