@tkeron/html-parser 1.1.2 → 1.3.0

package/tests/tree-construction-adoption01.test.ts

@@ -1,37 +1,46 @@
-import { expect, it, describe } from 'bun:test';
-import { parseHTML } from '../index';
-import { serializeToHtml5lib } from './helpers/tree-adapter';
-import { readFileSync } from 'fs';
+import { expect, it, describe } from "bun:test";
+import { parseHTML } from "../index";
+import { serializeToHtml5lib } from "./helpers/tree-adapter";
+import { readFileSync } from "fs";
 
-describe('Tree Construction Adoption01 Tests', () => {
-  const content = readFileSync('tests/html5lib-data/tree-construction/adoption01.dat', 'utf8');
-  const sections = content.split('#data\n').slice(1);
+describe("Tree Construction Adoption01 Tests", () => {
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/adoption01.dat",
+    "utf8",
+  );
+  const sections = content.split("#data\n").slice(1);
 
   sections.forEach((section, index) => {
-    const lines = section.trim().split('\n');
-    let data = '';
-    let document = '';
+    const lines = section.trim().split("\n");
+    let data = "";
+    let document = "";
     let inDocument = false;
     let inData = true; // Start with data since we split on #data\n
 
     for (const line of lines) {
-      if (line.startsWith('#document')) {
+      if (line.startsWith("#document")) {
         inDocument = true;
         inData = false;
-      } else if (line.startsWith('#errors')) {
+      } else if (line.startsWith("#errors")) {
         inData = false;
         inDocument = false;
       } else if (inDocument) {
-        document += line + '\n';
+        document += line + "\n";
       } else if (inData) {
         data += line;
       }
     }
 
-    it.skip(`Adoption test ${index + 1}`, () => {
+    const passingTests = [1, 2, 3, 4, 7, 8, 9, 16];
+    const testFn = passingTests.includes(index + 1) ? it : it.skip;
+
+    testFn(`Adoption test ${index + 1}`, () => {
       const doc = parseHTML(data);
-      const serialized = serializeToHtml5lib(doc);
+      const hasExplicitDoctype = data.toLowerCase().includes("<!doctype");
+      const serialized = serializeToHtml5lib(doc, {
+        skipImplicitDoctype: !hasExplicitDoctype,
+      });
       expect(serialized).toBe(document);
     });
   });
-});
+});

package/tests/tree-construction-adoption02.test.ts

@@ -1,34 +1,45 @@
-import { expect, it, describe } from 'bun:test';
-import { parseHTML } from '../index';
-import { serializeToHtml5lib } from './helpers/tree-adapter';
-import { readFileSync } from 'fs';
+import { expect, it, describe } from "bun:test";
+import { parseHTML } from "../index";
+import { serializeToHtml5lib } from "./helpers/tree-adapter";
+import { readFileSync } from "fs";
 
-describe('Tree Construction Adoption02 Tests', () => {
-  const content = readFileSync('tests/html5lib-data/tree-construction/adoption02.dat', 'utf8');
-  const sections = content.split('#data\n').slice(1);
+describe("Tree Construction Adoption02 Tests", () => {
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/adoption02.dat",
+    "utf8",
+  );
+  const sections = content.split("#data\n").slice(1);
+  const passingTests = [1];
 
   sections.forEach((section, index) => {
-    const lines = section.trim().split('\n');
-    let data = '';
-    let document = '';
+    const lines = section.trim().split("\n");
+    let data = "";
+    let document = "";
     let inDocument = false;
+    let inData = true;
 
     for (const line of lines) {
-      if (line.startsWith('#document')) {
+      if (line.startsWith("#document")) {
         inDocument = true;
-      } else if (line.startsWith('#data')) {
-        // next section
+        inData = false;
+      } else if (line.startsWith("#errors")) {
+        inData = false;
+        inDocument = false;
       } else if (inDocument) {
-        document += line.slice(2) + '\n';
-      } else if (!line.startsWith('#')) {
+        document += line + "\n";
+      } else if (inData) {
         data += line;
       }
     }
 
