fmt

Author: enitrat

ingesters/src/utils/RecursiveMarkdownSplitter.ts

@@ -219,7 +219,9 @@ export class RecursiveMarkdownSplitter {
    * Active source becomes the first URL and applies from the end of the block
    * until the start of the next Sources block (or end of document).
    */
-  private parseSourceRanges(markdown: string): Array<{ start: number; end: number; url: string }> {
+  private parseSourceRanges(
+    markdown: string,
+  ): Array<{ start: number; end: number; url: string }> {
     const lines = markdown.split('\n');
     const ranges: Array<{ start: number; end: number; url: string }> = [];
 
@@ -247,7 +249,11 @@ export class RecursiveMarkdownSplitter {
     };
 
     // Locate all source blocks (start/end + first URL)
-    const blocks: Array<{ blockStartLine: number; blockEndLine: number; firstUrl?: string }> = [];
+    const blocks: Array<{
+      blockStartLine: number;
+      blockEndLine: number;
+      firstUrl?: string;
+    }> = [];
     for (let i = 0; i < lines.length; i++) {
       if (!isDashLine(lines[i]!)) continue;
       // Scan ahead for Sources: header within the dashed block
@@ -789,7 +795,8 @@ export class RecursiveMarkdownSplitter {
       if (i > 0 && this.options.overlap > 0) {
         const prevSegment = segments[i - 1]!;
         const desired = Math.max(
-          prevSegment.end - Math.min(this.options.overlap, prevSegment.end - prevSegment.start),
+          prevSegment.end -
+            Math.min(this.options.overlap, prevSegment.end - prevSegment.start),
           prevSegment.start,
         );
         chunkStartAbs = desired;

ingesters/src/utils/__tests__/RecursiveMarkdownSplitter.noStartInsideCodeBlock.test.ts

@@ -1,6 +1,8 @@
 import { RecursiveMarkdownSplitter } from '../RecursiveMarkdownSplitter';
 
-function getCodeBlockRanges(text: string): Array<{ start: number; end: number }> {
+function getCodeBlockRanges(
+  text: string,
+): Array<{ start: number; end: number }> {
   const ranges: Array<{ start: number; end: number }> = [];
   const re = /```[\s\S]*?```/g;
   let m: RegExpExecArray | null;
@@ -10,13 +12,19 @@ function getCodeBlockRanges(text: string): Array<{ start: number; end: number }>
   return ranges;
 }
 
-function isInside(pos: number, ranges: Array<{ start: number; end: number }>): boolean {
+function isInside(
+  pos: number,
+  ranges: Array<{ start: number; end: number }>,
+): boolean {
   return ranges.some((r) => pos > r.start && pos < r.end);
 }
 
 describe('RecursiveMarkdownSplitter - No chunk starts inside code block', () => {
   it('ensures chunk starts are never within fenced code blocks even with overlap', () => {
-    const longCode = Array.from({ length: 60 }, (_, i) => `line ${i} of code`).join('\n');
+    const longCode = Array.from(
+      { length: 60 },
+      (_, i) => `line ${i} of code`,
+    ).join('\n');
     const md = `# Section One\n\nIntro paragraph text that will be part of the first section.\n\n\n## Subsection\n\nSome text before a large code block.\n\n
 \`\`\`cairo
 fn initializer(ref self: ContractState, owner: ContractAddress) {

ingesters/src/utils/__tests__/RecursiveMarkdownSplitter.sources.test.ts

@@ -39,7 +39,9 @@ More details here.`;
 
     // Find chunk under the first H1
     const introChunk = chunks.find((c) =>
-      c.content.includes('# The Cairo Book: Introduction and Learning Resources'),
+      c.content.includes(
+        '# The Cairo Book: Introduction and Learning Resources',
+      ),
     );
     expect(introChunk).toBeDefined();
     expect(introChunk!.meta.sourceLink).toBe(