Skip to content

Doc2FAQ v0.3.1 — Parsing and Draft Reliability Fixes

Choose a tag to compare

View all tags
@OKAN-SOFTWARE OKAN-SOFTWARE released this 26 Oct 22:15
· 7 commits to main since this release
v0.3.1
59696c1
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Changelog

All notable Doc2FAQ changes are recorded here. Version numbers follow semantic versioning; entries describe shipped source functionality and do not imply an external AI service.

0.3.1

Fixed

  • Avoid excessive regex backtracking on whitespace-heavy input in both FAQ modes.
  • Discard pending draft results and errors after the source, example, imported text, or mode changes.
  • Preserve separate client and provider responsibility suggestions while still filtering duplicates for the same party.

Changed

  • Replaced the document placeholder with neutral Q&A prompts and the Generate Mode sample with a team knowledge guide, removing payment and fee examples from the interface.
  • Added instructions for rebuilding and restarting the versioned JAR, handling occupied ports, and publishing the matching release artifact.

0.3.0

Added

  • Extract Mode for existing Markdown question headings and explicit Q:/A: pairs.
  • Generate Mode for source-grounded FAQ suggestions from document sections. Generation is local and deterministic, not powered by a paid AI provider.
  • In-browser FAQ editor for changing questions and answers, reordering items, removing duplicates, and adding new entries.
  • Markdown, JSON, and standalone HTML exports, with HTML content escaped for safety.
  • Windows Command Prompt instructions for installing Java 21 and Maven, configuring PATH, and troubleshooting local startup.
  • Java service and API integration tests for generation, limits, and validation, plus Node tests for export safety.

Improved

  • Expanded FAQ text validation to 50,000 characters, consistent with extraction and the browser.
  • Generate Mode filters common-topic duplicates, document-title headings, and existing FAQ appendices.
  • More natural question phrasing for client responsibilities, confidentiality, and intellectual-property ownership.
  • Existing Extract Mode behavior remains unchanged when a document contains explicit FAQ pairs.

Supported inputs and limits

  • TXT and Markdown uploads in the browser (256 KB file size).
  • PDF and DOCX uploads via the Java server (5 MB each).
  • Maximum 50 PDF pages, 50,000 extracted characters, and 30 FAQ items.

Known limitations

  • Scanned or image-only PDFs require OCR before import; OCR is not included.
  • Generate Mode is heuristic and may skip useful details or produce generic questions. Review every answer.
  • No accounts, paid plans, public-server hardening, or document persistence are included.
  • PDF and DOCX export are planned for a future release. The current outputs are Markdown, JSON, and HTML.

0.2.0

  • Added PDF and DOCX text import using Apache PDFBox and Apache POI.
  • Introduced format-signature validation and explicit errors for unsupported, empty, encrypted, or scanned documents.
  • Added document upload size and page/text extraction limits.
  • Preserved local, deterministic explicit-Q&A extraction.
Assets 3
Loading