slahiriandClaude Opus 4.5
737b8dddc5
- Add prompt_length dropdown (Short/Medium/Long/Free) to control output length - Default to "Free" (no constraint) for maximum flexibility - Updated prompts.toml with Z-Image best practices: - Structured format: composition → subject → clothing → lighting - Geometric face angles (front view, 45°, profile) - Specific lighting keywords (soft diffused daylight, rim lighting) - Cleanup constraints at end (no text, no watermark) - Added get_length_constraint() to config_loader.py - Length constraints applied to both single-shot and agentic pipelines 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
405 lines
13 KiB
Python
405 lines
13 KiB
Python
"""
|
|
TOML Configuration Loader for SID Photography Toolkit
|
|
|
|
Loads and caches configuration from TOML files for:
|
|
- Provider settings (tiers, stop strings, capabilities)
|
|
- Prompts (system/user prompts by tier and mode)
|
|
- Components (agentic analysis prompts)
|
|
- Filters (example text patterns to remove)
|
|
|
|
Uses Python 3.11+ tomllib or falls back to tomli package.
|
|
"""
|
|
|
|
import os
|
|
import re
|
|
import sys
|
|
from pathlib import Path
|
|
from typing import Any, Dict, List, Optional
|
|
|
|
# Try Python 3.11+ built-in, fall back to tomli
|
|
try:
|
|
import tomllib
|
|
except ImportError:
|
|
try:
|
|
import tomli as tomllib
|
|
except ImportError:
|
|
tomllib = None
|
|
|
|
|
|
# =============================================================================
|
|
# Configuration Cache
|
|
# =============================================================================
|
|
|
|
_config_cache: Dict[str, Dict] = {}
|
|
_CONFIG_DIR = Path(__file__).parent / "config"
|
|
|
|
|
|
def _load_toml(filename: str) -> Dict[str, Any]:
|
|
"""Load a TOML file from the config directory with caching."""
|
|
if filename in _config_cache:
|
|
return _config_cache[filename]
|
|
|
|
if tomllib is None:
|
|
print(f"[SID-Config] Warning: tomllib/tomli not available, using defaults")
|
|
return {}
|
|
|
|
filepath = _CONFIG_DIR / filename
|
|
if not filepath.exists():
|
|
print(f"[SID-Config] Warning: Config file not found: {filepath}")
|
|
return {}
|
|
|
|
try:
|
|
with open(filepath, "rb") as f:
|
|
data = tomllib.load(f)
|
|
_config_cache[filename] = data
|
|
return data
|
|
except Exception as e:
|
|
print(f"[SID-Config] Error loading {filename}: {e}")
|
|
return {}
|
|
|
|
|
|
def reload_config():
|
|
"""Clear cache and reload all configs (useful for development)."""
|
|
global _config_cache
|
|
_config_cache = {}
|
|
|
|
|
|
# =============================================================================
|
|
# Provider Configuration
|
|
# =============================================================================
|
|
|
|
def get_provider_tier(provider: str) -> str:
|
|
"""Get the tier for a provider (advanced/standard/basic)."""
|
|
config = _load_toml("providers.toml")
|
|
tiers = config.get("tiers", {})
|
|
return tiers.get(provider.lower(), "standard")
|
|
|
|
|
|
def get_stop_strings(provider: str) -> List[str]:
|
|
"""Get stop strings for a provider."""
|
|
config = _load_toml("providers.toml")
|
|
stop_strings = config.get("stop_strings", {})
|
|
return stop_strings.get(provider.lower(), stop_strings.get("default", []))
|
|
|
|
|
|
def get_provider_capabilities(tier: str) -> Dict[str, Any]:
|
|
"""Get capabilities for a tier."""
|
|
config = _load_toml("providers.toml")
|
|
capabilities = config.get("capabilities", {})
|
|
return capabilities.get(tier, capabilities.get("standard", {}))
|
|
|
|
|
|
def get_image_limit(provider: str) -> int:
|
|
"""Get max image size for a provider."""
