Tensionix ENDERU
GitHub28 RepositoriesDaily Tech3 Abonnentenaudion.devdas ProduktregalRSSVeröffentlichungen
← Alle NotizenProjects

Audion DocFlow (Portable)

A portable Windows toolkit for deterministic document cleanup, verification, and processing around DOCX, XLSX, CSV, and Markdown.

Audion DocFlow is built for local, repeatable office-document workflows:

  • CLI reads input\, while the GUI Workbench can directly select an external

folder or one file as Source;

  • processed files are written to output\;
  • reports are written to report\;
  • launcher temp files stay in ._runtime\;
  • originals are not modified in place.

The project intentionally does not include OCR or AI rewriting. OCR and AI-assisted proofreading belong in separate workflows unless the project scope changes explicitly.

Entry Points

  • launcher_gui.cmd - NiceGUI/pywebview GUI over the existing command layer
  • launcher_project.cmd - English CLI launcher
  • launcher_project_ru.cmd - Russian CLI launcher
  • launcher_tools.cmd - service/tools launcher
  • builder_main.cmd - build/release launcher

The user launchers share the same tool set and support FZF + CMD fallback. The first user screen is grouped by purpose: Rules audit, Styles and hygiene, Find and replace, DOCX control and cleanup, Compare and reconcile, WORD/EXCEL TABLES, Technical operations, and Checks.

The GUI passes the selected Source and Target directly to the backend. The audit section includes anchored audit, fixes/dry-run, and anchor removal; control-free leaf commands run as compact action buttons with separate descriptions.

Service and release actions stay in launcher_tools.cmd and builder_main.cmd; they are intentionally hidden from the first GUI screen.

Included Tools

DOCX

  • Style processing: scan/fix headings, appendix titles, list structures, table/figure captions, TOC titles, sections, and style consistency
  • Rules audit processor for deterministic corporate rules such as кв. м, куб. м, № 1, percent/degree spacing, date suffixes, table-caption typos, and scan-only РФ; it can write annotated DOCX copies with visible anchors and later remove those anchors after approval
  • Text hygiene scan/fix
  • Deep DOCX hygiene: a two-pass aggregate run over text hygiene,

deterministic audit rules, and optional document anomaly scan/correction, with one docx_deep_hygiene.md/.docx/.json report

  • Document anomalies: read-only scan and safe copied-file correction for

tables, sections, headers/footers, page numbering, fields, TOC, links, bookmarks, lists, blanks, and dangling headings

  • Conservative DOCX nonprinting-garbage cleanup: only U+00AD and w:softHyphen, preserving spaces, tabs, breaks, and line-wrapping controls
  • Comma-lowercase reporting/fix for suspicious uppercase words after commas inside DOCX table cells
  • Russian morphological DOCX/XLSX find/replace via pymorphy3
  • Quality Gate scan and hard-fail mode
  • Strip comments
  • Accept tracked changes in a simple best-effort mode
  • Finalize black + clean
  • One-click finalize + gate
  • Recursive similar-DOCX detection
  • Explicit two-DOCX diff: section matching by similarity, then body text and tables by heading structure, without TOC or package/media noise
  • Native two-DOCX comparison through Microsoft Word COM, producing a separate DOCX with Word revisions
  • DOCX merge through Microsoft Word COM
  • DOCX table stitching after PDF export, table unification, width optimization, orientation adaptation, table fitting, and table extraction to XLSX
  • Root Unify tables in document command for one-pass normalization of tables

inside an existing DOCX: fonts, sizes, borders, cell margins, column balance, and fitting to section margins

  • DOCX table extraction to XLSX
  • DOCX/PPTX media extraction
  • DOCX table-cell margin compaction to 0.1 cm

XLSX, CSV, and Markdown

  • XLSX values diff
  • DOCX/XLSX/CSV table reconcile into four lists
  • Markdown pipe tables to XLSX
  • Excel insignificant-row cleanup for fully empty or whitespace-only rows

In the Russian UI, pre-header is translated as Строки перед шапкой. The repeated table header is documented as повторяемая шапка.

Quick Start

  1. Extract the portable package.
  2. Put source files into input\.
  3. Run launcher_gui.cmd, launcher_project.cmd, or launcher_project_ru.cmd.
  4. Read processed files in output\ and reports in report\.

For environment setup or release packaging, use builder_main.cmd. For service/licensing operations, use launcher_tools.cmd.

Checks

Environment verification:

install\verify_portable_env.cmd

Quick smoke:

& '.\runtime\python.exe' '.\tests\smoke.py' --quick

Full smoke covers the scripted tool set:

& '.\runtime\python.exe' '.\tests\smoke.py' --full

The full smoke reports docx_word_compare.py and docx_merge.py as SKIP when Microsoft Word COM is not available in the current session.

Project Layout

  • .gitattributes - line-ending policy for portable Windows scripts
  • input\ - source files
  • output\ - generated outputs
  • report\ - user-facing reports
  • logs\ - logs
  • config\tool_manifest.yaml - GUI command tree
  • config\gui_settings.yaml - GUI language, theme, and GUI-only settings
  • config\ui_colors.yaml - GUI color palettes and CSS tokens
  • config\rules\rules.yaml - machine-readable audit rule map
  • config\rules\rules.md - human-readable rule split and anchor workflow
  • install\ - portable environment, verification, and release scripts
  • system_core\ - Python tools and internal helpers
  • system_core\ui_nicegui\ - NiceGUI shell
  • system_core\services\ - GUI adapters around CLI commands
  • system_core\word_com\ - Word COM helper scripts
  • system_core\core\cmd_encoding.py - CMD encoding checker
  • system_core\license\ - release licensing helpers
  • runtime\ - embedded Python runtime
  • wheelhouse\ - offline wheel cache
  • release\ - packaged release output
  • ._runtime\ - launcher temp files

Notes

  • User-facing reports and Russian GUI layout are Russian; technical identifiers such as OK, PASS, FAIL, diff, SHA-256, DOCX A/B, JSON/YAML keys, and CLI options remain technical.
  • DOCX comparison through Word COM and DOCX merge through Word COM require installed Microsoft Word and available PowerShell.
  • .cmd files must remain UTF-8 without BOM and CRLF-only. install\Check-CmdEncoding.cmd verifies this.
  • Complex DOCX revision histories and unusual package structures may still require a final visual review in Word.

Canonical Workbench labels

Workbench uses the same Audion Image Tools public vocabulary in every project. Its buttons always keep the same order and labels: Source, Add file..., Target, Reset, Delete, List.

Reset returns to project input/output and does not delete files; Delete clears the current Source and Target only after confirmation. The exact Russian labels are Источник, Добавить файл..., Назначение, Сбросить, Удалить, Список. The Workbench variants Destination, Clear, Цель, and Очистить are not used.

Bearbeitet 28.08.2026