Multi-line command argument parsing of MeTTaClaw

gistfile1.txt

import json

LLM_COMMANDS = {
    "pin",
    "remember",
    "query",
    "episodes",
    "search",
    "send",
    "promote",
    "demote",
    "metta",
    "shell",
    "read-file",
    "write-file",
    "append-file",
}

def quote_arg(x):
    return json.dumps(x, ensure_ascii=False)

def starts_command_line(line):
    s = line.lstrip()
    if not s:
        return False
    # allow "(send ...)" as command start too
    if s.startswith("("):
        s = s[1:].lstrip()
    if not s:
        return False
    first = s.split(maxsplit=1)[0].rstrip(")")
    return first in LLM_COMMANDS


def split_command_blocks(s):
    blocks = []
    cur = []
    for raw in s.splitlines():
        if not raw.strip():
            if cur:
                cur.append(raw)
            continue
        if starts_command_line(raw) and cur:
            blocks.append("\n".join(cur).strip())
            cur = [raw]
        else:
            cur.append(raw)
    if cur:
        blocks.append("\n".join(cur).strip())
    return blocks

def balance_parentheses(s):
    s = s.replace("_quote_", '"').replace("_newline_", "\n")
    sexprs = []
    special_two_arg_cmds = {"write-file", "append-file"}
    for line in split_command_blocks(s):
        line = line.strip()
        if not line:
            continue
        if line.startswith("(-"):
            line = "(pin -" + line[2:]
        elif line.startswith("-"):
            line = "pin " + line
        # remove one outer (...) if present
        if line.startswith("(") and line.endswith(")"):
            line = line[1:-1].strip()
        elif line.startswith("("):
            line = line[1:].strip()
        parts = line.split(maxsplit=1)
        if not parts:
            continue
        cmd = parts[0]
        rest = parts[1].strip() if len(parts) > 1 else ""
        if cmd in special_two_arg_cmds:
            if not rest:
                sexprs.append(f"({cmd})")
                continue
            # filename is first token unless already quoted
            if rest.startswith('"'):
                end = 1
                escaped = False
                while end < len(rest):
                    ch = rest[end]
                    if ch == '"' and not escaped:
                        break
                    escaped = (ch == '\\' and not escaped)
                    if ch != '\\':
                        escaped = False
                    end += 1
                if end < len(rest) and rest[end] == '"':
                    filename = rest[:end+1]
                    content = rest[end+1:].strip()
                else:
                    filename = quote_arg(rest[1:])
                    content = ""
            else:
                split_rest = rest.split(maxsplit=1)
                filename = quote_arg(split_rest[0])
                content = split_rest[1].strip() if len(split_rest) > 1 else ""
            if content:
                if content.startswith('"') and content.endswith('"') and "\n" not in content:
                    sexprs.append(f"({cmd} {filename} {content})")
                else:
                    sexprs.append(f"({cmd} {filename} {quote_arg(content)})")
            else:
                sexprs.append(f"({cmd} {filename})")
            continue
        if rest:
            if rest.startswith('"') and rest.endswith('"') and "\n" not in rest:
                sexprs.append(f"({cmd} {rest})")
            else:
                sexprs.append(f"({cmd} {quote_arg(rest)})")
        else:
            sexprs.append(f"({cmd})")
    ret = " ".join(sexprs)
    return "(" + ret + ")"