From 068f6f82cfee723e6c6213ea58d05c9e9708e131 Mon Sep 17 00:00:00 2001
From: Přemysl Eric Janouch <p@janouch.name>
Date: Thu, 21 Dec 2023 06:19:03 +0100
Subject: WIP: FS to DB sync

---
 initialize.sql | 17 +++++++++++++++-
 main.go        | 62 ++++++++++++++++++++++++++++++++++++++++++++--------------
 2 files changed, 63 insertions(+), 16 deletions(-)

diff --git a/initialize.sql b/initialize.sql
index c230f18..60b50f0 100644
--- a/initialize.sql
+++ b/initialize.sql
@@ -47,13 +47,28 @@ END;
 
 --
 
--- TODO: Add a trigger to delete from here when a node for the sha1 is added.
 CREATE TABLE IF NOT EXISTS orphan(
 	sha1 TEXT NOT NULL REFERENCES image(sha1)
 	path TEXT NOT NULL,
 	PRIMARY KEY (sha1)
 ) STRICT;
 
+-- Renaming/moving a file can result either in a (ref, unref) or a (unref, ref)
+-- sequence during sync, and I want to get at the same result.
+CREATE TRIGGER IF NOT EXISTS node__sha1__deorphan_insert
+AFTER INSERT ON node
+WHEN NEW.sha1 IS NOT NULL
+BEGIN
+	DELETE FROM orphan WHERE sha1 = NEW.sha1;
+END;
+
+CREATE TRIGGER IF NOT EXISTS node__sha1__deorphan_update
+AFTER UPDATE OF sha1 ON node
+WHEN NEW.sha1 IS NOT NULL
+BEGIN
+	DELETE FROM orphan WHERE sha1 = NEW.sha1;
+END;
+
 --
 
 CREATE TABLE IF NOT EXISTS tag_space(
diff --git a/main.go b/main.go
index 897ae1e..7f15b5c 100644
--- a/main.go
+++ b/main.go
@@ -831,6 +831,11 @@ type syncFile struct {
 	fsIsDir bool
 }
 
+type syncPair struct {
+	db *syncNode
+	fs *syncFile
+}
+
 // syncGetNodes returns direct children of a DB node, ordered by name.
 // SQLite, like Go, compares strings byte-wise by default.
 func syncGetNodes(tx *sql.Tx, dbParent int64) (nodes []syncNode, err error) {
@@ -885,9 +890,38 @@ func syncGetFiles(fsPath string) (files []syncFile, err error) {
 
 // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
 
-type syncPair struct {
-	db *syncNode
-	fs *syncFile
+func syncProcess(c *syncContext, info *syncFileInfo) error {
+	// Skip videos, which ImageMagick can process, but we don't want it to,
+	// so that they're not converted 1:1 to WebP.
+	pathIsImage, err := isImage(info.fsPath)
+	if err != nil {
+		return err
+	}
+	if !pathIsImage {
+		return nil
+	}
+
+	info.width, info.height, err = pingImage(info.fsPath)
+	if err != nil {
+		return err
+	}
+
+	f, err := os.Open(info.fsPath)
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+
+	// We could make this at least somewhat interruptible by c.ctx,
+	// though it would still work poorly.
+	hash := sha1.New()
+	_, err = io.CopyBuffer(hash, f, make([]byte, 65536))
+	if err != nil {
+		return err
+	}
+
+	info.sha1 = hex.EncodeToString(hash.Sum(nil))
+	return nil
 }
 
 // syncEnqueue runs file scanning, which can be CPU and I/O expensive,
@@ -899,8 +933,8 @@ func syncEnqueue(c *syncContext, info syncFileInfo) error {
 
 	go func(info syncFileInfo) {
 		defer taskSemaphore.release()
-
-		// TODO: Process the file and enqueue a result.
+		info.err = syncProcess(c, &info)
+		c.info <- info
 	}(info)
 	return nil
 }
@@ -921,22 +955,20 @@ func syncDequeue(c *syncContext) error {
 	}
 }
 
-// TODO: Implement.
-//
-//   - When collecting node subtrees, we need to delete bottom-up
-//     because of foreign key constraints,
-//     so maybe in reverse order of recursive CTE results.
-//
-//   - Sadly, this can't be done with a DB trigger. (What and why?)
-//
-//   - One of the inputs needs to be the FS path, for the orphan table.
-//
 // syncDispose creates orphan records for the entire subtree given by nodeID
 // as appropriate, then deletes all nodes within the subtree. The subtree root
 // node is not deleted if "keepNode" is true.
 //
 // Orphans keep their thumbnail files, as evidence.
 func syncDispose(c *syncContext, nodeID int64, keepNode bool) error {
+	// TODO: Implement.
+	//  - When collecting node subtrees, we need to delete bottom-up
+	//    because of foreign key constraints,
+	//    so maybe in reverse order of recursive CTE results.
+	//  - Sadly, this can't be done with a DB trigger. (What and why?)
+	//  - One of the inputs needs to be the FS path, for the orphan table.
+	//     - I may not have the FS path (symlink).
+	//     - I can just recursively select for the path based on nodeID.
 	return nil
 }
 
-- 
cgit v1.2.3-70-g09d2