release: integrate Use capabilities and prepare 0.1.2

.github/workflows/release.yml

@@ -42,7 +42,7 @@ jobs:
             test "${{ inputs.release_tag }}" = "v${version}"
           fi
       - run: cargo fmt --all -- --check
-      - run: cargo test --workspace --all-features --locked
+      - run: cargo test --workspace --all-features --locked -- --test-threads=1
       - run: cargo clippy --workspace --all-targets --all-features --locked -- -D warnings
 
   binaries:
@@ -84,12 +84,16 @@ jobs:
           version="${{ needs.validate.outputs.version }}"
           archive="a3s-use-${version}-${{ matrix.name }}.tar.gz"
           stage="${RUNNER_TEMP}/a3s-use-${version}-${{ matrix.name }}"
-          install -d "${stage}/skills" "${stage}/skill-data" "${stage}/office-skills" "${stage}/dashboard"
+          install -d "${stage}/skills" "${stage}/skill-data" "${stage}/office-skills" "${stage}/ocr-skills" "${stage}/dashboard"
           install -m 0755 "target/${{ matrix.target }}/release/a3s-use" "${stage}/a3s-use"
           install -m 0755 "target/${{ matrix.target }}/release/a3s-use-browser-driver" "${stage}/a3s-use-browser-driver"
+          A3S_USE_OCR_HOME="${stage}/ocr-models" \
+            "${stage}/a3s-use" component install ocr --json > "${RUNNER_TEMP}/ocr-model-install.json"
+          rm -f "${stage}/ocr-models/.install.lock"
           cp -R crates/browser-driver/skills/. "${stage}/skills/"
           cp -R crates/browser-driver/skill-data/. "${stage}/skill-data/"
           cp -R crates/office/skills/. "${stage}/office-skills/"
+          cp -R crates/ocr/skills/. "${stage}/ocr-skills/"
           cp -R crates/browser-driver/dashboard/out/. "${stage}/dashboard/"
           install -m 0644 LICENSE README.md THIRD_PARTY_NOTICES.md "${stage}/"
           install -m 0644 crates/browser-driver/LICENSE-APACHE-2.0 "${stage}/LICENSE-APACHE-2.0"
@@ -107,9 +111,15 @@ jobs:
           New-Item -ItemType Directory -Force -Path $stage | Out-Null
           Copy-Item "target/${{ matrix.target }}/release/a3s-use.exe" "$stage/a3s-use.exe"
           Copy-Item "target/${{ matrix.target }}/release/a3s-use-browser-driver.exe" "$stage/a3s-use-browser-driver.exe"
+          $env:A3S_USE_OCR_HOME = "$stage/ocr-models"
+          & "$stage/a3s-use.exe" component install ocr --json | Out-File "$env:RUNNER_TEMP/ocr-model-install.json"
+          if ($LASTEXITCODE -ne 0) { throw "Failed to install the pinned PP-OCRv6 release assets" }
+          Remove-Item Env:A3S_USE_OCR_HOME
+          Remove-Item "$stage/ocr-models/.install.lock" -ErrorAction SilentlyContinue
           Copy-Item -Recurse "crates/browser-driver/skills" "$stage/skills"
           Copy-Item -Recurse "crates/browser-driver/skill-data" "$stage/skill-data"
           Copy-Item -Recurse "crates/office/skills" "$stage/office-skills"
+          Copy-Item -Recurse "crates/ocr/skills" "$stage/ocr-skills"
           Copy-Item -Recurse "crates/browser-driver/dashboard/out" "$stage/dashboard"
           Copy-Item LICENSE,README.md,THIRD_PARTY_NOTICES.md $stage
           Copy-Item crates/browser-driver/LICENSE-APACHE-2.0 "$stage/LICENSE-APACHE-2.0"
@@ -128,6 +138,11 @@ jobs:
           test -x "${install_root}/a3s-use-browser-driver"
           test -f "${install_root}/skill-data/core/SKILL.md"
           test -f "${install_root}/office-skills/a3s-use-office/SKILL.md"
+          test -f "${install_root}/ocr-skills/a3s-use-ocr/SKILL.md"
