Skip to content

Commit 7f6c6ed

Browse files
sakshamarora1sakshamarora1
committed
Refactor(tests): Add more precise assertions
1 parent b27a854 commit 7f6c6ed

5 files changed

Lines changed: 274 additions & 108 deletions

File tree

‎cds_migrator_kit/rdm/comments/load.py‎

Lines changed: 67 additions & 37 deletions
Original file line numberDiff line numberDiff line change
@@ -11,7 +11,7 @@
1111

1212
import arrow
1313
from cds_rdm.legacy.resolver import get_pid_by_legacy_recid
14-
from flask import current_app, url_for
14+
from flask import current_app
1515
from invenio_access.permissions import system_identity
1616
from invenio_accounts.models import User
1717
from invenio_db.uow import UnitOfWork
@@ -49,7 +49,9 @@ def __init__(
4949

5050
def get_attached_files_for_comment(self, recid, comment_id):
5151
"""Get the attached files for the comment."""
52-
attached_files_directory = os.path.join(self.dirpath, str(recid), str(comment_id))
52+
attached_files_directory = os.path.join(
53+
self.dirpath, str(recid), str(comment_id)
54+
)
5355
if os.path.exists(attached_files_directory):
5456
return os.listdir(attached_files_directory)
5557
return []
@@ -59,19 +61,26 @@ def get_oldest_record(self, parent_pid_value):
5961
identity=system_identity, id_=parent_pid_value
6062
)
6163
search_result = current_rdm_records_service.scan_versions(
62-
system_identity, latest_record["id"],
64+
system_identity,
65+
latest_record["id"],
6366
)
6467
for hit in search_result.hits:
6568
self.all_record_versions[hit["versions"]["index"]] = hit
6669
oldest_version_index = min(self.all_record_versions.keys())
6770
return self.all_record_versions[oldest_version_index]
6871

6972
def create_event(
70-
self, request, data, community, record, uow, parent_comment_id=None
73+
self,
74+
request,
75+
data,
76+
community,
77+
uow,
78+
legacy_recid,
79+
parent_comment_id=None,
7180
):
7281
logger.info(
73-
"Creating event for record ID<{}> request ID<{}> comment ID<{}> parent_comment ID<{}>".format(
74-
record["id"],
82+
"Creating event for legacy recid ID<{}> request ID<{}> comment ID<{}> parent_comment ID<{}>".format(
83+
legacy_recid,
7584
request.id,
7685
data.get("comment_id"),
7786
"self" if not parent_comment_id else parent_comment_id,
@@ -89,20 +98,14 @@ def create_event(
8998
comment_status = data.get("status")
9099
event_type = CommentEventType
91100
if comment_status == "da":
92-
comment_payload["payload"].update(
93-
{
94-
"content": "comment was deleted by the author.",
95-
"event_type": "comment_deleted",
96-
}
97-
)
101+
comment_payload["payload"]["content"] = "comment was deleted by the author."
102+
comment_payload["payload"]["event"] = "comment_deleted"
98103
event_type = LogEventType
99104
elif comment_status == "dm":
100-
comment_payload["payload"].update(
101-
{
102-
"content": "comment was deleted by the moderator.",
103-
"event_type": "comment_deleted",
104-
}
105-
)
105+
comment_payload["payload"][
106+
"content"
107+
] = "comment was deleted by the moderator."
108+
comment_payload["payload"]["event"] = "comment_deleted"
106109
event_type = LogEventType
107110

108111
event = current_events_service.record_cls.create(
@@ -114,16 +117,17 @@ def create_event(
114117
file_relation = data.get("file_relation")
115118
file_id = file_relation.get("file_id")
116119
version = file_relation.get("version")
117-
record_version = self.all_record_versions.get(str(version), None)
120+
record_version = self.all_record_versions.get(version, None)
118121
if record_version:
119-
record_url = url_for(
120-
"invenio_app_rdm_records.record_detail",
121-
pid_value=record_version["id"],
122-
preview_file=file_id,
122+
record_url = (
123+
current_app.config["CDS_MIGRATOR_KIT_SITE_UI_URL"]
124+
+ "/records/"
125+
+ record_version["id"]
126+
+ f"?preview_file={file_id}"
123127
)
124-
version_link = f"<p><a href='{record_url}'>See related record version {version}</a></p>"
128+
version_link_html = f"<p><a href='{record_url}'>See related record version {version}</a></p>"
125129
comment_payload["payload"]["content"] = (
126-
version_link + "\n" + comment_payload["payload"]["content"]
130+
version_link_html + "\n" + comment_payload["payload"]["content"]
127131
)
128132

