net.html: create html parser module (#6076)

2025-09-15 23:42:28 +03:00 · 2020-08-08 23:13:34 -03:00 · 2020-08-08 23:13:34 -03:00 · f6a85d5305
commit f6a85d5305
parent 97fec78b40
7 changed files with 894 additions and 0 deletions
--- a/vlib/net/html/dom_test.v
+++ b/vlib/net/html/dom_test.v
@ -0,0 +1,60 @@
+module html
+
+fn generate_temp_html() string {
+	mut temp_html := '<!doctype html><html><head><title>Giant String</title></head><body>'
+	for counter := 0; counter < 4; counter++ {
+		temp_html += "<div id='name_$counter' "
+		temp_html += "class='several-$counter'>Look at $counter</div>"
+	}
+	temp_html += '</body></html>'
+	return temp_html
+}
+
+fn generate_dom(temp_html string) DocumentObjectModel {
+	mut parser := Parser{}
+	parser.parse_html(temp_html, false)
+	dom := parser.get_dom()
+	return dom
+}
+
+fn test_search_by_tag_type() {
+	dom := generate_dom(generate_temp_html())
+	assert dom.get_by_tag('div').len == 4
+	assert dom.get_by_tag('head').len == 1
+	assert dom.get_by_tag('body').len == 1
+}
+
+fn test_search_by_attribute_value() {
+	mut dom := generate_dom(generate_temp_html())
+	// println(temp_html)
+	print('Amount ')
+	println(dom.get_by_attribute_value('id', 'name_0'))
+	assert dom.get_by_attribute_value('id', 'name_0').len == 1
+}
+
+fn test_access_parent() {
+	mut dom := generate_dom(generate_temp_html())
+	div_tags := dom.get_by_tag('div')[0]
+	assert div_tags.get_parent() != C.NULL
+	parent := div_tags.get_parent()
+	for div_tag in div_tags {
+		assert div_tag.get_parent() == parent
+	}
+}
+
+fn test_search_by_attributes() {
+	dom := generate_dom(generate_temp_html())
+	assert dom.get_by_attribute('id').len == 4
+}
+
+fn test_tags_used() {
+	dom := generate_dom(generate_temp_html())
+	assert dom.get_all_tags().len == 9
+}
+
+fn test_access_tag_fields() {
+	dom := generate_dom(generate_temp_html())
+	id_tags := dom.get_by_attribute('id')
+	assert id_tags[0].get_name() == "div"
+	assert id_tags[1].get_attributes()['class'] == "several-1"
+}