From c5f4811d0ba74715c8e128da133248cc399a6a6a Mon Sep 17 00:00:00 2001
From: Aleksi Lindeman <dec05eba@protonmail.com>
Date: Sat, 25 May 2019 03:58:20 +0200
Subject: Allow reusing doc for multiple xpath searches

---
 include/quickmedia/HtmlSearch.h |  9 ++++++++-
 src/HtmlSearch.c                | 39 +++++++++++++++++++++++----------------
 tests/main.c                    | 12 +++++++++++-
 3 files changed, 42 insertions(+), 18 deletions(-)

diff --git a/include/quickmedia/HtmlSearch.h b/include/quickmedia/HtmlSearch.h
index e3bea33..568e101 100644
--- a/include/quickmedia/HtmlSearch.h
+++ b/include/quickmedia/HtmlSearch.h
@@ -13,6 +13,10 @@ typedef struct {
     void *text;
 } QuickMediaHtmlNode;
 
+typedef struct {
+    const void *doc;
+} QuickMediaHtmlSearch;
+
 /* Returns NULL if attribute doesn't exist or if it doesn't have any value */
 const char* quickmedia_html_node_get_attribute_value(QuickMediaHtmlNode *self, const char *attribute_name);
 
@@ -22,7 +26,10 @@ const QuickMediaStringView quickmedia_html_node_get_text(QuickMediaHtmlNode *sel
 /* @node is only valid within the callback function scope */
 typedef void (*QuickMediaHtmlSearchResultCallback)(QuickMediaHtmlNode *node, void *userdata);
 
-int quickmedia_html_find_nodes_xpath(const char *html_source, const char *xpath, QuickMediaHtmlSearchResultCallback result_callback, void *userdata);
+int quickmedia_html_search_init(QuickMediaHtmlSearch *self, const char *html_source);
+void quickmedia_html_search_deinit(QuickMediaHtmlSearch *self);
+
+int quickmedia_html_find_nodes_xpath(QuickMediaHtmlSearch *self, const char *xpath, QuickMediaHtmlSearchResultCallback result_callback, void *userdata);
 
 #ifdef __cplusplus
 }
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;
diff --git a/tests/main.c b/tests/main.c
index 5b697a9..4d16ad6 100644
--- a/tests/main.c
+++ b/tests/main.c
@@ -26,7 +26,17 @@ static void result_callback(QuickMediaHtmlNode *node, void *userdata) {
 
 int main(int argc, char **argv) {
     char *file_content = get_file_content("test_files/test.html");
-    int result = quickmedia_html_find_nodes_xpath(file_content, "//h3[class=\"story_name\"]//a", result_callback, NULL);
+    QuickMediaHtmlSearch html_search;
+
+    int result = quickmedia_html_search_init(&html_search, file_content);
+    if(result != 0)
+        goto cleanup;
+    result = quickmedia_html_find_nodes_xpath(&html_search, "//h3[class=\"story_name\"]//a", result_callback, NULL);
+    /* Test that the object can be reused without reloading html doc */
+    result = quickmedia_html_find_nodes_xpath(&html_search, "//h3[class=\"story_name\"]//a", result_callback, NULL);
+
+    cleanup:
+    quickmedia_html_search_deinit(&html_search);
     free(file_content);
     return result;
 }
-- 
cgit v1.2.3