Git status can be read directly

ogc/na/ingest_json.py

@@ -651,7 +651,7 @@ def process(input_files: str | Path | Sequence[str | Path],
         while remaining_fn:
             fn = str(remaining_fn.popleft())
 
-            if not fn or not os.path.isfile():
+            if not fn or not os.path.isfile(fn):
                 continue
 
             if re.match(r'.*\.ya?ml$', fn):
@@ -706,7 +706,7 @@ def _process_cmdln():
 
     parser.add_argument(
         "input",
-        nargs='+',
+        nargs='*',
         help="Source file (instead of service)",
     )
 
@@ -783,14 +783,25 @@ def _process_cmdln():
         help='Regular expression for URL whitelisting'
     )
 
+    parser.add_argument(
+        '--use-git-status',
+        action='store_true',
+        help='Use git status for obtaining batch filenames'
+    )
+
     args = parser.parse_args()
 
     if args.domain_config:
         domain_cfg = DomainConfiguration(args.domain_config)
     else:
         domain_cfg = None
 
-    result = process(args.input,
+    input_files = args.input
+    if args.batch and args.use_git_status:
+        git_status = util.git_status()
+        input_files = git_status['added'] + git_status['modified'] + [r[1] for r in git_status['renamed']]
+
+    result = process(input_files,
                      context_fn=args.context,
                      domain_cfg=domain_cfg,
                      jsonld_fn=args.json_ld_file if args.json_ld else False,

ogc/na/util.py

@@ -12,6 +12,7 @@
 
 import requests
 import rfc3987
+import git
 from rdflib import Graph
 from pyshacl import validate as shacl_validate
 from urllib.parse import urlparse
@@ -239,3 +240,24 @@ def deep_update(orig_dict: dict, with_dict: dict, replace: bool = False) -> dict
         else:
             dest[k] = v
     return dest
+
+
+def git_status(repo_path: str | Path = '.'):
+    repo = git.Repo(repo_path)
+    added = repo.untracked_files
+    modified = []
+    deleted = []
+    renamed = []
+    for diff in repo.head.commit.diff(None):
+        if diff.change_type == 'D':
+            deleted.append(diff.a_path)
+        elif diff.change_type == 'M':
+            modified.append(diff.a_path)
+        elif diff.chhange_type == 'R':
+            renamed.append((diff.a_path, diff.b_path))
+    return {
+        'added': added,
+        'modified': modified,
+        'deleted': deleted,
+        'renamed': renamed,
+    }