Redaction ToolsPDF benchmark v0.1.1 / SafeRedact
SafeRedact on extraction-conditions-1
Web app · published by Mykola Melnyk (Redaction Tools)
Leak rate
74.1%
95% interval 61.1–83.9% · n = 54
Self-scoredLeaked 40 of 54 sensitive values (74.1%).
- Over-redaction
- —
- Text kept (worst case)
- 0.0%
- Cases covered
- 1 of 1 cases
- Document damaged
- 1 cases
Is the document still usable?
- geometry failed in 1 of 1
- text_retained failed in 1 of 1
- not_rasterised failed in 1 of 1
Where the leaks are
| Layer | Leaked here | Only here | If it survives |
|---|---|---|---|
| ocr high | 55.6%42.4–68.0% 30 of 54 | 3.7%1.0–12.5% 2 of 54 | catches translucent or hairline covers |
| metadata high | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | invisible, survives forwarding |
| annotation high | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | shown by many viewers |
| attachment high | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | one toggle away |
| font_subset low | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | narrows the value; partial disclosure |
| image_xobject critical | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | a crop is a view, not a deletion |
| content_stream critical | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | copy-paste, pdftotext, any parser |
| prior_revision critical | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | 'save' instead of export; trivially recovered |
| rendered_pixels critical | 70.4%57.2–80.9% 38 of 54 | 18.5%10.4–30.8% 10 of 54 | visible to anyone who opens the file |
| optional_content high | 0.0%0.0–6.6% 0 of 54 | 0.0%0.0–6.6% 0 of 54 | one toggle away |
By category
| category | Leak rate | Over-redaction |
|---|---|---|
| PERSON | 74.1%61.1–83.9% 40 of 54 | — not measured |
By severity
| severity | Leak rate | Over-redaction |
|---|---|---|
| high | 74.1%61.1–83.9% 40 of 54 | — not measured |
By trap
| trap | Leak rate | Over-redaction |
|---|---|---|
| nan | 74.1%61.1–83.9% 40 of 54 | — not measured |
Removed, by where the value lives
- image_text
- 3.7%0.7–18.3%
1 of 27
- text_layer
- 48.1%30.7–66.0%
13 of 27
What was not measured
Every layer was read in every run.
Provenance
- Scored with
- pdfredeval (submitter’s copy)
- Engines
- ocr tesseract · pypdf 6.19.0 · pdfium 153.0.7999.0 · renderer pypdfium2 · pypdfium2 5.13.0 · tesseract tesseract 5.5.2
- Alignment
- page_box_verified, residual 1.12 px
- Run id
- 20260925T080508-saferedact-app-web-extraction-conditions-1-a1-39731d