-    it.skip(`Adoption02 test ${index + 1}`, () => {
+    const testFn = passingTests.includes(index + 1) ? it : it.skip;
+    testFn(`Adoption02 test ${index + 1}`, () => {
       const doc = parseHTML(data);
-      const serialized = serializeToHtml5lib(doc);
-      expect(serialized).toBe(document.trim());
+      const hasExplicitDoctype = data.toLowerCase().includes("<!doctype");
+      const serialized = serializeToHtml5lib(doc, {
+        skipImplicitDoctype: !hasExplicitDoctype,
+      });
+      expect(serialized).toBe(document);
     });
   });
-});
+});

package/tests/tree-construction-domjs-unsafe.test.ts

@@ -3,22 +3,24 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction DomjsUnsafe Tests", () => {
-  const data = readFileSync("tests/html5lib-data/tree-construction/domjs-unsafe.dat", "utf8");
+  const data = readFileSync(
+    "tests/html5lib-data/tree-construction/domjs-unsafe.dat",
+    "utf8",
+  );
   const sections = data.split("#data\n").slice(1);
 
   for (const section of sections) {
     const parts = section.split("#document\n");
     if (parts.length < 2) continue;
     const inputWithErrors = parts[0];
-    const expected = parts[1];
     const input = inputWithErrors.split("#errors\n")[0].trim();
 
     const testName = input.split("\n")[0] || "DomjsUnsafe test";
-    it.skip(testName, () => {
+    it(testName, () => {
       const doc = parse(input);
       // TODO: Implement DOM tree comparison with expected
       // For now, just ensure parsing doesn't throw
       expect(doc).toBeDefined();
     });
   }
-});
+});

package/tests/tree-construction-entities02.test.ts

@@ -1,33 +1,35 @@
-import { expect, it, describe } from 'bun:test';
-import { parse } from '../src/parser';
-import { readFileSync } from 'fs';
+import { expect, it, describe } from "bun:test";
+import { parse } from "../src/parser";
+import { readFileSync } from "fs";
 
-describe('Tree Construction Entities02 Tests', () => {
-  const content = readFileSync('tests/html5lib-data/tree-construction/entities02.dat', 'utf8');
-  const sections = content.split('#data\n').slice(1);
+describe("Tree Construction Entities02 Tests", () => {
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/entities02.dat",
+    "utf8",
+  );
+  const sections = content.split("#data\n").slice(1);
 
   sections.forEach((section, index) => {
-    const lines = section.trim().split('\n');
-    let data = '';
-    let document = '';
+    const lines = section.trim().split("\n");
+    let data = "";
+    let document = "";
     let inDocument = false;
 
     for (const line of lines) {
-      if (line.startsWith('#document')) {
+      if (line.startsWith("#document")) {
         inDocument = true;
-      } else if (line.startsWith('#data')) {
+      } else if (line.startsWith("#data")) {
         // next section
       } else if (inDocument) {
-        document += line + '\n';
-      } else if (!line.startsWith('#')) {
+        document += line + "\n";
+      } else if (!line.startsWith("#")) {
         data += line;
       }
     }
 
     it(`Entities02 test ${index + 1}`, () => {
-      const doc = parse(data);
-      // TODO: compare doc with expected document tree
+      parse(data);
       expect(true).toBe(true); // placeholder
     });
   });
-});
+});

package/tests/tree-construction-html5test-com.test.ts

@@ -4,7 +4,10 @@ import { parseHTML } from "../src/index.ts";
 import { serializeToHtml5lib } from "./helpers/tree-adapter";
 
 describe("Tree Construction Html5testCom Tests", () => {
-  const data = readFileSync("tests/html5lib-data/tree-construction/html5test-com.dat", "utf8");
+  const data = readFileSync(
+    "tests/html5lib-data/tree-construction/html5test-com.dat",
+    "utf8",
+  );
   const sections = data.split("#data\n").slice(1);
 
   for (const section of sections) {
@@ -12,21 +15,24 @@ describe("Tree Construction Html5testCom Tests", () => {
     if (parts.length < 2) continue;
     const inputWithErrors = parts[0];
     const expectedRaw = parts[1].split("\n#")[0];
-    const expected = expectedRaw.split("\n").filter(l => l.startsWith("|")).join("\n") + "\n";
+    const expected =
+      expectedRaw
+        .split("\n")
+        .filter((l) => l.startsWith("|"))
+        .join("\n") + "\n";
     const input = inputWithErrors.split("#errors\n")[0].trim();
     const hasDoctype = input.toLowerCase().startsWith("<!doctype");
 
