dump_things_pyclient/commands/dtc_plugins/delete_records.py

@@ -2,6 +2,7 @@ import json
 import logging
 import sys
 from functools import partial
+from typing import Iterable
 
 import rich_click as click
 from rich.progress import track
@@ -11,9 +12,12 @@ from ...communicate import (
     HTTPError,
     get_session,
     collection_delete_record,
+    collection_read_records_of_class,
     curated_delete_record,
+    curated_read_records_of_class,
     incoming_delete_record,
     incoming_read_labels,
+    incoming_read_records_of_class,
 )
 
 
@@ -56,6 +60,21 @@ console = Console(file=sys.stderr)
     is_flag=True,
     help='ignore errors when deleting a pid and continue with remaining pids',
 )
+@click.option(
+    '--class', '-C', 'class_name',
+    metavar='CLASS',
+    default=None,
+    help='delete ALL records of class CLASS from the collection\'s incoming '
+         'area that is associated with the token. Can be '
+         'combined with `-i/--incoming LABEL` or `-c/--curated` to delete all '
+         'records of class CLASS from the incoming area `LABEL` or from the '
+         'curated area. Note: if neither `-c/--curated` nor `-i/--incoming` is '
+         'specified, the command cannot reliably determine which records are '
+         'stored in the incoming area associated with a token and which '
+         'records are stored in the curated area of the collection. This can '
+         'lead to warnings about records that cannot be deleted. The command '
+         'will print a list of all PIDs that could not be deleted.',
+)
 def cli(
         obj,
         service_url,
@@ -64,17 +83,20 @@ def cli(
         curated,
         incoming,
         ignore_errors,
+        class_name,
 ):
     """Delete records from a collection on a dump-things-service
 
     This command delete the records given by PIDS from the collection COLLECTION
-    of the dump-things service SERVICE_URL. If no PIDS are provided on the
-    command line, the pids that should be deleted are read from stdin (one pid
+    of the dump-things service SERVICE_URL. If no pids are provided on the
+    command line, the pid that should be deleted are read from stdin (one pid
     per line, lines are stripped).
 
     By default, the records will be deleted from the inbox associated with the
-    token. If the option -c/--curated is given, the records are deleted from
+    token. If the option `-c/--curated` is given, the records are deleted from
     the curated area of the collection (this requires a token with
+    curator rights). If the option `-i/--incoming LABEL` is given, the records
+    are deleted from the inbox specified by `LABEL` (this requires a token with
     curator rights).
     """
     try:
@@ -86,6 +108,7 @@ def cli(
             curated,
             incoming,
             ignore_errors,
+            class_name,
         )
     except HTTPError as e:
         console.print(f'[red]Error[/red]: {e}: {e.response.text}')
@@ -93,23 +116,30 @@ def cli(
 
 
 def delete_records(
-        obj,
+        token,
         service_url,
         collection,
         pids,
         curated,
         incoming,
         ignore_errors,
+        class_name,
 ):
-    token = obj
-
-    if token is None:
-        click.echo(f'WARNING: no token provided', err=True)
-
     if incoming and curated:
         console.print('[red]Error[/red]: -i/--incoming and -c/--curated are mutually exclusive')
         return 1
 
+    if pids and class_name:
+        console.print('[red]Error[/red]: -C/--class cannot be combined with PID-specification')
+        return 1
+
+    if token is None:
+        console.print(f'[yellow]Warning[/yellow]: no token provided')
+        return 1
+
+    if class_name and pids:
+        console.print(f'[yellow]Warning[/yellow]: PIDs will be ignored because -C/--class was provided')
+
     session = get_session()
     kwargs = dict(
         service_url=service_url,
@@ -135,21 +165,67 @@ def delete_records(
     else:
         operation = collection_delete_record
 
-    if not pids:
-        pids = sys.stdin
+    # If a class name is specified, read the PIDs from the server
+    if class_name:
+        pids = _get_pids_for_class(
+            class_name=class_name,
+            incoming=incoming,
+            curated=curated,
+            kwargs=kwargs,
+        )
+    else:
+        if not pids:
+            pids = sys.stdin
 
+    result = []
+    explanation = False
     for pid in track(pids, console=console):
         try:
             operation(
-                service_url=service_url,
-                collection=collection,
                 pid=pid.strip(),
-                token=token,
-                session=session,
+                **kwargs,
             )
         except HTTPError as e:
-            console.print(f'[red]Error[/red]: while deleting pid {pid}: {e}, {e.response.text}')
+            if class_name and not (incoming or curated):
+                if not explanation:
+                    explanation = True
+                    console.print(
+                        f'[yellow]The record with pid {pid} could not be deleted, this is not an error, iff: '
+                        f'the record is only contained in the curated area of collection {collection}, and '
+                        f'not in the incoming area associated with the provided token. To delete such records, '
+                        f'use the option -c/--curated.[/yellow]'
+                    )
+                    result.append(pid)
+                console.print(f'[yellow]Warning[/yellow]: could not delete record with pid [green]{pid}[/green]: {e}: {e.response.text}')
+                continue
+            console.print(f'[red]Error[/red]: could not delete record with pid [green]{pid}[/green]: {e}, {e.response.text}')
             if ignore_errors:
                 continue
             return 1
+
+    if result:
+        click.echo('\n'.join(result))
     return 0
+
+
+def _get_pids_for_class(
+        class_name: str,
+        incoming: str | None,
+        curated: bool,
+        kwargs,
+) -> Iterable[str]:
+
+    if incoming:
+        operation = partial(incoming_read_records_of_class, label=incoming)
+    elif curated:
+        operation = curated_read_records_of_class
+    else:
+        operation = collection_read_records_of_class
+
+    return map(
+        lambda result: result[0]['pid'],
+        operation(
+            class_name=class_name,
+            **kwargs,
+        )
+    )