+          test -f "${install_root}/ocr-models/PP-OCRv6_small/det/inference.onnx"
+          test -f "${install_root}/ocr-models/PP-OCRv6_small/det/inference.yml"
+          test -f "${install_root}/ocr-models/PP-OCRv6_small/rec/inference.onnx"
+          test -f "${install_root}/ocr-models/PP-OCRv6_small/rec/inference.yml"
           test -f "${install_root}/dashboard/index.html"
           test -f "${install_root}/LICENSE-APACHE-2.0"
           test -f "${install_root}/UPSTREAM.md"
@@ -143,6 +158,49 @@ jobs:
               "agentcore", "core", "dogfood", "electron", "slack", "vercel-sandbox"
           }
           PY
+          "${install_root}/a3s-use" ocr doctor --json > "${RUNNER_TEMP}/ocr-doctor.json"
+          python3 - "${RUNNER_TEMP}/ocr-doctor.json" <<'PY'
+          import json, pathlib, sys
+          value = json.loads(pathlib.Path(sys.argv[1]).read_text())
+          assert value["ok"] is True
+          assert value["data"]["readiness"] == "ready"
+          assert value["data"]["provider"] == "pp-ocr-v6"
+          assert value["data"]["engine"] == "onnx-runtime"
+          assert value["data"]["model"] == "PP-OCRv6_small"
+          assert value["data"]["sendsSourceOffDevice"] is False
+          PY
+          python3 - "${RUNNER_TEMP}/ocr-white.bmp" <<'PY'
+          import base64, pathlib, sys
+          pathlib.Path(sys.argv[1]).write_bytes(base64.b64decode(
+              "Qk06AAAAAAAAADYAAAAoAAAAAQAAAAEAAAABABgAAAAAAAQAAAAAAAAAAAAAAAAAAAAAAAAA////AA=="
+          ))
+          PY
+          "${install_root}/a3s-use" ocr extract "${RUNNER_TEMP}/ocr-white.bmp" --json > "${RUNNER_TEMP}/ocr-extract.json"
+          python3 - "${RUNNER_TEMP}/ocr-extract.json" <<'PY'
+          import json, pathlib, sys
+          value = json.loads(pathlib.Path(sys.argv[1]).read_text())
+          assert value["ok"] is True
+          assert value["data"]["provider"] == "pp-ocr-v6"
+          assert value["data"]["engine"] == "onnx-runtime"
+          assert value["data"]["model"] == "PP-OCRv6_small"
+          assert value["data"]["source"]["mediaType"] == "image/bmp"
+          PY
+          "${install_root}/a3s-use" capability snapshot --json > "${RUNNER_TEMP}/capabilities.json"
+          python3 - "${RUNNER_TEMP}/capabilities.json" "${install_root}" <<'PY'
+          import json, pathlib, sys
+          value = json.loads(pathlib.Path(sys.argv[1]).read_text())
+          root = pathlib.Path(sys.argv[2]).resolve()
+          ocr = next(
+              capability
+              for capability in value["data"]["registry"]["capabilities"]
+              if capability["id"] == "use/ocr"
+          )
+          assert ocr["mcp"]["target"] == "ocr-native"
+          assert len(ocr["skills"]) == 1
+          assert pathlib.Path(ocr["skills"][0]["path"]).resolve() == (
+              root / "ocr-skills" / "a3s-use-ocr" / "SKILL.md"
+          )
+          PY
           A3S_OFFICECLI_EXECUTABLE="${install_root}/must-not-be-invoked" \
             "${install_root}/a3s-use" office skills list --json > "${RUNNER_TEMP}/office-skills.json"
           python3 - "${RUNNER_TEMP}/office-skills.json" <<'PY'
@@ -179,6 +237,11 @@ jobs:
             "$root/a3s-use-browser-driver.exe",
             "$root/skill-data/core/SKILL.md",
             "$root/office-skills/a3s-use-office/SKILL.md",
+            "$root/ocr-skills/a3s-use-ocr/SKILL.md",
+            "$root/ocr-models/PP-OCRv6_small/det/inference.onnx",
