feat: add rawContentTags options (#32)

* feat: add customTags options

* refactor

* add docs

Author: yeonjuan

README.md

@@ -68,6 +68,7 @@ parse(html: string, options?: Options): ParseResult;
 - `html`: HTML string to parse.
 - `options (optional)`
   - `tokenAdapter`: The adapter option for changing tokens information.
+  - `rawContentTags` (string[]) : Specifies tag names whose child contents should be treated as raw text, meaning the parser will not interpret characters like < and > as HTML syntax inside these tags.
 
 **Returns**
 

src/constants/tokenizer-context-types.ts

@@ -18,4 +18,5 @@ export enum TokenizerContextTypes {
   CommentOpen = "CommentOpen",
   CommentContent = "CommentContent",
   CommentClose = "CommentClose",
+  CustomTagRawContent = "CustomTagRawContent",
 }

src/parser/parse.ts

@@ -7,7 +7,7 @@ import { Options } from "../types/parse";
 
 export function parse(html: string, options?: Options): ParseResult {
   const tokenAdapter = (options && options.tokenAdapter) || defaultTokenAdapter;
-  const { tokens } = tokenize(html, tokenAdapter, options?.templateInfos);
+  const { tokens } = tokenize(html, tokenAdapter, options);
   const { ast } = constructTree(tokens);
   return {
     ast: clearParent(ast),

src/tokenizer/__tests__/__input__/custom-tag-raw-content.html

@@ -0,0 +1,17 @@
+<markdown attr="1">
+# Hello, world!
+
+```cpp{4-6,9}
+#include <iostream>
+
+class Example {
+    Example() {
+        std::cout << "Hello, world!" << std::endl;
+    }
+
+    Example(std::string name) {
+        std::cout << "Hello, " << name << std::endl;
+    }
+};
+```
+</markdown>

src/tokenizer/__tests__/__output__/custom-tag-raw-content.ts

@@ -0,0 +1,161 @@
+import { TokenTypes } from "../../../constants";
+import { AnyToken } from "../../../types";
+
+const OUTPUT: AnyToken[] = [
+  {
+    type: TokenTypes.OpenTagStart,
+    value: "<markdown",
+    range: [0, 9],
+    loc: {
+      start: {
+        column: 0,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 9,
+      },
+    },
+  },
+  {
+    type: TokenTypes.AttributeKey,
+    value: "attr",
+    loc: {
+      start: {
+        column: 10,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 14,
+      },
+    },
+    range: [10, 14],
+    parts: [],
+  },
+  {
+    type: TokenTypes.AttributeAssignment,
+    value: "=",
+    range: [14, 15],
+    loc: {
+      start: {
+        column: 14,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 15,
+      },
+    },
+  },
+  {
+    type: TokenTypes.AttributeValueWrapperStart,
+    value: '"',
+    range: [15, 16],
+    loc: {
+      start: {
+        column: 15,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 16,
+      },
+    },
+  },
+  {
+    type: TokenTypes.AttributeValue,
+    value: "1",
+    range: [16, 17],
+    loc: {
+      start: {
+        column: 16,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 17,
+      },
+    },
+    parts: [],
+  },
+  {
+    type: TokenTypes.AttributeValueWrapperEnd,
+    value: '"',
+    range: [17, 18],
+    loc: {
+      start: {
+        column: 17,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 18,
+      },
+    },
+  },
+  {
+    type: TokenTypes.OpenTagEnd,
+    value: ">",
+    range: [18, 19],
+    loc: {
+      start: {
+        column: 18,
+        line: 1,
+      },
+      end: {
+        line: 1,
+        column: 19,
+      },
+    },
+  },
+  {
+    type: TokenTypes.Text,
+    value: `
+# Hello, world!
+
+\`\`\`cpp{4-6,9}
+#include <iostream>
+
+class Example {
+    Example() {
+        std::cout << "Hello, world!" << std::endl;
+    }
+
+    Example(std::string name) {
+        std::cout << "Hello, " << name << std::endl;
+    }
+};
+\`\`\`
+`,
+    range: [19, 260],
+    loc: {
+      start: {
+        column: 19,
+        line: 1,
+      },
+      end: {
+        line: 17,
+        column: 0,
+      },
+    },
+    parts: [],
+  },
+  {
+    type: TokenTypes.CloseTag,
+    value: "</markdown>",
+    range: [260, 271],
+    loc: {
+      start: {
+        column: 0,
+        line: 17,
+      },
+      end: {
+        line: 17,
+        column: 11,
+      },
+    },
+  },
+];
+
+export default OUTPUT;

