Added MacOS support for pdfalto binary

.github/workflows/github-actions-ci.yml

@@ -111,6 +111,43 @@ jobs:
           path: ./dist-export/dist/
 
 
+  native-test:
+    strategy:
+      matrix:
+        include:
+          - runs-on: macos-latest
+            install-poppler: brew install poppler
+            pytest-ignore-args: --ignore=tests/cv_models
+          - runs-on: ubuntu-24.04-arm
+            install-poppler: sudo apt-get install -y poppler-utils
+            pytest-ignore-args: tests/external/pdfalto/url_test.py tests/external/pdfalto/parser_test.py tests/document tests/utils tests/config tests/lookup
+    runs-on: ${{ matrix.runs-on }}
+    steps:
+      - name: Check out repository code
+        uses: actions/checkout@v5
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install poppler (required by pdf2image)
+        run: ${{ matrix.install-poppler }}
+
+      - name: Install dependencies
+        run: uv sync --frozen --group dev --extra delft
+        if: runner.os != 'Linux' || runner.arch != 'ARM64'
+
+      - name: Install dependencies (Linux arm64 - skip delft extras with no arm64 wheels)
+        run: |
+          uv sync --frozen --group dev
+          uv pip install "sciencebeam-trainer-delft>=0.0.36"
+        if: runner.os == 'Linux' && runner.arch == 'ARM64'
+
+      - name: Run pytest
+        run: uv run python -m pytest -p no:cacheprovider ${{ matrix.pytest-ignore-args }}
+
+
   testpypi-publish:
     if: github.ref == 'refs/heads/main'
     needs: ["build-and-test"]

sciencebeam_parser/app/parser.py

@@ -14,7 +14,7 @@
 
 from sciencebeam_parser.app.context import AppContext
 from sciencebeam_parser.config.config import AppConfig, get_download_dir
-from sciencebeam_parser.external.pdfalto.wrapper import PdfAltoWrapper
+from sciencebeam_parser.external.pdfalto.wrapper import PdfAltoWrapper, get_default_pdfalto_url
 from sciencebeam_parser.external.pdfalto.parser import parse_alto_root
 from sciencebeam_parser.external.wapiti.wrapper import LazyWapitiBinaryWrapper
 from sciencebeam_parser.lookup.loader import load_lookup_from_config
@@ -177,7 +177,7 @@ def __init__(self, config: AppConfig):
         self.pdfalto_wrapper = PdfAltoWrapper(
             download_with_zip_path_support(
                 self.download_manager,
-                config['pdfalto']['path']
+                config['pdfalto'].get('path') or get_default_pdfalto_url()
             )
         )
         self.pdfalto_wrapper.ensure_executable()

sciencebeam_parser/external/pdfalto/wrapper.py

@@ -1,6 +1,8 @@
 import logging
 import os
+import platform
 import stat
+import sys
 from typing import Optional
 
 from sciencebeam_parser.utils.background_process import exec_with_logging
@@ -9,6 +11,26 @@
 LOGGER = logging.getLogger(__name__)
 
 
+PDFALTO_VERSION = 'v0.6.0'
+_PDFALTO_BASE_URL = (
+    f'https://github.com/kermitt2/pdfalto/releases/download/{PDFALTO_VERSION}'
+)
+
+
+def get_default_pdfalto_url() -> str:
+    machine = platform.machine().lower()
+    if sys.platform == 'darwin':
+        os_name = 'mac'
+    elif sys.platform.startswith('linux'):
+        os_name = 'linux'
+    else:
+        raise RuntimeError(f'Unsupported platform: {sys.platform!r}')
+    arch = 'arm64' if machine in ('arm64', 'aarch64') else '64'
+    zip_name = f'pdfalto-bin-{os_name}-{arch}.zip'
+    internal_path = f'pdfalto/{os_name}/{arch}/pdfalto'
+    return f'{_PDFALTO_BASE_URL}/{zip_name}!/{internal_path}'
+
+
 class PdfAltoWrapper:
     def __init__(self, binary_path: str):
         self.binary_path = binary_path