+            "$root/ocr-models/PP-OCRv6_small/det/inference.yml",
+            "$root/ocr-models/PP-OCRv6_small/rec/inference.onnx",
+            "$root/ocr-models/PP-OCRv6_small/rec/inference.yml",
             "$root/dashboard/index.html",
             "$root/LICENSE-APACHE-2.0",
             "$root/UPSTREAM.md",
@@ -194,6 +257,36 @@ jobs:
           if (-not $officeSkills.ok -or $officeSkills.data.Count -ne 1 -or $officeSkills.data[0].name -ne "a3s-use-office") {
             throw "Packaged Office Skill smoke failed"
           }
+          $ocr = (& "$root/a3s-use.exe" ocr doctor --json | ConvertFrom-Json)
+          if (
+            -not $ocr.ok -or
+            $ocr.data.readiness -ne "ready" -or
+            $ocr.data.provider -ne "pp-ocr-v6" -or
+            $ocr.data.engine -ne "onnx-runtime" -or
+            $ocr.data.model -ne "PP-OCRv6_small" -or
+            $ocr.data.sendsSourceOffDevice
+          ) { throw "Packaged PP-OCRv6 doctor smoke failed" }
+          $whiteBmp = [Convert]::FromBase64String("Qk06AAAAAAAAADYAAAAoAAAAAQAAAAEAAAABABgAAAAAAAQAAAAAAAAAAAAAAAAAAAAAAAAA////AA==")
+          [IO.File]::WriteAllBytes("$env:RUNNER_TEMP/ocr-white.bmp", $whiteBmp)
+          $ocrExtract = (& "$root/a3s-use.exe" ocr extract "$env:RUNNER_TEMP/ocr-white.bmp" --json | ConvertFrom-Json)
+          if (
+            -not $ocrExtract.ok -or
+            $ocrExtract.data.provider -ne "pp-ocr-v6" -or
+            $ocrExtract.data.engine -ne "onnx-runtime" -or
+            $ocrExtract.data.model -ne "PP-OCRv6_small" -or
+            $ocrExtract.data.source.mediaType -ne "image/bmp"
+          ) { throw "Packaged PP-OCRv6 extraction smoke failed" }
+          $capabilities = (& "$root/a3s-use.exe" capability snapshot --json | ConvertFrom-Json)
+          $ocrCapability = $capabilities.data.registry.capabilities |
+            Where-Object id -eq "use/ocr"
+          if ($ocrCapability.mcp.target -ne "ocr-native" -or $ocrCapability.skills.Count -ne 1) {
+            throw "Built-in OCR capability projection smoke failed"
+          }
+          $expectedOcrSkill = (Resolve-Path "$root/ocr-skills/a3s-use-ocr/SKILL.md").Path
+          $actualOcrSkill = (Resolve-Path $ocrCapability.skills[0].path).Path
+          if (-not [StringComparer]::OrdinalIgnoreCase.Equals($actualOcrSkill, $expectedOcrSkill)) {
+            throw "Built-in OCR Skill was not projected from the installed release"
+          }
           $requests = @(
             '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"release-smoke","version":"1"}}}',
             '{"jsonrpc":"2.0","method":"notifications/initialized","params":{}}',
@@ -223,7 +316,7 @@ jobs:
           ref: ${{ github.event_name == 'workflow_dispatch' && inputs.release_tag || github.ref }}
       - uses: dtolnay/rust-toolchain@stable
       - uses: Swatinem/rust-cache@v2
-      - name: Publish Core then Browser
+      - name: Publish Core, Extension, OCR, then Browser
         env:
           CARGO_REGISTRY_TOKEN: ${{ secrets.CARGO_TOKEN }}
           VERSION: ${{ needs.validate.outputs.version }}
@@ -262,6 +355,10 @@ jobs:
 
           publish_once a3s-use-core
           wait_until_visible a3s-use-core
+          publish_once a3s-use-extension
+          wait_until_visible a3s-use-extension
+          publish_once a3s-use-ocr
+          wait_until_visible a3s-use-ocr
           publish_once a3s-use-browser
 
   release: