Add structured logging for Docker; support LOG_LEVEL env and --log-level

scraibe/autotranscript.py

@@ -16,12 +16,15 @@ but ignored when not relevant.
 """
 
 import os
+import logging
 from typing import Union, Optional
 
 from .localai_client import LocalAIClient, LocalAIError
 from .summarizer import SummarizerClient, SummarizerError
 from .transcript_exporter import Transcript
 
+logger = logging.getLogger("scraibe.autotranscript")
+
 
 class Scraibe:
     """
@@ -68,6 +71,8 @@ class Scraibe:
         """
         self.verbose = verbose or kwargs.get("verbose", False)
 
+        logger.info("Initializing Scraibe.")
+
         try:
             self.client = LocalAIClient(
                 api_url=api_url,
@@ -75,6 +80,7 @@ class Scraibe:
                 model=model,
             )
         except LocalAIError as e:
+            logger.error("Failed to initialize LocalAI client: %s", e)
             raise LocalAIError(f"Failed to initialize LocalAI client: {e}")
 
         # Summarizer is lazy-initialized if needed
@@ -95,6 +101,7 @@ class Scraibe:
         if self._summarizer is not None:
             return self._summarizer
 
+        logger.info("Initializing SummarizerClient (lazy).")
         try:
             self._summarizer = SummarizerClient(
                 api_url=api_url,
@@ -102,6 +109,7 @@ class Scraibe:
                 model=model,
             )
         except SummarizerError as e:
+            logger.error("Failed to initialize Summarizer client: %s", e)
             raise SummarizerError(f"Failed to initialize Summarizer client: {e}")
 
         return self._summarizer
@@ -137,6 +145,7 @@ class Scraibe:
             )
 
         verbose = kwargs.get("verbose", self.verbose)
+        logger.info("transcribe called for: %s", audio_file)
 
         try:
             result = self.client.diarize_and_transcribe(
@@ -146,10 +155,13 @@ class Scraibe:
                 **kwargs,
             )
         except LocalAIError as e:
+            logger.error("Error during LocalAI transcription: %s", e)
             raise LocalAIError(f"Error during LocalAI transcription: {e}")
 
         transcripts = result.get("transcripts", [])
-        return " ".join(t.strip() for t in transcripts if t.strip())
+        text = " ".join(t.strip() for t in transcripts if t.strip())
+        logger.info("transcribe completed, length=%d chars", len(text))
+        return text
 
     def transcript_and_summarize(
         self,
@@ -182,6 +194,7 @@ class Scraibe:
             )
 
         verbose = kwargs.get("verbose", self.verbose)
+        logger.info("transcript_and_summarize called for: %s", audio_file)
 
         # 1) Get diarized + transcribed result
         try:
@@ -192,6 +205,7 @@ class Scraibe:
                 **kwargs,
             )
         except LocalAIError as e:
+            logger.error("Error during LocalAI transcription: %s", e)
             raise LocalAIError(f"Error during LocalAI transcription: {e}")
 
         segments = result.get("segments", [])
@@ -199,6 +213,7 @@ class Scraibe:
         transcripts = result.get("transcripts", [])
 
         if not segments:
+            logger.warning("No segments returned; returning empty transcript/summary.")
             return {
                 "transcript": "",
                 "summary": "No transcript content to summarize.",
@@ -213,6 +228,7 @@ class Scraibe:
             lines.append(line)
 
         full_transcript = "\n\n".join(lines)
+        logger.info("Built full transcript, length=%d chars", len(full_transcript))
 
         # 3) Summarize
         try:
@@ -222,13 +238,17 @@ class Scraibe:
                 model=summarizer_model,
             )
         except SummarizerError as e:
+            logger.error("Failed to initialize summarizer: %s", e)
             raise SummarizerError(f"Failed to initialize summarizer: {e}")
 
         try:
             summary = summarizer.summarize_transcript(full_transcript)
         except SummarizerError as e:
+            logger.error("Error during summarization: %s", e)
             raise SummarizerError(f"Error during summarization: {e}")
 
+        logger.info("transcript_and_summarize completed.")
+
         return {
             "transcript": full_transcript,
             "summary": summary,