src/tokenizer/__tests__/tokenize.spec.ts

@@ -1,6 +1,6 @@
 import * as fs from "fs";
 import * as path from "path";
-import { tokenize } from "../tokenize";
+import { tokenize, TokenizeOptions } from "../tokenize";
 import OPENING_CLOSING_TEXT from "./__output__/opening-closing-text";
 import NESTED_TAGS from "./__output__/nested-tags";
 import COMMENTS from "./__output__/comments";
@@ -30,7 +30,7 @@ import TEMPLATE_COMMENT from "./__output__/templates-comment";
 import TEMPLATE_SCRIPT_CONTENT from "./__output__/templates-script-content";
 import TEMPLATE_STYLE_CONTENT from "./__output__/templates-style-content";
 import TEMPLATE_CONTENT_END from "./__output__/templates-content-end";
-
+import CUSTOM_TAG_RAW_CONTENT from "./__output__/custom-tag-raw-content";
 import { defaultTokenAdapter } from "../../token-adapter";
 import { Range, TemplateInfo } from "../../types";
 
@@ -98,78 +98,107 @@ describe("tokenize", () => {
       "templates-attributes-key.html",
       TEMPLATE_ATTRIBUTES_KEY,
       null,
-      [[5, 11]] as Range[],
+      {
+        templateInfos: [[5, 11]] as Range[],
+      },
     ],
     [
       "Template Attributes Key (wrapper)",
       "templates-attributes-key.html",
       TEMPLATE_ATTRIBUTES_KEY_WRAPPER,
       null,
-      [
-        {
-          open: [5, 7],
-          close: [10, 11],
-        },
-      ] as TemplateInfo[],
+      {
+        templateInfos: [
+          {
+            open: [5, 7],
+            close: [10, 11],
+          },
+        ] as TemplateInfo[],
+      },
     ],
     [
       "Template Attributes Value Bare",
       "templates-attributes-value-bare.html",
       TEMPLATE_ATTRIBUTES_VALUE_BARE,
       null,
-      [[8, 13]] as Range[],
+      {
+        templateInfos: [[8, 13]] as Range[],
+      },
     ],
     [
       "Template Attributes Value Wrapped",
       "templates-attributes-value-wrapped.html",
       TEMPLATE_ATTRIBUTES_VALUE_WRAPPED,
       null,
-      [[9, 14]] as Range[],
+      {
+        templateInfos: [[9, 14]] as Range[],
+      },
     ],
     [
       "Template Attributes Value Wrapped 2",
       "templates-attributes-value-wrapped-2.html",
       TEMPLATE_ATTRIBUTES_VALUE_WRAPPED_2,
       null,
-      [
-        [16, 22],
-        [23, 31],
-      ] as Range[],
+      {
+        templateInfos: [
+          [16, 22],
+          [23, 31],
+        ] as Range[],
+      },
     ],
     [
       "Templates Data",
       "templates-data.html",
       TEMPLATE_DATA,
       null,
-      [[5, 16]] as Range[],
+      {
+        templateInfos: [[5, 16]] as Range[],
+      },
     ],
     [
       "Templates Comment",
       "templates-comment.html",
       TEMPLATE_COMMENT,
       null,
-      [[4, 14]] as Range[],
+      {
+        templateInfos: [[4, 14]] as Range[],
+      },
     ],
     [
       "Templates Script Content",
       "templates-script-content.html",
       TEMPLATE_SCRIPT_CONTENT,
       null,
-      [[8, 18]] as Range[],
+      {
+        templateInfos: [[8, 18]] as Range[],
+      },
     ],
     [
       "Templates Style Content",
       "templates-style-content.html",
       TEMPLATE_STYLE_CONTENT,
       null,
-      [[7, 17]] as Range[],
+      {
+        templateInfos: [[7, 17]] as Range[],
+      },
     ],
     [
       "Templates Content End",
       "templates-content-end.html",
       TEMPLATE_CONTENT_END,
       null,
-      [[0, 10]] as Range[],
+      {
+        templateInfos: [[0, 10]] as Range[],
+      },
+    ],
+    [
+      "Custom Tag Raw Content",
+      "custom-tag-raw-content.html",
+      CUSTOM_TAG_RAW_CONTENT,
+      null,
+      {
+        rawContentTags: ["markdown"],
+      },
     ],
   ])(
     "%s",
@@ -178,18 +207,14 @@ describe("tokenize", () => {
       input,
       output,
       process: null | ((html: string) => string) = null,
-      ranges: null | TemplateInfo[]
+      options: TokenizeOptions | null = null
     ) => {
       const inputPath = path.join(__dirname, "__input__", input);
       let html = fs.readFileSync(inputPath, "utf-8");
       if (process) {
         html = process(html);
       }
-      const { tokens } = tokenize(
-        html,
-        defaultTokenAdapter,
-        ranges ?? undefined
-      );
+      const { tokens } = tokenize(html, defaultTokenAdapter, options || {});
       expect(tokens).toEqual(output);
     }
   );