Simple ECL formatter

ecl_formatter.py

#!/usr/bin/env python3
"""
ECL Formatter - Formats SNOMED CT Expression Constraint Language queries
with consistent indentation and spacing.
Usage: cat ecl.txt | ./ecl_formatter.py > formatted_ecl.txt
"""

import re
import sys


def format_ecl(ecl):
    """Format ECL with consistent indentation and spacing."""
    # Remove extra whitespace
    ecl = re.sub(r'\s+', ' ', ecl).strip()
    
    # Track nesting level
    level = 0
    result = []
    i = 0
    in_concept_reference = False
    pipe_count = 0
    
    while i < len(ecl):
        char = ecl[i]
        
        # Handle pipes in concept references
        if char == '|':
            pipe_count += 1
            in_concept_reference = pipe_count % 2 == 1
            result.append(char)
            i += 1
            continue
            
        # Skip processing special characters when inside concept reference
        if in_concept_reference:
            result.append(char)
            i += 1
            continue
            
        if char == '(':
            result.append(char)
            level += 1
            result.append('\n' + '  ' * level)
            i += 1
        elif char == ')':
            level -= 1
            result.append('\n' + '  ' * level)
            result.append(char)
            i += 1
        elif char == '{':
            result.append(char)
            level += 1
            result.append('\n' + '  ' * level)
            i += 1
        elif char == '}':
            level -= 1
            result.append('\n' + '  ' * level)
            result.append(char)
            i += 1
        elif i + 2 <= len(ecl) and ecl[i:i+2] == 'OR':
            result.append('\n' + '  ' * level + 'OR')
            i += 2
        elif i + 3 <= len(ecl) and ecl[i:i+3] == 'AND':
            result.append('\n' + '  ' * level + 'AND')
            i += 3
        elif i + 5 <= len(ecl) and ecl[i:i+5] == 'MINUS':
            # If level is 0, we want to add extra newlines for separation
            if level == 0:
                result.append('\n) MINUS (\n' + '  ' * (level+1))
                i += 5
            else:
                result.append('\n' + '  ' * level + 'MINUS')
                i += 5
        else:
            result.append(char)
            i += 1
    
    formatted_ecl = ''.join(result)
    
    # Fix spacing around operators while preserving << and >> operators
    formatted_ecl = re.sub(r'(?<!<)\s*([:<>=])\s*(?!>)', r' \1 ', formatted_ecl)
    
    # Make sure << and >> operators are preserved (not split into < < or > >)
    formatted_ecl = re.sub(r'< <', r'<<', formatted_ecl)
    formatted_ecl = re.sub(r'> >', r'>>', formatted_ecl)
    
    # Special handling for top-level MINUS operation
    formatted_ecl = re.sub(r'\(\n\s+\)\s+MINUS\s+\(\n\s+', r'(\n  ', formatted_ecl)
    
    # Make sure the top level has proper wrapping
    if not formatted_ecl.startswith('('):
        formatted_ecl = '(\n  ' + formatted_ecl
    if not formatted_ecl.endswith(')'):
        formatted_ecl = formatted_ecl + '\n)'
    
    return formatted_ecl


if __name__ == "__main__":
    # Read from stdin
    ecl = sys.stdin.read()
    
    # Format the ECL
    formatted_ecl = format_ecl(ecl)
    
    # Write to stdout
    sys.stdout.write(formatted_ecl)