From bc0b7338ed2d4c126c15eaf57aeb84655e0e1cc3 Mon Sep 17 00:00:00 2001
From: dec05eba <dec05eba@protonmail.com>
Date: Sat, 25 May 2019 03:58:20 +0200
Subject: Allow reusing doc for multiple xpath searches

---
 src/HtmlSearch.c | 39 +++++++++++++++++++++++----------------
 1 file changed, 23 insertions(+), 16 deletions(-)

(limited to 'src')

diff --git a/src/HtmlSearch.c b/src/HtmlSearch.c
index c3608dd..7868f32 100644
--- a/src/HtmlSearch.c
+++ b/src/HtmlSearch.c
@@ -99,35 +99,42 @@ const QuickMediaStringView quickmedia_html_node_get_text(QuickMediaHtmlNode *sel
     return string_view;
 }
 
-static int quickmedia_html_find_nodes(const char *html_source, QuickMediaNodeSearch *search_data, QuickMediaHtmlSearchResultCallback result_callback, void *userdata) {
-    assert(html_source);
+static int quickmedia_html_find_nodes(QuickMediaHtmlSearch *self, QuickMediaNodeSearch *search_data, QuickMediaHtmlSearchResultCallback result_callback, void *userdata) {
     assert(search_data);
     assert(result_callback);
-    if(!html_source || !search_data || !result_callback)
+    if(!search_data || !result_callback)
         return -1;
+    
+    TidyNode root_node = tidyGetRoot(self->doc);
+    find_child_nodes(self->doc, root_node, search_data, result_callback, userdata);
+    return 0;
+}
 
-    TidyDoc tdoc = tidyCreate();
-    tidyOptSetBool(tdoc, TidyShowWarnings, no);
-    /* tidyOptSetBool(tdoc, TidyForceOutput, yes); */
-    int rc = tidyParseString( tdoc, html_source);
-    if(rc < 0) {
-        tidyRelease(tdoc);
-        return rc;
+int quickmedia_html_search_init(QuickMediaHtmlSearch *self, const char *html_source) {
+    self->doc = tidyCreate();
+    tidyOptSetBool(self->doc, TidyShowWarnings, no);
+    /* tidyOptSetBool(self->doc, TidyForceOutput, yes); */
+    if(tidyParseString(self->doc, html_source) < 0) {
+        tidyRelease(self->doc);
+        self->doc = NULL;
     }
-    
-    TidyNode root_node = tidyGetRoot(tdoc);
-    find_child_nodes(tdoc, root_node, search_data, result_callback, userdata);
-    tidyRelease(tdoc);
     return 0;
 }
 
-int quickmedia_html_find_nodes_xpath(const char *html_source, const char *xpath, QuickMediaHtmlSearchResultCallback result_callback, void *userdata) {
+void quickmedia_html_search_deinit(QuickMediaHtmlSearch *self) {
+    if(self->doc) {
+        tidyRelease(self->doc);
+        self->doc = NULL;
+    }
+}
+
+int quickmedia_html_find_nodes_xpath(QuickMediaHtmlSearch *self, const char *xpath, QuickMediaHtmlSearchResultCallback result_callback, void *userdata) {
     QuickMediaNodeSearch search_data;
     quickmedia_node_search_init(&search_data);
     int result = quickmedia_parse_xpath(xpath, &search_data);
     if(result != 0)
         goto cleanup;
-    result = quickmedia_html_find_nodes(html_source, &search_data, result_callback, userdata);
+    result = quickmedia_html_find_nodes(self, &search_data, result_callback, userdata);
     cleanup:
     quickmedia_node_search_deinit(&search_data);
     return result;
-- 
cgit v1.2.3