|
|
config = _load_toml("providers.toml")
|
|
limits = config.get("image_limits", {})
|
|
return limits.get(provider.lower(), limits.get("default", 1024))
|
|
|
|
|
|
# =============================================================================
|
|
# Prompt Configuration
|
|
# =============================================================================
|
|
|
|
def get_system_prompt(tier: str) -> str:
|
|
"""Get the base system prompt for a tier."""
|
|
config = _load_toml("prompts.toml")
|
|
system = config.get("system", {})
|
|
tier_config = system.get(tier, system.get("standard", {}))
|
|
return tier_config.get("base", "Describe this image for AI image generation.")
|
|
|
|
|
|
def get_style_addon(style: str, addon_type: str = "system_addon") -> str:
|
|
"""Get the style addon for system or user prompt."""
|
|
config = _load_toml("prompts.toml")
|
|
styles = config.get("styles", {})
|
|
|
|
# Normalize style name
|
|
style_key = style.lower().replace(" & ", "_").replace(" ", "_").replace("-", "_")
|
|
style_map = {
|
|
"auto_detect": "auto",
|
|
"portrait": "portrait",
|
|
"fashion_outfit": "fashion",
|
|
"fashion": "fashion",
|
|
"artistic_style": "artistic",
|
|
"artistic": "artistic",
|
|
"nsfw_detailed": "nsfw",
|
|
"nsfw": "nsfw",
|
|
}
|
|
style_key = style_map.get(style_key, "auto")
|
|
|
|
style_config = styles.get(style_key, {})
|
|
return style_config.get(addon_type, "")
|
|
|
|
|
|
def get_user_prompt(tier: str, mode: str) -> str:
|
|
"""Get the user prompt for a tier and mode."""
|
|
config = _load_toml("prompts.toml")
|
|
user = config.get("user", {})
|
|
tier_config = user.get(tier, user.get("standard", {}))
|
|
mode_config = tier_config.get(mode.lower(), {})
|
|
return mode_config.get("prompt", "Describe this image.")
|
|
|
|
|
|
def get_agentic_prompt(tier: str, prompt_type: str = "intro") -> str:
|
|
"""Get agentic prompts (intro/synthesis) for a tier."""
|
|
config = _load_toml("prompts.toml")
|
|
agentic = config.get("agentic", {})
|
|
tier_config = agentic.get(tier, agentic.get("standard", {}))
|
|
return tier_config.get(prompt_type, "")
|
|
|
|
|
|
def get_length_constraint(prompt_length: str) -> str:
|
|
"""
|
|
Get the length constraint instruction for a prompt_length setting.
|
|
|
|
Args:
|
|
prompt_length: "Short", "Medium", "Long", or "Free"
|
|
|
|
Returns:
|
|
Length constraint instruction string (empty for "Free")
|
|
"""
|
|
config = _load_toml("prompts.toml")
|
|
constraints = config.get("length_constraints", {})
|
|
# Normalize to lowercase for key lookup
|
|
key = prompt_length.lower()
|
|
return constraints.get(key, "")
|
|
|
|
|
|
# =============================================================================
|
|
# Component Configuration
|
|
# =============================================================================
|
|
|
|
def get_component_prompt(component: str, tier: str) -> str:
|
|
"""Get the prompt for a specific component and tier."""
|
|
config = _load_toml("components.toml")
|
|
comp_config = config.get(component, {})
|
|
tier_config = comp_config.get(tier, comp_config.get("standard", {}))
|
|
return tier_config.get("prompt", "")
|
|
|
|
|
|
def get_component_name(component: str) -> str:
|
|
"""Get the display name for a component."""
|
|
config = _load_toml("components.toml")
|
|
comp_config = config.get(component, {})
|
|
return comp_config.get("name", component.replace("_", " ").title())
|
|
|
|
|
|
def get_mode_config(mode: str) -> Dict[str, Any]:
|
|
"""Get the configuration for an analysis mode."""
|
|
config = _load_toml("components.toml")
|
|
modes = config.get("modes", {})
|
|
return modes.get(mode.lower(), modes.get("standard", {}))
|
|
|
|
|
|
def get_mode_components(mode: str) -> List[str]:
|
|
"""Get the list of components for an analysis mode."""
