From b5485da64c1f8086eb49a2c830ca75511f34abfd Mon Sep 17 00:00:00 2001
From: dec05eba <dec05eba@protonmail.com>
Date: Wed, 15 May 2024 22:28:05 +0200
Subject: Strip class attribute matching
---
src/HtmlSearch.c | 117 ++++++++++++++++++++++++++++---------------------------
1 file changed, 60 insertions(+), 57 deletions(-)
(limited to 'src')
diff --git a/src/HtmlSearch.c b/src/HtmlSearch.c
index 4ae27af..7d5c5c5 100644
--- a/src/HtmlSearch.c
+++ b/src/HtmlSearch.c
@@ -142,10 +142,65 @@ static char to_upper(char c) {
return c;
}
-static int string_views_equal_case_insensitive(const QuickMediaStringView str1, const QuickMediaStringView str2) {
+static int is_whitespace(int c) {
+ switch(c) {
+ case ' ':
+ case '\n':
+ case '\r':
+ case '\t':
+ case '\v':
+ return 1;
+ default:
+ return 0;
+ }
+}
+
+static int is_newline(int c) {
+ return c == '\n' || c == '\r';
+}
+
+static void lstrip(const char *str, size_t size, const char **output_str, size_t *output_size, int(*strip_filter_func)(int)) {
+ if(size == 0) {
+ *output_str = str;
+ *output_size = size;
+ return;
+ }
+
+ size_t i = 0;
+ while(i < size && strip_filter_func(str[i])) {
+ ++i;
+ }
+
+ *output_str = str + i;
+ *output_size = size - i;
+}
+
+static void rstrip(const char *str, size_t size, size_t *output_size, int(*strip_filter_func)(int)) {
+ if(size == 0) {
+ *output_size = size;
+ return;
+ }
+
+ ssize_t i = size - 1;
+ while(i >= 0 && strip_filter_func(str[i])) {
+ --i;
+ }
+
+ *output_size = i + 1;
+}
+
+static void strip(const char *str, size_t size, const char **output_str, size_t *output_size, int(*strip_filter_func)(int)) {
+ lstrip(str, size, output_str, output_size, strip_filter_func);
+ rstrip(*output_str, *output_size, output_size, strip_filter_func);
+}
+
+static int string_views_equal_case_insensitive_strip(QuickMediaStringView str1, QuickMediaStringView str2) {
if(str2.size != str1.size)
return 1;
+ strip(str1.data, str1.size, &str1.data, &str1.size, is_whitespace);
+ strip(str2.data, str2.size, &str2.data, &str2.size, is_whitespace);
+
for(size_t i = 0; i < str1.size; ++i) {
char c1 = str1.data[i];
char c2 = str2.data[i];
@@ -158,7 +213,7 @@ static int string_views_equal_case_insensitive(const QuickMediaStringView str1,
static QuickMediaHtmlAttribute* get_attribute_by_name(QuickMediaHtmlNode *node, QuickMediaStringView name) {
for(QuickMediaHtmlAttribute *attr = node->first_attribute; attr; attr = attr->next) {
- if(string_views_equal_case_insensitive(attr->key, name) == 0)
+ if(string_views_equal_case_insensitive_strip(attr->key, name) == 0)
return attr;
}
return NULL;
@@ -177,7 +232,7 @@ static int find_child_nodes(QuickMediaHtmlChildNode *node, const QuickMediaNodeS
continue;
/* Match without node name or node name matches */
- if(search_data->name.size == 0 || string_views_equal_case_insensitive(child->node.name, search_data->name) == 0) {
+ if(search_data->name.size == 0 || string_views_equal_case_insensitive_strip(child->node.name, search_data->name) == 0) {
#define on_match() do { \
if(search_data->child) { \
if(find_child_nodes(child->node.first_child, search_data->child, result_callback, userdata) != 0) \
@@ -409,72 +464,20 @@ QuickMediaHtmlNode* quickmedia_html_node_find_child(QuickMediaHtmlNode *self, co
if(!child->node.is_tag)
continue;
- if(string_views_equal_case_insensitive(child->node.name, tag) != 0)
+ if(string_views_equal_case_insensitive_strip(child->node.name, tag) != 0)
continue;
QuickMediaHtmlAttribute *attr = get_attribute_by_name(&child->node, attr_name);
if(!attr)
continue;
- if(string_views_equal_case_insensitive(attr->value, attr_value) == 0)
+ if(string_views_equal_case_insensitive_strip(attr->value, attr_value) == 0)
return &child->node;
}
return NULL;
}
-static int is_whitespace(int c) {
- switch(c) {
- case ' ':
- case '\n':
- case '\r':
- case '\t':
- case '\v':
- return 1;
- default:
- return 0;
- }
-}
-
-static int is_newline(int c) {
- return c == '\n' || c == '\r';
-}
-
-static void lstrip(const char *str, size_t size, const char **output_str, size_t *output_size, int(*strip_filter_func)(int)) {
- if(size == 0) {
- *output_str = str;
- *output_size = size;
- return;
- }
-
- size_t i = 0;
- while(i < size && strip_filter_func(str[i])) {
- ++i;
- }
-
- *output_str = str + i;
- *output_size = size - i;
-}
-
-static void rstrip(const char *str, size_t size, size_t *output_size, int(*strip_filter_func)(int)) {
- if(size == 0) {
- *output_size = size;
- return;
- }
-
- ssize_t i = size - 1;
- while(i >= 0 && strip_filter_func(str[i])) {
- --i;
- }
-
- *output_size = i + 1;
-}
-
-static void strip(const char *str, size_t size, const char **output_str, size_t *output_size, int(*strip_filter_func)(int)) {
- lstrip(str, size, output_str, output_size, strip_filter_func);
- rstrip(*output_str, *output_size, output_size, strip_filter_func);
-}
-
static int merge_inner_text(QuickMediaHtmlNode *node, QuickMediaString *str) {
if(node->is_tag) {
int newline = 0;
--
cgit v1.2.3-70-g09d2