style(parsers): apply biome formatting

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Author: waleedlatif1

apps/sim/lib/file-parsers/pdf-lines.test.ts

@@ -127,9 +127,7 @@ describe('joinLines', () => {
       const lines = paragraph(['we ship high-', 'quality builds'], 627.4)
       const words = collectWords(lines)
 
-      expect(joinLines(lines, { words, headingMarkers: false })).toBe(
-        'we ship high-quality builds'
-      )
+      expect(joinLines(lines, { words, headingMarkers: false })).toBe('we ship high-quality builds')
       expect(joinLines(lines, { headingMarkers: false })).toBe('we ship high-quality builds')
     })
 

apps/sim/scripts/parser-eval/bench-run.ts

@@ -26,21 +26,56 @@ for (const ext of readdirSync(filesRoot).sort()) {
     const bytes = readFileSync(path.join(dir, file))
     const started = performance.now()
     try {
-      const result = await parseBuffer(bytes, ext, { pdfTextMode: ext === 'pdf' ? 'complete' : undefined })
+      const result = await parseBuffer(bytes, ext, {
+        pdfTextMode: ext === 'pdf' ? 'complete' : undefined,
+      })
       const ms = performance.now() - started
       let chunkCount = -1
-      try { chunkCount = (await chunker.chunk(result.content)).length } catch {}
+      try {
+        chunkCount = (await chunker.chunk(result.content)).length
+      } catch {}
       const { html, sampledData, messages, headings, links, ...metadata } = result.metadata ?? {}
-      writeFileSync(path.join(OUT, `${label}.json`), JSON.stringify({ label, ext, file, bytes: bytes.length, ms, ok: true, content: result.content, metadata, chunkCount }))
+      writeFileSync(
+        path.join(OUT, `${label}.json`),
+        JSON.stringify({
+          label,
+          ext,
+          file,
+          bytes: bytes.length,
+          ms,
+          ok: true,
+          content: result.content,
+          metadata,
+          chunkCount,
+        })
+      )
       summary[ext].ok++
-      process.stdout.write(`ok   ${label} ${result.content.length}ch ${ms.toFixed(0)}ms${metadata.degraded ? ' DEGRADED' : ''}${metadata.truncated ? ' TRUNCATED' : ''}\n`)
+      process.stdout.write(
+        `ok   ${label} ${result.content.length}ch ${ms.toFixed(0)}ms${metadata.degraded ? ' DEGRADED' : ''}${metadata.truncated ? ' TRUNCATED' : ''}\n`
+      )
     } catch (error) {
       const ms = performance.now() - started
       const typed = error instanceof FileParserError
-      writeFileSync(path.join(OUT, `${label}.json`), JSON.stringify({ label, ext, file, bytes: bytes.length, ms, ok: false, typedError: typed, errorCode: typed ? error.code : undefined, errorName: (error as Error)?.name, error: String((error as Error)?.message ?? error).slice(0, 300) }))
+      writeFileSync(
+        path.join(OUT, `${label}.json`),
+        JSON.stringify({
+          label,
+          ext,
+          file,
+          bytes: bytes.length,
+          ms,
+          ok: false,
+          typedError: typed,
+          errorCode: typed ? error.code : undefined,
+          errorName: (error as Error)?.name,
+          error: String((error as Error)?.message ?? error).slice(0, 300),
+        })
+      )
       summary[ext].error++
       if (typed) summary[ext].typed++
-      process.stdout.write(`FAIL ${label} ${typed ? `typed:${error.code}` : `UNTYPED:${(error as Error)?.name}`} ${String((error as Error)?.message).slice(0, 80)}\n`)
+      process.stdout.write(
+        `FAIL ${label} ${typed ? `typed:${error.code}` : `UNTYPED:${(error as Error)?.name}`} ${String((error as Error)?.message).slice(0, 80)}\n`
+      )
     }
   }
 }

apps/sim/scripts/parser-eval/run-parsers.ts

@@ -13,19 +13,36 @@ const OUT = process.argv[2]
 const OUTPUTS = path.join(OUT, 'outputs')
 mkdirSync(OUTPUTS, { recursive: true })
 
