Factor out doIndex

README.md

@@ -44,3 +44,5 @@ By setting `$BASE_DIR` you can persist the database for later runs.
 - CI
 - Docker container (in flake)
 - try fourmolu
+- Test leap second handling in the `time` package
+- use queue to sync logging of indexer and main thread

app/Main.hs

@@ -10,14 +10,18 @@
 
 module Main (main) where
 
+import Control.Concurrent (forkIO)
+import Control.Monad (forM, forM_)
 import Control.Monad.IO.Class (liftIO)
-import Control.Monad.Trans.Except (ExceptT (..), runExceptT)
+import Control.Monad.Trans.Except (ExceptT (..), catchE, runExceptT)
+import Data.Bifunctor (Bifunctor (..))
 import Data.Proxy (Proxy (..))
 import Data.Text (Text)
-import Data.Time (Day, toGregorian)
+import Data.Time (Day, LocalTime (..), TimeOfDay (..), ZonedTime (..), addDays, toGregorian)
 import GHC.Generics (Generic)
 import Network.HTTP.Media ((//), (/:))
 import Network.HTTP.Types (Status (..), mkStatus, status200)
+import Network.Wai (Application, Request (..), Response, responseFile, responseLBS)
 import Servant
   ( Accept (..)
   , Capture
@@ -48,14 +52,16 @@ import qualified Data.ByteString.Lazy as LBS
 import qualified Data.Text as Text
 import qualified Data.Text.Encoding as Encoding
 import qualified Database.PostgreSQL.Opium as Opium
-import Network.Wai (Application, Request (..), Response, responseFile, responseLBS)
 import qualified Network.Wai.Handler.Warp as Warp
 import qualified Text.Blaze.Html5 as H
 import qualified Text.Blaze.Html5.Attributes as A
 
 import Envy (type (=@!), type (=@@), type (?))
-import Yore.DB (DayFile (..))
+import Yore.DB (DB, DayFile (..))
+import Yore.Download (downloadInto)
 import Yore.Error (Error (..))
+import Yore.Schedule (schedule)
+import Yore.Scrape (Issue (..), getIssuesByDay)
 import Yore.Time (addYears, getCurrentDay, getSecondsUntilMidnight)
 
 import qualified Envy
@@ -89,7 +95,38 @@ main = do
       Right c ->
         pure c
 
-  let
+  db <- DB.initDB cfg.yoreDb
+
+  _ <- runExceptT $ doIndex cfg db
+
+  _ <- forkIO $ runIndexer cfg db
+  runServer cfg db
+
+runIndexer :: Config -> DB -> IO ()
+runIndexer cfg db =
+  schedule shouldRunAt $ runExceptT $ doIndex cfg db
+  where
+    shouldRunAt zt = case localTimeOfDay $ zonedTimeToLocalTime zt of
+      TimeOfDay 3 0 _ -> True
+      _ -> False
+
+doIndex :: Config -> DB -> ExceptT Error IO ()
+doIndex cfg db =
+  catchE
+    ( DB.withTransaction db $ \conn -> do
+        dayThen <- getTodayWithOffset (-100) 0
+        indexDay cfg dayThen conn
+        dayThen2 <- getTodayWithOffset (-100) 1
+        indexDay cfg dayThen2 conn
+        dayThen3 <- getTodayWithOffset (-100) 2
+        indexDay cfg dayThen3 conn
+    )
+    (Log.error . show)
+
+runServer :: Config -> DB -> IO ()
+runServer cfg db =
+  Warp.runSettings settings $ serve (Proxy @API) $ hoistServer (Proxy @API) nt $ server cfg db
+  where
     settings =
       foldr
         ($)
@@ -100,9 +137,6 @@ main = do
         , Warp.setOnException onException
         ]
 
-  db <- DB.initDB cfg.yoreDb
-  Warp.runSettings settings $ serve (Proxy @API) $ hoistServer (Proxy @API) nt $ server cfg db
-  where
     logger req status _ = do
       Log.info $
         printf
@@ -158,20 +192,20 @@ handlerToRaw handler = Tagged $ \_ respond -> do
     Right response ->
       respond response
 
