"""Search/transform builtins: grep, sed. Pure builtins per windows-shell-workpackages-2026-07-19.md ยง2.2. Regex dialect is Python `re` (documented divergence from POSIX BRE/ERE). """ from __future__ import annotations import re from context import BuiltinContext from errors import UnsupportedConstruct from paths import resolve_request_path from shell_args import split_leading_short_options _GREP_FLAGS = set("ivnclwFE") def cmd_grep(ctx: BuiltinContext) -> int: args = ctx.argv[1:] flags: set[str] = set() option_args, remaining = split_leading_short_options(args) for option in option_args: for char in option[1:]: if char not in _GREP_FLAGS: raise UnsupportedConstruct( "unsupported-flag", f"grep: unsupported flag '-{char}'" ) flags.add(char) if not remaining: raise UnsupportedConstruct("unsupported-flag", "grep: PATTERN operand required") pattern = remaining[0] files = remaining[1:] ignore_case = "i" in flags invert = "v" in flags show_lineno = "n" in flags count_only = "c" in flags files_only = "l" in flags whole_word = "w" in flags fixed = "F" in flags if fixed: needle = pattern else: pat = pattern if whole_word: pat = r"\b" + pat + r"\b" re_flags = re.IGNORECASE if ignore_case else 0 try: compiled = re.compile(pat, re_flags) except re.error as exc: raise UnsupportedConstruct("malformed-syntax", f"grep: invalid pattern: {exc}") from exc def matches(line: str) -> bool: if fixed: hay = line.lower() if ignore_case else line n = needle.lower() if ignore_case else needle if whole_word: return re.search(r"(? list[str]: text = data.decode("utf-8", errors="replace") if text == "": return [] lines = text.split("\n") if lines and lines[-1] == "": lines.pop() return lines multi_file = len(files) > 1 any_match = False any_error = False out_lines: list[str] = [] def process(name: str, lines: list[str]) -> None: nonlocal any_match matched = [ln for ln in lines if matches(ln) != invert] if matched: any_match = True if count_only: prefix = f"{name}:" if multi_file else "" out_lines.append(f"{prefix}{len(matched)}") return if files_only: if matched: out_lines.append(name) return for i, ln in enumerate(lines): if matches(ln) == invert: continue prefix = "" if multi_file: prefix += f"{name}:" if show_lineno: prefix += f"{i + 1}:" out_lines.append(f"{prefix}{ln}") if not files: data = ctx.stdin.read() process("(standard input)", read_lines(data)) else: for name in files: try: with open(resolve_request_path(ctx.cwd, name), "rb") as fh: data = fh.read() except OSError as exc: out_lines_msg = f"grep: {name}: {exc.strerror or exc}" ctx.stdout.write((out_lines_msg + "\n").encode("utf-8")) any_error = True continue process(name, read_lines(data)) if out_lines: ctx.stdout.write(("\n".join(out_lines) + "\n").encode("utf-8")) if any_error: return 2 return 0 if any_match else 1 def _parse_sed_replacement(repl: str) -> str: """Translate sed REPL (\\1..\\9, &, \\&) into Python re replacement syntax.""" out: list[str] = [] i = 0 n = len(repl) while i < n: ch = repl[i] if ch == "\\" and i + 1 < n: nxt = repl[i + 1] if nxt.isdigit(): out.append("\\g<" + nxt + ">") i += 2 continue if nxt == "&": out.append("&") i += 2 continue if nxt == "\\": out.append("\\\\") i += 2 continue out.append(nxt) i += 2 continue if ch == "&": out.append("\\g<0>") i += 1 continue if ch == "\\": out.append("\\\\") i += 1 continue out.append(ch) i += 1 return "".join(out) def _parse_sed_script(script: str) -> tuple[str, str, bool, bool]: if len(script) < 2 or script[0] != "s": raise UnsupportedConstruct("unsupported-flag", "sed: only s/// scripts are supported") delim = script[1] if not delim or delim.isalnum() or delim == "\\": raise UnsupportedConstruct("malformed-syntax", "sed: invalid delimiter") parts: list[str] = [] current: list[str] = [] i = 2 n = len(script) while i < n: ch = script[i] if ch == "\\" and i + 1 < n and script[i + 1] == delim: current.append(delim) i += 2 continue if ch == delim: parts.append("".join(current)) current = [] i += 1 if len(parts) == 2: break continue current.append(ch) i += 1 if len(parts) != 2: raise UnsupportedConstruct("malformed-syntax", "sed: unterminated s/// script") remainder = script[i:] global_flag = "g" in remainder icase_flag = "i" in remainder for c in remainder: if c not in "gi": raise UnsupportedConstruct("unsupported-flag", f"sed: unsupported flag '{c}'") pattern, repl = parts return pattern, repl, global_flag, icase_flag def cmd_sed(ctx: BuiltinContext) -> int: args = ctx.argv[1:] if not args: raise UnsupportedConstruct("unsupported-flag", "sed: SCRIPT operand required") for a in args: if a.startswith("-") and a != "-": raise UnsupportedConstruct("unsupported-flag", f"sed: unsupported flag {a!r}") script = args[0] files = args[1:] pattern, repl, global_flag, icase_flag = _parse_sed_script(script) re_flags = re.IGNORECASE if icase_flag else 0 try: compiled = re.compile(pattern, re_flags) except re.error as exc: raise UnsupportedConstruct("malformed-syntax", f"sed: invalid pattern: {exc}") from exc py_repl = _parse_sed_replacement(repl) count = 0 if global_flag else 1 def transform(data: bytes) -> bytes: text = data.decode("utf-8", errors="replace") if text == "": return b"" had_trailing_newline = text.endswith("\n") body = text[:-1] if had_trailing_newline else text lines = body.split("\n") if body or had_trailing_newline else [""] if body == "" and not had_trailing_newline: lines = [""] result_lines = [compiled.sub(py_repl, ln, count=count) for ln in lines] out = "\n".join(result_lines) if had_trailing_newline: out += "\n" return out.encode("utf-8") if not files: ctx.stdout.write(transform(ctx.stdin.read())) return 0 for name in files: with open(resolve_request_path(ctx.cwd, name), "rb") as fh: data = fh.read() ctx.stdout.write(transform(data)) return 0