-interface Entry { file: string; dir: string; doc: string; format: string; tier: 'A' | 'B'; absence: string[]; variant?: string; firstSheetOnly?: boolean }
+interface Entry {
+  file: string
+  dir: string
+  doc: string
+  format: string
+  tier: 'A' | 'B'
+  absence: string[]
+  variant?: string
+  firstSheetOnly?: boolean
+}
 
 const entries: Entry[] = []
 for (const m of ['manifest-a.json', 'manifest-sheets.json']) {
   const p = path.join(OUT, m)
-  if (existsSync(p)) for (const e of JSON.parse(readFileSync(p, 'utf8'))) entries.push({ ...e, dir: 'files' })
+  if (existsSync(p))
+    for (const e of JSON.parse(readFileSync(p, 'utf8'))) entries.push({ ...e, dir: 'files' })
 }
 const realDir = path.join(OUT, 'real')
 if (existsSync(realDir)) {
   for (const file of readdirSync(realDir).sort()) {
     const ext = path.extname(file).slice(1).toLowerCase()
     if (!ext) continue
-    entries.push({ file, dir: 'real', doc: file.replace(/\.[^.]+$/, ''), format: ext, tier: 'B', absence: [] })
+    entries.push({
+      file,
+      dir: 'real',
+      doc: file.replace(/\.[^.]+$/, ''),
+      format: ext,
+      tier: 'B',
+      absence: [],
+    })
   }
 }
 
