From a10a8fb335e5a817e1a9add49ee179394eea67c4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Anton=20Luka=20=C5=A0ijanec?= <anton@sijanec.eu>
Date: Sun, 26 Dec 2021 19:52:31 +0100
Subject: fixed parser, fixed leak, O(log n) storage - tsearch(3) - 0.0.17

---
 src/api.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

(limited to 'src/api.c')

diff --git a/src/api.c b/src/api.c
index 5ab4083..a366882 100644
--- a/src/api.c
+++ b/src/api.c
@@ -87,7 +87,7 @@ htmlDocPtr sc_capix (struct sc_cache * c, char * body, char * headers, int isfmt
 char * sc_find_class (char * haystack, const char * definition) { /* you must free class after calling */
 	if (!haystack || !definition)
 		return NULL;
-	char * class = strstr(haystack, definition);
+	char * class = strcasestr(haystack, definition);
 	if (!class)
 		return NULL;
 	int found = 0;
@@ -117,7 +117,7 @@ int sc_fix_url (char ** h) { /* fixes a (result) URL in-place (removes tracking
 		urldecode(*h, *h);
 	}
 	char * c = NULL;
-	if ((c = strstr(*h, "googleweblight.com/fp?u="))) { /* stage 2: url may be "light web" tracking url by google results */
+	if ((c = strcasestr(*h, "googleweblight.com/fp?u="))) { /* stage 2: url may be "light web" tracking url by google results */
 		*h = c+strlen("googleweblight.com/fp?u="); /* we could disable this with a cookie but meh, this is easier and _stateless_ */
 		*strchrnul(*h, '&') = '\0';
 		urldecode(*h, *h);
@@ -156,6 +156,7 @@ struct sc_query * sc_query_google (const char * s, struct sc_cache * c, struct s
 		goto rc;
 	}
 	int qwasgiven = 0;
+	int sl = strlen(s);
 	if (!q)
 		q = sc_query_init();
 	else
@@ -214,6 +215,7 @@ struct sc_query * sc_query_google (const char * s, struct sc_cache * c, struct s
 					char * imgrefurl = NULL; /* easy, huh? */
 					SC_LOG(SC_LOG_DEBUG, c, "hreflink = %s", hreflink);
 					sscanf(hreflink, "/imgres?imgurl=%m[^&]&imgrefurl=%m[^&]", &imgurl, &imgrefurl);
+					xmlFree(hreflink);
 					if (!imgurl && !imgrefurl) {
 						SC_LOG(SC_LOG_ERROR, c, "!imgurl && !imgrefurl, txtdoc = %s", txtdoc);
 						/* rs = -6; */ /* we continue running not fail because of a single picture */
@@ -257,8 +259,9 @@ struct sc_query * sc_query_google (const char * s, struct sc_cache * c, struct s
 					if (hreflink) {
 						SC_GTR->url = malloc(strlen(hreflink)+1);
 						strcpy(SC_GTR->url, hreflink);
-						xmlFree(orig_hreflink_for_free);
 					} else SC_GTR->url = NULL;
+					if (orig_hreflink_for_free)
+						xmlFree(orig_hreflink_for_free);
 					cp = (char *) xmlNodeGetContent(descnode);
 					if (cp) {
 						SC_GTR->desc = malloc(strlen(cp)+1);
@@ -285,17 +288,20 @@ struct sc_query * sc_query_google (const char * s, struct sc_cache * c, struct s
 	}
 	q->cache = c;
 	q->lookup_time = time(NULL);
-	q->engines = SC_ENGINE_GOOGLE;
-	q->string = realloc(q->string, strlen(s)+1);
-	q->opt = opt;
+	q->string = realloc(q->string, sl+1);
+	q->opt |= opt | SC_ENGINE_GOOGLE;
 	strcpy(q->string, s);
 	if (!qwasgiven) {
 		SC_CWLE(c, c->queries_lock);
+#ifdef SC_OLD_STORAGE
 		if (c->queries_sizeof <= c->queries_length)
 			SC_BIGGER_ARRAY(c->queries, sc_query, 0);
 		c->queries_length++;
 #define SC_GTQ c->queries[c->queries_length-1]
 		SC_GTQ = q;
+#else /* we don't detect here if query is already stored, but it should not be ... */
+		tsearch(q, &c->qrp, SC_COMPAR_CAST sc_query_compar);
+#endif
 	}
 	SC_CUE(c, c->queries_lock);
 rc:
-- 
cgit v1.2.3