add cli command for the pdf parsing

sequestrae_engine/cli/commands.py

@@ -0,0 +1,32 @@
+import os
+from pathlib import Path
+
+from sequestrae_engine.document_parsing.parser import PDFToMarkdownParser
+
+
+def parse_pdfs_command(api_key, project_dir, limit=5):
+    if limit is None:
+        limit = 5
+    print(f"Max number of files to process: {limit}")
+
+    if not api_key:
+        print("Error: LLAMA_API_KEY is required")
+        return 1
+
+    parser = PDFToMarkdownParser(api_key=api_key)
+    project_path = Path(project_dir)
+
+    if not project_path.exists():
+        print(f"Error: Directory not found at {project_path}")
+        return 1
+
+    pdf_count = 0
+    for folder_path in project_path.iterdir():
+        if not folder_path.name.startswith(".") and folder_path.is_dir() and pdf_count < limit:
+            for pdf_path in folder_path.glob("*.pdf"):
+                if pdf_path.is_file() and "report" in pdf_path.stem.lower():
+                    pdf_count += 1
+                    parser.parse_pdf(pdf_path)
+
+    print(f"Successfully processed {pdf_count} PDF files")
+    return 0

sequestrae_engine/cli/main.py

@@ -0,0 +1,35 @@
+import argparse
+import os
+import sys
+
+from . import commands
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Sequestrae Engine CLI")
+    subparsers = parser.add_subparsers(dest="command")
+
+    # Parse PDFs command
+    parse_pdfs_parser = subparsers.add_parser(
+        "parse-pdf", help="Parse PDF audit report files to Markdown"
+    )
+    parse_pdfs_parser.add_argument(
+        "--api-key", help="Llama API key", default=os.environ.get("LLAMA_API_KEY")
+    )
+    parse_pdfs_parser.add_argument("--project-dir", help="Project data directory", required=True)
+    parse_pdfs_parser.add_argument("--limit", help="Maximum number of PDFs to process", type=int)
+    parse_pdfs_parser.set_defaults(
+        func=lambda args: commands.parse_pdfs_command(args.api_key, args.project_dir, args.limit)
+    )
+
+    args = parser.parse_args()
+    if hasattr(args, "func"):
+        exit_status = args.func(args)
+    else:
+        parser.print_help()
+        exit_status = 1
+    return exit_status
+
+
+if __name__ == "__main__":
+    main()