@@ -38,10 +55,29 @@ const robustness: Array<{ name: string; ext: string; bytes: Buffer }> = [
   { name: 'truncated-pdf', ext: 'pdf', bytes: fixture('memo.pdf').subarray(0, 3000) },
   { name: 'pdf-bytes-labelled-docx', ext: 'docx', bytes: fixture('memo.pdf') },
   { name: 'docx-bytes-labelled-pdf', ext: 'pdf', bytes: fixture('memo.docx') },
-  { name: 'png-labelled-doc', ext: 'doc', bytes: Buffer.concat([Buffer.from([0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a]), Buffer.from(Array.from({ length: 4000 }, (_, i) => (i * 7919) % 256))]) },
-  { name: 'random-bytes-labelled-ppt', ext: 'ppt', bytes: Buffer.from(Array.from({ length: 50000 }, (_, i) => (i * 104729 + 17) % 256)) },
-  { name: 'latin1-txt', ext: 'txt', bytes: Buffer.from('Caf\xe9 r\xe9sum\xe9 na\xefve \xa3 42', 'latin1') },
-  { name: 'utf16-txt', ext: 'txt', bytes: Buffer.concat([Buffer.from([0xff, 0xfe]), Buffer.from('Hello UTF-16 world', 'utf16le')]) },
+  {
+    name: 'png-labelled-doc',
+    ext: 'doc',
+    bytes: Buffer.concat([
+      Buffer.from([0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a]),
+      Buffer.from(Array.from({ length: 4000 }, (_, i) => (i * 7919) % 256)),
+    ]),
+  },
+  {
+    name: 'random-bytes-labelled-ppt',
+    ext: 'ppt',
+    bytes: Buffer.from(Array.from({ length: 50000 }, (_, i) => (i * 104729 + 17) % 256)),
+  },
+  {
+    name: 'latin1-txt',
+    ext: 'txt',
+    bytes: Buffer.from('Caf\xe9 r\xe9sum\xe9 na\xefve \xa3 42', 'latin1'),
+  },
+  {
+    name: 'utf16-txt',
+    ext: 'txt',
+    bytes: Buffer.concat([Buffer.from([0xff, 0xfe]), Buffer.from('Hello UTF-16 world', 'utf16le')]),
+  },
   { name: 'html-labelled-txt', ext: 'txt', bytes: fixture('memo.html') },
   { name: 'docx-labelled-xlsx', ext: 'xlsx', bytes: fixture('memo.docx') },
   { name: 'csv-labelled-xlsx', ext: 'xlsx', bytes: fixture('sheet-employees.csv') },
@@ -55,27 +91,70 @@ const chunker = new TextChunker({ chunkSize: 1024, chunkOverlap: 200, minCharact
 async function runOne(label: string, ext: string, bytes: Buffer, meta: Record<string, unknown>) {
   const started = performance.now()
   try {
-    const result = await parseBuffer(bytes, ext, { pdfTextMode: ext === 'pdf' ? 'complete' : undefined })
+    const result = await parseBuffer(bytes, ext, {
+      pdfTextMode: ext === 'pdf' ? 'complete' : undefined,
+    })
     const ms = performance.now() - started
     let chunks: string[] = []
-    try { chunks = (await chunker.chunk(result.content)).map((c) => c.text) } catch (e) { chunks = [`CHUNK_ERROR ${String(e)}`] }
+    try {
+      chunks = (await chunker.chunk(result.content)).map((c) => c.text)
+    } catch (e) {
+      chunks = [`CHUNK_ERROR ${String(e)}`]
+    }
     const { html, sampledData, messages, ...metadata } = result.metadata ?? {}
-    const record = { label, ext, bytes: bytes.length, ms, ok: true, content: result.content, metadata: { ...metadata, messageCount: Array.isArray(messages) ? messages.length : 0 }, chunks, ...meta }
+    const record = {
+      label,
+      ext,
+      bytes: bytes.length,
+      ms,
+      ok: true,
+      content: result.content,
+      metadata: { ...metadata, messageCount: Array.isArray(messages) ? messages.length : 0 },
+      chunks,
+      ...meta,
+    }
     writeFileSync(path.join(OUTPUTS, `${label}.json`), JSON.stringify(record, null, 1))
-    results.push({ ...record, content: undefined, chunks: undefined, contentLength: result.content.length, chunkCount: chunks.length })
-    process.stdout.write(`ok   ${label} ${result.content.length}ch ${ms.toFixed(0)}ms ${metadata.degraded ? 'DEGRADED' : ''} ${metadata.truncated ? 'TRUNCATED' : ''}\n`)
+    results.push({
+      ...record,
+      content: undefined,
+      chunks: undefined,
+      contentLength: result.content.length,
+      chunkCount: chunks.length,
+    })
+    process.stdout.write(
+      `ok   ${label} ${result.content.length}ch ${ms.toFixed(0)}ms ${metadata.degraded ? 'DEGRADED' : ''} ${metadata.truncated ? 'TRUNCATED' : ''}\n`
+    )
   } catch (error) {
     const ms = performance.now() - started
     const typed = error instanceof FileParserError
-    const record = { label, ext, bytes: bytes.length, ms, ok: false, typedError: typed, errorCode: typed ? (error as FileParserError).code : undefined, error: String((error as Error)?.message ?? error), ...meta }
+    const record = {
+      label,
+      ext,
+      bytes: bytes.length,
+      ms,
+      ok: false,
+      typedError: typed,
+      errorCode: typed ? (error as FileParserError).code : undefined,
+      error: String((error as Error)?.message ?? error),
+      ...meta,
+    }
     writeFileSync(path.join(OUTPUTS, `${label}.json`), JSON.stringify(record, null, 1))
     results.push(record)
-    process.stdout.write(`FAIL ${label} ${typed ? `typed:${record.errorCode}` : 'UNTYPED'} ${record.error.slice(0, 100)}\n`)
+    process.stdout.write(
+      `FAIL ${label} ${typed ? `typed:${record.errorCode}` : 'UNTYPED'} ${record.error.slice(0, 100)}\n`
+    )
   }
 }
 
 for (const e of entries) {
-  await runOne(e.file, e.format, readFileSync(path.join(OUT, e.dir, e.file)), { doc: e.doc, format: e.format, tier: e.tier, absence: e.absence, variant: e.variant, firstSheetOnly: e.firstSheetOnly })
+  await runOne(e.file, e.format, readFileSync(path.join(OUT, e.dir, e.file)), {
+    doc: e.doc,
+    format: e.format,
+    tier: e.tier,
+    absence: e.absence,
+    variant: e.variant,
+    firstSheetOnly: e.firstSheetOnly,
+  })
 }
 for (const r of robustness) await runOne(`robust__${r.name}`, r.ext, r.bytes, { tier: 'R' })