|
|
mode_config = get_mode_config(mode)
|
|
return mode_config.get("components", ["framing", "ethnicity", "clothing"])
|
|
|
|
|
|
def get_all_components() -> List[str]:
|
|
"""Get list of all available component keys."""
|
|
config = _load_toml("components.toml")
|
|
# Filter out the 'modes' section
|
|
return [k for k in config.keys() if k != "modes"]
|
|
|
|
|
|
# =============================================================================
|
|
# Filter Configuration
|
|
# =============================================================================
|
|
|
|
def get_example_patterns() -> List[str]:
|
|
"""Get regex patterns for example text to remove."""
|
|
config = _load_toml("filters.toml")
|
|
example = config.get("example_patterns", {})
|
|
return example.get("patterns", [])
|
|
|
|
|
|
def get_markdown_patterns() -> List[str]:
|
|
"""Get regex patterns for markdown cleanup."""
|
|
config = _load_toml("filters.toml")
|
|
markdown = config.get("markdown_patterns", {})
|
|
return markdown.get("patterns", [])
|
|
|
|
|
|
def get_whitespace_patterns() -> Dict[str, str]:
|
|
"""Get whitespace cleanup patterns."""
|
|
config = _load_toml("filters.toml")
|
|
return config.get("whitespace", {})
|
|
|
|
|
|
def get_provider_cleanup_patterns(provider: str) -> List[str]:
|
|
"""Get provider-specific cleanup patterns."""
|
|
config = _load_toml("filters.toml")
|
|
cleanup = config.get("provider_cleanup", {})
|
|
provider_config = cleanup.get(provider.lower(), cleanup.get("default", {}))
|
|
return provider_config.get("patterns", [])
|
|
|
|
|
|
# =============================================================================
|
|
# Output Cleaning Utility
|
|
# =============================================================================
|
|
|
|
def clean_output(text: str, provider: str = "default") -> str:
|
|
"""
|
|
Clean LLM output using patterns from filters.toml.
|
|
|
|
Args:
|
|
text: Raw LLM output
|
|
provider: Provider name for provider-specific cleanup
|
|
|
|
Returns:
|
|
Cleaned text
|
|
"""
|
|
if not text:
|
|
return ""
|
|
|
|
# Remove example text patterns
|
|
for pattern in get_example_patterns():
|
|
try:
|
|
text = re.sub(pattern, "", text, flags=re.IGNORECASE)
|
|
except re.error:
|
|
pass
|
|
|
|
# Remove markdown patterns
|
|
for pattern in get_markdown_patterns():
|
|
try:
|
|
# Handle patterns with capture groups (like bold/italic)
|
|
if "(" in pattern and ")" in pattern:
|
|
text = re.sub(pattern, r"\1", text, flags=re.MULTILINE)
|
|
else:
|
|
text = re.sub(pattern, "", text, flags=re.MULTILINE)
|
|
except re.error:
|
|
pass
|
|
|
|
# Provider-specific cleanup
|
|
for pattern in get_provider_cleanup_patterns(provider):
|
|
try:
|
|
text = re.sub(pattern, "", text, flags=re.MULTILINE)
|
|
except re.error:
|
|
pass
|
|
|
|
# Whitespace cleanup
|
|
ws = get_whitespace_patterns()
|
|
if ws.get("collapse_newlines"):
|
|
text = re.sub(ws["collapse_newlines"], " ", text)
|
|
if ws.get("collapse_spaces"):
|
|
text = re.sub(ws["collapse_spaces"], " ", text)
|
|
if ws.get("double_commas"):
|
|
text = re.sub(ws["double_commas"], ",", text)
|
|
if ws.get("leading_comma"):
|
|
text = re.sub(ws["leading_comma"], "", text)
|
|
if ws.get("trailing_comma"):
|
|
text = re.sub(ws["trailing_comma"], "", text)
|
|
if ws.get("double_periods"):
|
|
text = re.sub(ws["double_periods"], ".", text)
|
|
|
|
return text.strip()
|
|
|
|
|
|
# =============================================================================
|
|
# Convenience Functions
|
|
# =============================================================================
|
|
|
|
def build_system_prompt(provider: str, preset_style: str, user_guidance: str = "", prompt_length: str = "Free") -> str:
|
|
"""
|
|
Build complete system prompt based on provider tier and style.