-server :: Config -> DB.DB -> ServerT API (ExceptT Error IO)
+server :: Config -> DB -> ServerT API (ExceptT Error IO)
 server cfg db = rootR :<|> todayR :<|> apiTodayR
   where
     rootR = todayR 0
 
     todayR issue = do
       dateThen <- ExceptT get100YearsAgo
-      count <- ExceptT $ DB.withConn db $ DB.getNumberOfIssues dateThen
+      count <- DB.withConn db $ DB.getNumberOfIssues dateThen
-      dayFile <- ExceptT $ DB.withConn db $ DB.getDayFileByIssue dateThen issue
+      dayFile <- DB.withConn db $ DB.getDayFileByIssue dateThen issue
       pure $ RootModel dateThen dayFile issue count
 
     apiTodayR issue = handlerToRaw $ do
       dateThen <- ExceptT get100YearsAgo
-      dayFile <- ExceptT $ DB.withConn db $ DB.getDayFileByIssue dateThen issue
+      dayFile <- DB.withConn db $ DB.getDayFileByIssue dateThen issue
       let fullPath = cfg.yoreDownloadDir </> dayFile.relative_path
       secondsUntilMidnight <- liftIO getSecondsUntilMidnight
       pure $
@@ -223,5 +257,33 @@ instance Accept HTML where
   contentType _ = "text" // "html" /: ("charset", "utf-8")
 
 get100YearsAgo :: IO (Either Error Day)
-get100YearsAgo =
+get100YearsAgo = runExceptT $ getTodayWithOffset (-100) 0
-  either (Left . GenericError . ("can't go back 100 years: " <>)) Right . addYears (-100) <$> getCurrentDay
+
+getTodayWithOffset :: Integer -> Integer -> ExceptT Error IO Day
+getTodayWithOffset yearOffset dayOffset =
+  ExceptT $
+    first (GenericError . (Text.pack (printf "can't go back %d years and go forward %d days: " (-yearOffset) dayOffset) <>))
+      . addYears yearOffset
+      . addDays dayOffset
+      <$> getCurrentDay
+
+indexDay :: Config -> Day -> Opium.Connection -> ExceptT Error IO ()
+indexDay cfg dayThen conn = do
+  -- Transaction-level lock released automatically after transaction
+  DB.lift $ Opium.execute_ "SELECT pg_advisory_xact_lock(42);" conn
+  mbDi <- DB.readDayIndex dayThen conn
+  case mbDi of
+    Just _ ->
+      Log.info $ printf "index for %s already exists." (show dayThen)
+    Nothing -> do
+      Log.info $ printf "scraping issues for %s" (show dayThen)
+      issues <- liftIO $ getIssuesByDay dayThen
+      paths <- forM issues $ \issue -> do
+        Log.info $ printf "downloading %s" issue.url
+        path <- liftIO $ downloadInto cfg.yoreDownloadDir issue.url
+        pure (issue.label, path)
+      Log.info "creating DB entries"
+      dayIndex <- DB.createDayIndex dayThen conn
+      forM_ paths $ \(text, url) ->
+        DB.createDayFile dayIndex.day_index_id text url conn
+      Log.info "done."

scripts/dev

@@ -1,3 +0,0 @@
-#!/usr/bin/env bash
-
-YORE_DB="host=localhost port=${DB_PORT} user=${DB_USER} dbname=${DB_DBNAME}" cabal repl exe:yore

scripts/with-db

@@ -26,4 +26,6 @@ fi
 
 DATABASE_URL="postgres://${DB_USER}@localhost:${DB_PORT}/${DB_DBNAME}?sslmode=disable" dbmate up
 
