From 5ea1ef51f745a160b7751a14f12b5ccff5b15ddd Mon Sep 17 00:00:00 2001
From: mtrela <mtrela@syncad.com>
Date: Tue, 24 Nov 2020 12:22:22 +0100
Subject: [PATCH] Data from `hive_reblogs` is removed when a post is deleted
---
hive/db/schema.py | 2 +-
...elete.sql => delete_reblog_feed_cache.sql} | 6 ++-
hive/db/sql_scripts/hive_post_operations.sql | 49 ++++++++++++-------
hive/indexer/feed_cache.py | 29 -----------
hive/indexer/posts.py | 19 ++-----
hive/indexer/reblog.py | 3 +-
6 files changed, 41 insertions(+), 67 deletions(-)
rename hive/db/sql_scripts/{reblog_delete.sql => delete_reblog_feed_cache.sql} (72%)
delete mode 100644 hive/indexer/feed_cache.py
diff --git a/hive/db/schema.py b/hive/db/schema.py
index 9709fd39d..082ee6784 100644
--- a/hive/db/schema.py
+++ b/hive/db/schema.py
@@ -612,7 +612,7 @@ def setup(db):
"condenser_get_discussions_by_comments.sql",
"condenser_get_account_reputations.sql",
"update_follow_count.sql",
- "reblog_delete.sql"
+ "delete_reblog_feed_cache.sql"
]
from os.path import dirname, realpath
dir_path = dirname(realpath(__file__))
diff --git a/hive/db/sql_scripts/reblog_delete.sql b/hive/db/sql_scripts/delete_reblog_feed_cache.sql
similarity index 72%
rename from hive/db/sql_scripts/reblog_delete.sql
rename to hive/db/sql_scripts/delete_reblog_feed_cache.sql
index 332027849..6b26e2314 100644
--- a/hive/db/sql_scripts/reblog_delete.sql
+++ b/hive/db/sql_scripts/delete_reblog_feed_cache.sql
@@ -1,7 +1,9 @@
-DROP FUNCTION IF EXISTS delete_reblog(character varying,character varying,character varying, BOOLEAN)
+delete_reblog_feed_cache
+
+DROP FUNCTION IF EXISTS delete_reblog_feed_cache(character varying,character varying,character varying, BOOLEAN)
;
-CREATE OR REPLACE FUNCTION delete_reblog(
+CREATE OR REPLACE FUNCTION delete_reblog_feed_cache(
in _author hive_accounts.name%TYPE,
in _permlink hive_permlink_data.permlink%TYPE,
in _account hive_accounts.name%TYPE,
diff --git a/hive/db/sql_scripts/hive_post_operations.sql b/hive/db/sql_scripts/hive_post_operations.sql
index 2bf71e82f..800b28b11 100644
--- a/hive/db/sql_scripts/hive_post_operations.sql
+++ b/hive/db/sql_scripts/hive_post_operations.sql
@@ -109,32 +109,45 @@ END
$function$
;
-DROP FUNCTION if exists delete_hive_post(character varying,character varying,character varying, integer)
+DROP FUNCTION if exists delete_hive_post(character varying,character varying,character varying, integer, BOOLEAN)
;
CREATE OR REPLACE FUNCTION delete_hive_post(
in _author hive_accounts.name%TYPE,
in _permlink hive_permlink_data.permlink%TYPE,
- in _block_num hive_blocks.num%TYPE)
-RETURNS TABLE (id hive_posts.id%TYPE, depth hive_posts.depth%TYPE)
+ in _block_num hive_blocks.num%TYPE,
+ in _delete_feed_cache BOOLEAN)
+RETURNS VOID
LANGUAGE plpgsql
AS
$function$
+DECLARE
+ __post_id INT;
BEGIN
- RETURN QUERY UPDATE hive_posts AS hp
- SET counter_deleted =
- (
- SELECT max( hps.counter_deleted ) + 1
- FROM hive_posts hps
- INNER JOIN hive_accounts ha ON hps.author_id = ha.id
- INNER JOIN hive_permlink_data hpd ON hps.permlink_id = hpd.id
- WHERE ha.name = _author AND hpd.permlink = _permlink
- )
- , block_num = _block_num
- FROM hive_posts hp1
- INNER JOIN hive_accounts ha ON hp1.author_id = ha.id
- INNER JOIN hive_permlink_data hpd ON hp1.permlink_id = hpd.id
- WHERE hp.id = hp1.id AND ha.name = _author AND hpd.permlink = _permlink AND hp1.counter_deleted = 0
- RETURNING hp.id, hp.depth;
+
+ __post_id = find_comment_id( _author, _permlink, False );
+
+ IF __post_id = 0 THEN
+ RETURN;
+ END IF;
+
+ UPDATE hive_posts
+ SET counter_deleted =
+ (
+ SELECT max( counter_deleted ) + 1
+ FROM hive_posts
+ WHERE id = __post_id
+ )
+ ,block_num = _block_num
+ WHERE id = __post_id AND counter_deleted = 0;
+
+ DELETE FROM hive_reblogs
+ WHERE post_id = __post_id;
+
+ IF _delete_feed_cache THEN
+ DELETE FROM hive_feed_cache
+ WHERE post_id = __post_id;
+ END IF;
+
END
$function$
;
diff --git a/hive/indexer/feed_cache.py b/hive/indexer/feed_cache.py
deleted file mode 100644
index 3dc4d3b32..000000000
--- a/hive/indexer/feed_cache.py
+++ /dev/null
@@ -1,29 +0,0 @@
-"""Maintains feed cache (blogs + reblogs)"""
-
-import logging
-from hive.db.adapter import Db
-from hive.indexer.db_adapter_holder import DbAdapterHolder
-
-log = logging.getLogger(__name__)
-
-DB = Db.instance()
-
-class FeedCache(DbAdapterHolder):
- """Maintains `hive_feed_cache`, which merges posts and reports.
