feat: Use path info to complete partial Skaldleita results (#127)

CHANGELOG.md

@@ -2,7 +2,14 @@
 
 All notable changes to Library Manager will be documented in this file.
 
-## [0.9.0-beta.126] - 2026-02-16
+## [0.9.0-beta.127] - 2026-02-16
+
+### Added
+
+- **Issue #127: Path-based completion for partial results** - When Skaldleita returns truncated
+  names (e.g., "James S. A" instead of "James S. A. Corey"), the system now uses folder path
+  information to complete the full name. Also extracts series information from path structure
+  when missing from audio identification. Requires minimum 4-char prefix match for safety.
 
 ### Fixed
 

README.md

@@ -4,7 +4,7 @@
 
 **Smart Audiobook Library Organizer with Multi-Source Metadata & AI Verification**
 
-[![Version](https://img.shields.io/badge/version-0.9.0--beta.126-blue.svg)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-0.9.0--beta.127-blue.svg)](CHANGELOG.md)
 [![Docker](https://img.shields.io/badge/docker-ghcr.io-blue.svg)](https://ghcr.io/deucebucket/library-manager)
 [![License](https://img.shields.io/badge/license-AGPL--3.0-blue.svg)](LICENSE)
 

app.py

@@ -11,7 +11,7 @@
 - Multi-provider AI (Gemini, OpenRouter, Ollama)
 """
 
-APP_VERSION = "0.9.0-beta.126"
+APP_VERSION = "0.9.0-beta.127"
 GITHUB_REPO = "deucebucket/library-manager"  # Your GitHub repo
 
 # Versioning Guide:
@@ -701,7 +701,7 @@
             try:
                 with open(ERROR_REPORTS_PATH, 'r') as f:
                     reports = json.load(f)
             except:
                 reports = []
 
         # Add new report (keep last 100 reports to avoid file bloat)
@@ -725,7 +725,7 @@
         try:
             with open(ERROR_REPORTS_PATH, 'r') as f:
                 return json.load(f)
         except:
             return []
     return []
 
@@ -1680,7 +1680,7 @@
                     continue
                 result = call_gemini(prompt, merged_config)
                 if result:
                     logger.info(f"[PROVIDER CHAIN] Success with gemini")
                     return result
 
             elif provider == 'openrouter':
@@ -1689,13 +1689,13 @@
                     continue
                 result = call_openrouter(prompt, merged_config)
                 if result:
                     logger.info(f"[PROVIDER CHAIN] Success with openrouter")
                     return result
 
             elif provider == 'ollama':
                 result = call_ollama(prompt, merged_config)
                 if result:
                     logger.info(f"[PROVIDER CHAIN] Success with ollama")
                     return result
 
             else:
@@ -1797,7 +1797,7 @@
                         return result
                     elif result and result.get('transcript'):
                         # Got transcript but no match - still useful, return for potential AI fallback
                         logger.info(f"[AUDIO CHAIN] BookDB returned transcript only")
                         return result
                     elif result is None and attempt < max_retries - 1:
                         # Connection might be down, wait and retry
@@ -2129,11 +2129,11 @@
                 device = "cuda"
                 # int8 works on all CUDA devices including GTX 1080 (compute 6.1)
                 # float16 only works on newer GPUs (compute 7.0+)
                 logger.info(f"[WHISPER] Using CUDA GPU acceleration (10x faster)")
             else:
                 logger.info(f"[WHISPER] Using CPU (no CUDA GPU detected)")
         except ImportError:
             logger.info(f"[WHISPER] Using CPU (ctranslate2 not available)")
 
         _whisper_model = WhisperModel(model_name, device=device, compute_type=compute_type)
         _whisper_model_name = model_name
@@ -2340,7 +2340,7 @@
         if sample_path and os.path.exists(sample_path):
             try:
                 os.unlink(sample_path)
             except:
                 pass
 
     return result

library_manager/pipeline/layer_audio_id.py

@@ -14,6 +14,7 @@
 import json
 import logging
 import os
+import re
 import time
 from datetime import datetime, timedelta
 from pathlib import Path
@@ -146,7 +147,6 @@ def _validate_ai_result_against_path(result: Dict, folder_hint: str, book_path:
 
     # Clean up - remove common noise
     def clean_text(text):
-        import re
         # Remove brackets, hashes, special chars
         text = re.sub(r'\[[^\]]*\]', ' ', text)
         text = re.sub(r'[^a-z0-9\s]', ' ', text)
@@ -188,6 +188,136 @@ def clean_text(text):
     return result
 