+export YORE_DB="host=localhost port=${DB_PORT} user=${DB_USER} dbname=${DB_DBNAME}" 
+
 "$@"

src/Yore/DB.hs

@@ -9,6 +9,7 @@ module Yore.DB
   , Error (..)
   , initDB
   , withConn
+  , withTransaction
   , DayIndex (..)
   , DayFile (..)
   , createDayFile
@@ -17,9 +18,11 @@ module Yore.DB
   , readDayPaths
   , getDayFileByIssue
   , getNumberOfIssues
+  , lift
   ) where
 
 import Control.Concurrent (getNumCapabilities)
+import Control.Monad.Trans.Except (ExceptT (..), catchE, runExceptT, throwE)
 import Data.Functor.Identity (Identity (..))
 import Data.Pool (Pool, defaultPoolConfig, newPool, withResource)
 import Data.Text (Text)
@@ -28,6 +31,7 @@ import GHC.Generics (Generic)
 
 import qualified Database.PostgreSQL.Opium as Opium
 
+import Data.Bifunctor (Bifunctor (..))
 import Yore.Error (Error (..))
 
 newtype DB = DB (Pool (Either Opium.ConnectionError Opium.Connection))
@@ -45,13 +49,35 @@ initDB connString = do
           maxResources
       )
 
-withConn :: DB -> (Opium.Connection -> IO (Either Opium.Error a)) -> IO (Either Error a)
+-- TODO: This should probably also do something like bracket...
+-- For now let's assume no exceptions are thrown in f.
+withConn :: DB -> (Opium.Connection -> ExceptT Error IO a) -> ExceptT Error IO a
 withConn (DB connPool) f =
-  withResource connPool $ \case
+  ExceptT $ withResource connPool $ \case
     Left connectionError ->
       pure $ Left $ ConnectionError connectionError
     Right conn ->
-      either (Left . DBError) Right <$> f conn
+      runExceptT $ f conn
+
+withTransaction :: DB -> (Opium.Connection -> ExceptT Error IO a) -> ExceptT Error IO a
+withTransaction db f =
+  withConn db $ \conn ->
+    ( do
+        lift $ Opium.execute_ "BEGIN" conn
+        x <- f conn
+        lift $ Opium.execute_ "COMMIT" conn
+        pure x
+    )
+      `catchE` ( \err -> do
+                  -- rollback open transactions and release transaction level locks...
+                  lift $ Opium.execute_ "ROLLBACK" conn
+                  -- ...then rethrow the error so the app can handle it too.
+                  throwE err
+               )
+
+lift :: IO (Either Opium.Error a) -> ExceptT Error IO a
+lift f =
+  ExceptT $ first DBError <$> f
 
 data DayIndex = DayIndex
   { day_index_id :: Int
@@ -72,44 +98,47 @@ newtype CountResult = CountResult
   }
   deriving (Show, Generic, Opium.FromRow)
 
-createDayIndex :: Day -> Opium.Connection -> IO (Either Opium.Error DayIndex)
+createDayIndex :: Day -> Opium.Connection -> ExceptT Error IO DayIndex
 createDayIndex date =
   ex runIdentity
     . Opium.fetch
       "INSERT INTO yore.day_index (day) VALUES ($1) RETURNING *"
       (Identity date)
 
-createDayFile :: Int -> Text -> FilePath -> Opium.Connection -> IO (Either Opium.Error ())
+createDayFile :: Int -> Text -> FilePath -> Opium.Connection -> ExceptT Error IO ()
 createDayFile dayId label path =
-  Opium.execute
+  lift
-    "INSERT INTO yore.day_file (day_index_id, label, relative_path) VALUES ($1, $2, $3)"
+    . Opium.execute
-    (dayId, label, path)
+      "INSERT INTO yore.day_file (day_index_id, label, relative_path) VALUES ($1, $2, $3)"
+      (dayId, label, path)
 
-readDayIndex :: Day -> Opium.Connection -> IO (Either Opium.Error (Maybe DayIndex))
+readDayIndex :: Day -> Opium.Connection -> ExceptT Error IO (Maybe DayIndex)
 readDayIndex date =
-  Opium.fetch
+  lift
-    "SELECT * FROM yore.day_index WHERE day = $1"
+    . Opium.fetch
-    (Identity date)
+      "SELECT * FROM yore.day_index WHERE day = $1"
+      (Identity date)
 