     const testName = input.split("\n")[0] || "Html5testCom test";
-    
-    const isFosterParenting = input.includes('<table><form><input type=hidden><input></form><div></div></table>');
-    const isAdoptionAgency = input.includes('<i>A<b>B<p></i>C</b>D');
-    
-    const testFn = (isFosterParenting || isAdoptionAgency) ? it.skip : it;
-    
+
+    const testFn = it;
+
     testFn(testName, () => {
       const doc = parseHTML(input);
-      const actual = serializeToHtml5lib(doc, { skipImplicitDoctype: !hasDoctype });
+      const actual = serializeToHtml5lib(doc, {
+        skipImplicitDoctype: !hasDoctype,
+      });
       expect(actual).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-math.test.ts

@@ -1,18 +1,20 @@
-import { readFileSync } from 'fs';
-import { parse } from '../src/index.ts';
+import { readFileSync } from "fs";
+import { parse } from "../src/index.ts";
 
-describe('Tree Construction Math Tests', () => {
-  const content = readFileSync('tests/html5lib-data/tree-construction/math.dat', 'utf8');
-  const tests = content.split('#data\n').slice(1);
+describe("Tree Construction Math Tests", () => {
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/math.dat",
+    "utf8",
+  );
+  const tests = content.split("#data\n").slice(1);
 
   tests.forEach((test, index) => {
-    const parts = test.split('#document\n');
+    const parts = test.split("#document\n");
     const input = parts[0].trim();
-    const expected = parts[1]?.split('#errors\n')[0]?.trim() || '';
 
-    it.skip(`Math test ${index + 1}`, () => {
+    it(`Math test ${index + 1}`, () => {
       const doc = parse(input);
       expect(doc).toBeDefined();
     });
   });
-});
+});

package/tests/tree-construction-namespace-sensitivity.test.ts

@@ -1,18 +1,20 @@
-import { readFileSync } from 'fs';
-import { parse } from '../src/index.ts';
+import { readFileSync } from "fs";
+import { parse } from "../src/index.ts";
 
-describe('Tree Construction NamespaceSensitivity Tests', () => {
-  const content = readFileSync('tests/html5lib-data/tree-construction/namespace-sensitivity.dat', 'utf8');
-  const tests = content.split('#data\n').slice(1);
+describe("Tree Construction NamespaceSensitivity Tests", () => {
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/namespace-sensitivity.dat",
+    "utf8",
+  );
+  const tests = content.split("#data\n").slice(1);
 
   tests.forEach((test, index) => {
-    const parts = test.split('#document\n');
+    const parts = test.split("#document\n");
     const input = parts[0].trim();
-    const expected = parts[1]?.split('#errors\n')[0]?.trim() || '';
 
-    it.skip(`NamespaceSensitivity test ${index + 1}`, () => {
+    it(`NamespaceSensitivity test ${index + 1}`, () => {
       const doc = parse(input);
       expect(doc).toBeDefined();
     });
   });
-});
+});

package/tests/tree-construction-noscript01.test.ts

@@ -1,18 +1,20 @@
-import { readFileSync } from 'fs';
-import { parse } from '../src/index.ts';
+import { readFileSync } from "fs";
+import { parse } from "../src/index.ts";
 
-describe('Tree Construction Noscript01 Tests', () => {
-  const content = readFileSync('tests/html5lib-data/tree-construction/noscript01.dat', 'utf8');
-  const tests = content.split('#data\n').slice(1);
+describe("Tree Construction Noscript01 Tests", () => {
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/noscript01.dat",
+    "utf8",
+  );
+  const tests = content.split("#data\n").slice(1);
 
   tests.forEach((test, index) => {
-    const parts = test.split('#document\n');
+    const parts = test.split("#document\n");
     const input = parts[0].trim();
-    const expected = parts[1]?.split('#errors\n')[0]?.trim() || '';
 
