Add location information to parseFrontMatter() (#17948)

src/language-css/clean.js

@@ -1,3 +1,4 @@
+import cleanFrontMatter from "../utils/front-matter/clean.js";
 import isFrontMatter from "../utils/front-matter/is-front-matter.js";
 
 const ignoredProperties = new Set([
@@ -12,9 +13,7 @@ const ignoredProperties = new Set([
 ]);
 
 function clean(original, cloned, parent) {
-  if (isFrontMatter(original) && original.language === "yaml") {
-    delete cloned.value;
-  }
+  cleanFrontMatter(original, cloned);
 
   if (
     original.type === "css-comment" &&

src/language-css/embed.js

@@ -1,10 +1,11 @@
 import { hardline } from "../document/builders.js";
+import isFrontMatter from "../utils/front-matter/is-front-matter.js";
 import printFrontMatter from "../utils/front-matter/print.js";
 
 function embed(path) {
   const { node } = path;
 
-  if (node.type === "front-matter") {
+  if (isFrontMatter(node) && node.language === "yaml") {
     return async (textToDoc) => {
       const doc = await printFrontMatter(node, textToDoc);
       return doc ? [doc, hardline] : undefined;

src/language-css/parser-postcss.js

@@ -377,14 +377,12 @@ function parseNestedCSS(node, options) {
 }
 
 function parseWithParser(parse, text, options) {
-  const parsed = parseFrontMatter(text);
-  const { frontMatter } = parsed;
-  text = parsed.content;
+  const { frontMatter, content: textToParse } = parseFrontMatter(text);
 
   let result;
 
   try {
-    result = parse(text, {
+    result = parse(textToParse, {
       // Prevent file access https://github.com/postcss/postcss/blob/4f4e2932fc97e2c117e1a4b15f0272ed551ed59d/lib/previous-map.js#L18
       map: false,
     });
@@ -407,11 +405,13 @@ function parseWithParser(parse, text, options) {
   calculateLoc(result, text);
 
   if (frontMatter) {
-    frontMatter.source = {
-      startOffset: 0,
-      endOffset: frontMatter.raw.length,
+    result.frontMatter = {
+      ...frontMatter,
+      source: {
+        startOffset: frontMatter.start.index,
+        endOffset: frontMatter.end.index,
+      },
     };
-    result.frontMatter = frontMatter;
   }
 
   return result;

src/language-html/clean.js

@@ -1,4 +1,4 @@
-import isFrontMatter from "../utils/front-matter/is-front-matter.js";
+import cleanFrontMatter from "../utils/front-matter/clean.js";
 
 const ignoredProperties = new Set([
   "sourceSpan",
@@ -28,13 +28,10 @@ function clean(original, cloned) {
     return null;
   }
 
+  cleanFrontMatter(original, cloned);
+
   // may be formatted by multiparser
-  if (
-    isFrontMatter(original) ||
-    original.type === "yaml" ||
-    original.type === "toml"
-  ) {
-    delete cloned.raw;
+  if (original.type === "yaml") {
     delete cloned.value;
   }
 

src/language-html/embed.js

@@ -142,7 +142,10 @@ function embed(path, options) {
       return printAttribute(path, options);
 
     case "front-matter":
-      return (textToDoc) => printFrontMatter(node, textToDoc);
+      if (node.language === "yaml") {
+        return (textToDoc) => printFrontMatter(node, textToDoc);
+      }
+      break;
 
     case "angularControlFlowBlockParameters":
       if (!embeddedAngularControlFlowBlocks.has(path.parent.name)) {

src/language-html/parser-html.js

@@ -21,7 +21,7 @@ import HTML_TAGS from "./utils/html-tags.evaluate.js";
 import isUnknownNamespace from "./utils/is-unknown-namespace.js";
 
 /**
- * @import AngularHtmlParser, {ParseOptions as AngularHtmlParserParseOptions} from "angular-html-parser"
+ * @import {ParseOptions as AngularHtmlParserParseOptions} from "angular-html-parser"
  * @import {Node as AstNode, Attribute, Element} from "angular-html-parser/lib/compiler/src/ml_parser/ast.js"
  * @import {ParseTreeResult} from "angular-html-parser/lib/compiler/src/ml_parser/parser.js"
  */
@@ -348,7 +348,7 @@ function parse(
   options = {},
   shouldParseFrontMatter = true,
 ) {
-  const { frontMatter, content } = shouldParseFrontMatter
+  const { frontMatter, content: textToParse } = shouldParseFrontMatter
     ? parseFrontMatter(text)
     : { frontMatter: null, content: text };
 