 
+def _complete_result_from_path(result: Dict, folder_hint: str, book_path: str) -> Dict:
+    """
+    Use path info to complete partial/truncated Skaldleita results.
+
+    When SL returns a truncated name (e.g., "James S. A" instead of "James S. A. Corey"),
+    the file path often contains the full name. If the path version starts with the SL
+    version, use the longer path version.
+
+    Only completes - never replaces a longer SL result with a shorter path fragment.
+
+    Args:
+        result: The identification result dict (author, title, series, etc.)
+        folder_hint: "current_author - current_title" string from path parsing
+        book_path: Full filesystem path to the book folder
+
+    Returns:
+        The result dict with potentially completed fields.
+    """
+    if not result or not folder_hint:
+        return result
+
+    # Parse the folder hint into author and title components
+    # folder_hint format: "current_author - current_title"
+    hint_parts = folder_hint.split(' - ', 1)
+    path_author = hint_parts[0].strip() if len(hint_parts) >= 1 else ''
+    path_title = hint_parts[1].strip() if len(hint_parts) >= 2 else ''
+
+    def _is_truncated_version(shorter: str, longer: str) -> bool:
+        """Check if 'shorter' is a truncated prefix of 'longer'.
+
+        Returns True if the shorter string is a prefix of the longer string
+        (case-insensitive), and the longer string has meaningful additional content.
+        Requires the shorter version to be at least 4 characters to avoid false
+        matches on trivial prefixes.
+        """
+        if not shorter or not longer:
+            return False
+        s = shorter.strip()
+        l = longer.strip()
+        if len(s) < 4 or len(s) >= len(l):
+            return False
+        return l.lower().startswith(s.lower())
+
+    completed_any = False
+
+    # Complete author if truncated
+    sl_author = result.get('author', '') or ''
+    if sl_author and path_author and _is_truncated_version(sl_author, path_author):
+        logger.info(f"[PATH COMPLETE] Author: '{sl_author}' -> '{path_author}' (path has full name)")
+        result['author'] = path_author
+        result['path_completed_author'] = True
+        completed_any = True
+
+    # Complete title if truncated
+    sl_title = result.get('title', '') or ''
+    if sl_title and path_title and _is_truncated_version(sl_title, path_title):
+        logger.info(f"[PATH COMPLETE] Title: '{sl_title}' -> '{path_title}' (path has full name)")
+        result['title'] = path_title
+        result['path_completed_title'] = True
+        completed_any = True
+
+    # Try to extract series info from path if SL returned none
+    # Path often has patterns like "The Stormlight Archive 01" or "Series Name/Book 01"
+    sl_series = result.get('series', '') or ''
+    if not sl_series and path_title:
+        # Check for series number patterns in the path title
+        # e.g., "The Way of Kings (Stormlight Archive 01)" or "Book Title - Series Name 01"
+        # Look for series patterns in the full path (parent directories)
+        path_obj = Path(book_path) if book_path else None
+        if path_obj:
+            # Check parent directory names for series info not captured by SL
+            # Typical structure: library/Author/SeriesOrTitle/BookTitle
+            # Only check the last 3-4 relevant dirs, not filesystem root
+            parts = path_obj.parts
+            for part in parts[-4:-1]:
+                # Match patterns like "Series Name 01" or "Series Name - Book 01"
+                series_match = re.match(
+                    r'^(.+?)\s*[-–]\s*(?:Book\s+)?(\d+(?:\.\d+)?)\s*$',
+                    part, re.IGNORECASE
+                )
+                if not series_match:
+                    series_match = re.match(
+                        r'^(.+?)\s+(\d+(?:\.\d+)?)\s*$',
+                        part, re.IGNORECASE
+                    )
+                if series_match:
+                    potential_series = series_match.group(1).strip()
+                    potential_num = series_match.group(2).strip()
+                    # Min length 3 to avoid false matches like "The" or "No"
+                    if (len(potential_series) >= 3
+                            and potential_series.lower() != sl_title.lower()
+                            and potential_series.lower() != (result.get('author') or '').lower()):
+                        logger.info(f"[PATH COMPLETE] Series: '{potential_series}' #{potential_num} (extracted from path)")
+                        result['series'] = potential_series
+                        result['series_num'] = potential_num
+                        result['path_completed_series'] = True
+                        completed_any = True
+                        break
+
+    # If we completed anything, give a small confidence boost since path corroborates SL
+    if completed_any:
+        raw_conf = result.get('confidence', 0.7)
+        try:
+            if isinstance(raw_conf, str):
+                # SL can return string levels or numeric strings like "0.7"
+                try:
+                    numeric = float(raw_conf)
+                    if numeric <= 1:
+                        result['confidence'] = min(0.95, numeric + 0.05)
+                    else:
+                        result['confidence'] = min(95, numeric + 5)
+                except ValueError:
+                    # Named confidence levels - bump up one tier
+                    if raw_conf == 'low':
+                        result['confidence'] = 'medium'
+                    elif raw_conf == 'medium':
+                        result['confidence'] = 'high'
+                    # 'high' stays high
+            elif isinstance(raw_conf, (int, float)):
+                # Numeric confidence - small boost (5%) for path agreement, cap at 0.95
+                if raw_conf <= 1:
+                    result['confidence'] = min(0.95, raw_conf + 0.05)
+                else:
+                    result['confidence'] = min(95, raw_conf + 5)
+        except (ValueError, TypeError):
+            pass  # Leave confidence unchanged if we can't parse it
+
+    return result
+
+
 def process_layer_1_audio(
     config: Dict,
     get_db: Callable,
@@ -444,6 +574,8 @@ def process_layer_1_audio(
                         'sl_source': sl_source,
                         'requeue_suggested': True
                     }
+                    # Issue #127: Complete truncated SL results using path info
+                    result = _complete_result_from_path(result, folder_hint, book_path)
                     # Continue processing - let the normal flow create pending_fix
                     # The requeue flag will be used to schedule a future recheck
                 else:
@@ -475,6 +607,8 @@ def process_layer_1_audio(
                     transcript = bookdb_result.get('transcript')  # Keep transcript for AI
                     result = None  # Clear to trigger AI fallback
                 else:
+                    # Issue #127: Complete truncated SL results using path info
+                    bookdb_result = _complete_result_from_path(bookdb_result, folder_hint, book_path)
                     result = bookdb_result  # Passed sanity check
             else:
                 # Skaldleita didn't get a full match - might have a transcript though
@@ -513,6 +647,9 @@ def process_layer_1_audio(
             # This catches cases where AI completely misparses (e.g., narrator name as author)
             if result:
                 result = _validate_ai_result_against_path(result, folder_hint, book_path)
+                # Issue #127: Complete truncated AI results using path info
+                if result and not result.get('sanity_failed'):
+                    result = _complete_result_from_path(result, folder_hint, book_path)
 
         if result and result.get('author') and result.get('title') and result.get('confidence') != 'none':
             # Got identification from audio!