opendataloader-pdf 1.0.5__py3-none-any.whl → 1.0.6__py3-none-any.whl

opendataloader_pdf/__main__.py

@@ -0,0 +1,5 @@
+from .wrapper import main
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

opendataloader_pdf/wrapper.py

@@ -1,3 +1,4 @@
+import argparse
 import subprocess
 import sys
 import importlib.resources as importlib_resources
@@ -135,3 +136,84 @@ def run(
         if e.stdout:
             print(f"Stdout: {e.stdout}", file=sys.stderr)
         raise e
+
+
+def main(argv=None) -> int:
+    """CLI entry point for running the wrapper from the command line."""
+    parser = argparse.ArgumentParser(
+        description="Run the opendataloader-pdf CLI using the bundled JAR."
+    )
+    parser.add_argument("input_path", help="Path to the input PDF file or directory.")
+    parser.add_argument(
+        "-o",
+        "--output-dir",
+        dest="output_folder",
+        help="Directory where outputs are written.",
+    )
+    parser.add_argument("-p", "--password", help="Password for encrypted PDFs.")
+    parser.add_argument(
+        "--replace-invalid-chars",
+        help="Replacement character for invalid or unrecognized characters.",
+    )
+    parser.add_argument(
+        "--content-safety-off",
+        help="Disable content safety filtering (expects the desired mode).",
+    )
+    parser.add_argument(
+        "--markdown",
+        dest="generate_markdown",
+        action="store_true",
+        help="Generate Markdown output.",
+    )
+    parser.add_argument(
+        "--html",
+        dest="generate_html",
+        action="store_true",
+        help="Generate HTML output.",
+    )
+    parser.add_argument(
+        "--pdf",
+        dest="generate_annotated_pdf",
+        action="store_true",
+        help="Generate annotated PDF output.",
+    )
+    parser.add_argument(
+        "--keep-line-breaks",
+        action="store_true",
+        help="Preserve line breaks in text output.",
+    )
+    parser.add_argument(
+        "--markdown-with-html",
+        dest="html_in_markdown",
+        action="store_true",
+        help="Allow raw HTML within Markdown output.",
+    )
+    parser.add_argument(
+        "--markdown-with-images",
+        dest="add_image_to_markdown",
+        action="store_true",
+        help="Embed images in Markdown output.",
+    )
+    parser.add_argument(
+        "--no-json",
+        action="store_true",
+        help="Disable JSON output generation.",
+    )
+    parser.add_argument(
+        "--debug",
+        action="store_true",
+        help="Stream CLI logs directly to stdout.",
+    )
+    args = parser.parse_args(argv)
+
+    try:
+        run(**vars(args))
+    except FileNotFoundError as err:
+        print(err, file=sys.stderr)
+        return 1
+    except subprocess.CalledProcessError as err:
+        return err.returncode or 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())

{opendataloader_pdf-1.0.5.dist-info → opendataloader_pdf-1.0.6.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: opendataloader-pdf
-Version: 1.0.5
+Version: 1.0.6
 Summary: A Python wrapper for the opendataloader-pdf Java CLI.
 Home-page: https://github.com/opendataloader-project/opendataloader-pdf
 Author: opendataloader-project
@@ -109,9 +109,16 @@ opendataloader_pdf.run(
     generate_markdown=True,
     generate_html=True,
     generate_annotated_pdf=True,
+    debug=True,
 )
 ```
 
+- If you want to run it via CLI, you can use the following command:
+
+```sh
+opendataloader-pdf path/to/document.pdf --markdown --html --pdf
+```
+
 ### Function: run()
 
 The main function to process PDFs.
@@ -174,6 +181,24 @@ async function main() {
 main();
 ```
 
+If you want to run it via CLI, you can use the following command:
+
+```bash
+npx @opendataloader/pdf path/to/document.pdf -o path/to/output --markdown --html --pdf
+```
+
+or you can install it globally:
+
+```bash
+npm install -g @opendataloader/pdf
+```
+
+then run:
+
+```bash
+opendataloader-pdf path/to/document.pdf -o path/to/output --markdown --html --pdf
+```
+
 ### Function: run()
 
