My personal build of CMark ✏️
Optimize S_find_first_nonspace.
We were needlessly redoing things we'd already done. Now we skip the work if the first nonspace is greater than the current offset.
This fixes pathological slowdown with deeply nested lists (#255). For N = 3000, the time goes from over 17s to about 0.7s.
Thanks to @mity for diagnosing the problem.
1 file changed, 19 insertions, 14 deletions
| Status | File Name | N° Changes | Insertions | Deletions |
| Modified | src/blocks.c | 33 | 19 | 14 |
diff --git a/src/blocks.c b/src/blocks.c @@ -615,22 +615,24 @@ static void S_find_first_nonspace(cmark_parser *parser, cmark_chunk *input) { char c; int chars_to_tab = TAB_STOP - (parser->column % TAB_STOP); - parser->first_nonspace = parser->offset; - parser->first_nonspace_column = parser->column; - while ((c = peek_at(input, parser->first_nonspace))) { - if (c == ' ') { - parser->first_nonspace += 1; - parser->first_nonspace_column += 1; - chars_to_tab = chars_to_tab - 1; - if (chars_to_tab == 0) { + if (parser->first_nonspace <= parser->offset) { + parser->first_nonspace = parser->offset; + parser->first_nonspace_column = parser->column; + while ((c = peek_at(input, parser->first_nonspace))) { + if (c == ' ') { + parser->first_nonspace += 1; + parser->first_nonspace_column += 1; + chars_to_tab = chars_to_tab - 1; + if (chars_to_tab == 0) { + chars_to_tab = TAB_STOP; + } + } else if (c == '\t') { + parser->first_nonspace += 1; + parser->first_nonspace_column += chars_to_tab; chars_to_tab = TAB_STOP; + } else { + break; } - } else if (c == '\t') { - parser->first_nonspace += 1; - parser->first_nonspace_column += chars_to_tab; - chars_to_tab = TAB_STOP; - } else { - break; } } @@ -1160,6 +1162,9 @@ static void S_process_line(cmark_parser *parser, const unsigned char *buffer, parser->offset = 0; parser->column = 0; + parser->first_nonspace = 0; + parser->first_nonspace_column = 0; + parser->indent = 0; parser->blank = false; parser->partially_consumed_tab = false;