htex

commit 275940c7345d05564943a2bb14dfe0ccc306a95e
parent 0e33554f458efcf0b45d3985db29a333c2c35385
Author: Robin <kroekerrobin@gmail.com>
Date:   Tue,  8 Aug 2023 21:43:24 +0200

Support finding by just an attr value

Diffstat:
M
html.c
 | 
19
++++++++++---------
M
todo
 | 
2
+-

2 files changed, 11 insertions(+), 10 deletions(-)
diff --git a/html.c b/html.c
@@ -518,22 +518,19 @@ struct tag *findTag(struct tag *tag, struct tag_list *list, struct filter_opts *
 {
 	bool matchesTag = false;
 	bool matchesAttrKey = false;
-	bool matchesAttrKeyAndValue = false;
+	bool matchesAttrValue = false;
 	if (strcmp(tag->name, opt->tag) == 0)
 		matchesTag = true;
 	for (int i=0; i<tag->attrsLen; i++)
 	{
 		if (strcmp(tag->attrs[i]->name, opt->key) == 0)
 			matchesAttrKey = true;
-		if (matchesAttrKey)
-		{
-			if (strcmp(tag->attrs[i]->value, opt->attr) == 0)
-				matchesAttrKeyAndValue = true;
-		}
+		if (strcmp(tag->attrs[i]->value, opt->attr) == 0)
+			matchesAttrValue = true;
 	}
 	if (strlen(opt->tag) > 0 && strlen(opt->key) > 0 && strlen(opt->attr) > 0)
 	{
-		if (matchesTag && matchesAttrKeyAndValue)
+		if (matchesTag && matchesAttrKey && matchesAttrValue)
 			return tag;
 	}
 	else if (strlen(opt->tag) > 0 && strlen(opt->key) > 0)
@@ -548,7 +545,7 @@ struct tag *findTag(struct tag *tag, struct tag_list *list, struct filter_opts *
 	}
 	else if (strlen(opt->key) > 0 && strlen(opt->attr) > 0)
 	{
-		if (matchesAttrKeyAndValue)
+		if (matchesAttrKey && matchesAttrValue)
 			return tag;
 	}
 	else if (strlen(opt->key) > 0)
@@ -556,6 +553,11 @@ struct tag *findTag(struct tag *tag, struct tag_list *list, struct filter_opts *
 		if (matchesAttrKey)
 			return tag;
 	}
+	else if (strlen(opt->attr) > 0)
+	{
+		if (matchesAttrValue)
+			return tag;
+	}
 	for (int i=0; i<tag->childrenLen; i++)
 	{
 		struct tag *foundTag = findTag(tag->children[i], list, opt);
@@ -582,7 +584,6 @@ void filterHtml(char *text, struct filter_opts *opts)
 	else
 	{
 		printf("result: %s\n", result->name);
-		printf("%ld %ld\n", result->_outerHtmlBeginOffset, result->_outerHtmlEndOffset);
 		if (!result->_isVoidElement)
 		{
 			printf("innerHtml: %s\n", result->innerHtml);
diff --git a/todo b/todo
@@ -1,2 +1,2 @@
 strip beginning and ending whitespace of inner and outer html
-find element by attr value (something's still wrong)
+parse html comments right