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

scraibe/summarizer.py

@@ -6,8 +6,8 @@ Provides a client to summarize long transcripts via an LLM endpoint.
 
 Behavior:
 - Chunks transcript into 10,240-character segments.
-- Generates a summary for each chunk.
-- Combines all chunk summaries and produces a final, detailed summary.
+- Summarizes each chunk.
+- Summarizes the summaries into a final, detailed summary.
 
 Environment Variables:
 - SUMMARIZER_API_URL: (required) Base URL of the LLM API (e.g., http://localhost:8080)
@@ -17,10 +17,13 @@ Environment Variables:
 
 import os
 import json
+import logging
 from typing import Optional
 
 import httpx
 
+logger = logging.getLogger("scraibe.summarizer")
+
 
 class SummarizerError(Exception):
     """Raised when the summarization API call fails."""
@@ -53,6 +56,12 @@ class SummarizerClient:
                 "Provide the summarization LLM URL via environment or constructor."
             )
 
+        logger.info(
+            "Initializing SummarizerClient: url=%s model=%s",
+            self.api_url,
+            self.model,
+        )
+
         self._client = httpx.Client(
             base_url=self.api_url,
             timeout=self.timeout,
@@ -84,21 +93,40 @@ class SummarizerClient:
         - Next steps / action items
         """
         if not transcript.strip():
+            logger.warning("Empty transcript provided to summarize_transcript.")
             return "No transcript provided to summarize."
 
+        logger.info(
+            "Starting summarization for transcript length=%d chars",
+            len(transcript),
+        )
+
         # 1) Chunk the transcript
         chunks = self._chunk_text(transcript)
+        logger.info("Split transcript into %d chunks.", len(chunks))
 
         # 2) Summarize each chunk
         chunk_summaries = []
         for i, chunk in enumerate(chunks):
+            logger.info(
+                "Summarizing chunk %d/%d (length=%d)",
+                i + 1,
+                len(chunks),
+                len(chunk),
+            )
             summary = self._summarize_chunk(chunk, i, len(chunks))
             chunk_summaries.append(summary)
 
         # 3) Combine and summarize summaries
         combined = "\n\n".join(chunk_summaries)
+        logger.info(
+            "Combining %d chunk summaries (total length=%d) for final summary.",
+            len(chunk_summaries),
+            len(combined),
+        )
         final_summary = self._summarize_combined(combined)
 
+        logger.info("Summarization completed.")
         return final_summary
 
     def _chunk_text(self, text: str) -> list[str]:
@@ -183,13 +211,18 @@ class SummarizerClient:
         if self.api_key:
             headers["Authorization"] = f"Bearer {self.api_key}"
 
+        logger.info("Calling summarizer endpoint: /v1/chat/completions")
+
         resp = self._client.post(
             "/v1/chat/completions",
             json=payload,
             headers=headers,
         )
 
+        logger.info("Summarizer response status: %d", resp.status_code)
+
         if resp.status_code >= 400:
+            logger.error("Summarizer error response: %s", resp.text)
             raise SummarizerError(
                 f"Summarizer API error {resp.status_code}: {resp.text}"
             )
@@ -197,6 +230,7 @@ class SummarizerClient:
         try:
             data = resp.json()
         except json.JSONDecodeError:
+            logger.error("Failed to parse summarizer response as JSON.")
             raise SummarizerError(
                 "Failed to parse summarizer response as JSON."
             )
@@ -206,6 +240,10 @@ class SummarizerClient:
             content = data["choices"][0]["message"]["content"]
             return content.strip()
         except (KeyError, IndexError, TypeError):
+            logger.error(
+                "Unexpected summarizer response format: %s",
+                json.dumps(data, indent=2),
+            )
             raise SummarizerError(
                 "Unexpected summarizer response format: "
                 f"{json.dumps(data, indent=2)}"