PARSERGet Started
guidesparsersparser settingsParsing Configuration

PARSER GUIDES

Parsing Configuration

ParsersParser Settings

Configure document processing, PDF preparation, and extraction quality

In Parser Settings -> Document Processing, you can control which documents are accepted, how PDFs are prepared, and how extraction quality is handled.

The section is organized into:

  • Pre-processing - File type limits, page limits, document type rules, password-protected PDFs, and PDF splitting
  • Processing accuracy - Confidence scores, Markdown parsing, advanced table extraction, and data extraction mode
  • Post-processing - Document naming after extraction

Where to find it

1. Open your parser and go to Settings -> Parser Settings.

2. Scroll to Document Processing.

Screenshot placeholderguide-guides-parsers-parser-settings-parsing-configuration-01-parser-settings-processing-controls.webpRequired PARSER view: Parsing Configuration sectionFuture alt text: Parsing Configuration section

Pages to be parsed

By default, PARSER processes all pages in a document. You can limit which pages are processed (e.g. only odd or even pages, or specific page ranges). This applies to PDFs and email conversations. For email, page 1 is the newest visible message, and each message uses at least one page. A message longer than 5,000 visible characters continues onto additional pages. PARSER selects pages before extraction, and per-page charges apply only to the selected pages.

To set: In the Pages to be parsed section, turn the toggle on and choose Only odd pages, Only even pages, or Page ranges and enter a custom range (e.g. 1,2,8 or 4,7,12-16 or 2n-1). Then click Save Parser.

See Pages to be parsed for full details and examples.

Document type rules

Use this setting when files may enter the parser from email, uploads, upload pages, embeds, API imports, URL imports, or Zapier, but some document types should not reach extraction.

Choose how the rule should work:

  • Process only selected types processes matches and skips everything else.
  • Skip selected types skips matches and processes everything else.

Add document type labels such as Invoice, Purchase Order, Form W-2, or Form 1099. Skipped documents stay visible in the Document Inbox for review. Use Process anyway only when a skipped file should be processed once without these rules. Regular reprocess uses the current parser settings and checks the document type again.

Cost: 0.1 credit per document checked. This setting is available on Pro, Scale, and higher plans.

See Document Type Rules for setup steps and examples.

Document naming

Use Auto-rename documents when uploaded files have generic names and you want processed documents to include useful extracted values.

For example, a file named scan-001.pdf can become Invoice INV-1042 - Acme.pdf after PARSER extracts the invoice number and vendor name.

To set: Turn on Auto-rename documents, enter a Default template, use the variable button to add fields, then click Save document naming.

You can also open Advanced rules to use different templates based on conditions, such as the folder the document is assigned to.

See Document Naming for setup steps and examples.

Password-protected PDFs

Use Password-protected PDFs when incoming PDF files require a password before PARSER can read them.

Create rules based on the file name, folder, or a default password for the parser. When a matching password-protected PDF is received, PARSER unlocks it before extraction.

To set: In Password-protected PDFs, click Add rule, choose the match rule, enter the password, and save the rule.

See Password-protected PDFs for setup steps and examples.

Document Types

By default, PARSER processes all supported file types. You can restrict it to specific types.

Supported types:

  • PDF
  • Images: PNG, JPEG/JPG
  • Office: Word (.docx), Excel (.xlsx), PowerPoint (.pptx)
  • Text: Markdown, plain text, CSV, HTML, JSON

When to use:

  • PDFs only: If you only process PDFs and want to ignore images or Office files
  • Specific formats: If you only want to process certain file types (e.g. only PDFs and images)

To set: Use the Document Types dropdown to select which file types to process, then click Save Parser.

Calculate confidence scores

If enabled, PARSER shows how confident it is in each extracted value (e.g. as a percentage or color). This helps you spot values that might be wrong.

Cost: +1 credit per page on top of standard processing (2 credits per page total). Confidence scores may also increase compute time slightly.

When to use:

  • When you want to prioritize review on low-confidence documents
  • When you need to spot potential errors before export

To enable: Turn on Calculate confidence scores for extracted fields and click Save Parser.

Enable advanced table data extraction

This setting helps PARSER read larger or more complex tables more accurately.

When to use:

  • Your documents include long line-item tables (for example 100+ rows)
  • Table rows or columns are sometimes missed or misaligned
  • You process documents with mixed table layouts from different vendors

Options:

  • Auto: PARSER decides when advanced table extraction is needed
  • Enabled: Always use advanced table extraction
  • Disabled: Turn advanced table extraction off

To set: In Enable advanced table data extraction, choose Auto, Enabled, or Disabled, then click Save Parser.

See Advanced table extraction for detailed guidance and best practices.

Enable Markdown parsing

Turn on Enable Markdown parsing to create a Markdown version of each processed document. PARSER completes structured data extraction first. It then creates and stores the Markdown file before post-processing and exports run.

After processing finishes, open the document and select the Markdown tab. You can view the Markdown, view HTML tables, download the .md file, or download detected tables.

Cost: +1 credit per processed page. This cost also applies when your team uses its own AI provider key for structured extraction.

For merged email attachments, PARSER parses each included attachment in its stored order and combines the results into one Markdown file. The file includes a heading with the filename before each attachment. Email subject and body text are not added to this Markdown file.

If Pages to be parsed selects only part of a PDF, the Markdown file contains the same selected pages. If Markdown parsing fails, the complete document run fails and PARSER does not save the new extraction result or charge processing credits.

Reprocessing can reuse a valid stored Markdown file, but it still uses the extra Markdown credit for each processed page. If the stored file is missing, PARSER creates it again. Existing documents do not get Markdown until you process or reprocess them with this setting enabled.

This version uses Retab for Markdown parsing. Gemini and OpenAI Markdown parsing are not available yet.

The public parser API returns this setting as settings.markdownParsing.enabled. You can enable it in a parser update request:

{
  "markdownParsing": {
    "enabled": true
  }
}

After a document finishes, its public detail response and supported automation payloads include metadata.markdownUrl. Webhook payloads use the same field. This is a temporary read URL, so request the document details again when you need a new URL. PARSER does not expose the internal storage key in public responses.

Data extraction mode

This controls the speed vs quality balance when extracting data.

Options:

  • Fast: Prioritizes speed
  • Accurate: High accuracy with balanced speed
  • Critical: Maximum accuracy for important workflows

Credit notes:

  • Fast and Accurate use standard credits (1 credit per page)
  • Critical adds +2 credits per page (3 credits per page total)

To set: In Data extraction mode, choose Fast, Accurate, or Critical, then click Save Parser.

See Data extraction modes for mode selection guidance and detailed examples.

Tips

  • Start with defaults: Use default settings (all pages, all file types) unless you have a specific need
  • Use document type rules for noisy inboxes: If unrelated attachments sometimes arrive in the same parser, enable Document type rules
  • Add password rules for repeat senders: If the same sender regularly sends encrypted PDFs, add a matching PDF password rule
  • Test page ranges: If limiting pages, test on a few documents first to make sure you're not missing data
  • Advanced table extraction: Start with Auto and switch to Enabled only if complex tables need extra help
  • Data extraction mode: Start with Accurate, then adjust to Fast or Critical by business need
  • Confidence scores: Enable when you need better visibility into extraction quality (+1 credit per page)
  • Markdown parsing: Enable when you need a stored Markdown file or HTML table view (+1 credit per page)

Next steps

General SettingsConfigure parser name, description, and basic informationOpen guide → Document Type RulesChoose which document types a parser should process or skipOpen guide →