sciencebeam_parser/resources/default_config/config.yml

@@ -43,7 +43,10 @@ logging:
 download_dir: '~/.cache/sciencebeam-parser/downloads'
 
 pdfalto:
-  path: https://github.com/kermitt2/pdfalto/releases/download/v0.6.0/pdfalto-bin-linux-64.zip!/pdfalto/linux/64/pdfalto
+  # path is auto-detected based on the current platform if not specified
+  # (override with an explicit URL if needed, e.g.:)
+  # path: https://github.com/kermitt2/pdfalto/releases/download/v0.6.0/pdfalto-bin-linux-64.zip!/pdfalto/linux/64/pdfalto
+  path:
 wapiti:
   install_source: 'https://github.com/kermitt2/Wapiti/archive/a9c25d2bcccd60f1a54a7019689bd5229e866f00.tar.gz'
 xslt:

tests/external/pdfalto/url_test.py

@@ -0,0 +1,44 @@
+from unittest.mock import patch
+
+from sciencebeam_parser.external.pdfalto.wrapper import (
+    PDFALTO_VERSION,
+    get_default_pdfalto_url
+)
+
+
+class TestGetDefaultPdfaltoUrl:
+    def test_linux_x86_64(self):
+        with patch('sys.platform', 'linux'), \
+                patch('platform.machine', return_value='x86_64'):
+            url = get_default_pdfalto_url()
+        assert url == (
+            f'https://github.com/kermitt2/pdfalto/releases/download/{PDFALTO_VERSION}'
+            f'/pdfalto-bin-linux-64.zip!/pdfalto/linux/64/pdfalto'
+        )
+
+    def test_linux_aarch64(self):
+        with patch('sys.platform', 'linux'), \
+                patch('platform.machine', return_value='aarch64'):
+            url = get_default_pdfalto_url()
+        assert url == (
+            f'https://github.com/kermitt2/pdfalto/releases/download/{PDFALTO_VERSION}'
+            f'/pdfalto-bin-linux-arm64.zip!/pdfalto/linux/arm64/pdfalto'
+        )
+
+    def test_macos_x86_64(self):
+        with patch('sys.platform', 'darwin'), \
+                patch('platform.machine', return_value='x86_64'):
+            url = get_default_pdfalto_url()
+        assert url == (
+            f'https://github.com/kermitt2/pdfalto/releases/download/{PDFALTO_VERSION}'
+            f'/pdfalto-bin-mac-64.zip!/pdfalto/mac/64/pdfalto'
+        )
+
+    def test_macos_arm64(self):
+        with patch('sys.platform', 'darwin'), \
+                patch('platform.machine', return_value='arm64'):
+            url = get_default_pdfalto_url()
+        assert url == (
+            f'https://github.com/kermitt2/pdfalto/releases/download/{PDFALTO_VERSION}'
+            f'/pdfalto-bin-mac-arm64.zip!/pdfalto/mac/arm64/pdfalto'
+        )

tests/external/pdfalto/wrapper_test.py

@@ -6,7 +6,8 @@
 
 from sciencebeam_parser.config.config import get_download_dir
 from sciencebeam_parser.external.pdfalto.wrapper import (
-    PdfAltoWrapper
+    PdfAltoWrapper,
+    get_default_pdfalto_url
 )
 from sciencebeam_parser.utils.download import download_with_zip_path_support
 
@@ -22,7 +23,7 @@ def _pdfalto_wrapper(sciencebeam_parser_config: dict) -> PdfAltoWrapper:
     pdfalto_wrapper = PdfAltoWrapper(
         download_with_zip_path_support(
             download_manager,
-            sciencebeam_parser_config['pdfalto']['path']
+            sciencebeam_parser_config['pdfalto'].get('path') or get_default_pdfalto_url()
         )
     )
     pdfalto_wrapper.ensure_executable()