environment/parser.py

"""
Free-form command parser for the CI/CD Doctor environment.
Converts raw command strings into structured ParsedCommand objects.
"""

import re
from dataclasses import dataclass
from typing import Optional


@dataclass
class ParsedCommand:
    type: str  # "cat" | "echo_append" | "sed" | "pipeline_run" | "pipeline_logs" | "pipeline_status" | "unknown"
    filename: Optional[str] = None
    content: Optional[str] = None    # for echo >>
    pattern: Optional[str] = None    # for sed: old value
    replacement: Optional[str] = None  # for sed: new value
    stage: Optional[str] = None      # for pipeline logs: which stage


def parse_command(command: str) -> ParsedCommand:
    command = command.strip()

    # cat <filename>
    if command.startswith("cat "):
        return ParsedCommand(type="cat", filename=command[4:].strip())

    # echo "text" >> filename
    m = re.match(r'echo\s+"([^"]+)"\s*>>\s*(\S+)', command)
    if m:
        return ParsedCommand(type="echo_append", content=m.group(1), filename=m.group(2))

    # sed -i 's/old/new/' filename
    m = re.match(r"sed\s+-i\s+'s/([^/]+)/([^/]*)/'[\s]+(\S+)", command)
    if m:
        return ParsedCommand(type="sed", pattern=m.group(1), replacement=m.group(2), filename=m.group(3))

    # pipeline run
    if command == "pipeline run":
        return ParsedCommand(type="pipeline_run")

    # pipeline logs [stage]
    m = re.match(r"pipeline\s+logs(?:\s+(\S+))?\s*$", command)
    if m:
        return ParsedCommand(type="pipeline_logs", stage=m.group(1))

    # pipeline status
    if command == "pipeline status":
        return ParsedCommand(type="pipeline_status")

    return ParsedCommand(type="unknown")