From 2011f2b198871b6acb4bbc50012b8bd6801b6a7f Mon Sep 17 00:00:00 2001
From: Přemysl Eric Janouch <p@janouch.name>
Date: Thu, 21 Dec 2023 04:48:56 +0100
Subject: WIP: FS to DB sync

---
 main.go | 79 +++++++++++++++++++++++++++++++++++++++++++++++------------------
 1 file changed, 57 insertions(+), 22 deletions(-)

diff --git a/main.go b/main.go
index 8a3de6d..897ae1e 100644
--- a/main.go
+++ b/main.go
@@ -134,6 +134,11 @@ func (pb *progressBar) Stop() {
 }
 
 func (pb *progressBar) update() {
+	if pb.target < 0 {
+		fmt.Printf("\r%d/?", pb.current)
+		return
+	}
+
 	var fraction int
 	if pb.target != 0 {
 		fraction = int(float32(pb.current) / float32(pb.target) * 100)
@@ -806,6 +811,7 @@ type syncContext struct {
 	ctx  context.Context
 	tx   *sql.Tx
 	info chan syncFileInfo
+	pb   *progressBar
 }
 
 // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
@@ -915,17 +921,12 @@ func syncDequeue(c *syncContext) error {
 	}
 }
 
-// TODO: [[B]]: A function to collect references.
-//
-//   - This is almost always combined with [[A1]] or [[A2]],
-//     so perhaps combine these.
+// 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.
 //
-//   - Orphans will keep their GD/thumbs file, as evidence.
-//
 //   - 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.
@@ -933,20 +934,51 @@ func syncDequeue(c *syncContext) error {
 // 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 {
 	return nil
 }
 
+func syncImage(c *syncContext, info syncFileInfo) error {
+	if _, err := c.tx.Exec(`INSERT INTO image(sha1, width, height)
+		VALUES (?, ?, ?) ON CONFLICT(sha1) DO NOTHING`,
+		info.sha1, info.width, info.height); err != nil {
+		return err
+	}
+
+	// Fast path: it may already there, and not be a dead symlink.
+	path := imagePath(info.sha1)
+	if _, err := os.Stat(path); err == nil {
+		return nil
+	}
+
+	dirname, _ := filepath.Split(path)
+	if err := os.MkdirAll(dirname, 0755); err != nil {
+		return err
+	}
+
+	for {
+		err := os.Symlink(info.fsPath, path)
+		if !errors.Is(err, fs.ErrExist) {
+			return err
+		}
+
+		// Try to remove anything standing in the way, and try again.
+		if err = os.Remove(path); err != nil {
+			return err
+		}
+	}
+}
+
 func syncPostProcess(c *syncContext, info syncFileInfo) error {
-	// TODO:
-	//  - When replacing a file, which needn't become orphaned,
-	//    we could offer copying all tags, but this needs another table
-	//    to track it (if it's equivalent enough, the dhash will stay the same,
-	//    so user can resolve this through the duplicates feature).
-	//  - When creating or updating a node,
-	//    try inserting an image record first as well.
-	//    And test that we have a symlink, and that it works.
-	//    (Just do os.Stat() on it, which fails on both dead and missing links.)
+	defer c.pb.Step()
+
+	// TODO: When replacing an image node (whether it has or doesn't have
+	// other links to keep it alive), we could offer copying all tags,
+	// though this needs another table to track it.
+	// (If it's equivalent enough, the dhash will stay the same,
+	// so user can resolve this through the duplicates feature.)
 	switch {
 	case info.err != nil:
 		// * → error
@@ -963,8 +995,9 @@ func syncPostProcess(c *syncContext, info syncFileInfo) error {
 
 	case info.dbID == 0:
 		// 0 → F
-		// TODO: Add an image entry.
-
+		if err := syncImage(c, info); err != nil {
+			return err
+		}
 		if _, err := c.tx.Exec(`INSERT INTO node(parent, name, mtime, sha1)
 			VALUES (?, ?, ?, ?)`,
 			info.dbParent, info.dbName, info.fsMtime, info.sha1); err != nil {
@@ -978,11 +1011,10 @@ func syncPostProcess(c *syncContext, info syncFileInfo) error {
 			return err
 		}
 
-		// TODO: Ideally, do nothing if the hash doesn't change.
-		// But that will be a rare situation.
-
-		// TODO: Add an image entry.
-
+		// Even if the hash didn't change, we may fix any broken symlinks.
+		if err := syncImage(c, info); err != nil {
+			return err
+		}
 		if _, err := c.tx.Exec(`UPDATE node SET mtime = ?, sha1 = ?
 			WHERE id = ?`, info.fsMtime, info.sha1, info.dbID); err != nil {
 			return err
@@ -1098,6 +1130,9 @@ func syncDirectory(c *syncContext, dbParent int64, fsPath string) error {
 func syncArgument(ctx context.Context, tx *sql.Tx, path string) error {
 	c := syncContext{ctx: ctx, tx: tx}
 
+	c.pb = newProgressBar(-1)
+	defer c.pb.Stop()
+
 	// Info tasks take a position in the task semaphore channel.
 	// then fill the info channel.
 	//
-- 
cgit v1.2.3-70-g09d2