Upload prompt_builder.py

llm/prompt_builder.py

@@ -0,0 +1,354 @@
+"""
+Flare – Prompt Builder (Refactored with Multi-language Support)
+==============================================================
+"""
+
+from typing import List, Dict, Any
+from datetime import datetime, timedelta
+from config_provider import ConfigProvider
+from logger import log_info, log_error, log_warning, log_debug
+from locale_manager import LocaleManager
+
+# Date helper for locale-aware date expressions
+def _get_date_context(locale_code: str = "tr") -> Dict[str, str]:
+    """Generate date context for date expressions"""
+    now = datetime.now()
+    
+    # Get locale data
+    locale_data = LocaleManager.get_locale(locale_code)
+    weekdays = locale_data.get("weekdays", ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"])
+    
+    today_weekday = weekdays[now.weekday()]
+    
+    # Calculate various dates
+    dates = {
+        "today": now.strftime("%Y-%m-%d"),
+        "tomorrow": (now + timedelta(days=1)).strftime("%Y-%m-%d"),
+        "day_after_tomorrow": (now + timedelta(days=2)).strftime("%Y-%m-%d"),
+        "this_weekend_saturday": (now + timedelta(days=(5-now.weekday())%7)).strftime("%Y-%m-%d"),
+        "this_weekend_sunday": (now + timedelta(days=(6-now.weekday())%7)).strftime("%Y-%m-%d"),
+        "next_week_same_day": (now + timedelta(days=7)).strftime("%Y-%m-%d"),
+        "two_weeks_later": (now + timedelta(days=14)).strftime("%Y-%m-%d"),
+        "today_weekday": today_weekday,
+        "today_day": now.day,
+        "today_month": now.month,
+        "today_year": now.year,
+        "locale_code": locale_code
+    }
+    
+    return dates
+
+def _build_locale_aware_date_prompt(param, date_ctx: Dict, locale_data: Dict, locale_code: str) -> str:
+    """Build locale-aware date extraction prompt"""
+    
+    # Get locale-specific date patterns
+    date_patterns = locale_data.get("date_patterns", {})
+    months = locale_data.get("months", [])
+    
+    prompt_parts = [f"• {param.name}: {param.extraction_prompt}"]
+    prompt_parts.append(f"  IMPORTANT DATE RULES for {locale_data.get('name', 'this language')}:")
+    
+    # Add locale-specific patterns
+    if locale_code == "tr":
+        prompt_parts.extend([
+            f"  - 'bugün' = {date_ctx['today']}",
+            f"  - 'yarın' = {date_ctx['tomorrow']}",
+            f"  - 'öbür gün' = {date_ctx['day_after_tomorrow']}",
+            f"  - 'bu hafta sonu' = {date_ctx['this_weekend_saturday']} or {date_ctx['this_weekend_sunday']}",
+            f"  - 'bu cumartesi' = {date_ctx['this_weekend_saturday']}",
+            f"  - 'bu pazar' = {date_ctx['this_weekend_sunday']}",
+            f"  - 'haftaya' or 'gelecek hafta' = add 7 days to current date",
+            f"  - 'haftaya bugün' = {date_ctx['next_week_same_day']}",
+            f"  - 'iki hafta sonra' = {date_ctx['two_weeks_later']}",
+            f"  - '15 gün sonra' = add 15 days to today",
+            f"  - '10 Temmuz' = {date_ctx['today_year']}-07-10",
+            f"  - Turkish months: Ocak=01, Şubat=02, Mart=03, Nisan=04, Mayıs=05, Haziran=06, "
+            f"Temmuz=07, Ağustos=08, Eylül=09, Ekim=10, Kasım=11, Aralık=12"
+        ])
+    elif locale_code == "en":
+        prompt_parts.extend([
+            f"  - 'today' = {date_ctx['today']}",
+            f"  - 'tomorrow' = {date_ctx['tomorrow']}",
+            f"  - 'day after tomorrow' = {date_ctx['day_after_tomorrow']}",
+            f"  - 'this weekend' = {date_ctx['this_weekend_saturday']} or {date_ctx['this_weekend_sunday']}",
+            f"  - 'this Saturday' = {date_ctx['this_weekend_saturday']}",
+            f"  - 'this Sunday' = {date_ctx['this_weekend_sunday']}",
