From 086a4a293327062052029850ded9aa38c628408d Mon Sep 17 00:00:00 2001
From: Keith Edmunds <kae@tiger-computing.co.uk>
Date: Tue, 23 Mar 2021 08:54:02 +0000
Subject: [PATCH] Store all track metadata in db

---
 model.py  |  6 ++++-
 songdb.py | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 74 insertions(+), 4 deletions(-)

diff --git a/model.py b/model.py
index d7e9a2c..6b7a27f 100644
--- a/model.py
+++ b/model.py
@@ -42,8 +42,12 @@ class Tracks(Base):
     id = Column(Integer, primary_key=True, autoincrement=True)
     title = Column(String(256), index=True)
     artist = Column(String(256), index=True)
-    length = Column(Integer, index=True)
+    duration = Column(Integer, index=True)
+    start_gap = Column(Integer, index=False)
+    fade_at = Column(Integer, index=False)
+    silence_at = Column(Integer, index=False)
     path = Column(String(2048), index=False, nullable=False)
+
     mtime = Column(Float, index=True)
     lastplayed = Column(DateTime, index=True, default=None)
 
diff --git a/songdb.py b/songdb.py
index 57b7cf6..6524d91 100755
--- a/songdb.py
+++ b/songdb.py
@@ -8,6 +8,7 @@ import os
 import pytiger.logging.config
 
 from tinytag import TinyTag
+from pydub import AudioSegment
 
 from model import Tracks, session
 
@@ -38,6 +39,66 @@ def main():
     log.info("Finished")
 
 
+def get_audio_segment(path):
+    try:
+        if path.endswith('.mp3'):
+            return AudioSegment.from_mp3(path)
+        elif path.endswith('.flac'):
+            return AudioSegment.from_file(path, "flac")
+    except AttributeError:
+        return None
+
+
+def leading_silence(audio_segment, silence_threshold=-50.0,
+                    chunk_size=10):
+    """
+    Returns the millisecond/index that the leading silence ends.
+    audio_segment - the segment to find silence in
+    silence_threshold - the upper bound for how quiet is silent in dFBS
+    chunk_size - chunk size for interating over the segment in ms
+
+    https://github.com/jiaaro/pydub/blob/master/pydub/silence.py
+    """
+
+    trim_ms = 0  # ms
+    assert chunk_size > 0  # to avoid infinite loop
+    while (
+        audio_segment[trim_ms:trim_ms + chunk_size].dBFS <
+            silence_threshold and trim_ms < len(audio_segment)):
+        trim_ms += chunk_size
+
+    # if there is no end it should return the length of the segment
+    return min(trim_ms, len(audio_segment))
+
+
+def fade_point(audio_segment, fade_threshold=-20.0, chunk_size=10):
+    """
+    Returns the millisecond/index of the point where the fade is down to
+    fade_threshold and doesn't get louder again.
+    audio_segment - the sdlg_search_database_uiegment to find silence in
+    fade_threshold - the upper bound for how quiet is silent in dFBS
+    chunk_size - chunk size for interating over the segment in ms
+    """
+
+    assert chunk_size > 0  # to avoid infinite loop
+
+    segment_length = audio_segment.duration_seconds * 1000  # ms
+    trim_ms = segment_length - chunk_size
+    while (
+        audio_segment[trim_ms:trim_ms + chunk_size].dBFS < fade_threshold
+            and trim_ms > 0):
+        trim_ms -= chunk_size
+
+    # if there is no trailing silence, return lenght of track (it's less
+    # the chunk_size, but for chunk_size = 10ms, this may be ignored)
+    return int(trim_ms)
+
+
+def trailing_silence(audio_segment, silence_threshold=-50.0,
+                     chunk_size=10):
+    return fade_point(audio_segment, silence_threshold, chunk_size)
+
+
 def update_db():
     """
     Repopulate database
@@ -54,15 +115,20 @@ def update_db():
             if ext in [".flac", ".mp3"]:
                 track = Tracks.get_or_create(os.path.relpath(path, ROOT))
                 tag = TinyTag.get(path)
+                audio = get_audio_segment(path)
+
                 track.title = tag.title
                 track.artist = tag.artist
-                track.length = int(tag.duration * 1000)
+                track.duration = int(tag.duration * 1000)
+                track.start_gap = leading_silence(audio)
+                track.fade_at = fade_point(audio)
+                track.silence_at = trailing_silence(audio)
                 track.mtime = os.path.getmtime(path)
+                session.commit()
+
             elif ext not in [".jpg"]:
                 print(f"Unrecognised file type: {path}")
 
-    session.commit()
-
     print(f"{count} files processed")