Import
- Accepted input: PDF files. Other file types are rejected and named in the import report.
- Ways to add: multi-select in the file dialog, or drag a folder onto the window.
- Per-file facts: file name, size, page count, the first 12 characters of the SHA-256, and whether a text layer exists.
- Scanned PDFs: imported and marked
Needs OCRon the pages that have no text. They cannot be matched by text rules. - Originals: opened read-only for hashing and text extraction. Nothing is written back to them.
Rules
A rule is a name plus one condition. Conditions compare against the file as a whole:
| Condition | Example | What has to match |
|---|---|---|
| File name contains | invoice |
Substring of the file name, case-insensitive |
| Text contains | purchase order |
Substring found on a page of extracted text |
| Page count | more than 20 |
A comparison against the page count |
| File size | under 2 MB |
A comparison against the file size |
Priority. Rules have an order. When more than one rule matches, the highest-priority match becomes the suggestion and the others remain attached as evidence. Moving a rule up or down changes which suggestion wins, and nothing else.
Explanation. Every suggestion records the rule name, the file, the matched page where there is one, and the keyword or comparison that matched. If a rule stops firing, the explanation tells you which part failed to match.
Review states
| State | How it is set | What the export says |
|---|---|---|
| Accepted | You accepted the proposal as it stood | The category you accepted |
| Changed | You chose a different category | Your category, with the original rule hit kept as evidence |
| Ignored | You dismissed the proposal | No category; the file stays in the list |
| Needs Review | Default for anything you did not touch | Needs Review, never the unconfirmed suggestion |
An unconfirmed suggestion is never exported as if it were a decision. That is why a first export of a new folder usually contains a lot of Needs Review rows.
Export
You pick an output folder; the app writes three outputs there.
CSV manifest
One row per file: name, size, page count, SHA-256, category, review state, matched rule, matched page and matched keyword. Open it in any spreadsheet program; CSV is the only spreadsheet format in 0.2.0.
Index PDF
A contents tree built from the confirmed categories, with the source page number next to each entry, so a reader can go from the index back to the page in the original document.
ZIP of the originals
The PDFs themselves, unmodified, plus the manifest so the hashes travel with the files. Originals are copied, not moved.
Checking the hashes afterwards
The manifest’s SHA-256 column is the same value shown in the import list (in full, not just the 12-character prefix). Recompute it on any file to confirm the documented original still matches what you exported:
Get-FileHash .\some-file.pdf -Algorithm SHA256
Known limits
- No OCR. Pages without a text layer stay unread.
- No merge, split, encryption, decryption or repair.
- No XLSX export and no multi-language interface.
- No cloud storage, no account and no sharing links.