21 — SWEEP

Create a clean visual substrate by removing text glyph ink while preserving accepted page structure and media.

PARTIAL Partly built

The owner's own correction records that the current system fails here: line-local fills produce pale strips, and whole line boxes are removed instead of glyphs.

Deterministic Reconstruction 2 archived step 21 of 26

Where this output sits

Source evidence›Candidate evidence›Human decision›Audit acceptance›Release›Production

Source evidence, candidate evidence, human decision, audit acceptance, release and production are six separate things. This step produces the highlighted one and nothing further.

Scope

Removes text glyphs — not whole rectangular line boxes — from the original page, filling the removed pixels from the enclosing region's measured background, so that reconstructed text can be placed on a clean substrate while gradients, fills, borders, header, footer and media survive untouched.

This role is allowed to

Remove glyph components rather than line rectangles
Fill from measured enclosing background

How it works

fed by 20 PROTECTION 01 CAP

Inputs

Original pixelsCAP token/line masksProtection manifestSTYLE background evidenceCompleteness ledger

Deterministic Deterministic CV/OpenCV.

Read the frozen protection manifest — which pixel classes may never be touched
Build the glyph mask from frozen CAP polygons
Remove glyphs, not whole rectangular line boxes
Fill the removed pixels from the enclosing region's measured background
Audit residual ink and background continuity
Reject the sweep if any foreground component survives beneath reconstructed text

Must pass

Protected-pixel changes exactly zero
No visible source text beneath reconstruction
Continuous backgrounds remain continuous

passes →

Outputs

Swept substrateRemoved-ink and protected-pixel masksResidual-ink and background-continuity reportsArtifact hashes

hands off to ASSEMBLE · PDF · QA

fails →

HOLD

SWEEP_PROTECTED_PIXEL_CHANGEDSWEEP_RESIDUAL_TEXTSWEEP_BACKGROUND_DISCONTINUITYSWEEP_MASK_INCOMPLETE

Nothing continues on a failed gate. Uncertainty becomes an explicit HOLD, and no later step may read an unanswered item as an accepted one.

Non-scope — what this role does NOT own

Feeding any evidence detector
Semantic decisions
Starting before CAP and the protection evidence are frozen

Explicitly forbidden

Erase protected media/tables/rules/header/footer
Create pale per-line strips
Feed swept pixels back into MAP/CAP/ROLE/TAG

Dependencies and position

Starts: Only after CAP polygons and PROTECTION are frozen.
Previous step: 20 PROTECTION MANIFEST
Next step: 22 ASSEMBLE / GROUP
Hands off to: ASSEMBLE, PDF, QA

Exact inputs

Original pixels
CAP token/line masks
Protection manifest
STYLE background evidence
Completeness ledger

Exact outputs

Swept substrate
Removed-ink and protected-pixel masks
Residual-ink and background-continuity reports
Artifact hashes

Performer and AI/ML boundary

Performer: Deterministic CV/OpenCV.
Class: Deterministic
AI boundary: None.

Training information

Not a trained component. No model, no weights, no training corpus. Behaviour is fixed by code and configuration, and identical inputs must produce identical outputs.

Deterministic validation and acceptance gates

Protected-pixel changes exactly zero
No visible source text beneath reconstruction
Continuous backgrounds remain continuous

HOLD and failure behaviour

Failure codes this step may emit:

SWEEP_PROTECTED_PIXEL_CHANGED
SWEEP_RESIDUAL_TEXT
SWEEP_BACKGROUND_DISCONTINUITY
SWEEP_MASK_INCOMPLETE

Uncertainty becomes an explicit HOLD. Omission never converts uncertainty into acceptance, and no downstream step may treat an unanswered item as an accepted one.