+            f"  - 'next week' = add 7 days to current date",
+            f"  - 'next {date_ctx['today_weekday']}' = {date_ctx['next_week_same_day']}",
+            f"  - 'in two weeks' = {date_ctx['two_weeks_later']}",
+            f"  - 'July 10' or '10th of July' = {date_ctx['today_year']}-07-10",
+            f"  - English months: January=01, February=02, March=03, April=04, May=05, June=06, "
+            f"July=07, August=08, September=09, October=10, November=11, December=12"
+        ])
+    
+    # Add month mappings if available
+    if months:
+        month_mapping = [f"{month}={i+1:02d}" for i, month in enumerate(months)]
+        prompt_parts.append(f"  - Month names: {', '.join(month_mapping)}")
+    
+    return "\n".join(prompt_parts)
+
+# ─────────────────────────────────────────────────────────────────────────────
+# INTENT PROMPT
+# ─────────────────────────────────────────────────────────────────────────────
+def build_intent_prompt(version: Any,  # VersionConfig
+                       conversation: List[Dict[str, str]], 
+                       project_locale: str) -> str:
+    """Build intent detection prompt with enhanced detection prompts and examples"""
+    
+    # Get config
+    cfg = ConfigProvider.get()
+    
+    # Get internal prompt from LLM provider settings
+    internal_prompt = ""
+    if cfg.global_config.llm_provider and cfg.global_config.llm_provider.settings:
+        internal_prompt = cfg.global_config.llm_provider.settings.get("internal_prompt", "")
+    
+    # Extract intent names and captions
+    intent_names = [it.name for it in version.intents]
+    intent_captions = [str(it.caption) if it.caption else it.name for it in version.intents]
+    
+    # Get project language name
+    locale_data = LocaleManager.get_locale(project_locale)
+    project_language = locale_data.get("name", "Turkish")
+    current_language_name = project_language
+    
+    # Replace placeholders in internal prompt
+    if internal_prompt:
+        # Intent names - quoted and comma-separated
+        intent_names_str = ', '.join([f'"{name}"' for name in intent_names])
+        internal_prompt = internal_prompt.replace("<intent names>", intent_names_str)
+        
+        # Intent captions - quoted and comma-separated
+        intent_captions_str = ', '.join([f'"{caption}"' for caption in intent_captions])
+        internal_prompt = internal_prompt.replace("<intent captions>", intent_captions_str)
+        
+        # Project language
+        internal_prompt = internal_prompt.replace("<project language>", project_language)
+        internal_prompt = internal_prompt.replace("{{current_language_name}}", current_language_name)
+        internal_prompt = internal_prompt.replace("{{project_language}}", project_language)
+
+    # === ENHANCED INTENT INDEX WITH DETECTION PROMPTS AND EXAMPLES ===
+    lines = ["### INTENT DETECTION RULES ###"]
+    for it in version.intents:
+        lines.append(f"\n{it.name}:")
+        
+        # Add detection prompt
+        if it.detection_prompt:
+            lines.append(f"  Detection Rule: {it.detection_prompt}")
+        
+        # Add examples to enhance detection
+        examples = it.get_examples_for_locale(project_locale)
+        if examples:
+            # Combine detection prompt with examples
+            examples_str = " | ".join([f'"{ex}"' for ex in examples[:5]])  # Max 5 examples
+            lines.append(f"  Examples that match this intent: {examples_str}")
+        
+        # If we have both detection prompt and examples, make it clear
+        if it.detection_prompt and examples:
+            lines.append(f"  → This intent should be detected when user says something similar to the examples above AND matches the detection rule.")
