chore: import upstream snapshot with attribution

build-scripts/fetch_shaded_jar.py

@@ -0,0 +1,80 @@
+"""
+Finds and copies the latest shaded JAR from the Java build to the Python package source.
+
+This script is intended to be run from the monorepo root, typically as part of a
+CI/CD pipeline, before the Python package is built.
+"""
+
+import argparse
+import logging
+import re
+import shutil
+import sys
+from pathlib import Path
+from typing import Optional
+
+# Requires 'packaging' library (pip install packaging)
+from packaging.version import parse as parse_version
+
+def find_latest_jar_by_semver(target_dir: Path) -> Optional[Path]:
+    """Finds the shaded JAR with the highest semantic version in its filename."""
+
+    # Example filename: opendataloader-pdf-runtime-0.1.0.jar
+    jar_pattern = "opendataloader-pdf-runtime-*.jar"
+    version_regex = re.compile(r"opendataloader-pdf-runtime-(.+?)\.jar")
+
+    latest_version = parse_version("0.0.0")
+    latest_jar_path = None
+
+    # Exclude Maven's 'original' JARs to ensure we get the shaded (fat) JAR.
+    potential_jars = [p for p in target_dir.glob(jar_pattern) if 'original' not in p.name]
+
+    if not potential_jars:
+        return None
+
+    # Iterate through potential JARs to find the one with the highest version number.
+    for jar_path in potential_jars:
+        match = version_regex.search(jar_path.name)
+        if match:
+            try:
+                current_version = parse_version(match.group(1))
+                if current_version > latest_version:
+                    latest_version = current_version
+                    latest_jar_path = jar_path
+            except Exception:
+                # Ignore files with non-parseable version strings.
+                continue
+
+    return latest_jar_path
+
+def main():
+    """Parse command-line arguments and orchestrate the copy process."""
+
+    logging.basicConfig(level=logging.INFO, format='%(levelname)s: %(message)s', stream=sys.stdout)
+
+    parser = argparse.ArgumentParser(description="Copies the latest shaded JAR to the Python source tree.")
+    parser.add_argument("java_target_dir", type=Path, help="Path to the Java module's 'target' directory.")
+    parser.add_argument("python_jars_dir", type=Path, help="Path to the Python package's destination directory for JARs.")
+    args = parser.parse_args()
+
+    java_target_path: Path = args.java_target_dir.resolve()
+    python_jars_path: Path = args.python_jars_dir.resolve()
+
+    if not java_target_path.is_dir():
+        parser.error(f"Java target directory not found: {java_target_path}")
+
+    # Ensure the destination directory exists.
+    python_jars_path.mkdir(parents=True, exist_ok=True)
+
+    source_jar_path = find_latest_jar_by_semver(java_target_path)
+    if not source_jar_path:
+        parser.error(f"No versioned shaded JAR found in: {java_target_path}")
+
+    # Standardize the destination name for consistent access within the Python package.
+    destination_jar_path = python_jars_path / 'runtime.jar'
+
+    shutil.copy2(source_jar_path, destination_jar_path)
+    logging.info(f"Copied '{source_jar_path.name}' to '{destination_jar_path}'")
+
+if __name__ == "__main__":
+    main()

build-scripts/set_version.py

@@ -0,0 +1,45 @@
+# build-scripts/set_version.py
+
+import os
+import re
+import sys
+
+def set_version(version_file, pom_file, pyproject_toml_file):
+    with open(version_file, 'r') as f:
+        version = f.read().strip()
+
+    # Update Maven POM
+    with open(pom_file, 'r') as f:
+        pom_content = f.read()
+    pom_content = re.sub(r'<version>.*</version>', f'<version>{version}</version>', pom_content, count=1)
+    with open(pom_file, 'w') as f:
+        f.write(pom_content)
+    print(f"Updated Maven POM version to {version}")
+
+    # Update Python pyproject.toml
+    with open(pyproject_toml_file, 'r') as f:
+        pyproject_content = f.read()
+    pyproject_content = re.sub(r'version = ".*"', f'version = "{version}"', pyproject_content, count=1)
+    with open(pyproject_toml_file, 'w') as f:
+        f.write(pyproject_content)
+    print(f"Updated Python pyproject.toml version to {version}")
+
+if __name__ == "__main__":
+    # Paths are relative to the monorepo root
+    root_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))
+    
+    version_path = os.path.join(root_dir, 'VERSION')
+    java_pom_path = os.path.join(root_dir, 'java', 'pom.xml')
+    python_pyproject_path = os.path.join(root_dir, 'python', 'packages', 'opendataloader_pdf', 'pyproject.toml')
+
+    if not os.path.exists(version_path):
+        print(f"Error: VERSION file not found at {version_path}")
+        sys.exit(1)
+    if not os.path.exists(java_pom_path):
+        print(f"Error: Java pom.xml not found at {java_pom_path}")
+        sys.exit(1)
+    if not os.path.exists(python_pyproject_path):
+        print(f"Error: Python pyproject.toml not found at {python_pyproject_path}")
+        sys.exit(1)
+
+    set_version(version_path, java_pom_path, python_pyproject_path)