 `run(inputPath: string, options?: RunOptions): Promise<string>`

{opendataloader_pdf-1.0.5.dist-info → opendataloader_pdf-1.0.6.dist-info}/RECORD

@@ -1,7 +1,8 @@
 opendataloader_pdf/LICENSE,sha256=rxdbnZbuk8IaA2FS4bkFsLlTBNSujCySHHYJEAuo334,15921
 opendataloader_pdf/NOTICE.md,sha256=Uxc6sEbVz2hfsDinzzSNMtmsjx9HsQUod0yy0cswUwg,562
 opendataloader_pdf/__init__.py,sha256=T5RV-dcgjNCm8klNy_EH-IgOeodcPg6Yc34HHXtuAmQ,44
-opendataloader_pdf/wrapper.py,sha256=WL7qTsX214L0jXxlSDesYadRVpdrsLQd2Hgum5BdD1s,4962
+opendataloader_pdf/__main__.py,sha256=lmla4yz3SaYBfRJXOXnwO_8ID31-Ja20aQmomiz1eEc,84
+opendataloader_pdf/wrapper.py,sha256=Dsvw5un_HROLcy2xX0WqoKKRnOjL081LEYC6YfpViLE,7331
 opendataloader_pdf/THIRD_PARTY/THIRD_PARTY_LICENSES.md,sha256=QRYYiXFS2zBDGdmWRo_SrRfGhrdRBwhiRo1SdUKfrQo,11235
 opendataloader_pdf/THIRD_PARTY/THIRD_PARTY_NOTICES.md,sha256=pB2ZitFM1u0x3rIDpMHsLxOe4OFNCZRqkzeR-bfpFzE,8911
 opendataloader_pdf/THIRD_PARTY/licenses/Apache-2.0.txt,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
@@ -13,8 +14,9 @@ opendataloader_pdf/THIRD_PARTY/licenses/LICENSE-JJ2000.txt,sha256=itSesIy3XiNWgJ
 opendataloader_pdf/THIRD_PARTY/licenses/MIT.txt,sha256=JPCdbR3BU0uO_KypOd3sGWnKwlVHGq4l0pmrjoGtop8,1078
 opendataloader_pdf/THIRD_PARTY/licenses/MPL-2.0.txt,sha256=CGF6Fx5WV7DJmRZJ8_6w6JEt2N9bu4p6zDo18fTHHRw,15818
 opendataloader_pdf/THIRD_PARTY/licenses/Plexus Classworlds License.txt,sha256=ZQuKXwVz4FeC34ApB20vYg8kPTwgIUKRzEk5ew74-hU,1937
-opendataloader_pdf/jar/opendataloader-pdf-cli.jar,sha256=Z9WU68Tw5ckOTgnlUPJs_Jub_C6ZGyQ-0sqjjSNMYYk,20477542
-opendataloader_pdf-1.0.5.dist-info/METADATA,sha256=RNIDw03Rwl4wGRSPIhbHR6VyTzhc7cnlYHEEIajZBTk,25452
-opendataloader_pdf-1.0.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-opendataloader_pdf-1.0.5.dist-info/top_level.txt,sha256=xee0qFQd6HPfS50E2NLICGuR6cq9C9At5SJ81yv5HkY,19
-opendataloader_pdf-1.0.5.dist-info/RECORD,,
+opendataloader_pdf/jar/opendataloader-pdf-cli.jar,sha256=HmcxP25ZCOJNRV9U1IXy-beAN243_iCWTBIV6JB-6S8,20477911
+opendataloader_pdf-1.0.6.dist-info/METADATA,sha256=2BWSSScAW3mmpWum3N7g-01fMZITHsmQcDBqSmGSkU0,25966
+opendataloader_pdf-1.0.6.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+opendataloader_pdf-1.0.6.dist-info/entry_points.txt,sha256=Tupa9pVNF6nXD9sqzCLI8PCHbSu0jKkL3SYyTkQy0dc,71
+opendataloader_pdf-1.0.6.dist-info/top_level.txt,sha256=xee0qFQd6HPfS50E2NLICGuR6cq9C9At5SJ81yv5HkY,19
+opendataloader_pdf-1.0.6.dist-info/RECORD,,

opendataloader_pdf-1.0.6.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+opendataloader-pdf = opendataloader_pdf.wrapper:main