working consistent list and remove with filtering

archivebox/cli/archivebox_list.py

@@ -5,12 +5,11 @@ __command__ = 'archivebox list'
 __description__ = 'List all the URLs currently in the archive.'
 
 import sys
-import json
 import argparse
 
 
-from ..legacy.util import reject_stdin, ExtendedEncoder
-from ..legacy.main import list_archive_data, csv_format
+from ..legacy.util import reject_stdin, to_json, to_csv
+from ..legacy.main import list_archive_data
 
 
 def main(args=None):
@@ -33,16 +32,10 @@ def main(args=None):
         action='store_true',
         help="Print the output in JSON format with all columns included.",
     )
-    parser.add_argument(
-        '--filter', #'-f',
-        type=str,
-        help="List only URLs matching the given regex pattern.",
-        default=None,
-    )
     parser.add_argument(
         '--sort', #'-s',
         type=str,
-        help="List the links sorted using the given key, e.g. timestamp or updated",
+        help="List the links sorted using the given key, e.g. timestamp or updated.",
         default=None,
     )
     parser.add_argument(
@@ -57,11 +50,26 @@ def main(args=None):
         help="List only URLs bookmarked after the given timestamp.",
         default=None,
     )
+    parser.add_argument(
+        '--filter-type',
+        type=str,
+        choices=('exact', 'substring', 'domain', 'regex'),
+        default='exact',
+        help='Type of pattern matching to use when filtering URLs',
+    )
+    parser.add_argument(
+        'patterns',
+        nargs='*',
+        type=str,
+        default=None,
+        help='List only URLs matching these filter patterns.'
+    )
     command = parser.parse_args(args)
     reject_stdin(__command__)
 
     links = list_archive_data(
-        filter_regex=command.filter,
+        filter_patterns=command.patterns,
+        filter_type=command.filter_type,
         before=command.before,
         after=command.after,
     )
@@ -69,10 +77,9 @@ def main(args=None):
         links = sorted(links, key=lambda link: getattr(link, command.sort))
 
     if command.csv:
-        print(command.csv)
-        print('\n'.join(csv_format(link, command.csv) for link in links))
+        print(to_csv(links, csv_cols=command.csv.split(','), header=True))
     elif command.json:
-        print(json.dumps(list(links), indent=4, cls=ExtendedEncoder))
+        print(to_json(links, indent=4, sort_keys=True))
     else:
         print('\n'.join(link.url for link in links))
     

archivebox/cli/archivebox_remove.py

@@ -0,0 +1,87 @@
+#!/usr/bin/env python3
+
+__package__ = 'archivebox.cli'
+__command__ = 'archivebox remove'
+__description__ = 'Remove the specified URLs from the archive.'
+
+import sys
+import argparse
+
+
+from ..legacy.main import list_archive_data, remove_archive_links
+from ..legacy.util import reject_stdin, to_csv, TimedProgress
+from ..legacy.config import ANSI
+
+
+def main(args=None):
+    args = sys.argv[1:] if args is None else args
+
+    parser = argparse.ArgumentParser(
+        prog=__command__,
+        description=__description__,
+        add_help=True,
+    )
+    parser.add_argument(
+        '--yes', # '-y',
+        action='store_true',
+        help='Remove links instantly without prompting to confirm.',
+    )
+    parser.add_argument(
+        '--delete', # '-r',
+        action='store_true',
+        help=(
+            "In addition to removing the link from the index, "
+            "also delete its archived content and metadata folder."
+        ),
+    )
+    parser.add_argument(
+        '--before', #'-b',
+        type=float,
+        help="List only URLs bookmarked before the given timestamp.",
+        default=None,
+    )
+    parser.add_argument(
+        '--after', #'-a',
+        type=float,
+        help="List only URLs bookmarked after the given timestamp.",
+        default=None,
+    )
+    parser.add_argument(
+        '--filter-type',
+        type=str,
+        choices=('exact', 'substring', 'domain', 'regex'),
+        default='exact',
+        help='Type of pattern matching to use when filtering URLs',
+    )
+    parser.add_argument(
+        'pattern',
+        nargs='?',
+        type=str,
+        default=None,
+        help='URLs matching this filter pattern will be removed from the index.'
+    )
+    command = parser.parse_args(args)
+    reject_stdin(__command__)
+
+    if not sys.stdin.isatty():
+        stdin_raw_text = sys.stdin.read()
+        if stdin_raw_text and command.url:
+            print(
+                '[X] You should pass either a pattern as an argument, '
+                'or pass a list of patterns via stdin, but not both.\n'
+            )
+            raise SystemExit(1)
+
+        patterns = [pattern.strip() for pattern in stdin_raw_text.split('\n')]
+    else:
+        patterns = [command.pattern]
+
+    remove_archive_links(
+        filter_patterns=patterns, filter_type=command.filter_type,
+        before=command.before, after=command.after,
+        yes=command.yes, delete=command.delete,
+    )
+    
+
+if __name__ == '__main__':
+    main()