moving to not-inotify based file discovery

dockerfile

@@ -1,4 +1,4 @@
-FROM python
+FROM python:3.9.18-bookworm
 
 RUN mkdir /app
 COPY notescriber /app/notescriber
@@ -9,4 +9,6 @@ COPY load_model.py /app/load_model.py
 WORKDIR /app
 
 RUN pip install -r requirements.txt
-RUN /app/load_model.py
+RUN /app/load_model.py
+
+RUN apt update && apt install -y ffmpeg

notescriber/dispatcher.py

@@ -20,8 +20,8 @@ class Dispatcher:
         # TODO: we need to have a way to wipe old threads, otherwise we may have a memory leak
 
     def _execute_thread(self, audio_file):
-        filename = os.path.basename(audio_file.pathname)
-        note = self.audioProcessor.execute(audio_file.pathname)
+        filename = os.path.basename(audio_file)
+        note = self.audioProcessor.execute(audio_file)
         self.noteGenerator.generate_note(filename, note)
 
     def halt(self):

notescriber/picker.py

@@ -1,34 +1,39 @@
-import pyinotify
+import time
 import os
 
-class EventHandler(pyinotify.ProcessEvent):
-    def process_IN_CREATE(self, event):
-        print(f'New file: {event.pathname}')
-        return event.pathname
-    
-    def apply_filter(self, audio_file):
-        # TODO: return True if audio file is valid, based on name, extension, and other criteria
-        if os.path.isdir(audio_file):
-            return False
-        return True
-
 class Picker:
 
     directory = None
     dispatcher = None
-    notifier = None
+    observer = None
+    last_file_time = None
 
     def __init__(self, directory, dispatcher):
         self.directory = directory
         self.dispatcher = dispatcher
 
+    def identify_time_window(self, file_time):
+        if self.last_file_time is None:
+            self.last_file_time = time.time()
+        if file_time > self.last_file_time:
+            return True
+        return False
+
+    def list_files(self):
+        files = []
+        for filename in os.listdir(self.directory):
+            filepath = os.path.join(self.directory, filename)
+            if os.path.isfile(filepath):
+                files.append(filepath)
+        return files
+
+
     def observe(self):
-        wm = pyinotify.WatchManager()
-        self.notifier = pyinotify.Notifier(wm, EventHandler())
-        wm.add_watch(self.directory, pyinotify.IN_CREATE, self.dispatcher.process_audio_file)
-
-        print(f'Watching {self.directory} for changes...')
-        self.notifier.loop()
-
-    def halt(self):
-        self.notifier.stop()
+        while True:
+            files = self.list_files()
+            for file in files:
+                file_time = os.path.getmtime(file)
+                if self.identify_time_window(file_time):
+                    self.dispatcher.process_audio_file(file)
+            self.last_file_time = time.time()
+            time.sleep(10)

requirements.txt

@@ -1,5 +1,4 @@
 openai-whisper==20231117
-pyinotify==0.9.6
 nextcloud_notes_api==1.0.0
 PyYaml==6.0.1
 pyasyncore==1.0.3