Michael
b8aff862ed
Phase 1/6 of the PDF Extractor tool. Pure module — no Streamlit,
no user-config I/O — that turns a PDF blob plus a template dict
into a ``pandas.DataFrame`` of transaction rows. Primary use case
is accountant-style extraction of bank-statement transactions,
where each bank's format is encoded as a reusable template.
Pipeline:
1. ``extract_pages(pdf_bytes)`` reads with pdfplumber and surfaces
words with bounding boxes.
2. ``cluster_rows(words)`` groups words into rows by ``top``
tolerance — no reliance on PDF table-line detection (most bank
statements have no visible cell borders).
3. ``assign_columns(row_words, boundaries)`` buckets each word by
its horizontal midpoint into N+1 columns defined by N interior
x-boundaries.
4. ``_within_table_window`` slices to the band between the header
line and the end-marker (e.g. "Closing balance").
5. ``apply_template`` orchestrates the above, handling:
- parens-style negative amounts, currency stripping, custom
decimal/thousands separators
- separate debit + credit columns combined into a single signed
``amount`` (credit positive, debit negative — accounting
register convention; matches QuickBooks/Xero imports)
- multi-line description wrapping (rows with empty date column
attach to the previous row's description)
- row-level regex skip filters (e.g., "Total", "Subtotal")
- page-range filters ("all", "2-", "1,3-5")
Optional OCR fallback for scanned statements:
- ``page_has_extractable_text`` heuristic flags pages with <5
words as likely-scanned.
- ``ocr_available()`` checks both the ``pytesseract`` Python
binding and the Tesseract binary; surfaces a clear reason
string when either is missing.
- ``extract_pages_auto`` does text-first, OCR-the-blanks, and
returns warnings the UI can surface.
29 unit tests cover the parsing pipeline against synthetic
WordBox/Page data — no fixture PDFs required, runs in 0.1s. Real
PDF extraction is exercised by hand on the user's statements.
Dependencies added:
- ``pdfplumber>=0.10,<1`` — text + position extraction
- ``pypdfium2>=4,<6`` — page rasterization for OCR + visual picker
- ``streamlit-drawable-canvas>=0.9,<1`` — visual region picker
(used in commit 5)
- ``pytesseract>=0.3,<1`` — OCR (used in commit 6; system
Tesseract binary required separately)
- ``cryptography>=41,<49`` — bumped upper bound; pdfminer.six
transitively requires a recent release. Internal ed25519
license-signing usage is API-stable across the bump.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
🌐 Language: English · Español
DataTools
Local CSV / Excel cleaning. CLI + browser GUI, no cloud, no install ceremony. GUI ships with English and Spanish language packs.
Tools
| # | Tool | Status |
|---|---|---|
| 01 | Find Duplicates — exact + fuzzy match, 5 normalizers, survivor rules, audit | Ready |
| 02 | Clean Text — whitespace, smart chars, BOM, line endings, case ops | Ready |
| 03 | Standardize Formats — dates, phones, emails, addresses, names, currencies, booleans | Ready |
| 04 | Fix Missing Values — disguised-null detection, profile, mean/median/mode/ffill/bfill/interpolate, drop strategies | Ready |
| 05 | Map Columns — fuzzy auto-rename, target schema with type coercion, required fields with defaults, drop/reorder | Ready |
| 06 | Find Unusual Values | Coming Soon |
| 07 | Combine Files | Coming Soon |
| 08 | Quality Check | Coming Soon |
| 09 | Automated Workflows — chain tools with recommended (not forced) order, save/load JSON, automate weekly cleanups | Ready |
Download (non-technical users)
Pre-built installers — no Python required:
| Platform | Download | First-launch note |
|---|---|---|
| macOS | DataTools-X.Y.Z-mac.dmg |
Drag DataTools.app into /Applications, then double-click. |
| Windows | DataTools-X.Y.Z-win-setup.exe |
Run the installer; launches from Start Menu. |
| Linux | DataTools-X.Y.Z-linux-x86_64.AppImage |
chmod +x the file, then double-click. |
Latest release: see GitHub Releases (or the Gumroad listing). The installers are ~150–200 MB; the launcher boots a local server at http://127.0.0.1:8501 and opens your browser. Nothing is sent to the cloud.
Install from source (developers)
pip install -r requirements.txt
Python 3.10+ required.
Run
GUI (recommended):
streamlit run src/gui/app.py
CLI — seven entry points:
python -m src.cli customers.csv [--apply] # dedup
python -m src.cli_text_clean messy.csv [--apply] # text clean
python -m src.cli_format intl.csv [--apply] # format standardize (auto-streams >100 MB)
python -m src.cli_missing holes.csv [--apply] # missing values
python -m src.cli_column_map vendor.csv [--apply] # column mapper
python -m src.cli_pipeline any_file.csv [--apply] # chain tools end-to-end
python -m src.cli_analyze any_file.csv [--json] # scan only
Every CLI runs preview-only by default; add --apply to write output.
Language
The GUI sidebar has a language picker. Packs ship for English and Español (src/i18n/packs/); the choice persists for the session. Adding a language: drop a <code>.json next to en.json mirroring its key tree, then list it in LANGUAGES. See Developer Guide §i18n.
Review & Normalize gate
Every uploaded file passes through a CSV-normalization gate before any tool sees it. The analyzer flags ~15 issue types (whitespace, NBSP / zero-width chars, BOM, encoding, smart punct, dirty headers, null sentinels, mojibake, …) tagged by confidence (high / medium / low) and fix action. The GUI shows each finding with Auto-fix / Skip / Customize, a live before/after preview, and an encoding-override picker. Tool pages refuse to load until the gate passes.
Output
Every run writes:
{input}_<tool>.csv— the cleaned data{input}_changes.csv(text cleaner) or{input}_match_groups.csv(dedup) — audit traillogs/<tool>_YYYYMMDD_HHMMSS.log— debug-level run log
Original input file is never modified.
Docs
- User Guide — install, GUI workflow, gate
- CLI Reference — every flag with recipes
- Requirements — file sizes, encodings, detectors, perf targets
- Technical — architecture, gate internals, fix registry
- Developer Guide — adding fixes / detectors / standardizers
Dependencies
pandas, openpyxl, rapidfuzz, phonenumbers, typer, loguru, charset-normalizer, streamlit. Optional: ftfy for mojibake repair.
License
Proprietary.