From 103bf8601718176ce227de3b11a6d3c2a2a13293 Mon Sep 17 00:00:00 2001
From: towards-a-new-leftypol <paul_cockshott@protonmail.com>
Date: Tue, 16 Jan 2024 17:13:55 -0500
Subject: [PATCH] WIP: actually adding attachments

    - attachments table was already defined
    - change the hash type to sha256
    - define sql function to upsert attachment rows
---
 backfill_settings.json |  1 +
 sql/initialize.sql     | 32 +++++++++++++++++++++++++++++++-
 src/AttachmentType.hs  | 19 +++++++++++++++++++
 src/JSONSettings.hs    |  1 +
 4 files changed, 52 insertions(+), 1 deletion(-)
 create mode 100644 src/AttachmentType.hs

diff --git a/backfill_settings.json b/backfill_settings.json
index 31ff7cf..cfe5fc2 100644
--- a/backfill_settings.json
+++ b/backfill_settings.json
@@ -2,6 +2,7 @@
   "postgrest_url": "http://localhost:3000",
   "jwt": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJyb2xlIjoiY2hhbl9hcmNoaXZlciJ9.rGIKZokTDKTuQLIv8138bUby5PELfDipYYIDpJzH02c",
   "backup_read_root": "/home/phil/linixy/tmp/leftypol_back/lainchan.leftypol.org",
+  "media_root_path": "/home/phil/linixy/tmp/chan_archive_media",
   "site_name": "leftychan",
   "site_url": "https://leftychan.net"
 }
diff --git a/sql/initialize.sql b/sql/initialize.sql
index c5d7dc4..71f0ba2 100644
--- a/sql/initialize.sql
+++ b/sql/initialize.sql
@@ -103,7 +103,7 @@ CREATE TABLE IF NOT EXISTS attachments
     ( attachment_id bigserial primary key
     , mimetype text NOT NULL
     , creation_time timestamp with time zone NOT NULL
-    , sha256_hash text NOT NULL
+    , sha256_hash text NOT NULL UNIQUE
     , phash bigint
     , illegal boolean NOT NULL DEFAULT false
     , post_id bigint NOT NULL
@@ -181,6 +181,36 @@ $$ LANGUAGE sql;
 -- 1:21 for full db (nothing inserted)
 
 
+CREATE OR REPLACE FUNCTION insert_attachments_and_return_ids(
+    attachments_payload attachments[])
+RETURNS TABLE (attachment_id bigint, post_id bigint, sha256_hash text) AS $$
+WITH
+selected AS (
+    SELECT attachment_id, post_id, sha256_hash
+    FROM attachments
+    WHERE sha256_hash IN (
+        SELECT sha256_hash FROM unnest(attachments_payload)
+    )
+),
+to_insert AS (
+    SELECT new_a.*
+    FROM unnest(attachments_payload) AS new_a
+    LEFT OUTER JOIN selected s
+        ON new_a.sha256_hash = s.sha256_hash
+    WHERE s.attachment_id IS NULL
+),
+inserted AS (
+    INSERT INTO attachments (mimetype, creation_time, sha256_hash, phash, illegal, post_id)
+    SELECT mimetype, creation_time, sha256_hash, phash, illegal, post_id
+    FROM to_insert
+    RETURNING attachment_id, post_id, sha256_hash
+)
+SELECT * FROM inserted
+UNION ALL
+SELECT * FROM selected;
+$$ LANGUAGE sql;
+
+
 CREATE OR REPLACE FUNCTION fetch_top_threads(
     p_start_time TIMESTAMPTZ,
     lookback INT DEFAULT 10000
diff --git a/src/AttachmentType.hs b/src/AttachmentType.hs
new file mode 100644
index 0000000..6ef5c74
--- /dev/null
+++ b/src/AttachmentType.hs
@@ -0,0 +1,19 @@
+{-# LANGUAGE DeriveAnyClass #-}
+module AttachmentType
+( Attachment (..)
+) where
+
+import GHC.Generics
+import Data.Int (Int64)
+import Data.Aeson (FromJSON)
+import Data.Text (Text)
+
+data Attachment = Attachment
+    { attachment_id   :: Maybe Int64
+    , mimetype        :: Text
+    , creation_time   :: UTCTime
+    , sha256_hash     :: Int
+    , phash           :: Int64
+    , phash           :: Bool
+    , post_id         :: Int64
+    } deriving (Show, Generic, FromJSON)
diff --git a/src/JSONSettings.hs b/src/JSONSettings.hs
index 6e794db..144b52d 100644
--- a/src/JSONSettings.hs
+++ b/src/JSONSettings.hs
@@ -9,6 +9,7 @@ data JSONSettings = JSONSettings
   { postgrest_url :: String
   , jwt :: String
   , backup_read_root :: FilePath
+  , media_root_path :: FilePath
   , site_name :: String
   , site_url :: String
   } deriving (Show, Generic)