From 4894f005f0d484d5203c8d5e141fa1bfd15a71c4 Mon Sep 17 00:00:00 2001
From: roadscape <roadscape@users.noreply.github.com>
Date: Wed, 18 Sep 2019 10:13:34 -0500
Subject: [PATCH] test basic mentions match
---
hive/indexer/cached_post.py | 6 ++++-
hive/utils/post.py | 42 ++++++++--------------------------
tests/utils/test_utils_post.py | 13 ++++-------
3 files changed, 20 insertions(+), 41 deletions(-)
diff --git a/hive/indexer/cached_post.py b/hive/indexer/cached_post.py
index 48fb797ef..afa15c91c 100644
--- a/hive/indexer/cached_post.py
+++ b/hive/indexer/cached_post.py
@@ -560,7 +560,11 @@ class CachedPost:
Notify(notif_type, src_id=author_id, dst_id=parent_id,
post_id=pid, when=post['last_update'], score=score).write()
if level in ('insert', 'update'):
- mentions(post)
+ accts = mentions(post['body'])
+ for acct in accts:
+ if not Accounts.exists(acct):
+ url = '@' + post['author'] + '/' + post['permlink']
+ log.warning("bad mention [%s] in %s", acct, url)
@classmethod
diff --git a/hive/utils/post.py b/hive/utils/post.py
index 6a7ffd130..c91e0f03c 100644
--- a/hive/utils/post.py
+++ b/hive/utils/post.py
@@ -11,39 +11,17 @@ from hive.utils.normalize import sbd_amount, rep_log10, safe_img_url, parse_time
log = logging.getLogger(__name__)
-def mentions(post):
- """Given a post, return proper @-mentioned account names."""
- # pylint: disable=invalid-name
- detected = text_mentions(post['body'])
- provided = _post_users(post)
-
- d1 = detected - provided
- d2 = provided - detected
-
- url = '@' + post['author'] + '/' + post['permlink']
- if d1: log.warning("%s detected - provided: %s", url, d1)
- if d2: log.warning("%s provided - detected: %s", url, d2)
-
- return detected & provided
-
-def text_mentions(body):
+def mentions(body):
"""Given a post body, return proper @-mentioned account names."""
- matches = re.findall('(?:^|[^a-zA-Z0-9_!#$%&*@])(:?@)([a-z\\d\\-.]+)', body)
- return {grp[1] for grp in matches}
-
-def _post_users(post):
- """Retrieve `users` key from json_metadata."""
- md = {}
- try:
- md = json.loads(post['json_metadata'])
- if not isinstance(md, dict):
- md = {}
- except Exception:
- pass
-
- if 'users' in md and isinstance(md['users'], list):
- return {user.strip('.@') for user in md['users'] if user and isinstance(user, str)}
- return set()
+ # condenser:
+ # /(^|[^a-zA-Z0-9_!#$%&*@ï¼ \/]|(^|[^a-zA-Z0-9_+~.-\/#]))[@ï¼ ]([a-z][-\.a-z\d]+[a-z\d])/gi,
+
+ matches = re.findall(
+ '(?:^|[^a-zA-Z0-9_!#$%&*@\\/])'
+ '(:?@)'
+ '([a-zA-Z0-9][a-zA-Z0-9\\-.]{1,14}[a-zA-Z0-9])'
+ '(?![a-z])', body)
+ return {grp[1].lower() for grp in matches}
def post_basic(post):
"""Basic post normalization: json-md, tags, and flags."""
diff --git a/tests/utils/test_utils_post.py b/tests/utils/test_utils_post.py
index a604cbe27..a17a6bd45 100644
--- a/tests/utils/test_utils_post.py
+++ b/tests/utils/test_utils_post.py
@@ -3,7 +3,6 @@ from decimal import Decimal
from hive.utils.post import (
mentions,
- text_mentions,
post_basic,
post_legacy,
post_payout,
@@ -140,19 +139,17 @@ POST_2 = {
}
def test_mentions():
- post = {'body': 'Who is @abc and @foo and @bar',
- 'json_metadata': '{"users":["foo","bar"]}'}
- assert mentions(post) == {'foo', 'bar'}
-
-def test_text_mentions():
# pylint: disable=invalid-name
- m = text_mentions
+ m = mentions
assert m('Hi @abc, meet @bob') == {'abc', 'bob'}
assert m('Hi @abc, meet @abc') == {'abc'}
assert not m('')
assert not m('@')
- assert m('steemit.com/@apple') == {'apple'}
+ assert not m('steemit.com/@apple')
assert not m('joe@apple.com')
+ assert m('@longestokaccount') == {'longestokaccount'}
+ assert not m('@longestokaccountx')
+ assert m('@abc- @-foo @bar.') == {'abc', 'bar'}
def test_post_basic():
ret = post_basic(POST_1)
--
GitLab