moderation: added query match rule

site/zenodo_rdm/moderation/config.py

@@ -7,7 +7,8 @@
 
 """Moderation config."""
 
-from .rules import files_rule, links_rule, text_sanitization_rule, verified_user_rule
+from .rules import (files_rule, links_rule, match_query_rule,
+                    text_sanitization_rule, verified_user_rule)
 
 MODERATION_BANNED_LINK_DOMAINS = []
 """Banned domains for links."""
@@ -46,12 +47,14 @@
     links_rule,
     files_rule,
     text_sanitization_rule,
+    match_query_rule,
 ]
 """Scoring rules for record moderation."""
 
 MODERATION_COMMUNITY_SCORE_RULES = [
     links_rule,
     text_sanitization_rule,
     verified_user_rule,
+    match_query_rule,
 ]
 """Scoring rules for communtiy moderation."""

site/zenodo_rdm/moderation/handlers.py

@@ -17,7 +17,8 @@
 from invenio_requests.tasks import request_moderation
 from invenio_users_resources.proxies import current_users_service
 from invenio_users_resources.records.api import UserAggregate
-from invenio_users_resources.services.users.tasks import execute_moderation_actions
+from invenio_users_resources.services.users.tasks import \
+    execute_moderation_actions
 from werkzeug.utils import cached_property
 
 from .errors import UserBlockedException

site/zenodo_rdm/moderation/models.py

@@ -0,0 +1,68 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2024 CERN.
+#
+# ZenodoRDM is free software; you can redistribute it and/or modify
+# it under the terms of the MIT License; see LICENSE file for more details.
+
+"""Moderation models."""
+
+from invenio_db import db
+from invenio_search import current_search_client
+
+
+class ModerationQuery(db.Model):
+    """Moderation queries model."""
+
+    __tablename__ = "moderation_queries"
+
+    id = db.Column(db.Integer, primary_key=True, autoincrement=True)
+    """Primary key identifier for the moderation query."""
+
+    score = db.Column(db.Integer, default=0)
+    """Score associated with the query."""
+
+    query_string = db.Column(db.Text, nullable=False)
+    """Query string containing the filter criteria."""
+
+    notes = db.Column(db.Text, nullable=True)
+    """Additional notes or comments regarding the moderation query."""
+
+    active = db.Column(db.Boolean, default=True)
+    """Indicates whether the moderation query is currently active."""
+
+    @classmethod
+    def create(cls, query_string, notes=None, score=0, active=True):
+        """Create a new moderation query."""
+        query = cls(query_string=query_string, notes=notes, score=score, active=active)
+        db.session.add(query)
+
+        try:
+            current_search_client.index(
+                index="moderation-queries",
+                # id=query.id,
+                body={
+                    "query": {"query_string": {"query": query_string}},
+                    "active": active,
+                    "score": score,
+                    "notes": notes,
+                },
+            )
+        except Exception as e:
+            print(query_string)
+            print(f"Failed to index query: {e}")
+
+        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}>"
+        )

site/zenodo_rdm/moderation/rules.py

@@ -10,7 +10,7 @@
 import re
 
 from flask import current_app
-
+from invenio_search import current_search_client
 from zenodo_rdm.moderation.proxies import current_domain_tree
 
 from .proxies import current_domain_tree, current_scores
@@ -123,3 +123,20 @@ def files_rule(identity, draft=None, record=None):
         score += current_scores.ham_files
 
     return score
+
+
+def match_query_rule(identity, draft=None, record=None):
+    """Calculate a score based on matched percolate queries against the record."""
+    breakpoint()
+    matched_queries = current_search_client.search(
+        index="moderation-queries",
+        body={"query": {"percolate": {"field": "query", "document": record.dumps()}}},
+    )
+    breakpoint()
+    score = 0
+
+    for hit in matched_queries["hits"]["hits"]:
+        query_score = hit["_source"].get("score", 0)
+        score += query_score
+
+    return score