@tkeron/html-parser 0.1.5 → 1.0.0

package/tests/selectors.test.ts

@@ -62,4 +62,67 @@ describe('CSS Selectors', () => {
             expect(nonExistent).toBeNull();
         });
     });
-});
+
+    describe('Element.matches', () => {
+        it('should match by tag name', () => {
+            const p = querySelector(doc, 'p');
+            expect(p?.matches('p')).toBe(true);
+            expect(p?.matches('div')).toBe(false);
+        });
+
+        it('should match by id', () => {
+            const intro = querySelector(doc, '#intro');
+            expect(intro?.matches('#intro')).toBe(true);
+            expect(intro?.matches('#other')).toBe(false);
+        });
+
+        it('should match by class', () => {
+            const first = querySelector(doc, '.first');
+            expect(first?.matches('.first')).toBe(true);
+            expect(first?.matches('.second')).toBe(false);
+        });
+
+        it('should match by multiple classes', () => {
+            const doc2 = parseHTML('<div class="foo bar baz">Test</div>');
+            const div = doc2.querySelector('div');
+            expect(div?.matches('.foo')).toBe(true);
+            expect(div?.matches('.bar')).toBe(true);
+            expect(div?.matches('.foo.bar')).toBe(true);
+            expect(div?.matches('.foo.baz')).toBe(true);
+            expect(div?.matches('.foo.bar.baz')).toBe(true);
+            expect(div?.matches('.foo.missing')).toBe(false);
+        });
+
+        it('should match by attribute', () => {
+            const intro = querySelector(doc, '#intro');
+            expect(intro?.matches('[id]')).toBe(true);
+            expect(intro?.matches('[id="intro"]')).toBe(true);
+            expect(intro?.matches('[class]')).toBe(true);
+            expect(intro?.matches('[title]')).toBe(false);
+        });
+
+        it('should match complex selectors', () => {
+            const intro = querySelector(doc, '#intro');
+            expect(intro?.matches('p#intro')).toBe(true);
+            expect(intro?.matches('p.first')).toBe(true);
+            expect(intro?.matches('div#intro')).toBe(false);
+        });
+
+        it('should match descendant selectors', () => {
+            const span = querySelector(doc, 'span');
+            expect(span?.matches('p span')).toBe(true);
+            expect(span?.matches('body span')).toBe(true);
+            expect(span?.matches('div span')).toBe(false);
+        });
+
+        it('should return false for invalid selector', () => {
+            const p = querySelector(doc, 'p');
+            expect(p?.matches('')).toBe(false);
+        });
+
+        it('should work with universal selector', () => {
+            const p = querySelector(doc, 'p');
+            expect(p?.matches('*')).toBe(true);
+        });
+    });
+});

package/tests/serializer-core.test.ts

@@ -0,0 +1,16 @@
+import { expect, it, describe } from 'bun:test';
+import { serializeTokens } from '../src/serializer';
+import { readFileSync } from 'fs';
+
+describe('Serializer Core Tests', () => {
+  const content = readFileSync('tests/serializer-data/core.test', 'utf8');
+  const data = JSON.parse(content);
+  const tests = data.tests;
+
+  tests.forEach((test: any, index: number) => {
+    it(test.description, () => {
+      const result = serializeTokens(test.input);
+      expect(result).toBe(test.expected[0]);
+    });
+  });
+});

