Sync starts by getting latest post for each board

This commit is contained in:
towards-a-new-leftypol 2025-01-30 18:00:42 -05:00
parent 518467c7eb
commit 1113539321
8 changed files with 57 additions and 20 deletions

View File

@ -80,6 +80,7 @@ executable chan-delorean
Data.WordUtil
Network.DataClient
Network.DataClientTypes
Network.GetLatestPostsPerBoardResponse
Common.Server.JSONSettings
Common.Server.ConsumerSettings

View File

@ -13,5 +13,5 @@
"jwt": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJyb2xlIjoiY2hhbl9hcmNoaXZlciJ9.rGIKZokTDKTuQLIv8138bUby5PELfDipYYIDpJzH02c",
"media_root_path": "/home/phil/linixy/tmp/chan_archive_media_repaired/archive",
"http_fill_all": false,
"http_sync_continously": false
"http_sync_continously": true
}

View File

@ -290,6 +290,7 @@ SELECT DISTINCT ON (b.board_id)
JOIN posts p ON p.thread_id = t.thread_id
ORDER BY b.board_id, p.creation_time DESC;
CREATE OR REPLACE FUNCTION get_latest_posts_per_board()
RETURNS TABLE (
board_id int,
@ -298,7 +299,6 @@ RETURNS TABLE (
post_id bigint,
board_post_id bigint,
creation_time timestamp with time zone,
body text,
thread_id bigint,
board_thread_id bigint
) AS $$
@ -309,7 +309,6 @@ RETURNS TABLE (
p.post_id,
p.board_post_id,
p.creation_time,
p.body,
t.thread_id,
t.board_thread_id
FROM boards b
@ -318,4 +317,4 @@ RETURNS TABLE (
ORDER BY b.board_id, p.creation_time DESC;
$$ LANGUAGE sql STABLE;
SELECT * FROM get_latest_posts_per_board();
SELECT * FROM get_latest_posts_per_board();

View File

@ -409,7 +409,6 @@ RETURNS TABLE (
post_id bigint,
board_post_id bigint,
creation_time timestamp with time zone,
body text,
thread_id bigint,
board_thread_id bigint
) AS $$
@ -420,7 +419,6 @@ RETURNS TABLE (
p.post_id,
p.board_post_id,
p.creation_time,
p.body,
t.thread_id,
t.board_thread_id
FROM boards b

View File

@ -1,5 +1,7 @@
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE DeriveAnyClass #-}
{-# OPTIONS_GHC -Wno-unrecognised-pragmas #-}
{-# HLINT ignore "Use <&>" #-}
module Network.DataClient
( HttpError(..)
@ -19,6 +21,7 @@ module Network.DataClient
, postAttachments
, getJSON
, getFile
, getLatestPostsPerBoard
) where
import Control.Monad (forM)
@ -49,6 +52,7 @@ import qualified Common.AttachmentType as Attachments
import qualified Common.PostsType as Posts
import Common.Network.HttpClient
import qualified Network.DataClientTypes as T
import qualified Network.GetLatestPostsPerBoardResponse as GLPPBR
data PostId = PostId
@ -233,14 +237,20 @@ getFile url = do
case result of
Left (err :: HttpError) -> do
putStrLn $ "getFile " ++ url ++ " Error!"
putStrLn $ show err
print err
return Nothing
Right lbs -> do
putStrLn $ "getFile " ++ url ++ " SUCCESS!"
tmp_root <- getCanonicalTemporaryDirectory
(tmp_filepath, tmp_filehandle) <- openBinaryTempFile tmp_root "chan.attachment"
putStrLn $ "Created " ++ tmp_filepath
putStrLn $ "Writing attachment..."
putStrLn "Writing attachment..."
LBS.hPut tmp_filehandle lbs
hClose tmp_filehandle
return $ Just tmp_filepath
-- | Function to handle each chunk.
getLatestPostsPerBoard :: T.JSONSettings -> IO (Either HttpError [ GLPPBR.GetLatestPostsPerBoardResponse ])
getLatestPostsPerBoard settings =
post settings "/rpc/get_latest_posts_per_board" mempty False >>= return . eitherDecodeResponse

View File

@ -10,4 +10,3 @@ data ThreadMaxIdx = ThreadMaxIdx
{ thread_id :: Int64
, max_idx :: Int
} deriving (Show, Generic, FromJSON)

View File

@ -0,0 +1,20 @@
{-# LANGUAGE DeriveAnyClass #-}
module Network.GetLatestPostsPerBoardResponse
where
import Data.Int (Int64)
import Data.Time.Clock (UTCTime)
import Data.Aeson (FromJSON)
import GHC.Generics
data GetLatestPostsPerBoardResponse = GetLatestPostsPerBoardResponse
{ board_id :: Int
, site_id :: Int
, pathpart :: String
, post_id :: Maybe Int64
, board_post_id :: Int64
, creation_time :: UTCTime
, thread_id :: Int64
, board_thread_id :: Integer
} deriving (Show, Generic, FromJSON)

View File

@ -1,21 +1,31 @@
{-# LANGUAGE RecordWildCards #-}
module Sync where
import Common.Server.ConsumerSettings
import Lib (getBoards, toClientSettings)
import SitesType (Site)
import BoardsType (Board)
import Common.Server.ConsumerSettings as Settings
import Common.Server.JSONSettings as JSONSettings
import Network.DataClient (getLatestPostsPerBoard)
getSiteBoards :: ConsumerJSONSettings -> JSONSiteSettings -> IO (Site, [ Board ])
getSiteBoards settings site_settings =
let client_settings = toClientSettings settings site_settings
in getBoards
client_settings
(boards site_settings)
consumerSettingsToPartialJSONSettings :: Settings.ConsumerJSONSettings -> JSONSettings.JSONSettings
consumerSettingsToPartialJSONSettings ConsumerJSONSettings {..} =
JSONSettings
{ JSONSettings.postgrest_url = postgrest_url
, JSONSettings.jwt = jwt
, backup_read_root = undefined
, JSONSettings.media_root_path
, site_name = undefined
, site_url = undefined
}
syncWebsites :: ConsumerJSONSettings -> IO ()
syncWebsites _ = do
syncWebsites consumer_settings = do
putStrLn "Starting channel web synchronization."
let json_settings = consumerSettingsToPartialJSONSettings consumer_settings
asdf <- getLatestPostsPerBoard json_settings
print asdf
-- first we need all the (Site, Board) tuples
-- perhaps we even want all (Site, Board, Thread) pairs
-- But then we don't load the posts of each thread, instead only do