+
+    intent_index = "\n".join(lines)
+
+    # === HISTORY ===
+    history_block = "\n".join(
+        f"{m['role'].upper()}: {m['content']}" for m in conversation[-10:]
+    )
+
+    # Combine prompts
+    combined_prompt = internal_prompt + "\n\n" + version.general_prompt if internal_prompt else version.general_prompt
+
+    # Get last user message
+    user_input = conversation[-1]['content'] if conversation else ""
+
+    prompt = (
+        f"{combined_prompt}\n\n"
+        f"{intent_index}\n\n"
+        f"Conversation so far:\n{history_block}\n\n"
+        f"USER: {user_input.strip()}"
+    )
+    
+    log_info("✅ Intent prompt built with enhanced detection prompts and examples")
+    return prompt
+
+# ─────────────────────────────────────────────────────────────────────────────
+# PARAMETER PROMPT
+# ─────────────────────────────────────────────────────────────────────────────
+_FMT = """#PARAMETERS:{"extracted":[{"name":"<param>","value":"<val>"},...],"missing":["<param>",...]}"""
+
+def build_parameter_prompt(
+    version: Any,  # VersionConfig
+    intent_config: Any,  # IntentConfig
+    chat_history: List[Dict[str, str]],
+    collected_params: Dict[str, Any],
+    missing_params: List[str],
+    params_to_ask: List[str],
+    max_params: int,
+    project_locale: str,
+    unanswered_params: List[str] = None
+) -> str:
+    """Build parameter collection prompt with approval support"""
+    
+    # Check if we're asking for approval parameter
+    is_approval_question = len(params_to_ask) == 1 and params_to_ask[0] == "is_approved"
+    
+    if is_approval_question:
+        # For approval, use special format without LLM collection prompt
+        # This will be handled in chat_handler with custom approval_question
+        return ""  # Return empty, chat_handler will use custom question
+    
+    # Normal parameter collection
+    cfg = ConfigProvider.get()
+    collection_config = cfg.global_config.llm_provider.settings.get("parameter_collection_config", {})
+    collection_prompt = collection_config.get("collection_prompt", "")
+    
+    if not collection_prompt:
+        # Fallback prompt
+        collection_prompt = "Ask for the missing parameters naturally."
+    
+    # Build conversation history string
+    history_str = "\n".join([
+        f"{msg['role'].upper()}: {msg['content']}" 
+        for msg in chat_history[-5:]  # Last 5 messages
+    ])
+    
+    # Build collected params string
+    collected_str = "\n".join([
+        f"- {k}: {v}" for k, v in collected_params.items()
+    ])
+    
+    # Build missing params string with captions
+    missing_str = []
+    for param_name in missing_params:
+        param = next((p for p in intent_config.parameters if p.name == param_name), None)
+        if param:
+            caption = param.get_caption_for_locale(project_locale) if hasattr(param, 'get_caption_for_locale') else param_name
+            missing_str.append(f"- {param_name} ({caption})")
+        else:
+            missing_str.append(f"- {param_name}")
+    missing_str = "\n".join(missing_str)
+    
+    # Build unanswered params string
+    unanswered_str = "\n".join([f"- {p}" for p in (unanswered_params or [])])
+    
+    # Replace placeholders
+    prompt = collection_prompt
+    prompt = prompt.replace("{{conversation_history}}", history_str)
+    prompt = prompt.replace("{{intent_name}}", intent_config.name)
+    prompt = prompt.replace("{{intent_caption}}", str(intent_config.caption))
+    prompt = prompt.replace("{{collected_params}}", collected_str)
+    prompt = prompt.replace("{{missing_params}}", missing_str)
+    prompt = prompt.replace("{{unanswered_params}}", unanswered_str)
+    prompt = prompt.replace("{{max_params}}", str(max_params))
+    prompt = prompt.replace("{{project_language}}", project_locale)
+    prompt = prompt.replace("{{current_language_name}}", project_locale)  # For compatibility
+    
+    return prompt
+
+# ─────────────────────────────────────────────────────────────────────────────
+# SMART PARAMETER COLLECTION PROMPT
+# ─────────────────────────────────────────────────────────────────────────────
+def build_smart_parameter_question_prompt(intent_cfg,
+                                        missing_params: List[str],
+                                        conversation: List[Dict[str, str]],
+                                        collection_config: Dict[str, Any],
+                                        project_language: str) -> str:
+    """Build prompt for smart parameter collection"""
+    
+    # Get the collection prompt template
+    template = collection_config.get("collection_prompt", "Ask for the missing parameters.")