package/tests/serializer-data/core.test

@@ -0,0 +1,125 @@
+{"tests": [
+
+{"description": "proper attribute value escaping",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "test \"with\" ""}]]],
+ "expected": ["<span title='test \"with\" &amp;quot;'>"]
+},
+
+{"description": "proper attribute value non-quoting",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo"}]]],
+ "expected": ["<span title=foo>"],
+ "xhtml":    ["<span title=\"foo\">"]
+},
+
+{"description": "proper attribute value non-quoting (with <)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo<bar"}]]],
+ "expected": ["<span title=foo<bar>"],
+ "xhtml":    ["<span title=\"foo&lt;bar\">"]
+},
+
+{"description": "proper attribute value quoting (with =)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo=bar"}]]],
+ "expected": ["<span title=\"foo=bar\">"]
+},
+
+{"description": "proper attribute value quoting (with >)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo>bar"}]]],
+ "expected": ["<span title=\"foo>bar\">"]
+},
+
+{"description": "proper attribute value quoting (with \")",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo\"bar"}]]],
+ "expected": ["<span title='foo\"bar'>"]
+},
+
+{"description": "proper attribute value quoting (with ')",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo'bar"}]]],
+ "expected": ["<span title=\"foo'bar\">"]
+},
+
+{"description": "proper attribute value quoting (with both \" and ')",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo'bar\"baz"}]]],
+ "expected": ["<span title=\"foo'bar&quot;baz\">"]
+},
+
+{"description": "proper attribute value quoting (with space)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo bar"}]]],
+ "expected": ["<span title=\"foo bar\">"]
+},
+
+{"description": "proper attribute value quoting (with tab)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo\tbar"}]]],
+ "expected": ["<span title=\"foo\tbar\">"]
+},
+
+{"description": "proper attribute value quoting (with LF)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo\nbar"}]]],
+ "expected": ["<span title=\"foo\nbar\">"]
+},
+
+{"description": "proper attribute value quoting (with CR)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo\rbar"}]]],
+ "expected": ["<span title=\"foo\rbar\">"]
+},
+
+{"description": "proper attribute value non-quoting (with linetab)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo\u000Bbar"}]]],
+ "expected": ["<span title=foo\u000Bbar>"],
+ "xhtml": ["<span title=\"foo\u000Bbar\">"]
+},
+
+{"description": "proper attribute value quoting (with form feed)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "span", [{"namespace": null, "name": "title", "value": "foo\u000Cbar"}]]],
+ "expected": ["<span title=\"foo\u000Cbar\">"]
+},
+
+{"description": "void element (as EmptyTag token)",
+ "input": [["EmptyTag", "img", {}]],
+ "expected": ["<img>"],
+ "xhtml":    ["<img />"]
+},
+
+{"description": "void element (as StartTag token)",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "img", {}]],
+ "expected": ["<img>"],
+ "xhtml":    ["<img />"]
+},
+
+{"description": "doctype in error",
+ "input": [["Doctype", "foo"]],
+ "expected": ["<!DOCTYPE foo>"]
+},
+
+{"description": "character data",
+ "options": {"encoding":"utf-8"},
+ "input": [["Characters", "a<b>c&d"]],
+ "expected": ["a&lt;b&gt;c&amp;d"]
+},
+
+{"description": "rcdata",
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "script", {}], ["Characters", "a<b>c&d"]],
+ "expected": ["<script>a<b>c&d"],
+ "xhtml": ["<script>a&lt;b&gt;c&amp;d"]
+},
+
+{"description": "doctype",
+ "input": [["Doctype", "HTML"]],
+ "expected": ["<!DOCTYPE HTML>"]
+},
+
+{"description": "HTML 4.01 DOCTYPE",
+ "input": [["Doctype", "HTML",  "-//W3C//DTD HTML 4.01//EN", "http://www.w3.org/TR/html4/strict.dtd"]],
+ "expected": ["<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\" \"http://www.w3.org/TR/html4/strict.dtd\">"]
+},
+
+{"description": "HTML 4.01 DOCTYPE without system identifier",
+ "input": [["Doctype", "HTML",  "-//W3C//DTD HTML 4.01//EN"]],
+ "expected": ["<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01//EN\">"]
+},
+
+{"description": "IBM DOCTYPE without public identifier",
+ "input": [["Doctype", "html",  "", "http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd"]],
+ "expected": ["<!DOCTYPE html SYSTEM \"http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd\">"]
+}
+
+]}

