Architecture upgrade: multi-layer viewer, derived revisions, and processing pipeline scaffolding

README.md

@@ -1,43 +1,56 @@
-# Gitplant Pass 1 (Desktop-first)
+# Gitplant Pass 1.5 (Desktop-first architecture upgrade)
 
-This repository now contains the Pass 1 desktop-first vertical slice defined in `docs/PDF_REVIEW_ARCHITECTURE_PACKAGE.md`.
+This repository now includes the architectural upgrade for multi-layer viewing, document transformations, and processing/indexing scaffolding.
 
-## Workspace structure
+Design authority: `docs/PDF_REVIEW_ARCHITECTURE_PACKAGE.md`.
 
-- `apps/desktop` – Tauri desktop shell + React/TypeScript UI
-- `apps/desktop/src-tauri` – Rust native commands, SQLite schema, managed file storage
-- `packages/shared-types` – shared contracts
-- `packages/viewer-core` – renderer abstraction interfaces
-- `packages/viewer-pdfjs` – PDF.js renderer adapter implementation
-- `packages/persistence-core` – persistence gateway interfaces used by UI
+## Repo/package structure
 
-## Prerequisites
+- `apps/desktop` – Tauri shell + React UI
+- `apps/desktop/src-tauri` – Rust commands, SQLite schema, managed storage, transform + processing proof paths
+- `packages/shared-types` – domain/shared records for revisions, jobs, extracted text
+- `packages/viewer-core` – renderer abstraction + render-scene/layer model
+- `packages/viewer-pdfjs` – PDF.js adapter implementation
+- `packages/persistence-core` – UI persistence gateway contracts
+- `packages/document-transform-core` – transform command/result contracts
+- `packages/document-transform-pdflib` – adapter slot (desktop implementation currently in Rust service)
+- `packages/processing-core` – processing job and OCR provider contracts
+- `packages/text-extraction-core` – text extraction provider contracts
 
-- Node.js 20+
-- Rust stable toolchain
-- Tauri system dependencies: https://v2.tauri.app/start/prerequisites/
+## What changed
+
+- Viewer architecture now models stacked render layers (`base_pdf`, `overlay_pdf`, future markup/selection overlays).
+- Transformations are isolated behind dedicated command boundary and create **derived revisions**.
+- Processing pipeline tracks jobs and stores extracted page text by revision/page.
+- Schema supports immutable originals, revision lineage, processing jobs, extracted text, and audit events.
+- Desktop app includes minimal dev scaffolding buttons to trigger text extraction and extract-page transformation proof path.
+
+## Implemented now (real paths)
+
+1. Import + open PDFs (existing behavior still supported).
+2. Extract page range to a new derived revision (managed storage + metadata persistence).
+3. Trigger text extraction job for current revision and persist per-page text.
+
+## Scaffolded for later
+
+- OCR provider implementation (boundary and job type already in place)
+- Full transform UI for delete/insert/reorder/combine
+- Native renderer adapter implementation
+- Full markup/selection overlay rendering
 
 ## Commands
 
 From repository root:
 
-- Dev (one command): `npm run desktop:dev`
+- Dev (desktop): `npm run desktop:dev`
 - Desktop build: `npm run desktop:build`
 - Tests: `npm test`
 - Typecheck: `npm run typecheck`
 
-## Pass 1 manual verification checklist
+## Manual verification checklist
 
 1. Run `npm run desktop:dev` and confirm desktop window opens.
-2. Click **Import PDF**, pick a local `.pdf` through native picker.
-3. Confirm document appears in viewer and page renders.
-4. Use **Prev/Next** and **Zoom In/Zoom Out/Fit Width**.
-5. Confirm imported document appears in **Recent Documents**.
-6. Restart app and confirm recent item reopens.
-
-## Deferred to Pass 2+
-
-- Markup tools/redlines/comments/workflow
-- Collaboration/sync/export pipeline
-- Native renderer adapter
-- Advanced performance optimization and tile rendering
+2. Import a PDF and confirm page rendering still works.
+3. Click **Extract page 1 to derived revision** and confirm no crash + viewer shows comparison-capable scene layer info.
+4. Click **Trigger text extraction** and confirm extracted row count updates.
+5. Restart app and confirm recent document still opens.

apps/desktop/package.json

@@ -17,7 +17,11 @@
     "@gitplant/viewer-pdfjs": "0.1.0",
     "@tauri-apps/api": "^2.0.0",
     "react": "^18.3.1",
-    "react-dom": "^18.3.1"
+    "react-dom": "^18.3.1",
+    "@gitplant/viewer-core": "0.1.0",
+    "@gitplant/processing-core": "0.1.0",
+    "@gitplant/document-transform-core": "0.1.0",
+    "@gitplant/text-extraction-core": "0.1.0"
   },
   "devDependencies": {
     "@tauri-apps/cli": "^2.0.0",
@@ -31,4 +35,4 @@
     "vite": "^6.0.5",
     "vitest": "^2.1.8"
   }
-}
+}

apps/desktop/src-tauri/Cargo.toml

@@ -15,6 +15,11 @@ serde = { version = "1", features = ["derive"] }
 tauri = { version = "2", features = [] }
 uuid = { version = "1", features = ["v4", "serde"] }
 rfd = "0.15"
+lopdf = "0.35"
+serde_json = "1"
 
 [build-dependencies]
 tauri-build = { version = "2", features = [] }
+
+[dev-dependencies]
+tempfile = "3"