moderation:added cli command to add queries

site/tests/moderation/test_moderation_queries.py

@@ -8,10 +8,9 @@
 """Test ModerationQuery model class."""
 
 from invenio_db import db
-from invenio_search import current_search_client
+
 from zenodo_rdm.moderation.models import ModerationQuery
-from zenodo_rdm.moderation.rules import match_query_rule
-from zenodo_rdm.api import ZenodoRDMRecord
+
 
 def test_moderation_query_creation(app):
     """Test to create and index a ModerationQuery."""
@@ -22,7 +21,10 @@ def test_moderation_query_creation(app):
         active = True
 
         query = ModerationQuery.create(
-            query_string, ZenodoRDMRecord, notes=notes, score=score, active=active
+            query_string,
+            notes=notes,
+            score=score,
+            active=active,
         )
         db.session.commit()
 
@@ -36,4 +38,5 @@ def test_moderation_query_creation(app):
             ]
         )
 
-#TODO: Add test for matching query
+
+# TODO: Add test for matching query

site/zenodo_rdm/cli.py

@@ -6,6 +6,9 @@
 # under the terms of the MIT License; see LICENSE file for more details.
 """Zenodo RDM cli commands."""
 
+import csv
+import os
+
 import click
 from flask.cli import with_appcontext
 from invenio_access.permissions import system_identity
@@ -27,9 +30,11 @@
 from invenio_requests.records.models import RequestMetadata
 
 from zenodo_rdm.api import ZenodoRDMRecord
+from zenodo_rdm.moderation.models import ModerationQuery
 from zenodo_rdm.moderation.percolator import (
     create_percolator_index,
     get_percolator_index,
+    index_percolate_query,
 )
 
 
@@ -279,3 +284,89 @@ def create_index(record_cls):
         click.secho(f"Percolator index '{index_name}' created successfully.")
     except Exception as e:
         click.secho(f"Error creating percolator index: {e}")
+
+
+@moderation.command("add-query")
+@click.option(
+    "-r",
+    "--record-cls",
+    type=click.Choice(["records", "communities"], case_sensitive=False),
+    default="records",
+    help="Record class to base the query on (default: records).",
+)
+@click.option(
+    "-q",
+    "--query-string",
+    help="The query string for the moderation query (optional if loading from CSV).",
+)
+@click.option(
+    "-n",
+    "--notes",
+    default="Example note",
+    help="Additional notes for the moderation query (optional if loading from CSV).",
+)
+@click.option(
+    "-s",
+    "--score",
+    default=10,
+    type=int,
+    help="The score for the moderation query (optional if loading from CSV).",
+)
+@click.option(
+    "-a",
+    "--active",
+    default=True,
+    type=bool,
+    help="Whether the query is active (optional if loading from CSV).",
+)
+@click.option(
+    "-f",
+    "--file",
+    type=click.Path(exists=True, readable=True),
+    help="Path to CSV file containing queries.",
+)
+@with_appcontext
+def add_query(record_cls, query_string, notes, score, active, file):
+    """Command to add a moderation query from CSV or directly and index it."""
+    record_cls = ZenodoRDMRecord if record_cls == "records" else Community
+
+    try:
+        if file:
+            add_queries_from_csv(file, record_cls)
+        else:
+            create_and_index_query(record_cls, query_string, notes, score, active)
+
+        click.secho("Queries added and indexed successfully.")
+    except Exception as e:
+        click.secho(f"Error adding or indexing query: {e}")
+
+
+def add_queries_from_csv(file_path, record_cls=ZenodoRDMRecord):
+    """Load queries from a CSV file, add them to the database, and index them."""
+    with open(file_path, mode="r", newline="", encoding="utf-8") as csvfile:
+        csvreader = csv.reader(csvfile)
+
+        for row in csvreader:
+            if row:
+                query_string = row[0].strip().strip("'")
+                notes = row[1].strip().strip("'") if len(row) > 1 else None
+                score = int(row[2].strip()) if len(row) > 2 else 10  # Default score 10
+                active = (
+                    row[3].strip().lower() == "true" if len(row) > 3 else True
+                )  # Default to True
+
+                # Ensure to add query only if there's a query string
+                if query_string:
+                    create_and_index_query(
+                        record_cls, query_string, notes, score, active
+                    )
+
+
+def create_and_index_query(record_cls, query_string, notes, score, active):
+    """Create and index a single moderation query."""
+    query = ModerationQuery.create(
+        query_string=query_string, notes=notes, score=score, active=active
+    )
+
+    db.session.commit()
+    index_percolate_query(record_cls, query.id, query_string, active, score, notes)

site/zenodo_rdm/moderation/models.py

@@ -10,15 +10,9 @@
 import enum
 from urllib.parse import urlparse
 
-from flask import current_app
 from invenio_db import db
-from invenio_search import current_search_client
 from sqlalchemy_utils import ChoiceType, Timestamp
 
-from zenodo_rdm.api import ZenodoRDMRecord
-
-from .percolator import index_percolate_query
-
 
 class LinkDomainStatus(enum.Enum):
     """Link domain status."""
@@ -86,7 +80,7 @@ class ModerationQuery(db.Model):
 
     __tablename__ = "moderation_queries"
 
-    id = db.Column(db.Integer, primary_key=True, autoincrement=True)
+    id = db.Column(db.Integer, primary_key=True)
     """Primary key identifier for the moderation query."""
 
     score = db.Column(db.Integer, default=0)
@@ -102,26 +96,19 @@ class ModerationQuery(db.Model):
     """Indicates whether the moderation query is currently active."""
 
     @classmethod
-    def create(
-        cls, query_string, record_cls=ZenodoRDMRecord, notes=None, score=0, active=True
-    ):
+    def create(cls, query_string, notes=None, score=0, active=True):
         """Create a new moderation query with a configurable record class."""
         query = cls(query_string=query_string, notes=notes, score=score, active=active)
         db.session.add(query)
 
-        index_percolate_query(record_cls, query_string, active, score, notes)
-
         return query
 
     @classmethod
     def get(cls, query_id=None):
         """Retrieve a moderation query by ID or return all queries if no ID is provided."""
         if query_id is not None:
             return cls.query.filter_by(id=query_id).one_or_none()
-        return cls.query.all()
 
     def __repr__(self):
         """Get a string representation of the moderation query."""
-        return (
-            f"<ModerationQuery id={self.id}, score={self.score}, active={self.active}>"
-        )
+        return f"<ModerationQuery id={self.id}, query_string={self.query_string}, score={self.score}, active={self.active}>"

site/zenodo_rdm/moderation/percolator.py

@@ -1,8 +1,7 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of Invenio.
-# Copyright (C) 2017-2024 CERN.
-# Copyright (C) 2022 Graz University of Technology.
+# Copyright (C) 2024 CERN.
 #
 # Invenio is free software; you can redistribute it and/or modify it
 # under the terms of the MIT License; see LICENSE file for more details.
@@ -74,12 +73,15 @@ def create_percolator_index(record_cls):
             current_app.logger.exception(e)
 
 
-def index_percolate_query(record_cls, query_string, active=True, score=1, notes=None):
+def index_percolate_query(
+    record_cls, query_id, query_string, active=True, score=1, notes=None
+):
     """Index a percolate query."""
     try:
         current_search_client.index(
             index=get_percolator_index(record_cls),
             body={
+                "id": query_id,
                 "query": {"query_string": {"query": query_string}},
                 "active": active,
                 "score": score,