@@ -358,15 +358,24 @@ function parse(
   const rawAst = {
     type: "root",
     sourceSpan: new ParseSourceSpan(start, end),
-    children: ngHtmlParser(content, parseOptions, options),
+    children: ngHtmlParser(textToParse, parseOptions, options),
   };
 
   if (frontMatter) {
-    const start = new ParseLocation(file, 0, 0, 0);
-    const end = start.moveBy(frontMatter.raw.length);
-    frontMatter.sourceSpan = new ParseSourceSpan(start, end);
+    const [start, end] = [frontMatter.start, frontMatter.end].map(
+      (location) =>
+        new ParseLocation(
+          file,
+          location.index,
+          location.line - 1,
+          location.column,
+        ),
+    );
     // @ts-expect-error -- not a real AstNode
-    rawAst.children.unshift(frontMatter);
+    rawAst.children.unshift({
+      ...frontMatter,
+      sourceSpan: new ParseSourceSpan(start, end),
+    });
   }
 
   const ast = new Node(rawAst);

src/language-markdown/clean.js

@@ -1,4 +1,5 @@
 import collapseWhiteSpace from "collapse-white-space";
+import cleanFrontMatter from "../utils/front-matter/clean.js";
 import isFrontMatter from "../utils/front-matter/is-front-matter.js";
 import { FORMAT_PRAGMAS } from "../utils/pragma/pragma.evaluate.js";
 import { startWithPragma } from "./pragma.js";
@@ -10,7 +11,6 @@ const ignoredProperties = new Set([
 function clean(original, cloned, parent) {
   // for codeblock
   if (
-    original.type === "front-matter" ||
     original.type === "code" ||
     original.type === "yaml" ||
     original.type === "import" ||
@@ -20,6 +20,8 @@ function clean(original, cloned, parent) {
     delete cloned.value;
   }
 
+  cleanFrontMatter(original, cloned);
+
   if (original.type === "list") {
     delete cloned.isAligned;
   }

src/language-markdown/printer-markdown.js

@@ -90,10 +90,7 @@ function genericPrint(path, options, print) {
 
   switch (node.type) {
     case "front-matter":
-      return options.originalText.slice(
-        node.position.start.offset,
-        node.position.end.offset,
-      );
+      return node.raw;
     case "root":
       /* c8 ignore next 3 */
       if (node.children.length === 0) {

src/language-markdown/unified-plugins/front-matter.js

@@ -11,10 +11,10 @@ const frontMatter = function () {
   proto.blockTokenizers.frontMatter = tokenizer;
 
   function tokenizer(eat, value) {
-    const parsed = parseFrontMatter(value);
+    const { frontMatter } = parseFrontMatter(value);
 
-    if (parsed.frontMatter) {
-      return eat(parsed.frontMatter.raw)(parsed.frontMatter);
+    if (frontMatter) {
+      return eat(frontMatter.raw)(frontMatter);
     }
   }
   tokenizer.onlyAtStart = true;

src/utils/front-matter/clean.js

@@ -0,0 +1,11 @@
+import isFrontMatter from "./is-front-matter.js";
+
+function clean(original, cloned) {
+  if (isFrontMatter(original) && original.language === "yaml") {
+    delete cloned.end;
+    delete cloned.raw;
+    delete cloned.value;
+  }
+}
+
+export default clean;

src/utils/front-matter/parse.js

@@ -1,5 +1,24 @@
 const DELIMITER_LENGTH = 3;
 
+/**
+@typedef {{index: number, line: number, column: number}} Position
+@typedef {{
+  type: "front-matter",
+  language: string,
+  explicitLanguage: string | null,
+  value: string,
+  startDelimiter: string,
+  endDelimiter: string,
+  raw: string,
+  start: Position,
+  end: Position,
+}} FrontMatter
+*/
+
+/**
+@param {string} text
+@returns {FrontMatter | undefined}
+*/
 function getFrontMatter(text) {
   const startDelimiter = text.slice(0, DELIMITER_LENGTH);
 
@@ -48,15 +67,29 @@ function getFrontMatter(text) {
   }
 
   const raw = text.slice(0, frontMatterEndIndex);
+  /** @type {string[]} */
+  let lines;
 
   return {
     type: "front-matter",
     language,
-    explicitLanguage,
+    explicitLanguage: explicitLanguage || null,
     value: text.slice(firstLineBreakIndex + 1, endDelimiterIndex),
     startDelimiter,
     endDelimiter: raw.slice(-DELIMITER_LENGTH),
     raw,
+    start: { line: 1, column: 0, index: 0 },
+    end: {
+      index: raw.length,
+      get line() {
+        lines ??= raw.split("\n");
+        return lines.length;
+      },
+      get column() {
+        lines ??= raw.split("\n");
+        return lines.at(-1).length;
+      },
+    },
   };
 }
 
@@ -67,11 +100,12 @@ function parse(text) {
     return { content: text };
   }
 
-  const { raw } = frontMatter;
-
   return {
     frontMatter,
-    content: raw.replaceAll(/[^\n]/gu, " ") + text.slice(raw.length),
+    get content() {
+      const { raw } = frontMatter;
+      return raw.replaceAll(/[^\n]/gu, " ") + text.slice(raw.length);
+    },
   };
 }
 

src/utils/front-matter/print.js

@@ -6,7 +6,7 @@ async function print(node, textToDoc) {
     const doc = value ? await textToDoc(value, { parser: "yaml" }) : "";
     return markAsRoot([
       node.startDelimiter,
-      node.explicitLanguage,
+      node.explicitLanguage ?? "",
       hardline,
       doc,
       doc ? hardline : "",