Merge pull request #1235 from brandonrobertz/add-text-output-mode

Add a --format text option
jsvine · Dec 16, 2024 · e0ee254 · e0ee254
2 parents 28d4f50 + f3f2b57
commit e0ee254
Showing 1 changed file with 4 additions and 1 deletion.
diff --git a/pdfplumber/cli.py b/pdfplumber/cli.py
@@ -37,7 +37,7 @@ def parse_args(args_raw: List[str]) -> argparse.Namespace:
         action="store_true",
     )
 
-    parser.add_argument("--format", choices=["csv", "json"], default="csv")
+    parser.add_argument("--format", choices=["csv", "json", "text"], default="csv")
 
     parser.add_argument("--types", nargs="+")
 
@@ -109,6 +109,9 @@ def main(args_raw: List[str] = sys.argv[1:]) -> None:
                 include_attrs=args.include_attrs,
                 exclude_attrs=args.exclude_attrs,
             )
+        elif args.format == "text":
+            for page in pdf.pages:
+                print(page.extract_text(layout=True))
         else:
             pdf.to_json(
                 sys.stdout,