-readDayPaths :: Day -> Opium.Connection -> IO (Either Opium.Error [(DayIndex, DayFile)])
+readDayPaths :: Day -> Opium.Connection -> ExceptT Error IO [(DayIndex, DayFile)]
 readDayPaths date =
-  Opium.fetch
+  lift
-    "SELECT * FROM yore.day_index LEFT JOIN yore.day_file USING (day_index_id) WHERE day = $1"
+    . Opium.fetch
-    (Identity date)
+      "SELECT * FROM yore.day_index LEFT JOIN yore.day_file USING (day_index_id) WHERE day = $1"
+      (Identity date)
 
-getDayFileByIssue :: Day -> Int -> Opium.Connection -> IO (Either Opium.Error DayFile)
+getDayFileByIssue :: Day -> Int -> Opium.Connection -> ExceptT Error IO DayFile
 getDayFileByIssue date issue =
   ex runIdentity
     . Opium.fetch
       "SELECT yore.day_file.* FROM yore.day_index LEFT JOIN yore.day_file USING (day_index_id) WHERE day = $1 ORDER BY day_file_id LIMIT 1 OFFSET $2"
       (date, issue)
 
-getNumberOfIssues :: Day -> Opium.Connection -> IO (Either Opium.Error Int)
+getNumberOfIssues :: Day -> Opium.Connection -> ExceptT Error IO Int
 getNumberOfIssues date =
   ex (count . runIdentity)
     . Opium.fetch
       "SELECT count(*) FROM yore.day_index LEFT JOIN yore.day_file USING (day_index_id) WHERE day = $1"
       (Identity date)
 
-ex :: (a -> b) -> IO (Either e a) -> IO (Either e b)
+ex :: (a -> b) -> IO (Either Opium.Error a) -> ExceptT Error IO b
-ex = fmap . fmap
+ex f = fmap f . lift

src/Yore/Index.hs

@@ -1,24 +0,0 @@
-{-# LANGUAGE OverloadedRecordDot #-}
-{-# LANGUAGE OverloadedStrings #-}
-
-module Yore.Index (createEntries) where
-
-import Control.Exception (Exception, throwIO)
-import Control.Monad (forM_)
-import Data.Text (Text)
-import Data.Time (Day)
-
-import qualified Database.PostgreSQL.Opium as Opium
-
-import qualified Yore.DB as DB
-
-createEntries :: Day -> [(Text, FilePath)] -> Opium.Connection -> IO ()
-createEntries date urls conn = do
-  confidently $ Opium.execute_ "BEGIN" conn
-  dayIndex <- confidently $ DB.createDayIndex date conn
-  forM_ urls $ \(text, url) ->
-    confidently $ DB.createDayFile dayIndex.day_index_id text url conn
-  confidently $ Opium.execute_ "COMMIT" conn
-
-confidently :: (Exception e) => IO (Either e a) -> IO a
-confidently action = action >>= either throwIO pure

src/Yore/Log.hs

@@ -1,18 +1,19 @@
 module Yore.Log (Yore.Log.error, info) where
 
+import Control.Monad.IO.Class (MonadIO (..))
 import Data.Time (ZonedTime, getZonedTime)
 import Data.Time.Format (defaultTimeLocale, formatTime)
 import GHC.Stack (HasCallStack, SrcLoc (..), callStack, getCallStack)
 import Text.Printf (printf)
 
-info :: (HasCallStack) => String -> IO ()
+info :: (HasCallStack, MonadIO m) => String -> m ()
 info = doLog "INF"
 
-error :: (HasCallStack) => String -> IO ()
+error :: (HasCallStack, MonadIO m) => String -> m ()
 error = doLog "ERR"
 