-
- The feed cache allows for efficient querying of posts + reblogs,
- savings us from expensive queries. Effectively a materialized view.
- """
- @classmethod
- def delete(cls, post_id, account_id=None):
- """Remove a post from feed cache.
-
- If `account_id` is specified, we remove a single entry (e.g. a
- singular un-reblog). Otherwise, we remove all instances of the
- post (e.g. a post was deleted; its entry and all reblogs need
- to be removed.
- """
- sql = "DELETE FROM hive_feed_cache WHERE post_id = :id"
- if account_id:
- sql = sql + " AND account_id = :account_id"
- DB.query(sql, account_id=account_id, id=post_id)
diff --git a/hive/indexer/posts.py b/hive/indexer/posts.py
index 08edcbb58..732d205f2 100644
--- a/hive/indexer/posts.py
+++ b/hive/indexer/posts.py
@@ -10,7 +10,7 @@ from diff_match_patch import diff_match_patch
from hive.db.adapter import Db
from hive.db.db_state import DbState
-from hive.indexer.feed_cache import FeedCache
+from hive.indexer.reblog import Reblog
from hive.indexer.community import Community
from hive.indexer.notify import Notify
from hive.indexer.post_data_cache import PostDataCache
@@ -400,21 +400,10 @@ class Posts(DbAdapterHolder):
def delete(cls, op):
"""Marks a post record as being deleted."""
- sql = """
- SELECT id, depth
- FROM delete_hive_post((:author)::varchar, (:permlink)::varchar, (:block_num)::int);
- """
- row = DB.query_row(sql, author=op['author'], permlink = op['permlink'], block_num=op['block_num'])
-
- result = dict(row)
- pid = result['id']
-
- depth = result['depth']
+ delete_feed_cache = not DbState.is_initial_sync()
- if depth == 0 and not DbState.is_initial_sync():
- # TODO: delete from hive_reblogs -- otherwise feed cache gets
- # populated with deleted posts somwrimas
- FeedCache.delete(pid)
+ sql = "SELECT delete_hive_post((:author)::varchar, (:permlink)::varchar, (:block_num)::int, (:delete_feed_cache)::BOOLEAN );"
+ DB.query_no_return(sql, author=op['author'], permlink = op['permlink'], block_num=op['block_num'], delete_feed_cache=delete_feed_cache)
@classmethod
def _verify_post_against_community(cls, op, community_id, is_valid, is_muted):
diff --git a/hive/indexer/reblog.py b/hive/indexer/reblog.py
index a0023fe63..0647e29ec 100644
--- a/hive/indexer/reblog.py
+++ b/hive/indexer/reblog.py
@@ -5,7 +5,6 @@ import logging
from hive.db.db_state import DbState
from hive.indexer.accounts import Accounts
-from hive.indexer.feed_cache import FeedCache
from hive.indexer.db_adapter_holder import DbAdapterHolder
from hive.utils.normalize import escape_characters
@@ -54,7 +53,7 @@ class Reblog(DbAdapterHolder):
"""
delete_feed_cache = not DbState.is_initial_sync()
- sql = "SELECT delete_reblog( (:author)::VARCHAR, (:permlink)::VARCHAR, (:permlink)::VARCHAR, (:delete_feed_cache)::BOOLEAN );"
+ sql = "SELECT delete_reblog_feed_cache( (:author)::VARCHAR, (:permlink)::VARCHAR, (:permlink)::VARCHAR, (:delete_feed_cache)::BOOLEAN );"
status = await db.query_col(sql, author=author, permlink=permlink, account=account, delete_feed_cache=delete_feed_cache);
assert status is not None
if status == 0:
--
GitLab