PLC_Generation/PLC Data Generator/classifiers.py

import pandas as pd
import re

def _is_hub_tagname_pattern(tagname: str) -> bool:
    """Check if TAGNAME matches hub patterns like FIO1H, FIO2H3, FIOH1, etc."""
    # Match patterns like FIO1H, FIO2H3, FIOH1, etc.
    hub_pattern = re.compile(r'FIO(\d+)?H\d*', re.IGNORECASE)
    return bool(hub_pattern.search(tagname))

def _is_master_tagname_pattern(tagname: str) -> bool:
    """Check if TAGNAME matches master patterns like FIO2M1, FIO3M, etc."""
    # Match patterns like FIO2M1, FIO3M, etc.
    master_pattern = re.compile(r'FIO\d+M\d*', re.IGNORECASE)
    return bool(master_pattern.search(tagname))

def _is_hub_desca_pattern(desca: str) -> bool:
    """Check if DESCA matches hub patterns like FIO1H, PDP12_FIO1H, FIO2H3, etc."""
    # Match patterns like FIO1H, PDP12_FIO1H, FIO2H3, etc. in DESCA
    # Allow optional prefix, FIO + digits + H + optional digits  
    hub_pattern = re.compile(r'FIO\d+H\d*', re.IGNORECASE)
    return bool(hub_pattern.search(desca))

def classify_signal(desca, tagname, descb=None, excel_filename=None):
    """
    Classify signal based on DESCA content, TAGNAME, DESCB, and Excel filename
    Priority order matters: PB_LT before PB, FIOH before FIO, SOL+DIVERT before SOL
    
    Special rules:
    - If Excel filename contains MCM05, SOL signals are classified as 'O' instead of 'IOLink'
    """
    if pd.isna(desca):
        return 'UNKNOWN'
    
    desca_str = str(desca).upper()
    tagname_str = str(tagname).upper()
    descb_str = str(descb).upper() if pd.notna(descb) else ''
    
    # Check if Excel filename contains MCM05 (special rule for solenoids)
    is_mcm05 = False
    if excel_filename:
        is_mcm05 = 'MCM05' in str(excel_filename).upper()
    
    # Check for SPARE first
    if re.search(r'SPARE', desca_str):
        return 'SPARE'
    
    # Signal O patterns (check higher priority first)
    if re.search(r'BCN\d+_[AGBR]', desca_str):  # e.g., BCN3_B, BCN1_A
        return 'O'
    elif re.search(r'PB_LT', desca_str):
        return 'O'
    
    # Check for PR sensors (user rule: PR is I) - moved after specific patterns
    elif re.search(r'_PR\d+|PR\d+', desca_str):  # Match PR followed by digits (proximity sensors)
        return 'I'
    elif re.search(r'STO', desca_str):
        return 'O'
    elif re.search(r'BCN', desca_str) and 'FIOH' in tagname_str:
        return 'O'
    elif re.search(r'_H', desca_str):
        return 'O'
    
    # IOLink patterns (check these first before general I patterns to avoid conflicts)
    elif re.search(r'SOL', desca_str) and re.search(r'DIVERT', descb_str):
        # MCM05 rule: SOL signals should be 'O', not 'IOLink'
        if is_mcm05:
            return 'O'
        return 'IOLink'
    elif re.search(r'LPE', desca_str):  # LPE must be checked before PE patterns
        return 'IOLink'
    
    # Signal I patterns (check these after IOLink patterns to avoid conflicts)
    elif re.search(r'TS\d*|BDS\d*', desca_str):  # TS and BDS patterns
        return 'I'
    elif re.search(r'LRPE', desca_str):  # LRPE patterns
        return 'I'
    elif re.search(r'PWM', desca_str):
        return 'I'
    elif re.search(r'CB', desca_str):
        return 'I'
    elif re.search(r'FPE', desca_str):
        return 'I'
    elif re.search(r'EN', desca_str):
        return 'I'
    elif re.search(r'PS', desca_str):
        return 'I'
    elif re.search(r'EPC', desca_str):
        return 'I'
    elif re.search(r'PX', desca_str):
        return 'I'
    elif re.search(r'_PE\d+', desca_str):  # _PE followed by digits
        # Check if DESCB contains SEND - if so, it's output
        if re.search(r'SEND', descb_str):
            return 'O'
        return 'I'
    elif re.search(r'PB', desca_str):  # PB patterns (already had this but moved up)
        return 'I'
    elif re.search(r'MDR', desca_str):  # PB patterns (already had this but moved up)
        return 'O'

    # Check SOL patterns first (before general FIOH patterns)
    elif re.search(r'SOL', desca_str):
        # SOL signals are always 'O' (including MCM05 rule)
        return 'O'
    
    # Remaining IOLink patterns
    elif re.search(r'FIOH', desca_str):
        return 'IOLink'
    elif _is_hub_desca_pattern(desca_str):  # Check for FIO1H1, FIO2H3 etc patterns in DESCA
        return 'IOLink'
    elif _is_hub_tagname_pattern(tagname_str):  # Check for FIO1H, FIO2H3 etc patterns (lower priority)
        return 'IOLink'
    elif re.search(r'BCN', desca_str) and 'FIO' in tagname_str:
        return 'IOLink'
    elif re.search(r'DISC', desca_str):
        return 'I'
    elif re.search(r'ESTOP', desca_str):
        return 'I'
    elif 'IB16' in tagname_str or 'IB16S' in tagname_str:
        return 'I'
    elif 'OB16E' in tagname_str:
        return 'O'
    
    return 'UNKNOWN'

def get_device_type(tagname):
    """Determine device type from TAGNAME"""
    tagname_str = str(tagname).upper()
    
    if 'VFD' in tagname_str:
        return 'APF'
    elif 'FIOH' in tagname_str:
        return 'Hub'
    elif _is_hub_tagname_pattern(tagname_str):  # Check for FIO1H, FIO2H3 etc patterns
        return 'Hub'
    elif _is_master_tagname_pattern(tagname_str):  # Check for FIO2M1, FIO3M etc patterns
        return 'M12DR'
    elif 'FIO' in tagname_str:
        return 'M12DR'
    elif 'SIO' in tagname_str:
        return 'SIO'
    elif 'OB16E' in tagname_str:
        return 'OB16E'
    elif 'IB16S' in tagname_str:
        return 'IB16S'
    elif 'IB16' in tagname_str:
        return 'IB16'
    return 'UNKNOWN'