Add count commands

ISI-MIP · Jul 11, 2024 · a94efe6 · a94efe6
1 parent e03a01e
commit a94efe6
Show file tree

Hide file tree

Showing 2 changed files with 48 additions and 0 deletions.
diff --git a/isimip_publisher/commands.py b/isimip_publisher/commands.py
@@ -111,6 +111,48 @@ def match_public_links():
             print(file.path)
 
 
+def count_remote():
+    remote_files = files.list_files(settings.REMOTE_PATH, settings.PATH,
+                                    remote_dest=settings.REMOTE_DEST, suffix=settings.PATTERN['suffix'])
+    datasets = patterns.match_datasets(settings.PATTERN, settings.REMOTE_PATH, remote_files,
+                                       include=settings.INCLUDE, exclude=settings.EXCLUDE)
+    validation.validate_datasets(settings.SCHEMA, settings.PATH, datasets)
+    print(len([file for dataset in datasets for file in dataset.files]))
+
+
+def count_remote_links():
+    remote_links = files.list_links(settings.REMOTE_PATH, settings.PATH,
+                                    remote_dest=settings.REMOTE_DEST, suffix=settings.PATTERN['suffix'])
+    datasets = patterns.match_datasets(settings.PATTERN, settings.REMOTE_PATH, remote_links,
+                                       include=settings.INCLUDE, exclude=settings.EXCLUDE)
+    validation.validate_datasets(settings.SCHEMA, settings.PATH, datasets)
+    print(len([file for dataset in datasets for file in dataset.files]))
+
+
+def count_local():
+    local_files = files.list_files(settings.LOCAL_PATH, settings.PATH)
+    datasets = patterns.match_datasets(settings.PATTERN, settings.LOCAL_PATH, local_files,
+                                       include=settings.INCLUDE, exclude=settings.EXCLUDE)
+    validation.validate_datasets(settings.SCHEMA, settings.PATH, datasets)
+    print(len([file for dataset in datasets for file in dataset.files]))
+
+
+def count_public():
+    public_files = files.list_files(settings.PUBLIC_PATH, settings.PATH)
+    datasets = patterns.match_datasets(settings.PATTERN, settings.PUBLIC_PATH, public_files,
+                                       include=settings.INCLUDE, exclude=settings.EXCLUDE)
+    validation.validate_datasets(settings.SCHEMA, settings.PATH, datasets)
+    print(len([file for dataset in datasets for file in dataset.files]))
+
+
+def count_public_links():
+    public_links = files.list_links(settings.PUBLIC_PATH, settings.PATH)
+    datasets = patterns.match_datasets(settings.PATTERN, settings.PUBLIC_PATH, public_links,
+                                       include=settings.INCLUDE, exclude=settings.EXCLUDE)
+    validation.validate_datasets(settings.SCHEMA, settings.PATH, datasets)
+    print(len([file for dataset in datasets for file in dataset.files]))
+
+
 def fetch_files():
     remote_files = files.list_files(settings.REMOTE_PATH, settings.PATH,
                                     remote_dest=settings.REMOTE_DEST, suffix=settings.PATTERN['suffix'])

diff --git a/isimip_publisher/main.py b/isimip_publisher/main.py
@@ -7,6 +7,11 @@
     check,
     check_doi,
     clean,
+    count_local,
+    count_public,
+    count_public_links,
+    count_remote,
+    count_remote_links,
     diff_remote,
     diff_remote_links,
     fetch_files,
@@ -99,6 +104,7 @@ def get_parser(add_path=False, add_subparsers=False):
         # add a subparser for each subcommand
         for func in [list_remote, list_remote_links, list_local, list_public, list_public_links,
                      match_remote, match_remote_links, match_local, match_public, match_public_links,
+                     count_remote, count_remote_links, count_local, count_public, count_public_links,
                      fetch_files, write_local_jsons, write_public_jsons,
                      insert_datasets, update_datasets, publish_datasets, archive_datasets,
                      diff_remote, diff_remote_links, check, clean, update_search, update_tree, run]: