Skip to content

CHANGELOG.md

Latest commit

 

History

History
73 lines (53 loc) · 3.28 KB

CHANGELOG.md

File metadata and controls

73 lines (53 loc) · 3.28 KB
Raw

Changelog

All notable public-facing changes to PDF TO ESX AGENT should be documented here.

This root changelog is the public release summary. Deeper architectural and decision history lives in docs_repo/07_PROJECT_HISTORY/.

[0.2.0] - 2026-04-12

Windows packaging and release-hardening milestone.

Added

  • PyInstaller-based Windows onedir packaging with a checked-in spec file
  • repeatable build scripts for PowerShell and batch launch
  • packaged-app validation documentation and packaged release guidance
  • frozen-mode runtime tests covering user-writable logs and output paths

Changed

  • frozen-mode path handling now prefers Windows known folders before environment fallbacks
  • packaged logging is file-only and writes to %LOCALAPPDATA%\PDF-TO-ESX-Agent\logs\pdf_to_esx_agent.log
  • packaged builds no longer create the default output folder at startup
  • packaging cleanup removes transient metadata directories from the final bundle
  • root and release docs now describe how to distribute the full onedir folder safely

Validation

  • packaged GUI startup smoke
  • packaged conversion smoke with explicit output path
  • packaged conversion smoke with default output path
  • copied-release-folder validation from a temp path outside the repo
  • no new conhost.exe process observed during packaged GUI launch
  • unit tests and compile validation remained green

Known Limitations At This Release

  • distribution is still a Windows onedir folder, not an installer
  • the executable is not yet signed
  • some shells do not report useful exit codes for the windowed validation mode

[0.1.0] - 2026-04-12

Initial public foundation release.

Added

  • runnable Windows desktop app for converting one or more insurance estimate PDFs
  • drag-and-drop upload, file list, output folder selection, preview/status panels, and clear success or failure states
  • local PDF ingestion with text extraction and selective OCR routing
  • structured extraction for metadata, totals, line items, and roof-related measurements
  • canonical estimate model and multi-PDF merge layer
  • deterministic XML and .esx package generation
  • package validation before success is reported
  • automated tests for core output and conversion failure paths
  • docs_repo/ knowledge base for architecture, mapping, testing, contribution, and release guidance
  • public-repo governance files, GitHub templates, and lightweight validation workflows
  • PyInstaller-based Windows executable build with repeatable scripts and a checked-in spec file

Improved Before Public Release

  • parsing reliability across multiple real estimate layouts
  • error messages for bad PDFs, empty parses, and invalid output paths
  • warning visibility for scanned pages, totals fallback, and weak extraction cases
  • logging consistency across validation, parsing, OCR, merge, and export stages
  • contributor onboarding and release-readiness documentation
  • frozen-mode runtime handling for logs, output paths, and packaged validation

Known Limitations At This Release

  • .esx output is a standards-based package, not a native proprietary XACTDOC.ZIPXML writer
  • parser coverage is useful but not universal
  • OCR-heavy layouts remain the weakest extraction area
  • the packaged distribution is a Windows onedir artifact, not an installer