129133
if parent_comment_id:
@@ -145,20 +149,29 @@ def create_event(
145149
event.id,
146150
)
147151
)
148-
deep_link = f"<p><a href='{current_app.config['CDS_MIGRATOR_KIT_SITE_UI_URL']}/communities/{community.slug}/requests/{request.id}#commentevent-{parent_comment_id}_{mentioned_event_id}'>Link to the reply</a></p>"
152+
deep_link = (
153+
current_app.config["CDS_MIGRATOR_KIT_SITE_UI_URL"]
154+
+ f"/communities/{community.slug}/requests/{request.id}#commentevent-{parent_comment_id}_{mentioned_event_id}"
155+
)
156+
deep_link_html = f"<p><a href='{deep_link}'>Link to the reply</a></p>"
149157
comment_payload["payload"]["content"] = (
150-
deep_link + "\n" + comment_payload["payload"]["content"]
158+
deep_link_html + "\n" + comment_payload["payload"]["content"]
151159
)
152160

153161
# TODO: Add attached files to the event
154162
# https://github.com/CERNDocumentServer/cds-migrator-kit/issues/381
155163
# For now, if attached files are found, raise ManualImportRequired error
156164
attached_files = self.get_attached_files_for_comment(
157-
record["id"], data.get("comment_id")
165+
legacy_recid, data.get("comment_id")
158166
)
159167
if attached_files:
160168
raise ManualImportRequired(
161-
f"Attached files found for comment ID<{data.get('comment_id')}>."
169+
message=f"Attached files found.",
170+
field=data.get("comment_id"),
171+
value=attached_files,
172+
stage="load",
173+
recid=legacy_recid,
174+
priority="critical",
162175
)
163176

164177
event.update(comment_payload)
@@ -169,7 +182,12 @@ def create_event(
169182
)
170183
else:
171184
raise ManualImportRequired(
172-
f"User not found for email: {data.get('user_email')}"
185+
message=f"User not found.",
186+
field=data.get("comment_id"),
187+
value=data.get("user_email"),
188+
stage="load",
189+
recid=legacy_recid,
190+
priority="critical",
173191
)
174192
created_at = arrow.get(data.get("created_at")).datetime.replace(tzinfo=None)
175193
event.model.created = created_at
@@ -182,6 +200,7 @@ def create_event(
182200

183201
def create_accepted_community_submission_request(
184202
self,
203+
legacy_recid,
185204
record,
186205
community,
187206
creator_user_id,
@@ -228,34 +247,37 @@ def create_accepted_community_submission_request(
228247

229248
for comment_data in comments:
230249
comment_event = self.create_event(
231-
request, comment_data, community, record, uow
250+
request, comment_data, community, uow, legacy_recid
232251
)
233252
for reply in comment_data.get("replies", []):
234253
reply_event = self.create_event(
235254
request,
236255
reply,
237256
community,
238-
record,
239257
uow,
258+
legacy_recid,
240259
parent_comment_id=comment_event.id,
241260
)
242261
self.LEGACY_REPLY_LINK_MAP[reply.get("comment_id")] = reply_event.id
243262

244-
# Commit at the end to rollback if any error occurs not only for the request but also for the comments
263+
# Commit at the end so that rollback can be done if any error occurs not only for the request but also for the comments in the middle
245264
uow.commit()
246265

247266
return request
248267

249-
def _process_legacy_comments_for_recid(self, recid, comments):
268+
def _process_legacy_comments_for_recid(self, recid, comments, dry_run):
250269
"""Process the legacy comments for the record."""
251270
logger.info(f"Processing legacy comments for recid: {recid}")
252271
parent_pid = get_pid_by_legacy_recid(recid)
253272
oldest_record = self.get_oldest_record(parent_pid.pid_value)
254273
parent = RDMParent.pid.resolve(parent_pid.pid_value)
255274
community = parent.communities.default
256275
record_owner_id = parent.access.owned_by.owner_id
276+
if dry_run:
277+
logger.info(f"Dry loading legacy comments for recid: {recid}")
278+
return None
257279
request = self.create_accepted_community_submission_request(
258-
oldest_record, community, record_owner_id, comments
280+
recid, oldest_record, community, record_owner_id, comments
259281
)
260282
return request
261283