-    it.skip(`Noscript01 test ${index + 1}`, () => {
+    it(`Noscript01 test ${index + 1}`, () => {
       const doc = parse(input);
       expect(doc).toBeDefined();
     });
   });
-});
+});

package/tests/tree-construction-ruby.test.ts

@@ -3,13 +3,15 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Ruby Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/ruby.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/ruby.dat",
+    "utf8",
+  );
   const sections = content.split(/^#data$/gm).slice(1);
 
   for (const section of sections) {
-    const [data, document] = section.split(/^#document$/gm);
+    const [data] = section.split(/^#document$/gm);
     const input = data.trim();
-    const expected = document.trim();
 
     it(`Ruby test: ${input.slice(0, 50)}${input.length > 50 ? "..." : ""}`, () => {
       const doc = parse(input);
@@ -18,4 +20,4 @@ describe("Tree Construction Ruby Tests", () => {
       // expect(serialize(doc)).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-scriptdata01.test.ts

@@ -3,13 +3,15 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Scriptdata01 Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/scriptdata01.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/scriptdata01.dat",
+    "utf8",
+  );
   const sections = content.split(/^#data$/gm).slice(1);
 
   for (const section of sections) {
-    const [data, document] = section.split(/^#document$/gm);
+    const [data] = section.split(/^#document$/gm);
     const input = data.trim();
-    const expected = document.trim();
 
     it(`Scriptdata01 test: ${input.slice(0, 50)}${input.length > 50 ? "..." : ""}`, () => {
       const doc = parse(input);
@@ -18,4 +20,4 @@ describe("Tree Construction Scriptdata01 Tests", () => {
       // expect(serialize(doc)).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-svg.test.ts

@@ -3,13 +3,15 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction SVG Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/svg.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/svg.dat",
+    "utf8",
+  );
   const sections = content.split(/^#data$/gm).slice(1);
 
   for (const section of sections) {
-    const [data, document] = section.split(/^#document$/gm);
+    const [data] = section.split(/^#document$/gm);
     const input = data.trim();
-    const expected = document.trim();
 
     it(`SVG test: ${input.slice(0, 50)}${input.length > 50 ? "..." : ""}`, () => {
       const doc = parse(input);
@@ -18,4 +20,4 @@ describe("Tree Construction SVG Tests", () => {
       // expect(serialize(doc)).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-template.test.ts

@@ -3,13 +3,15 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Template Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/template.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/template.dat",
+    "utf8",
+  );
   const sections = content.split(/^#data$/gm).slice(1);
 
   for (const section of sections) {
-    const [data, document] = section.split(/^#document$/gm);
+    const [data] = section.split(/^#document$/gm);
     const input = data.trim();
-    const expected = document.trim();
 
     it(`Template test: ${input.slice(0, 50)}${input.length > 50 ? "..." : ""}`, () => {
       const doc = parse(input);
@@ -18,4 +20,4 @@ describe("Tree Construction Template Tests", () => {
       // expect(serialize(doc)).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-tests10.test.ts

@@ -3,13 +3,15 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests10 Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/tests10.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/tests10.dat",
+    "utf8",
+  );
   const sections = content.split(/^#data$/gm).slice(1);
 
   for (const section of sections) {
-    const [data, document] = section.split(/^#document$/gm);
+    const [data] = section.split(/^#document$/gm);
     const input = data.trim();
-    const expected = document.trim();
 
     it(`Tests10 test: ${input.slice(0, 50)}${input.length > 50 ? "..." : ""}`, () => {
       const doc = parse(input);
@@ -18,4 +20,4 @@ describe("Tree Construction Tests10 Tests", () => {
       // expect(serialize(doc)).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-tests11.test.ts

@@ -3,13 +3,15 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests11 Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/tests11.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/tests11.dat",
+    "utf8",
+  );
   const sections = content.split(/^#data$/gm).slice(1);
 
   for (const section of sections) {
-    const [data, document] = section.split(/^#document$/gm);
+    const [data] = section.split(/^#document$/gm);
     const input = data.trim();
-    const expected = document.trim();
 
     it(`Tests11 test: ${input.slice(0, 50)}${input.length > 50 ? "..." : ""}`, () => {
       const doc = parse(input);
@@ -18,4 +20,4 @@ describe("Tree Construction Tests11 Tests", () => {
       // expect(serialize(doc)).toBe(expected);
     });
   }
-});
+});

package/tests/tree-construction-tests20.test.ts

@@ -2,17 +2,20 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests20 Tests", () => {
-  const data = readFileSync("tests/html5lib-data/tree-construction/tests20.dat", "utf8");
+  const data = readFileSync(
+    "tests/html5lib-data/tree-construction/tests20.dat",
+    "utf8",
+  );
   const tests = data.split("#data\n").slice(1);
 
   for (const test of tests) {
-    const [input, expected] = test.split("#document\n");
+    const [input] = test.split("#document\n");
     const title = input.trim().split("\n")[0] || "Unnamed test";
     const html = input.trim();
 
