ML-pipeline/config.py

import os


class Config:
    # === System Architecture ===
    USER_ID = "ION606"
    LOG_DIR = "logs"

    # === Code Execution ===
    CODE_TIMEOUT = 15                # Seconds before killing process
    CODE_MEMORY_LIMIT_MB = 100       # Memory limit for subprocesses
    OUTPUT_CHAR_LIMIT = 2000         # Max characters for stdout/stderr
    MAX_CODE_RETRIES = 3             # Max code fix attempts
    CORE_DUMP_LIMIT = 0              # Disable core dumps
    MAX_CODE_LENGTH = 10000          # Character limit for code inputs

    # === Orchestration ===
    MAX_ORCHESTRATION_ITERATIONS = 5  # Max loops in orchestrate()
    SEARCH_CONTEXT_LIMIT = 3000      # Characters for web context
    LOCAL_CONTEXT_ITEMS = 3          # File chunks to include

    # === Text Processing ===
    CHUNK_SIZE = 1000                # Document chunking size
    CHUNK_OVERLAP = 200              # Document chunk overlap
    RESPONSE_TOKEN_LIMIT = 4096      # Max LLM response tokens

    # === Search ===
    MAX_SEARCH_RESULTS = 3           # Web results per query
    MIN_SEARCH_QUERY_LEN = 3         # Minimum search terms

    # === Conversation Store ===
    CONVERSATION_DB_PATH = os.path.join(
        os.path.dirname(__file__), "data/conversations.db")
    MAX_QUERY_LENGTH = 2000          # Characters for stored queries
    MAX_RESPONSE_LENGTH = 10000       # Characters for stored responses

    # === Model Settings ===
    MODEL_TEMPERATURE = {
        "simple": 0.3,
        "medium": 0.6,
        "complex": 0.7
    }

    MAX_CLASSIFY_ATTEMPTS = 3        # Task classification retries

    # === Safety Limits ===
    MAX_INPUT_LENGTH = 15000         # Absolute input character limit
    MAX_CONTEXT_DEPTH = 5            # Max nested context references

    # === Formatting ===
    @staticmethod
    def valid_range(num_options: int) -> str:
        return f"0-{num_options - 1}"

    @staticmethod
    def code_block_regex():
        return r'```python(.*?)```'

    # === Path Helpers ===
    @staticmethod
    def chroma_path():
        return "./data/chroma_db"

    @staticmethod
    def debug_log_path():
        return f"{Config.LOG_DIR}/debug.txt"