|
|
|
|
Args:
|
|
provider: LLM provider name
|
|
preset_style: Selected style preset
|
|
user_guidance: Optional user guidance text
|
|
prompt_length: Output length setting (Short/Medium/Long/Free)
|
|
|
|
Returns:
|
|
Complete system prompt string
|
|
"""
|
|
tier = get_provider_tier(provider)
|
|
base = get_system_prompt(tier)
|
|
|
|
style_addon = get_style_addon(preset_style, "system_addon")
|
|
if style_addon:
|
|
base += f"\n\n{style_addon}"
|
|
|
|
# Add length constraint
|
|
length_constraint = get_length_constraint(prompt_length)
|
|
if length_constraint:
|
|
base += f"\n\n{length_constraint}"
|
|
|
|
if user_guidance and user_guidance.strip():
|
|
base += f"""
|
|
|
|
USER MODIFICATION REQUEST (CRITICAL - MUST APPLY):
|
|
"{user_guidance.strip()}"
|
|
|
|
This is a MODIFICATION directive. You MUST:
|
|
1. Apply this modification to your description, overriding what's literally visible
|
|
2. Describe the subject AS IF this modification is true
|
|
3. For clothing/nudity changes: describe the MODIFIED state (e.g., if "make topless" and image shows bikini, describe as topless)
|
|
4. Do NOT describe the original state that contradicts the user request
|
|
5. Start your description with the modified element"""
|
|
|
|
return base
|
|
|
|
|
|
def build_user_prompt(provider: str, analysis_mode: str, preset_style: str, prompt_length: str = "Free") -> str:
|
|
"""
|
|
Build user prompt based on provider tier and analysis mode.
|
|
|
|
Args:
|
|
provider: LLM provider name
|
|
analysis_mode: Analysis mode (Quick/Standard/Detailed/Extreme)
|
|
preset_style: Selected style preset
|
|
prompt_length: Output length setting (Short/Medium/Long/Free)
|
|
|
|
Returns:
|
|
Complete user prompt string
|
|
"""
|
|
tier = get_provider_tier(provider)
|
|
prompt = get_user_prompt(tier, analysis_mode)
|
|
|
|
style_addon = get_style_addon(preset_style, "user_addon")
|
|
if style_addon:
|
|
prompt += f"\n\n{style_addon}"
|
|
|
|
# Reinforce length constraint in user prompt (already in system, but reinforce)
|
|
length_constraint = get_length_constraint(prompt_length)
|
|
if length_constraint:
|
|
prompt += f"\n\nREMEMBER: {length_constraint}"
|
|
|
|
return prompt
|
|
|
|
|
|
# =============================================================================
|
|
# Initialization Check
|
|
# =============================================================================
|
|
|
|
def check_config_files() -> Dict[str, bool]:
|
|
"""Check which config files exist and are loadable."""
|
|
files = ["providers.toml", "prompts.toml", "components.toml", "filters.toml"]
|
|
status = {}
|
|
for f in files:
|
|
filepath = _CONFIG_DIR / f
|
|
status[f] = filepath.exists()
|
|
if status[f]:
|
|
try:
|
|
_load_toml(f)
|
|
except Exception:
|
|
status[f] = False
|
|
return status
|
|
|
|
|
|
# Print status on import (for debugging)
|
|
if __name__ == "__main__":
|
|
print("Config file status:")
|
|
for file, exists in check_config_files().items():
|
|
status = "OK" if exists else "MISSING"
|
|
print(f" {file}: {status}")
|
|
|
|
print("\nProvider tiers:")
|
|
for p in ["anthropic", "openai", "ollama", "lmstudio"]:
|
|
print(f" {p}: {get_provider_tier(p)}")
|
|
|
|
print("\nSample system prompt (basic tier):")
|
|
print(get_system_prompt("basic"))
|