@@ -264,9 +286,17 @@ def _load(self, entry):
264286
if entry:
265287
recid, comments = entry
266288
try:
267-
self._process_legacy_comments_for_recid(recid, comments)
289+
self._process_legacy_comments_for_recid(recid, comments, self.dry_run)
290+
except ManualImportRequired as ex:
291+
error_message = (
292+
f"Error: {ex.message} | "
293+
f"Value: {ex.value} | "
294+
f"Recid: {recid} | "
295+
f"Comment ID: {ex.field}"
296+
)
297+
logger.error(error_message)
268298
except Exception as ex:
269-
logger.error(ex)
299+
logger.error(f"Error: {ex} | Recid: {recid}")
270300

271301
def _cleanup(self, *args, **kwargs):
272302
"""Cleanup the entries."""

‎tests/cds-rdm/conftest.py‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -165,6 +165,8 @@ def app_config(app_config):
165165
app_config["CDS_MIGRATOR_KIT_LOGS_PATH"] = logs_dir
166166
app_config["RDM_CUSTOM_FIELDS"] = CUSTOM_FIELDS
167167
app_config["CDS_MIGRATOR_KIT_ENV"] = "test"
168+
app_config["CDS_MIGRATOR_KIT_SITE_UI_URL"] = "https://localhost:5000"
169+
app_config["CDS_MIGRATOR_KIT_SITE_API_URL"] = "https://localhost:5000/api"
168170

169171
return app_config
170172

‎tests/cds-rdm/data/comments/23456/4/content.pdf‎

13 KB
Binary file not shown.

‎tests/cds-rdm/data/comments/comments_metadata.json‎

Lines changed: 62 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -5,17 +5,77 @@
55
"content": "This is a test comment",
66
"status": "ok",
77
"user_email": "submitter13@cern.ch",
8-
"created_at": "2026-01-10 10:00:00",
8+
"created_at": "2025-01-10 10:00:00",
99
"file_relation": {},
1010
"replies": [
1111
{
1212
"comment_id": 2,
1313
"content": "This is a reply",
1414
"status": "ok",
1515
"user_email": "submitter10@gmail.com",
16-
"created_at": "2026-02-10 11:00:00",
16+
"created_at": "2025-02-10 11:00:00",
1717
"reply_to_id": 1,
1818
"file_relation": {}
19+
},
20+
{
21+
"comment_id": 3,
22+
"content": "This is a deleted comment",
23+
"status": "da",
24+
"user_email": "submitter10@gmail.com",
25+
"created_at": "2025-02-10 11:00:00",
26+
"reply_to_id": 1,
27+
"file_relation": {}
28+
}
29+
]
30+
}
31+
],
32+
"23456": [
33+
{
34+
"comment_id": 4,
35+
"content": "This is a test comment with attached files",
36+
"status": "ok",
37+
"user_email": "submitter13@cern.ch",
38+
"created_at": "2024-01-10 10:00:00",
39+
"file_relation": {}
40+
}
41+
],
42+
"34567": [
43+
{
44+
"comment_id": 5,
45+
"content": "This is a comment with an unknown user",
46+
"status": "ok",
47+
"user_email": "unknown@example.com",
48+
"created_at": "2022-01-10 10:00:00",
49+
"file_relation": {}
50+
}
51+
],
52+
"45678": [
53+
{
54+
"comment_id": 6,
55+
"content": "This is a comment related to a file",
56+
"status": "ok",
57+
"user_email": "submitter13@cern.ch",
58+
"created_at": "2021-01-10 10:00:00",
59+
"file_relation": {
60+
"file_id": "example.pdf",
61+
"version": 1
62+
},
63+
"replies": [
64+
{
65+
"comment_id": 7,
66+
"content": "This is a reply to a comment.",
67+
"status": "ok",
68+
"user_email": "submitter13@cern.ch",
69+
"created_at": "2021-01-10 10:00:00",
70+
"reply_to_id": 6
71+
},
72+
{
73+
"comment_id": 8,
74+
"content": "This is a reply to a reply.",
75+
"status": "ok",
76+
"user_email": "submitter13@cern.ch",
77+
"created_at": "2021-01-10 10:00:00",
78+
"reply_to_id": 7
1979
}
2080
]
2181
}

0 commit comments

Comments
 (0)