-    it.skip(title, () => {
+    it(title, () => {
       const doc = parse(html);
       expect(doc).toBeDefined();
     });
   }
-});
+});

package/tests/tree-construction-tests21.test.ts

@@ -2,17 +2,20 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests21 Tests", () => {
-  const data = readFileSync("tests/html5lib-data/tree-construction/tests21.dat", "utf8");
+  const data = readFileSync(
+    "tests/html5lib-data/tree-construction/tests21.dat",
+    "utf8",
+  );
   const tests = data.split("#data\n").slice(1);
 
   for (const test of tests) {
-    const [input, expected] = test.split("#document\n");
+    const [input] = test.split("#document\n");
     const title = input.trim().split("\n")[0] || "Unnamed test";
     const html = input.trim();
 
-    it.skip(title, () => {
+    it(title, () => {
       const doc = parse(html);
       expect(doc).toBeDefined();
     });
   }
-});
+});

package/tests/tree-construction-tests23.test.ts

@@ -2,17 +2,20 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests23 Tests", () => {
-  const data = readFileSync("tests/html5lib-data/tree-construction/tests23.dat", "utf8");
+  const data = readFileSync(
+    "tests/html5lib-data/tree-construction/tests23.dat",
+    "utf8",
+  );
   const tests = data.split("#data\n").slice(1);
 
   for (const test of tests) {
-    const [input, expected] = test.split("#document\n");
+    const [input] = test.split("#document\n");
     const title = input.trim().split("\n")[0] || "Unnamed test";
     const html = input.trim();
 
-    it.skip(title, () => {
+    it(title, () => {
       const doc = parse(html);
       expect(doc).toBeDefined();
     });
   }
-});
+});

package/tests/tree-construction-tests24.test.ts

@@ -2,17 +2,20 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests24 Tests", () => {
-  const data = readFileSync("tests/html5lib-data/tree-construction/tests24.dat", "utf8");
+  const data = readFileSync(
+    "tests/html5lib-data/tree-construction/tests24.dat",
+    "utf8",
+  );
   const tests = data.split("#data\n").slice(1);
 
   for (const test of tests) {
-    const [input, expected] = test.split("#document\n");
+    const [input] = test.split("#document\n");
     const title = input.trim().split("\n")[0] || "Unnamed test";
     const html = input.trim();
 
-    it.skip(title, () => {
+    it(title, () => {
       const doc = parse(html);
       expect(doc).toBeDefined();
     });
   }
-});
+});

package/tests/tree-construction-tests5.test.ts

@@ -2,15 +2,16 @@ import { readFileSync } from "fs";
 import { parse } from "../src/index.ts";
 
 describe("Tree Construction Tests5 Tests", () => {
-  const content = readFileSync("tests/html5lib-data/tree-construction/tests5.dat", "utf8");
+  const content = readFileSync(
+    "tests/html5lib-data/tree-construction/tests5.dat",
+    "utf8",
+  );
   const sections = content.split("#data\n");
 
   for (let i = 1; i < sections.length; i++) {
     const section = sections[i];
-    const [dataPart, documentPart] = section.split("#document\n");
+    const [dataPart] = section.split("#document\n");
     const data = dataPart.trim();
-    const expectedDocument = documentPart ? documentPart.split("#errors\n")[0].trim() : "";
-    const errors = documentPart && documentPart.includes("#errors\n") ? documentPart.split("#errors\n")[1].trim() : "";
 
     it(`Tests5 test ${i}`, () => {
       const doc = parse(data);
@@ -18,4 +19,4 @@ describe("Tree Construction Tests5 Tests", () => {
       // TODO: Implement DOM serialization and comparison
     });
   }
-});
+});