From cc3b801abf9a28946fffb64312bbad4d14030c70 Mon Sep 17 00:00:00 2001 From: CIC4DA Date: Sun, 14 Jan 2024 20:25:57 +0530 Subject: [PATCH] Zerver: Improve topic names for Slack threads --- zerver/data_import/slack.py | 14 +++++++++----- zerver/tests/test_slack_importer.py | 23 ++++++++++++++--------- 2 files changed, 23 insertions(+), 14 deletions(-) diff --git a/zerver/data_import/slack.py b/zerver/data_import/slack.py index 8f98f6cb32..16c1f9a4d3 100644 --- a/zerver/data_import/slack.py +++ b/zerver/data_import/slack.py @@ -57,6 +57,7 @@ from zerver.models import ( Recipient, UserProfile, ) +from zerver.models.constants import MAX_TOPIC_NAME_LENGTH SlackToZulipUserIDT: TypeAlias = Dict[str, int] AddedChannelsT: TypeAlias = Dict[str, Tuple[str, int]] @@ -976,20 +977,23 @@ def channel_message_to_zerver_message( # Slack's unthreaded messages go into a single topic, while # threads each generate a unique topic labeled by the date and - # a counter among topics on that day. + # message content. topic_name = "imported from Slack" if convert_slack_threads and "thread_ts" in message: thread_ts = datetime.fromtimestamp(float(message["thread_ts"]), tz=timezone.utc) thread_ts_str = thread_ts.strftime(r"%Y/%m/%d %H:%M:%S") - # The topic name is "2015-08-18 Slack thread 2", where the counter at the end is to disambiguate - # threads with the same date. + thread_message = content[: min(MAX_TOPIC_NAME_LENGTH, len(content))] + + # The topic name is "2015-08-18 content[:49]" or "2015-08-18 content[:48]…" + # using the unicode ellipsis character (…) -- it's only one character long if thread_ts_str in thread_map: topic_name = thread_map[thread_ts_str] else: thread_date = thread_ts.strftime(r"%Y-%m-%d") thread_counter[thread_date] += 1 - count = thread_counter[thread_date] - topic_name = f"{thread_date} Slack thread {count}" + topic_name = f"{thread_date} {thread_message}" + if len(topic_name) > MAX_TOPIC_NAME_LENGTH: + topic_name = topic_name[: (MAX_TOPIC_NAME_LENGTH - 1)] + "…" thread_map[thread_ts_str] = topic_name zulip_message = build_message( diff --git a/zerver/tests/test_slack_importer.py b/zerver/tests/test_slack_importer.py index 45ad1430d5..83be9bbccc 100644 --- a/zerver/tests/test_slack_importer.py +++ b/zerver/tests/test_slack_importer.py @@ -1131,7 +1131,7 @@ class SlackImporter(ZulipTestCase): "channel_name": "random", }, { - "text": "random", + "text": "A random text smaller than 60 characters", "user": "U061A5N1G", "ts": "1439868294.000008", # A different Thread! @@ -1139,7 +1139,7 @@ class SlackImporter(ZulipTestCase): "channel_name": "random", }, { - "text": "random", + "text": "A random text which is greater than 60 characters is used", "user": "U061A5N1G", "ts": "1439868295.000008", # Another different Thread! @@ -1190,13 +1190,18 @@ class SlackImporter(ZulipTestCase): self.assertEqual(zerver_message[0]["content"], "@**Jane**: hey!") self.assertEqual(zerver_message[0]["has_link"], False) self.assertEqual(zerver_message[1]["content"], "random") - self.assertEqual(zerver_message[1][EXPORT_TOPIC_NAME], "2015-06-12 Slack thread 1") - self.assertEqual(zerver_message[2][EXPORT_TOPIC_NAME], "2015-06-12 Slack thread 1") - # A new thread with a different date from 2015-06-12, starts the counter from 1. - self.assertEqual(zerver_message[3][EXPORT_TOPIC_NAME], "2015-08-18 Slack thread 1") - # A new thread with a different timestamp, but the same date as 2015-08-18, starts the - # counter from 2. - self.assertEqual(zerver_message[4][EXPORT_TOPIC_NAME], "2015-08-18 Slack thread 2") + self.assertEqual(zerver_message[1][EXPORT_TOPIC_NAME], "2015-06-12 random") + self.assertEqual(zerver_message[2][EXPORT_TOPIC_NAME], "2015-06-12 random") + # A new thread with a different date from 2015-06-12, with content length smaller than 60 characters. + self.assertEqual( + zerver_message[3][EXPORT_TOPIC_NAME], + "2015-08-18 A random text smaller than 60 characters", + ) + # A new thread with a different timestamp, but the same date as 2015-08-18, with content length greater than 60 characters. + self.assertEqual( + zerver_message[4][EXPORT_TOPIC_NAME], + "2015-08-18 A random text which is greater than 60 character…", + ) self.assertEqual( zerver_message[1]["recipient"], slack_recipient_name_to_zulip_recipient_id["random"] )