Downstream handoff

ASSEMBLE, PDF, QA

Active tool

Status: Partial
Run / inspect: python tools/mechanical_pipeline.py
Input: original pixels + CAP masks + protection manifest
Output: swept substrate + removed-ink mask + residual report

Dependencies

Python 3.12 (NOT installed on this machine)
OpenCV
Frozen CAP polygons
Protection masks — see step 20, incomplete

Why this is the active version

The V4 mechanical sweep is the code-bearing base of the sweep lineage (V4 to V5 swept substrate to V6 alignment to V7 solver). Later sweep work exists (V32 SWEEP_ENGINE_REWRITE, V33 HEADER_BOUNDARY_COMPLEX_SWEEP) and is newer by date, but those packages contain evidence and reports only — their code is not in the archive. PARTIAL for that reason, and because the owner's own correction records that the current system fails at this step: line-local fills producing pale strips, and removal of whole rectangular line boxes rather than glyphs.

Copied files — source receipt

File	Original SHA-256	Bytes	Copy
`mechanical_pipeline.py`	a8822c71887f0783…	13,507	verified
`audit_v4.py`	44de136f555194b3…	8,105	verified
`validate_package.py`	d5c73279ec7b0706…	5,511	verified
`MECHANICAL_QA_REPORT.md`	73ba8ed8e44fa8c6…	980	verified

4 file(s), all hash-verified against the original. Full detail in tool/SOURCE_RECEIPT.json. Copy-only: the historical source is never modified.

Archived versions

Archived version	Why superseded	Retained value	Bytes
MARSOOM_REV6_PHASE1_2015_SWEEP_ENGINE_REWRITE_APPEND_ONLY_V32_2026-09-02	Newer by date but evidence-only; no code present in the archive	Sweep engine rewrite findings	referenced
MARSOOM_REV6_PHASE1_2015_HEADER_BOUNDARY_COMPLEX_SWEEP_APPEND_ONLY_V33_2026-09-02	Newer by date but evidence-only; no code present in the archive	Header-boundary sweep findings	referenced

2 archived version(s). Historical packages are never deleted or mutated; large ones are referenced with verified paths rather than copied, because the source archive is 11 GB.

Known limitations

The owner's own correction records that the current system fails here: line-local fills produce pale strips, and whole line boxes are removed instead of glyphs.
Code is present only up to the V7 generation. Two later sweep packages (V32 engine rewrite, V33 header-boundary) contain evidence and reports but no code in the archive.
Sweep cannot be trusted while step 20 emits no protection manifest.
The swept substrate is a renderer derivative only. It must never feed an evidence detector, and there is no feedback edge back into MAP, CAP, ROLE, TYPE, PIC or TABLE.

Security and privacy

Operates on frozen local artifacts. No credentials required. No data leaves the machine.

Cost behaviour

No provider calls. No metered cost. Compute is local.

Copying a tool into this repository does not authorise running it, retraining it, calling a model, or processing a new issue. No paid call may be made without the owner's explicit authorisation and a hard cost cap.

Provenance

Registry: registry/roles.json (schema marsoom.e2t.roles.v1)
Derivation: 28 legacy roles - VER - HUMAN + REV = 27 active steps
Legacy role id: SWEEP
Legacy source SHA-256: 6f5d5d7ed8869e45307424c2193f95ffab84abc44a6a188f5c7f98c2a48ec64d
Generated: 2026-09-04T22:13:12.433Z

Feedback and decisions

Feedback is recorded per source and never merged into an invented consensus. Where sources disagree, both positions stand and the owner decides.

Source	Events
Naser / owner Final authority. Overrides every other source.	none recorded
Codex / orchestrator Architecture and sequencing.	none recorded
Builder / designer Implementation reality and constraints.	none recorded
Independent reviewer Adversarial review of claims.	none recorded
Auditor Evidence verification against artifacts.	none recorded
Human REV reviewer Page-level truth from the review site.	none recorded

Recorded events

No feedback events recorded yet. The ledger exists and is append-only: feedback/FEEDBACK_LEDGER.jsonl.

Editing feedback is not possible: a change is a new event whose supersedes names the one it replaces, and the original stays exactly as written.

Generated from registry/roles.json by tools/build-reports.mjs. Do not hand-edit — edit the registry and rebuild.
Original page pixels are the visual authority. CAP owns the exact captured text.