From f7ee691b4dcdb6f5ce383ec4cc1e342184b10a55 Mon Sep 17 00:00:00 2001
From: "Roncero Blanco, Edgar" <eroncero@elisava.net>
Date: Mon, 26 May 2025 20:17:05 +0200
Subject: [PATCH] Better split algorithm

---
 Program/app.py | 52 +++++++++++++++++++++++---------------------------
 1 file changed, 24 insertions(+), 28 deletions(-)

diff --git a/Program/app.py b/Program/app.py
index 1a822ec..dc8f103 100644
--- a/Program/app.py
+++ b/Program/app.py
@@ -3,10 +3,10 @@ import shutil
 from pathlib import Path
 from mutagen.id3 import ID3, TBPM, ID3NoHeaderError
 from datetime import datetime
-from collections import defaultdict
 import configparser
 import sys
 import librosa
+import warnings
 
 GROUP_SIZE = 5
 
@@ -20,6 +20,7 @@ def load_config():
             print("config.ini created from def_config.ini with default settings.")
         else:
             print("Error: def_config.ini not found! Please create it and rerun.")
+            input("Press Enter to exit...")
             sys.exit(1)
 
     default_config = configparser.ConfigParser()
@@ -36,7 +37,9 @@ def load_config():
 
 def analyze_bpm_librosa(file_path):
     try:
-        y, sr = librosa.load(file_path, mono=True)
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore")
+            y, sr = librosa.load(file_path, mono=True)
         tempo, _ = librosa.beat.beat_track(y=y, sr=sr)
         return int(round(float(tempo)))
     except Exception as e:
@@ -49,7 +52,6 @@ def write_bpm_tag(file_path, bpm):
             tags = ID3(file_path)
         except ID3NoHeaderError:
             tags = ID3()
-
         tags.delall("TBPM")
         tags.add(TBPM(encoding=3, text=str(bpm)))
         tags.save(file_path)
@@ -67,14 +69,11 @@ def get_bpm(file_path):
         pass
     return None
 
-def split_evenly(lst, n):
-    k, m = divmod(len(lst), n)
-    return [lst[i*k + min(i, m):(i+1)*k + min(i+1, m)] for i in range(n)]
-
 def main():
     config = load_config()
 
-    CD_SIZE = config.getint("Settings", "SplitFolderMB") * 1024 * 1024
+    SAFETY_MARGIN = 1024 * 1024 * 2  # 2 MB buffer
+    CD_SIZE = config.getint("Settings", "SplitFolderMB") * 1024 * 1024 - SAFETY_MARGIN
     bWriteNonPresentBPM = config.getboolean("Settings", "bWriteNonPresentBPM")
     bCheckAllTracksBPM = config.getboolean("Settings", "bCheckAllTracksBPM")
 
@@ -85,6 +84,7 @@ def main():
 
     if not source_media_path.exists() or not source_media_path.is_dir():
         print(f"Error: {source_media_path} is not a valid directory.")
+        input("Press Enter to exit...")
         sys.exit(1)
 
     all_tracks = []
@@ -103,14 +103,12 @@ def main():
                 write_bpm_tag(file, bpm)
 
         size = file.stat().st_size
-        if bpm is None:
-            bpm_range = "[Unknown BPM]"
-        else:
-            bpm_range = f"{(bpm // GROUP_SIZE) * GROUP_SIZE}-to-{((bpm // GROUP_SIZE) * GROUP_SIZE) + GROUP_SIZE - 1}"
+        bpm_range = f"{(bpm // GROUP_SIZE) * GROUP_SIZE}-to-{((bpm // GROUP_SIZE) * GROUP_SIZE) + GROUP_SIZE - 1}" if bpm else "[Unknown BPM]"
         all_tracks.append({"file": file, "size": size, "bpm_range": bpm_range})
 
     if not all_tracks:
         print("No MP3 files found.")
+        input("Press Enter to exit...")
         sys.exit(0)
 
     parent = source_media_path.parent
@@ -118,27 +116,24 @@ def main():
     dest_media_path = parent / f"[CDs-{run_date}]{folder_name}"
     dest_media_path.mkdir(parents=True, exist_ok=True)
 
-    total_size = sum(t["size"] for t in all_tracks)
-    num_cds = max(1, (total_size + CD_SIZE - 1) // CD_SIZE)
+    all_tracks.sort(key=lambda x: -x["size"])  # Sort by size descending for better packing
 
-    print(f"Total size: {total_size / 1024**2:.2f} MB, splitting into {num_cds} CDs")
+    cd_contents = []
+    current_cd = []
+    current_size = 0
 
-    bpm_groups = defaultdict(list)
     for track in all_tracks:
-        bpm_groups[track["bpm_range"]].append(track)
+        if current_size + track["size"] > CD_SIZE:
+            cd_contents.append(current_cd)
+            current_cd = []
+            current_size = 0
+        current_cd.append(track)
+        current_size += track["size"]
 
-    for bpm_range in bpm_groups:
-        bpm_groups[bpm_range].sort(key=lambda x: x["file"].name)
+    if current_cd:
+        cd_contents.append(current_cd)
 
-    bpm_chunks = {}
-    for bpm_range, tracks in bpm_groups.items():
-        bpm_chunks[bpm_range] = split_evenly(tracks, num_cds)
-
-    cd_contents = [[] for _ in range(num_cds)]
-
-    for bpm_range in bpm_chunks:
-        for i, chunk in enumerate(bpm_chunks[bpm_range]):
-            cd_contents[i].extend(chunk)
+    print(f"Total size: {sum(t['size'] for t in all_tracks) / 1024**2:.2f} MB, splitting into {len(cd_contents)} CDs")
 
     for i, tracks in enumerate(cd_contents, start=1):
         cd_folder = dest_media_path / f"CD-{i:02}"
@@ -152,6 +147,7 @@ def main():
         print(f"[WRITE] CD-{i:02}: {len(tracks)} tracks, approx {size_accum / 1024**2:.2f} MB")
 
     print("\n✓ Done!")
+    input("Press Enter to exit...")
 
 if __name__ == "__main__":
     main()