Flesh out audio conversion script and add initial network sync

src/music/convert.py

@@ -4,38 +4,58 @@ import logging
 import shutil
 from pathlib import Path
 import subprocess
+import uuid
 from multiprocessing import Pool
 
 logger = logging.getLogger(__name__)
 
-def launch_task(cmd):
-    subprocess.run(cmd, shell=True, check=True)
-
-def convert(input_dir: Path, output_dir: Path):
-
-    logger.info("Converting files in %s", input_dir)
-    logger.info("Writing output to: %s", output_dir)
-
-    os.makedirs(output_dir, exist_ok=True)
-
-    flac_files = list(input_dir.resolve().rglob("*.flac"))
-    tasks = []
+def convert_audio_file(args):
+    cmd, output_tmp, output_path = args
     
-    for idx, path in enumerate(flac_files):
+    subprocess.run(cmd, shell=True)
+    os.makedirs(output_path.parent, exist_ok=True)
+    shutil.move(output_tmp, output_path)
+ 
+class AudioFileConverter():
 
-        logger.info("Converting file %d of %d", idx, len(flac_files))
-        
-        relative_path = path.relative_to(input_dir)
-        output_filename = str(path.stem) + ".mp3"
+    def __init__(self, input_dir: Path, output_dir: Path):
+        self.input_dir = input_dir
+        self.output_dir = output_dir
+        self.output_fmt = "mp3"
+        self.input_fmt = "flac"
+
+    def get_converted_path(self, input_path: Path):
+        relative_path = input_path.relative_to(self.input_dir)
+        output_filename = str(input_path.stem) + f".{self.output_fmt}"
         output_filename.replace("'", "")
+        output_path = self.output_dir / relative_path.parent / output_filename
+        return output_path
+        
+    def convert(self):
 
-        output_path = output_dir / relative_path.parent / output_filename
-    
-        cmd = f"ffmpeg -i '{path}' -ab 320k -map_metadata 0 -id3v2_version 3 '{output_path}'"
-        tasks.append(cmd)
+        logger.info("Converting files in %s", self.input_dir)
+        logger.info("Writing output to: %s", self.output_dir)
 
-    with Pool(10) as p:
-        p.map(launch_task, tasks)
+        os.makedirs(self.output_dir, exist_ok=True)
+
+        input_files = list(self.input_dir.resolve().rglob(f"*.{self.input_fmt}"))
+        output_files = []
+        for input_file in input_files:
+            candidate_output = self.get_converted_path(input_file)
+            if not candidate_output.exists():
+                output_files.append(candidate_output)
+            
+        tasks = []
+        for idx, (input_path, output_path) in enumerate(zip(input_files, output_files)):
+
+            logger.info("Converting file %d of %d", idx, len(input_files))
+
+            output_tmp = self.output_dir / (str(uuid.uuid4()) + f".{self.output_fmt}")
+            cmd = f"ffmpeg -i '{input_path}' -ab 320k -map_metadata 0 -id3v2_version 3 '{output_tmp}'"
+            tasks.append((cmd, output_tmp, output_path))
+
+        with Pool(10) as p:
+            p.map(convert_audio_file, tasks)
 
 
 def move(input_dir: Path, output_dir: Path):
@@ -50,7 +70,6 @@ def move(input_dir: Path, output_dir: Path):
         output_path = output_dir / relative_path
         os.makedirs(output_path.parent, exist_ok=True)
         shutil.move(path, output_path)
-
            
 if __name__ == "__main__":
 
@@ -71,7 +90,8 @@ if __name__ == "__main__":
     logging.basicConfig()
     logging.getLogger().setLevel(logging.INFO)
 
-    # convert(args.input_dir.resolve(), args.output_dir.resolve())
-    move(args.input_dir.resolve(), args.output_dir.resolve())
+    converter = AudioFileConverter(args.input_dir.resolve(), args.output_dir.resolve())
+    converter.convert()
+    #move(args.input_dir.resolve(), args.output_dir.resolve())
   
 

src/network/sync.py

@@ -0,0 +1,70 @@
+import argparse
+import os
+import logging
+import shutil
+from pathlib import Path
+import subprocess
+import uuid
+
+from fabric import Connection
+
+logger = logging.getLogger(__name__)
+
+
+def sync(source_dir: Path, target_dir: Path, host: str):
+
+    logger.info("Syncing files in %s to %s on % s", source_dir, target_dir, host)
+
+    cmd = f"find {target_dir} -type f"
+
+    if host:
+        result = Connection(host).run(cmd, hide=True)
+        target_files = result.stdout.splitlines()
+    else:
+        result = subprocess.run(cmd, shell=True, capture_output=True, text=True)
+        target_files = result.stdout.splitlines()
+
+    target_files = [Path(f).relative_to(target_dir) for f in target_files]
+    source_files = []
+    for dir_path, _, files in os.walk(source_dir):
+        for each_file in files:
+            source_files.append(Path(dir_path + "/" + each_file).relative_to(source_dir))
+            
+    sync_files = []
+    for source_file in source_files:
+        if source_file not in target_files:
+            sync_files.append(source_file)
+
+    sync_dir = Path(os.getcwd())/"sync"
+    os.makedirs(sync_dir, exist_ok=True)
+    for sync_file in sync_files:
+        os.makedirs((sync_dir/sync_file).parent, exist_ok=True)
+        shutil.copy(source_dir / sync_file, sync_dir / sync_file)
+         
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument('--source_dir',
+                        type=Path,
+                        default=Path(),
+        help="Directory with source files to sync from.")
+
+    parser.add_argument('--host',
+                        type=str,
+                         default = "",
+                        help="Name of host system to sync with.")
+   
+    parser.add_argument('--target_dir',
+                        type=Path,
+                        default=Path(),
+                        help="Directory with source files to sync to.")
+
+    args = parser.parse_args()
+
+    logging.basicConfig()
+    logging.getLogger().setLevel(logging.INFO)
+
+    sync(args.source_dir.resolve(), args.target_dir.resolve(), args.host)
+ 
+