Issue #1882: when compiling code from a string, encoding cookies in the
second line of code were not always recognized correctly.
diff --git a/Parser/tokenizer.c b/Parser/tokenizer.c
index 0aaec19..bbfbe7d 100644
--- a/Parser/tokenizer.c
+++ b/Parser/tokenizer.c
@@ -586,6 +586,7 @@
{
PyObject* utf8 = NULL;
const char *s;
+ char *newl[2] = {NULL, NULL};
int lineno = 0;
tok->enc = NULL;
tok->str = str;
@@ -604,13 +605,23 @@
for (s = str;; s++) {
if (*s == '\0') break;
else if (*s == '\n') {
+ newl[lineno] = s;
lineno++;
if (lineno == 2) break;
}
}
tok->enc = NULL;
- if (!check_coding_spec(str, s - str, tok, buf_setreadl))
- return error_ret(tok);
+ /* need to check line 1 and 2 separately since check_coding_spec
+ assumes a single line as input */
+ if (newl[0]) {
+ if (!check_coding_spec(str, newl[0] - str, tok, buf_setreadl))
+ return error_ret(tok);
+ if (tok->enc == NULL && newl[1]) {
+ if (!check_coding_spec(newl[0]+1, newl[1] - newl[0],
+ tok, buf_setreadl))
+ return error_ret(tok);
+ }
+ }
#ifdef Py_USING_UNICODE
if (tok->enc != NULL) {
assert(utf8 == NULL);