feat(linter/plugins): implement SourceCode#lines property (#14290)

Add support for obtaining source text split into lines via `context.sourceCode.lines`.

Author: overlookmotel

apps/oxlint/src-js/plugins/source_code.ts

@@ -16,6 +16,9 @@ const require = createRequire(import.meta.url);
 
 const { max } = Math;
 
+// Pattern for splitting source text into lines
+const LINE_BREAK_PATTERN = /\r\n|[\r\n\u2028\u2029]/gu;
+
 // Text decoder, for decoding source text from buffer
 const textDecoder = new TextDecoder('utf-8', { ignoreBOM: true });
 
@@ -31,6 +34,10 @@ let sourceText: string | null = null;
 let sourceByteLen: number = 0;
 let ast: Program | null = null;
 
+// Lazily populated when `SOURCE_CODE.lines` is accessed.
+const lines: string[] = [],
+  lineStartOffsets: number[] = [];
+
 // Lazily populated when `SOURCE_CODE.visitorKeys` is accessed.
 let visitorKeys: { [key: string]: string[] } | null = null;
 
@@ -75,6 +82,35 @@ export function getAst(): Program {
   return ast;
 }
 
+/**
+ * Split source text into lines.
+ */
+function initLines(): void {
+  if (sourceText === null) initSourceText();
+
+  // This implementation is based on the one in ESLint.
+  // TODO: Investigate if using `String.prototype.matchAll` is faster.
+  // This comment is above ESLint's implementation:
+  /*
+   * Previously, this was implemented using a regex that
+   * matched a sequence of non-linebreak characters followed by a
+   * linebreak, then adding the lengths of the matches. However,
+   * this caused a catastrophic backtracking issue when the end
+   * of a file contained a large number of non-newline characters.
+   * To avoid this, the current implementation just matches newlines
+   * and uses match.index to get the correct line start indices.
+   */
+
+  lineStartOffsets.push(0);
+  let lastOffset = 0, offset, match;
+  while ((match = LINE_BREAK_PATTERN.exec(sourceText))) {
+    offset = match.index;
+    lines.push(sourceText.slice(lastOffset, offset));
+    lineStartOffsets.push(lastOffset = offset + match[0].length);
+  }
+  lines.push(sourceText.slice(lastOffset));
+}
+
 /**
  * Reset source after file has been linted, to free memory.
  *
@@ -89,6 +125,8 @@ export function resetSource(): void {
   buffer = null;
   sourceText = null;
   ast = null;
+  lines.length = 0;
+  lineStartOffsets.length = 0;
 }
 
 // `SourceCode` object.
@@ -136,7 +174,8 @@ export const SOURCE_CODE = Object.freeze({
 
   // Get source text as array of lines, split according to specification's definition of line breaks.
   get lines(): string[] {
-    throw new Error('`sourceCode.lines` not implemented yet'); // TODO
+    if (lines.length === 0) initLines();
+    return lines;
   },
 
   /**

apps/oxlint/test/fixtures/sourceCode/files/1.js

@@ -1 +1,4 @@
 let foo, bar;
+
+// x
+// y

apps/oxlint/test/fixtures/sourceCode/output.snap.md

@@ -4,44 +4,51 @@
 # stdout
 ```
   x source-code-plugin(create): create:
-  | text: "let foo, bar;\n"
-  | getText(): "let foo, bar;\n"
+  | text: "let foo, bar;\n\n// x\n// y\n"
+  | getText(): "let foo, bar;\n\n// x\n// y\n"
+  | lines: ["let foo, bar;","","// x","// y",""]
   | ast: "foo"
   | visitorKeys: left, right
    ,-[files/1.js:1:1]
  1 | let foo, bar;
    : ^
+ 2 | 
    `----
 
   x source-code-plugin(create-once): after:
-  | source: "let foo, bar;\n"
+  | source: "let foo, bar;\n\n// x\n// y\n"
    ,-[files/1.js:1:1]
  1 | let foo, bar;
    : ^
+ 2 | 
    `----
 
   x source-code-plugin(create-once): before:
-  | text: "let foo, bar;\n"
-  | getText(): "let foo, bar;\n"
+  | text: "let foo, bar;\n\n// x\n// y\n"
+  | getText(): "let foo, bar;\n\n// x\n// y\n"
+  | lines: ["let foo, bar;","","// x","// y",""]
   | ast: "foo"
   | visitorKeys: left, right
    ,-[files/1.js:1:1]
  1 | let foo, bar;
    : ^
+ 2 | 
    `----
 
   x source-code-plugin(create): var decl:
   | source: "let foo, bar;"
    ,-[files/1.js:1:1]
  1 | let foo, bar;
    : ^^^^^^^^^^^^^
+ 2 | 
    `----
 
   x source-code-plugin(create-once): var decl:
   | source: "let foo, bar;"
    ,-[files/1.js:1:1]
  1 | let foo, bar;
    : ^^^^^^^^^^^^^
+ 2 | 
    `----
 
   x source-code-plugin(create): ident "foo":
@@ -52,6 +59,7 @@
    ,-[files/1.js:1:5]
  1 | let foo, bar;
    :     ^^^
+ 2 | 
    `----
 
   x source-code-plugin(create-once): ident "foo":
@@ -62,6 +70,7 @@
    ,-[files/1.js:1:5]
  1 | let foo, bar;
    :     ^^^
+ 2 | 
    `----
 
   x source-code-plugin(create): ident "bar":
@@ -72,6 +81,7 @@
    ,-[files/1.js:1:10]
  1 | let foo, bar;
    :          ^^^
+ 2 | 
    `----
 
   x source-code-plugin(create-once): ident "bar":
@@ -82,11 +92,13 @@
    ,-[files/1.js:1:10]
  1 | let foo, bar;
    :          ^^^
+ 2 | 
    `----
 
   x source-code-plugin(create): create:
   | text: "let qux;\n"
   | getText(): "let qux;\n"
+  | lines: ["let qux;",""]
   | ast: "qux"
   | visitorKeys: left, right
    ,-[files/2.js:1:1]
@@ -104,6 +116,7 @@
   x source-code-plugin(create-once): before:
   | text: "let qux;\n"
   | getText(): "let qux;\n"
+  | lines: ["let qux;",""]
   | ast: "qux"
   | visitorKeys: left, right
    ,-[files/2.js:1:1]

apps/oxlint/test/fixtures/sourceCode/plugin.ts

@@ -13,6 +13,7 @@ const createRule: Rule = {
       message: 'create:\n' +
         `text: ${JSON.stringify(context.sourceCode.text)}\n` +
         `getText(): ${JSON.stringify(context.sourceCode.getText())}\n` +
+        `lines: ${JSON.stringify(context.sourceCode.lines)}\n` +
         // @ts-ignore
         `ast: "${ast.body[0].declarations[0].id.name}"\n` +
         `visitorKeys: ${context.sourceCode.visitorKeys.BinaryExpression.join(', ')}`,
@@ -55,6 +56,7 @@ const createOnceRule: Rule = {
           message: 'before:\n' +
             `text: ${JSON.stringify(context.sourceCode.text)}\n` +
             `getText(): ${JSON.stringify(context.sourceCode.getText())}\n` +
+            `lines: ${JSON.stringify(context.sourceCode.lines)}\n` +
             // @ts-ignore
             `ast: "${ast.body[0].declarations[0].id.name}"\n` +
             `visitorKeys: ${context.sourceCode.visitorKeys.BinaryExpression.join(', ')}`,