dedupe before transcoding

- copy worker over

note: originally removed in commit 641abcdd90

src/layers/__init__.py

@@ -1,2 +1,4 @@
 from .layer import Layer
+from .worker import Worker
 from .transcoder import Transcoder
+from .dedupe import Dedupe

src/layers/transcoder.py

@@ -0,0 +1,45 @@
+from pathlib import Path
+from multiprocessing import Pool, Manager, set_start_method
+from log import Log, LogCat
+from .worker import Worker
+from dir import Root
+from .layer import Layer
+
+
+class Transcoder(Layer):
+    def __init__(self, encoder: Path, extension: str, output_root: Path, log: Log, log_path: Path):
+        super().__init__(log, 'TCD')
+        self.encoder = encoder
+        self.extension = extension
+        self.output_root = output_root
+
+        self.log_path = log_path
+
+    def _process(self, root: Root):
+        transcode_list = root.all_files
+        self._transcode(transcode_list, self.encoder)
+
+    def _transcode(self, transcode_list: list, encoder: Path, workers=16):
+        manager = Manager()
+        queue = manager.Queue()
+        log = Log(self.log_path, queue)
+        logcat = LogCat(log.queue, "TCD")
+        args = [(str(self.output_root), self.extension, track, encoder, logcat) for track in transcode_list]
+        with Pool(workers) as pool:
+            pool.starmap(self.worker, args)
+            pool.close()
+            pool.join()
+        log.stop()
+
+    def _transcode_single_thread(self, transcode_list: list, encoder: Path):
+        log = Log(self.log_path)
+        logcat = LogCat(log.queue, "TCD")
+        worker_args = [(track, encoder) for track in transcode_list]
+        for track, encoder in worker_args:
+            self.worker(str(self.output_root), self.extension, track, encoder, logcat)
+        log.stop()
+
+    @staticmethod
+    def worker(output_root, extension, track, encoder, log):
+        w = Worker(output_root, extension)
+        w.transcode_worker(track, encoder, log)

src/layers/worker.py

@@ -0,0 +1,70 @@
+import shutil
+import subprocess
+from pathlib import Path
+from dir.file import File
+
+class Worker:
+    def __init__(self, output_root, extension):
+        self.output_root = Path(output_root)
+        self.extension = extension
+
+    def transcode_worker(self, track: File, encoder, log):
+        if track.is_art:
+            return self.copy_album_art(track, log)
+        elif track.is_audio:
+            return self.transcode_audio(track, encoder, log)
+        else:
+            log.info("WRK", f"File {track.path} ignored")
+
+    def copy_album_art(self, file: File, log):
+        self.create_directories(file)
+        output_path = file.output_file(self.output_root, file.extension[1:])
+        if output_path.exists():
+            log.info("ART", f"Skipped {output_path}")
+            return
+        try:
+            shutil.copy(file.path, output_path)
+            log.info("ART", f"Successfully copied {output_path}")
+        except FileExistsError:
+            log.warning("ART", f"File exists: {output_path}")
+        except Exception:
+            log.error("ART", f"{file.path} failed to copy")
+
+    def transcode_audio(self, track, encoder, log):
+        track_output = track.output_file(self.output_root, self.extension)
+        if track_output.exists():
+            log.info("AUD", f"Skipped {track_output}")
+            return
+        self.create_directories(track)
+        encoder_args = self.encoder_args(encoder, track, log)
+        try:
+            subprocess.run(encoder_args, capture_output=True, text=True, check=True)
+            log.info("AUD", f"Transcoded '{track}' successfully.")
+        except Exception:
+            log.error("AUD", f"ERROR: Transcoding of '{track}' failed.")
+
+    def encoder_args(self, encoder, track, log):
+        track_output = track.output_file(self.output_root, self.extension)
+        enc_filename = encoder.parts[-1]
+        encoder_args = ""
+        if enc_filename == "opusenc.exe":
+            additional_args = ('--bitrate', '128', '--music')
+            encoder_args = (str(encoder),) + additional_args + (str(track.path), str(track_output))
+        elif enc_filename == "qaac64.exe":
+            additional_args = (str(track), '-o', str(track_output), '--threading')
+            encoder_args = (str(encoder),) + additional_args
+        else:
+            log.error("AUD", f"Encoder {encoder} not recognised.")
+        return encoder_args
+
+    def create_directories(self, track: File):
+        if not track.artist_out(self.output_root).exists():
+            try:
+                track.artist_out(self.output_root).mkdir()
+            except FileExistsError:
+                pass
+        if not track.album_out(self.output_root).exists():
+            try:
+                track.album_out(self.output_root).mkdir()
+            except FileExistsError:
+                pass

src/main.py

@@ -1,8 +1,9 @@
 import argparse
 from os.path import realpath
 from pathlib import Path
-from transcode import Transcoder
+from dir import Root
 from log import Log
+from layers import Dedupe, Transcoder
 
 def get_args():
     parser = argparse.ArgumentParser()
@@ -17,9 +18,17 @@ def main(input_dir: Path, output_dir: Path, encoder: Path, out_extension: str =
     log_path = wd / "logs"
     if encoder.parts[-1] == "qaac64.exe":
         out_extension = "m4a"
-    transcoder = Transcoder(encoder, out_extension, input_dir, output_dir, log_path)
+    log = Log(log_path)
-    transcoder.transcode()
+    input_root = Root(input_dir, log)
+    output_root = Root(output_dir, log)
 
+    dedupe = Dedupe(output_root, log)
+    dedupe.process(input_root)
+
+    transcoder = Transcoder(encoder, out_extension, output_dir, log, log_path)
+    transcoder.process(input_root)
+
+    log.stop()
 
 if __name__ == '__main__':
     args = get_args()