+    
+    # Replace placeholders
+    prompt = template.replace("{{max_params}}", str(collection_config.get("max_params_per_question", 2)))
+    prompt = prompt.replace("{{project_language}}", project_language)
+    
+    # Add parameter information
+    param_info = []
+    for param_name in missing_params[:collection_config.get("max_params_per_question", 2)]:
+        param = next((p for p in intent_cfg.parameters if p.name == param_name), None)
+        if param:
+            # Get caption for default locale (first supported locale)
+            caption = param.get_caption_for_locale("tr")  # Default to Turkish
+            param_info.append(f"- {param_name}: {caption}")
+    
+    # Add context
+    parts = [
+        prompt,
+        "",
+        "Missing parameters:",
+        "\n".join(param_info),
+        "",
+        "Recent conversation:"
+    ]
+    
+    # Add recent conversation
+    for msg in conversation[-5:]:
+        parts.append(f"{msg['role'].upper()}: {msg['content']}")
+    
+    return "\n".join(parts)
+
+# ─────────────────────────────────────────────────────────────────────────────
+# PARAMETER EXTRACTION FROM QUESTION
+# ─────────────────────────────────────────────────────────────────────────────
+def extract_params_from_question(question: str, param_names: List[str]) -> Dict[str, str]:
+    """Extract parameters that might be embedded in the question itself"""
+    extracted = {}
+    
+    # Simple pattern matching for common cases
+    # This is a basic implementation - can be enhanced with NLP
+    
+    # Example: "İstanbul'dan Ankara'ya ne zaman gitmek istiyorsunuz?"
+    # Could extract origin=İstanbul, destination=Ankara
+    
+    # For now, return empty dict
+    # This function can be enhanced based on specific use cases
+    
+    return extracted
+
+# ─────────────────────────────────────────────────────────────────────────────
+# API RESPONSE PROMPT
+# ─────────────────────────────────────────────────────────────────────────────
+def build_api_response_prompt(api_config, api_response: Dict) -> str:
+    """Build prompt for API response with mappings"""
+    
+    response_prompt = api_config.response_prompt
+    if not response_prompt:
+        return "İşlem başarıyla tamamlandı."
+    
+    # Apply response mappings if available
+    mapped_data = {}
+    if hasattr(api_config, 'response_mappings'):
+        for mapping in api_config.response_mappings:
+            field_path = mapping.get('field_path', '')
+            display_name = mapping.get('display_name', field_path)
+            
+            # Extract value from response
+            value = _extract_value_from_path(api_response, field_path)
+            if value is not None:
+                mapped_data[display_name] = value
+    
+    # Replace placeholders in response prompt
+    for key, value in mapped_data.items():
+        response_prompt = response_prompt.replace(f"{{{key}}}", str(value))
+    
+    # Also try direct field replacement
+    for key, value in api_response.items():
+        response_prompt = response_prompt.replace(f"{{{key}}}", str(value))
+    
+    return response_prompt
+
+def _extract_value_from_path(data: Dict, path: str) -> Any:
+    """Extract value from nested dict using dot notation"""
+    try:
+        parts = path.split('.')
+        value = data
+        for part in parts:
+            if isinstance(value, dict):
+                value = value.get(part)
+            elif isinstance(value, list) and part.isdigit():
+                value = value[int(part)]
+            else:
+                return None
+        return value
+    except:
+        return None