nostrdb

commit d08532738aa0a122685eb6922ff1ad49e7d53b95
parent fefb9df7cd51e3233401aad9c9508f302e903297
Author: William Casarin <jb55@jb55.com>
Date:   Wed, 29 Nov 2023 12:01:17 -0800

search: fix infinite loop when parsing some notes

Our word parser gets stuck on some notes with utf8 chars. Make sure we
are always advancing so we don't get stuck.

Diffstat:
M
nostrdb.c
 | 
9
+++++++++

1 file changed, 9 insertions(+), 0 deletions(-)
diff --git a/nostrdb.c b/nostrdb.c
@@ -2218,6 +2218,7 @@ static inline int consume_until_boundary(struct cursor *cur) {
 					// TODO: We should work towards
 					// handling all UTF-8 characters.
 					//printf("Invalid UTF-8 code point: %x\n", c);
+					return 0;
 				}
 			}
 		}
@@ -2305,6 +2306,14 @@ static int ndb_parse_words(struct cursor *cur, void *ctx, ndb_word_parser_fn fn)
 		if (word_len == 0 && cur->p >= cur->end)
 			break;
 
+		if (word_len == 0) {
+			if (!cursor_skip(cur, 1))
+				break;
+			continue;
+		}
+
+		//ndb_debug("writing word index '%.*s'\n", word_len, word);
+
 		if (!fn(ctx, word, word_len, words))
 			continue;