diff options
author | Victor Stinner <victor.stinner@haypocalc.com> | 2011-04-04 22:39:01 (GMT) |
---|---|---|
committer | Victor Stinner <victor.stinner@haypocalc.com> | 2011-04-04 22:39:01 (GMT) |
commit | 7f2fee36401f7b987a368fe043637b3ae7116600 (patch) | |
tree | d21cec17c440b8255cf9bc4c79269d6bfb3253f0 /Parser/tokenizer.c | |
parent | 9bdb43e43f9f5d80699e297c2f73c106179b33d5 (diff) | |
download | cpython-7f2fee36401f7b987a368fe043637b3ae7116600.zip cpython-7f2fee36401f7b987a368fe043637b3ae7116600.tar.gz cpython-7f2fee36401f7b987a368fe043637b3ae7116600.tar.bz2 |
Issue #10785: Store the filename as Unicode in the Python parser.
Diffstat (limited to 'Parser/tokenizer.c')
-rw-r--r-- | Parser/tokenizer.c | 35 |
1 files changed, 22 insertions, 13 deletions
diff --git a/Parser/tokenizer.c b/Parser/tokenizer.c index 3f6be2f..5edd958 100644 --- a/Parser/tokenizer.c +++ b/Parser/tokenizer.c @@ -128,7 +128,6 @@ tok_new(void) tok->prompt = tok->nextprompt = NULL; tok->lineno = 0; tok->level = 0; - tok->filename = NULL; tok->altwarning = 1; tok->alterror = 1; tok->alttabsize = 1; @@ -140,6 +139,7 @@ tok_new(void) tok->encoding = NULL; tok->cont_line = 0; #ifndef PGEN + tok->filename = NULL; tok->decoding_readline = NULL; tok->decoding_buffer = NULL; #endif @@ -545,7 +545,6 @@ decoding_fgets(char *s, int size, struct tok_state *tok) { char *line = NULL; int badchar = 0; - PyObject *filename; for (;;) { if (tok->decoding_state == STATE_NORMAL) { /* We already have a codec associated with @@ -586,16 +585,12 @@ decoding_fgets(char *s, int size, struct tok_state *tok) if (badchar) { /* Need to add 1 to the line number, since this line has not been counted, yet. */ - filename = PyUnicode_DecodeFSDefault(tok->filename); - if (filename != NULL) { - PyErr_Format(PyExc_SyntaxError, - "Non-UTF-8 code starting with '\\x%.2x' " - "in file %U on line %i, " - "but no encoding declared; " - "see http://python.org/dev/peps/pep-0263/ for details", - badchar, filename, tok->lineno + 1); - Py_DECREF(filename); - } + PyErr_Format(PyExc_SyntaxError, + "Non-UTF-8 code starting with '\\x%.2x' " + "in file %U on line %i, " + "but no encoding declared; " + "see http://python.org/dev/peps/pep-0263/ for details", + badchar, tok->filename, tok->lineno + 1); return error_ret(tok); } #endif @@ -853,6 +848,7 @@ PyTokenizer_Free(struct tok_state *tok) #ifndef PGEN Py_XDECREF(tok->decoding_readline); Py_XDECREF(tok->decoding_buffer); + Py_XDECREF(tok->filename); #endif if (tok->fp != NULL && tok->buf != NULL) PyMem_FREE(tok->buf); @@ -1247,8 +1243,13 @@ indenterror(struct tok_state *tok) return 1; } if (tok->altwarning) { - PySys_WriteStderr("%s: inconsistent use of tabs and spaces " +#ifdef PGEN + PySys_WriteStderr("inconsistent use of tabs and spaces " + "in indentation\n"); +#else + PySys_FormatStderr("%U: inconsistent use of tabs and spaces " "in indentation\n", tok->filename); +#endif tok->altwarning = 0; } return 0; @@ -1718,6 +1719,11 @@ PyTokenizer_FindEncoding(int fd) fclose(fp); return NULL; } +#ifndef PGEN + tok->filename = PyUnicode_FromString("<string>"); + if (tok->filename == NULL) + goto error; +#endif while (tok->lineno < 2 && tok->done == E_OK) { PyTokenizer_Get(tok, &p_start, &p_end); } @@ -1727,6 +1733,9 @@ PyTokenizer_FindEncoding(int fd) if (encoding) strcpy(encoding, tok->encoding); } +#ifndef PGEN +error: +#endif PyTokenizer_Free(tok); return encoding; } |