File size: 7,832 Bytes
a8b3f00
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
import datetime
import time

import click
from sqlalchemy import func
from werkzeug.exceptions import NotFound

import app
from configs import dify_config
from core.rag.index_processor.index_processor_factory import IndexProcessorFactory
from extensions.ext_database import db
from extensions.ext_redis import redis_client
from models.dataset import Dataset, DatasetQuery, Document
from services.feature_service import FeatureService


@app.celery.task(queue="dataset")
def clean_unused_datasets_task():
    click.echo(click.style("Start clean unused datasets indexes.", fg="green"))
    plan_sandbox_clean_day_setting = dify_config.PLAN_SANDBOX_CLEAN_DAY_SETTING
    plan_pro_clean_day_setting = dify_config.PLAN_PRO_CLEAN_DAY_SETTING
    start_at = time.perf_counter()
    plan_sandbox_clean_day = datetime.datetime.now() - datetime.timedelta(days=plan_sandbox_clean_day_setting)
    plan_pro_clean_day = datetime.datetime.now() - datetime.timedelta(days=plan_pro_clean_day_setting)
    page = 1
    while True:
        try:
            # Subquery for counting new documents
            document_subquery_new = (
                db.session.query(Document.dataset_id, func.count(Document.id).label("document_count"))
                .filter(
                    Document.indexing_status == "completed",
                    Document.enabled == True,
                    Document.archived == False,
                    Document.updated_at > plan_sandbox_clean_day,
                )
                .group_by(Document.dataset_id)
                .subquery()
            )

            # Subquery for counting old documents
            document_subquery_old = (
                db.session.query(Document.dataset_id, func.count(Document.id).label("document_count"))
                .filter(
                    Document.indexing_status == "completed",
                    Document.enabled == True,
                    Document.archived == False,
                    Document.updated_at < plan_sandbox_clean_day,
                )
                .group_by(Document.dataset_id)
                .subquery()
            )

            # Main query with join and filter
            datasets = (
                db.session.query(Dataset)
                .outerjoin(document_subquery_new, Dataset.id == document_subquery_new.c.dataset_id)
                .outerjoin(document_subquery_old, Dataset.id == document_subquery_old.c.dataset_id)
                .filter(
                    Dataset.created_at < plan_sandbox_clean_day,
                    func.coalesce(document_subquery_new.c.document_count, 0) == 0,
                    func.coalesce(document_subquery_old.c.document_count, 0) > 0,
                )
                .order_by(Dataset.created_at.desc())
                .paginate(page=page, per_page=50)
            )

        except NotFound:
            break
        if datasets.items is None or len(datasets.items) == 0:
            break
        page += 1
        for dataset in datasets:
            dataset_query = (
                db.session.query(DatasetQuery)
                .filter(DatasetQuery.created_at > plan_sandbox_clean_day, DatasetQuery.dataset_id == dataset.id)
                .all()
            )
            if not dataset_query or len(dataset_query) == 0:
                try:
                    # remove index
                    index_processor = IndexProcessorFactory(dataset.doc_form).init_index_processor()
                    index_processor.clean(dataset, None)

                    # update document
                    update_params = {Document.enabled: False}

                    Document.query.filter_by(dataset_id=dataset.id).update(update_params)
                    db.session.commit()
                    click.echo(click.style("Cleaned unused dataset {} from db success!".format(dataset.id), fg="green"))
                except Exception as e:
                    click.echo(
                        click.style("clean dataset index error: {} {}".format(e.__class__.__name__, str(e)), fg="red")
                    )
    page = 1
    while True:
        try:
            # Subquery for counting new documents
            document_subquery_new = (
                db.session.query(Document.dataset_id, func.count(Document.id).label("document_count"))
                .filter(
                    Document.indexing_status == "completed",
                    Document.enabled == True,
                    Document.archived == False,
                    Document.updated_at > plan_pro_clean_day,
                )
                .group_by(Document.dataset_id)
                .subquery()
            )

            # Subquery for counting old documents
            document_subquery_old = (
                db.session.query(Document.dataset_id, func.count(Document.id).label("document_count"))
                .filter(
                    Document.indexing_status == "completed",
                    Document.enabled == True,
                    Document.archived == False,
                    Document.updated_at < plan_pro_clean_day,
                )
                .group_by(Document.dataset_id)
                .subquery()
            )

            # Main query with join and filter
            datasets = (
                db.session.query(Dataset)
                .outerjoin(document_subquery_new, Dataset.id == document_subquery_new.c.dataset_id)
                .outerjoin(document_subquery_old, Dataset.id == document_subquery_old.c.dataset_id)
                .filter(
                    Dataset.created_at < plan_pro_clean_day,
                    func.coalesce(document_subquery_new.c.document_count, 0) == 0,
                    func.coalesce(document_subquery_old.c.document_count, 0) > 0,
                )
                .order_by(Dataset.created_at.desc())
                .paginate(page=page, per_page=50)
            )

        except NotFound:
            break
        if datasets.items is None or len(datasets.items) == 0:
            break
        page += 1
        for dataset in datasets:
            dataset_query = (
                db.session.query(DatasetQuery)
                .filter(DatasetQuery.created_at > plan_pro_clean_day, DatasetQuery.dataset_id == dataset.id)
                .all()
            )
            if not dataset_query or len(dataset_query) == 0:
                try:
                    features_cache_key = f"features:{dataset.tenant_id}"
                    plan = redis_client.get(features_cache_key)
                    if plan is None:
                        features = FeatureService.get_features(dataset.tenant_id)
                        redis_client.setex(features_cache_key, 600, features.billing.subscription.plan)
                        plan = features.billing.subscription.plan
                    if plan == "sandbox":
                        # remove index
                        index_processor = IndexProcessorFactory(dataset.doc_form).init_index_processor()
                        index_processor.clean(dataset, None)

                        # update document
                        update_params = {Document.enabled: False}

                        Document.query.filter_by(dataset_id=dataset.id).update(update_params)
                        db.session.commit()
                        click.echo(
                            click.style("Cleaned unused dataset {} from db success!".format(dataset.id), fg="green")
                        )
                except Exception as e:
                    click.echo(
                        click.style("clean dataset index error: {} {}".format(e.__class__.__name__, str(e)), fg="red")
                    )
    end_at = time.perf_counter()
    click.echo(click.style("Cleaned unused dataset from db success latency: {}".format(end_at - start_at), fg="green"))