From c57102011e29ef81c714f29bc9f849c7e2c08c5a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?David=20Majdand=C5=BEi=C4=87?=
 <david.majdandzic@hotmail.com>
Date: Fri, 2 Feb 2024 11:15:05 +0100
Subject: [PATCH] Refactor file scanning and add file counting

---
 main.go | 38 ++++++++++++++++++++++++++++----------
 1 file changed, 28 insertions(+), 10 deletions(-)

diff --git a/main.go b/main.go
index c922be1..a2c7041 100644
--- a/main.go
+++ b/main.go
@@ -22,6 +22,9 @@ var timeUnits = map[string]int64{
 	"y":  1000 * 60 * 60 * 24 * 365,
 }
 
+var numFilesArchived = 0
+var numFilesDeleted = 0
+
 var valueRegex, _ = regexp.Compile(`\d+`)
 var unitRegex, _ = regexp.Compile(`[a-zA-Z]+`)
 
@@ -78,10 +81,11 @@ func scanRoot() {
 			return nil
 		}
 
-		if info.IsDir() {
-			log.Printf("Skipping directory %s...\n", path)
-			return filepath.SkipDir
-		}
+		// I forgot why this code was here... It doesn't make sense to me now
+		// if info.IsDir() {
+		// 	log.Printf("Skipping directory %s...\n", path)
+		// 	return filepath.SkipDir
+		// }
 
 		// We hope that IGNORED_DIRECTORIES is a small list, so we can afford to iterate over it
 		// In fact iteration should be faster for small lists rather than hashing
@@ -121,6 +125,7 @@ func processFile(path string, info os.FileInfo) {
 	var now = time.Now().UnixMilli()
 
 	var fileATime int64 = times.Get(info).AccessTime().UnixMilli()
+	// var fileATime int64 = times.Get(info).ModTime().UnixMilli()
 	var accessTimeDelta = now - fileATime
 	log.Printf("File %s last accessed at %d, %dms ago\n", path, fileATime, accessTimeDelta)
 	if accessTimeDelta > constants.ARCHIVE_THRESHOLD {
@@ -133,6 +138,7 @@ func processArchiveFile(path string, info os.FileInfo) {
 	var now = time.Now().UnixMilli()
 
 	var fileATime int64 = times.Get(info).AccessTime().UnixMilli()
+	// var fileATime int64 = times.Get(info).ModTime().UnixMilli()
 	var accessTimeDelta = now - fileATime
 	log.Printf("File %s last accessed at %d, %dms ago\n", path, fileATime, accessTimeDelta)
 	if accessTimeDelta > constants.DELETE_THRESHOLD {
@@ -152,6 +158,7 @@ func archiveFile(path string) {
 		log.Printf("Error archiving file %s: %s\n", path, err)
 		return
 	}
+	numFilesArchived++
 }
 
 func deleteFile(path string) {
@@ -162,6 +169,7 @@ func deleteFile(path string) {
 		log.Printf("Error deleting file %s: %s\n", path, err)
 		return
 	}
+	numFilesDeleted++
 }
 
 func cleanRoot() {
@@ -218,10 +226,24 @@ type Constants struct {
 	SCAN_INTERVAL       time.Duration
 }
 
+func doRun() {
+	scanRoot()
+	scanArchive()
+	cleanRoot()
+	log.Printf("Archived %d files, deleted %d files\n", numFilesArchived, numFilesDeleted)
+	numFilesArchived = 0
+	numFilesDeleted = 0
+}
+
 var constants = Constants{}
 
 func main() {
 	log.SetFlags(0b111)
+	// Important: Access times don’t accumulate.
+	// This implies that archiving the file won't alter its access time.
+	// Therefore, assign X as the ARCHIVE_TIME and X + Y as the DELETE_TIME,
+	// where X represents the duration it can exist in the folder,
+	// and Y represents the duration it can exist in the archive.
 
 	var ROOT = filepath.ToSlash(strings.TrimSpace(getEnv("ROOT", "/tmp")))
 	var ROOT_ARCHIVE = filepath.ToSlash(strings.TrimSpace(getEnv("ROOT_ARCHIVE", ROOT+"/archive")))
@@ -255,14 +277,10 @@ func main() {
 	log.Printf("DELETE_THRESHOLD: %d\n", DELETE_THRESHOLD)
 	log.Printf("SCAN_INTERVAL: %d\n", SCAN_INTERVAL.Milliseconds())
 
-	scanRoot()
-	scanArchive()
-	cleanRoot()
+	doRun()
 	for {
 		log.Printf("Running at %d", time.Now().UnixMilli())
 		time.Sleep(SCAN_INTERVAL)
-		scanRoot()
-		scanArchive()
-		cleanRoot()
+		doRun()
 	}
 }