package/tests/serializer-data/injectmeta.test

@@ -0,0 +1,66 @@
+{"tests": [
+
+{"description": "no encoding",
+ "options": {"inject_meta_charset": true},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": [""],
+ "xhtml": ["<head></head>"]
+},
+
+{"description": "empytag head",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta charset=utf-8>"],
+ "xhtml":    ["<head><meta charset=\"utf-8\" /></head>"]
+},
+
+{"description": "head w/title",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["StartTag", "http://www.w3.org/1999/xhtml","title",{}], ["Characters", "foo"],["EndTag", "http://www.w3.org/1999/xhtml", "title"], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta charset=utf-8><title>foo</title>"],
+ "xhtml":    ["<head><meta charset=\"utf-8\" /><title>foo</title></head>"]
+},
+
+{"description": "head w/meta-charset",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EmptyTag","meta",[{"namespace": null, "name": "charset", "value": "ascii"}]], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta charset=utf-8>"],
+ "xhtml":    ["<head><meta charset=\"utf-8\" /></head>"]
+},
+
+{"description": "head w/ two meta-charset",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EmptyTag","meta",[{"namespace": null, "name": "charset", "value": "ascii"}]], ["EmptyTag","meta",[{"namespace": null, "name": "charset", "value": "ascii"}]], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta charset=utf-8><meta charset=utf-8>", "<head><meta charset=utf-8><meta charset=ascii>"],
+ "xhtml": ["<head><meta charset=\"utf-8\" /><meta charset=\"utf-8\" /></head>", "<head><meta charset=\"utf-8\" /><meta charset=\"ascii\" /></head>"]
+},
+
+{"description": "head w/robots",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EmptyTag","meta",[{"namespace": null, "name": "name", "value": "robots"},{"namespace": null, "name": "content", "value": "noindex"}]], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta charset=utf-8><meta content=noindex name=robots>"],
+ "xhtml":    ["<head><meta charset=\"utf-8\" /><meta content=\"noindex\" name=\"robots\" /></head>"]
+},
+
+{"description": "head w/robots & charset",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EmptyTag","meta",[{"namespace": null, "name": "name", "value": "robots"},{"namespace": null, "name": "content", "value": "noindex"}]], ["EmptyTag","meta",[{"namespace": null, "name": "charset", "value": "ascii"}]], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta content=noindex name=robots><meta charset=utf-8>"],
+ "xhtml":    ["<head><meta content=\"noindex\" name=\"robots\" /><meta charset=\"utf-8\" /></head>"]
+},
+
+{"description": "head w/ charset in http-equiv content-type",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EmptyTag","meta",[{"namespace": null, "name": "http-equiv", "value": "content-type"}, {"namespace": null, "name": "content", "value": "text/html; charset=ascii"}]], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta content=\"text/html; charset=utf-8\" http-equiv=content-type>"],
+ "xhtml":    ["<head><meta content=\"text/html; charset=utf-8\" http-equiv=\"content-type\" /></head>"]
+},
+
+{"description": "head w/robots & charset in http-equiv content-type",
+ "options": {"inject_meta_charset": true, "encoding":"utf-8"},
+ "input": [["StartTag", "http://www.w3.org/1999/xhtml", "head", {}], ["EmptyTag","meta",[{"namespace": null, "name": "name", "value": "robots"},{"namespace": null, "name": "content", "value": "noindex"}]], ["EmptyTag","meta",[{"namespace": null, "name": "http-equiv", "value": "content-type"}, {"namespace": null, "name": "content", "value": "text/html; charset=ascii"}]], ["EndTag", "http://www.w3.org/1999/xhtml", "head"]],
+ "expected": ["<meta content=noindex name=robots><meta content=\"text/html; charset=utf-8\" http-equiv=content-type>"],
+ "xhtml": ["<head><meta content=\"noindex\" name=\"robots\" /><meta content=\"text/html; charset=utf-8\" http-equiv=\"content-type\" /></head>"]
+}
+
+]}