feat: Support xmp tag parsing (#1790)

Author: nati-elmaliach

src/Tokenizer.spec.ts

@@ -37,6 +37,9 @@ describe("Tokenizer", () => {
         it("for self-closing textarea tag", () => {
             expect(tokenize("<textarea /><div></div>")).toMatchSnapshot();
         });
+        it("for self-closing xmp tag", () => {
+            expect(tokenize("<xmp /><div></div>")).toMatchSnapshot();
+        });
     });
 
     describe("should support standard special tags", () => {
@@ -54,6 +57,9 @@ describe("Tokenizer", () => {
                 tokenize("<textarea></textarea><div></div>"),
             ).toMatchSnapshot();
         });
+        it("for normal xmp tag", () => {
+            expect(tokenize("<xmp></xmp><div></div>")).toMatchSnapshot();
+        });
     });
 
     describe("should treat html inside special tags as text", () => {
@@ -71,6 +77,9 @@ describe("Tokenizer", () => {
                 tokenize("<textarea><div></div></textarea>"),
             ).toMatchSnapshot();
         });
+        it("for div inside xmp tag", () => {
+            expect(tokenize("<xmp><div></div></xmp>")).toMatchSnapshot();
+        });
     });
 
     describe("should correctly mark attributes", () => {

src/Tokenizer.ts

@@ -138,6 +138,7 @@ const Sequences = {
     TextareaEnd: new Uint8Array([
         0x3c, 0x2f, 0x74, 0x65, 0x78, 0x74, 0x61, 0x72, 0x65, 0x61,
     ]), // `</textarea`
+    XmpEnd: new Uint8Array([0x3c, 0x2f, 0x78, 0x6d, 0x70]), // `</xmp`
 };
 
 export default class Tokenizer {
@@ -391,7 +392,10 @@ export default class Tokenizer {
                 this.state = State.InTagName;
             } else if (lower === Sequences.ScriptEnd[2]) {
                 this.state = State.BeforeSpecialS;
-            } else if (lower === Sequences.TitleEnd[2]) {
+            } else if (
+                lower === Sequences.TitleEnd[2] ||
+                lower === Sequences.XmpEnd[2]
+            ) {
                 this.state = State.BeforeSpecialT;
             } else {
                 this.state = State.InTagName;
@@ -593,13 +597,26 @@ export default class Tokenizer {
 
     private stateBeforeSpecialT(c: number): void {
         const lower = c | 0x20;
-        if (lower === Sequences.TitleEnd[3]) {
-            this.startSpecial(Sequences.TitleEnd, 4);
-        } else if (lower === Sequences.TextareaEnd[3]) {
-            this.startSpecial(Sequences.TextareaEnd, 4);
-        } else {
-            this.state = State.InTagName;
-            this.stateInTagName(c); // Consume the token again
+        switch (lower) {
+            case Sequences.TitleEnd[3]: {
+                this.startSpecial(Sequences.TitleEnd, 4);
+
+                break;
+            }
+            case Sequences.TextareaEnd[3]: {
+                this.startSpecial(Sequences.TextareaEnd, 4);
+
+                break;
+            }
+            case Sequences.XmpEnd[3]: {
+                this.startSpecial(Sequences.XmpEnd, 4);
+
+                break;
+            }
+            default: {
+                this.state = State.InTagName;
+                this.stateInTagName(c); // Consume the token again
+            }
         }
     }
 

src/__snapshots__/Tokenizer.spec.ts.snap

@@ -487,6 +487,37 @@ exports[`Tokenizer > should support self-closing special tags > for self-closing
 ]
 `;
 
+exports[`Tokenizer > should support self-closing special tags > for self-closing xmp tag 1`] = `
+[
+  [
+    "onopentagname",
+    1,
+    4,
+  ],
+  [
+    "onselfclosingtag",
+    6,
+  ],
+  [
+    "onopentagname",
+    8,
+    11,
+  ],
+  [
+    "onopentagend",
+    11,
+  ],
+  [
+    "onclosetag",
+    14,
+    17,
+  ],
+  [
+    "onend",
+  ],
+]
+`;
+
 exports[`Tokenizer > should support standard special tags > for normal script tag 1`] = `
 [
   [
@@ -631,6 +662,42 @@ exports[`Tokenizer > should support standard special tags > for normal textarea
 ]
 `;
 
+exports[`Tokenizer > should support standard special tags > for normal xmp tag 1`] = `
+[
+  [
+    "onopentagname",
+    1,
+    4,
+  ],
+  [
+    "onopentagend",
+    4,
+  ],
+  [
+    "onclosetag",
+    7,
+    10,
+  ],
+  [
+    "onopentagname",
+    12,
+    15,
+  ],
+  [
+    "onopentagend",
+    15,
+  ],
+  [
+    "onclosetag",
+    18,
+    21,
+  ],
+  [
+    "onend",
+  ],
+]
+`;
+
 exports[`Tokenizer > should treat html inside special tags as text > for div inside script tag 1`] = `
 [
   [
@@ -738,3 +805,30 @@ exports[`Tokenizer > should treat html inside special tags as text > for div ins
   ],
 ]
 `;
+
+exports[`Tokenizer > should treat html inside special tags as text > for div inside xmp tag 1`] = `
+[
+  [
+    "onopentagname",
+    1,
+    4,
+  ],
+  [
+    "onopentagend",
+    4,
+  ],
+  [
+    "ontext",
+    5,
+    16,
+  ],
+  [
+    "onclosetag",
+    18,
+    21,
+  ],
+  [
+    "onend",
+  ],
+]
+`;