-doLog :: (HasCallStack) => String -> String -> IO ()
+doLog :: (HasCallStack, MonadIO m) => String -> String -> m ()
-doLog level msg = do
+doLog level msg = liftIO $ do
   now <- getZonedTime
   let location = getLocation $ getCallStack callStack
   printf "(%s) (%s) (%s) %s\n" (iso8601Show now) level location msg

src/Yore/Repl.hs

@@ -1,59 +0,0 @@
-{-# LANGUAGE LambdaCase #-}
-{-# LANGUAGE OverloadedRecordDot #-}
-{-# LANGUAGE OverloadedStrings #-}
-
-module Yore.Repl (connect, exec, getToday, indexDay, addYears) where
-
-import Control.Exception (bracket, throwIO)
-import Control.Monad (forM)
-import Data.Time (Day, getZonedTime, localDay, zonedTimeToLocalTime)
-import Text.Printf (printf)
-
-import qualified Database.PostgreSQL.Opium as Opium
-
-import Yore.DB (DayIndex)
-import Yore.Download (downloadInto)
-import Yore.Index (createEntries)
-import Yore.Scrape (Issue (..), getIssuesByDay)
-import Yore.Time (addYears)
-
-import qualified Yore.DB as DB
-
-connect :: IO Opium.Connection
-connect =
-  Opium.connect "host=localhost port=5433 user=yore-test" >>= \case
-    Left e -> error $ show e
-    Right c -> pure c
-
-exec :: (Opium.Connection -> IO a) -> IO a
-exec = bracket connect Opium.close
-
-getToday :: IO [Issue]
-getToday = do
-  now <- getZonedTime
-  print now
-  let currentDay = localDay $ zonedTimeToLocalTime now
-  case addYears (-100) currentDay of
-    Right then_ ->
-      getIssuesByDay then_
-    Left _ ->
-      pure []
-
-indexDay :: Day -> IO ()
-indexDay date = exec $ \conn -> do
-  either throwIO pure =<< Opium.execute_ "SELECT pg_advisory_lock(42);" conn
-  printf "Checking whether %s already exists...\n" (show date)
-  mbDi <- either throwIO pure =<< DB.readDayIndex date conn
-  case mbDi of
-    Just (_ :: DayIndex) ->
-      putStrLn "Nothing to do."
-    Nothing -> do
-      putStrLn "Retrieving URLs..."
-      issues <- getIssuesByDay date
-      paths <- forM issues $ \issue -> do
-        printf "Downloading %s...\n" issue.url
-        path <- downloadInto "./download" issue.url
-        pure (issue.label, path)
-      putStrLn "Creating DB entries..."
-      createEntries date paths conn
-      putStrLn "Done."

src/Yore/Schedule.hs

@@ -0,0 +1,22 @@
+module Yore.Schedule (schedule) where
+
+import Control.Concurrent (forkIO, threadDelay)
+import Control.Monad (when)
+import Data.Functor (($>))
+import Data.Time (ZonedTime, getZonedTime, nominalDiffTimeToSeconds, secondsToNominalDiffTime)
+import Data.Time.Clock.POSIX (getPOSIXTime)
+
+schedule :: (ZonedTime -> Bool) -> IO a -> IO ()
+schedule shouldRunAt f = everyMinute $ \now -> when (shouldRunAt now) (f $> ())
+
+everyMinute :: (ZonedTime -> IO ()) -> IO ()
+everyMinute f = do
+  -- Use POSIX time to avoid having to handle leap seconds
+  now <- getPOSIXTime
+  let posixSeconds :: Int
+      posixSeconds = floor $ nominalDiffTimeToSeconds now
+      lastFullMinute = secondsToNominalDiffTime $ fromIntegral $ posixSeconds - posixSeconds `mod` 60
+      secondsSinceLastFullMinute = now - lastFullMinute
+  threadDelay $ ceiling $ (60 - secondsSinceLastFullMinute) * 1_000_000
+  _ <- forkIO $ f =<< getZonedTime
+  everyMinute f

yore.cabal

@@ -37,9 +37,8 @@ library
     , Yore.DB
     , Yore.Download
     , Yore.Error
-    , Yore.Index
     , Yore.Log
-    , Yore.Repl
+    , Yore.Schedule
     , Yore.Scrape
     , Yore.Time
   hs-source-dirs: