Add semantic tagging, search, favourite/NSFW filtering, and LLM job queue

Replaces old list-format tags (which duplicated prompt content) with structured
dict tags per category (origin_series, outfit_type, participants, style_type,
scene_type, etc.). Tags are now purely organizational metadata — removed from
the prompt pipeline entirely.

Adds is_favourite and is_nsfw columns to all 8 resource models. Favourite is
DB-only (user preference); NSFW is mirrored in JSON tags for rescan persistence.
All library pages get filter controls and favourites-first sorting.

Introduces a parallel LLM job queue (_enqueue_task + _llm_queue_worker) for
background tag regeneration, with the same status polling UI as ComfyUI jobs.
Fixes call_llm() to use has_request_context() fallback for background threads.

Adds global search (/search) across resources and gallery images, with navbar
search bar. Adds gallery image sidecar JSON for per-image favourite/NSFW metadata.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

services/generation.py

@@ -11,13 +11,14 @@ from services.sync import _resolve_preset_entity, _resolve_preset_fields
 logger = logging.getLogger('gaze')
 
 
-def generate_from_preset(preset, overrides=None):
+def generate_from_preset(preset, overrides=None, save_category='presets'):
     """Execute preset-based generation.
 
     Args:
         preset: Preset ORM object
         overrides: optional dict with keys:
             checkpoint, extra_positive, extra_negative, seed, width, height, action
+        save_category: upload sub-directory ('presets' or 'generator')
 
     Returns:
         job dict from _enqueue_job()
@@ -52,6 +53,19 @@ def generate_from_preset(preset, overrides=None):
     detailer_obj = _resolve_preset_entity('detailer', detailer_cfg.get('detailer_id'))
     look_obj = _resolve_preset_entity('look', look_cfg.get('look_id'))
 
+    # Build sidecar metadata with resolved entity slugs
+    resolved_meta = {
+        'preset_slug': preset.slug,
+        'preset_name': preset.name,
+        'character_slug': character.slug if character else None,
+        'outfit_slug': outfit.slug if outfit else None,
+        'action_slug': action_obj.slug if action_obj else None,
+        'style_slug': style_obj.slug if style_obj else None,
+        'scene_slug': scene_obj.slug if scene_obj else None,
+        'detailer_slug': detailer_obj.slug if detailer_obj else None,
+        'look_slug': look_obj.slug if look_obj else None,
+    }
+
     # Checkpoint: override > preset config > default
     checkpoint_override = overrides.get('checkpoint', '').strip() if overrides.get('checkpoint') else ''
     if checkpoint_override:
@@ -71,14 +85,31 @@ def generate_from_preset(preset, overrides=None):
         else:
             ckpt_path, ckpt_data = _get_default_checkpoint()
 
+    resolved_meta['checkpoint_path'] = ckpt_path
+
     # Resolve selected fields from preset toggles
     selected_fields = _resolve_preset_fields(data)
 
+    # Check suppress_wardrobe: preset override > action default
+    suppress_wardrobe = False
+    preset_suppress = action_cfg.get('suppress_wardrobe')
+    if preset_suppress == 'random':
+        suppress_wardrobe = random.choice([True, False])
+    elif preset_suppress is not None:
+        suppress_wardrobe = bool(preset_suppress)
+    elif action_obj:
+        suppress_wardrobe = action_obj.data.get('suppress_wardrobe', False)
+
+    if suppress_wardrobe:
+        selected_fields = [f for f in selected_fields if not f.startswith('wardrobe::')]
+
     # Build combined data for prompt building
     active_wardrobe = char_cfg.get('fields', {}).get('wardrobe', {}).get('outfit', 'default')
     wardrobe_source = outfit.data.get('wardrobe', {}) if outfit else None
     if wardrobe_source is None:
         wardrobe_source = character.get_active_wardrobe() if character else {}
+    if suppress_wardrobe:
+        wardrobe_source = {}
 
     combined_data = {
         'character_id': character.character_id if character else 'unknown',
@@ -88,7 +119,6 @@ def generate_from_preset(preset, overrides=None):
         'styles': character.data.get('styles', {}) if character else {},
         'lora': (look_obj.data.get('lora', {}) if look_obj
                  else (character.data.get('lora', {}) if character else {})),
-        'tags': (character.data.get('tags', []) if character else []) + data.get('tags', []),
     }
 
     # Build extras prompt from secondary resources
@@ -108,7 +138,6 @@ def generate_from_preset(preset, overrides=None):
             trg = action_obj.data.get('lora', {}).get('lora_triggers', '')
             if trg:
                 extras_parts.append(trg)
-        extras_parts.extend(action_obj.data.get('tags', []))
     if style_obj:
         s = style_obj.data.get('style', {})
         if s.get('artist_name'):
@@ -133,7 +162,6 @@ def generate_from_preset(preset, overrides=None):
             trg = scene_obj.data.get('lora', {}).get('lora_triggers', '')
             if trg:
                 extras_parts.append(trg)
-        extras_parts.extend(scene_obj.data.get('tags', []))
     if detailer_obj:
         prompt_val = detailer_obj.data.get('prompt', '')
         if isinstance(prompt_val, list):
@@ -195,6 +223,7 @@ def generate_from_preset(preset, overrides=None):
     )
 
     label = f"Preset: {preset.name} – {action}"
-    job = _enqueue_job(label, workflow, _make_finalize('presets', preset.slug, Preset, action))
+    db_model = Preset if save_category == 'presets' else None
+    job = _enqueue_job(label, workflow, _make_finalize(save_category, preset.slug, db_model, action, metadata=resolved_meta))
 
     return job