From eca74088a0185aac7b7ca56b1b14128daacc6fe0 Mon Sep 17 00:00:00 2001 From: Ali Mohammad Pur Date: Sat, 10 Jul 2021 13:20:44 +0430 Subject: [PATCH] grep: Use Basic POSIX regexps by default and make -E not the default --- Userland/Utilities/grep.cpp | 195 ++++++++++++++++++------------------ 1 file changed, 99 insertions(+), 96 deletions(-) diff --git a/Userland/Utilities/grep.cpp b/Userland/Utilities/grep.cpp index 3210412f61..ceb15fe4d6 100644 --- a/Userland/Utilities/grep.cpp +++ b/Userland/Utilities/grep.cpp @@ -42,7 +42,7 @@ int main(int argc, char** argv) Vector files; bool recursive { false }; - bool use_ere { true }; + bool use_ere { false }; const char* pattern = nullptr; BinaryFileMode binary_mode { BinaryFileMode::Binary }; bool case_insensitive = false; @@ -50,7 +50,7 @@ int main(int argc, char** argv) Core::ArgsParser args_parser; args_parser.add_option(recursive, "Recursively scan files starting in working directory", "recursive", 'r'); - args_parser.add_option(use_ere, "Extended regular expressions (default)", "extended-regexp", 'E'); + args_parser.add_option(use_ere, "Extended regular expressions", "extended-regexp", 'E'); args_parser.add_option(pattern, "Pattern", "regexp", 'e', "Pattern"); args_parser.add_option(case_insensitive, "Make matches case-insensitive", nullptr, 'i'); args_parser.add_option(invert_match, "Select non-matching lines", "invert-match", 'v'); @@ -93,9 +93,6 @@ int main(int argc, char** argv) args_parser.add_positional_argument(files, "File(s) to process", "file", Core::ArgsParser::Required::No); args_parser.parse(argc, argv); - if (!use_ere) - return 0; - // mock grep behaviour: if -e is omitted, use first positional argument as pattern if (pattern == nullptr && files.size()) pattern = files.take_first(); @@ -104,104 +101,110 @@ int main(int argc, char** argv) if (case_insensitive) options |= PosixFlags::Insensitive; - Regex re(pattern, options); - if (re.parser_result.error != Error::NoError) { - return 1; - } - - auto matches = [&](StringView str, StringView filename = "", bool print_filename = false, bool is_binary = false) { - size_t last_printed_char_pos { 0 }; - if (is_binary && binary_mode == BinaryFileMode::Skip) - return false; - - auto result = re.match(str, PosixFlags::Global); - if (result.success ^ invert_match) { - if (is_binary && binary_mode == BinaryFileMode::Binary) { - outln("binary file \x1B[34m{}\x1B[0m matches", filename); - } else { - if ((result.matches.size() || invert_match) && print_filename) { - out("\x1B[34m{}:\x1B[0m", filename); - } - - for (auto& match : result.matches) { - - out("{}\x1B[32m{}\x1B[0m", - StringView(&str[last_printed_char_pos], match.global_offset - last_printed_char_pos), - match.view.to_string()); - last_printed_char_pos = match.global_offset + match.view.length(); - } - outln("{}", StringView(&str[last_printed_char_pos], str.length() - last_printed_char_pos)); - } - - return true; + auto grep_logic = [&](auto&& re) { + if (re.parser_result.error != Error::NoError) { + return 1; } - return false; - }; - - auto handle_file = [&matches, binary_mode](StringView filename, bool print_filename) -> bool { - auto file = Core::File::construct(filename); - if (!file->open(Core::OpenMode::ReadOnly)) { - warnln("Failed to open {}: {}", filename, file->error_string()); - return false; - } - - while (file->can_read_line()) { - auto line = file->read_line(); - auto is_binary = memchr(line.characters(), 0, line.length()) != nullptr; - - if (matches(line, filename, print_filename, is_binary) && is_binary && binary_mode == BinaryFileMode::Binary) - return true; - } - return true; - }; - - auto add_directory = [&handle_file](String base, Optional recursive, auto handle_directory) -> void { - Core::DirIterator it(recursive.value_or(base), Core::DirIterator::Flags::SkipDots); - while (it.has_next()) { - auto path = it.next_full_path(); - if (!Core::File::is_directory(path)) { - auto key = path.substring_view(base.length() + 1, path.length() - base.length() - 1); - handle_file(key, true); - } else { - handle_directory(base, path, handle_directory); - } - } - }; - - bool did_match_something = false; - if (!files.size() && !recursive) { - char* line = nullptr; - size_t line_len = 0; - ssize_t nread = 0; - ScopeGuard free_line = [line] { free(line); }; - while ((nread = getline(&line, &line_len, stdin)) != -1) { - VERIFY(nread > 0); - if (line[nread - 1] == '\n') - --nread; - StringView line_view(line, nread); - bool is_binary = line_view.contains(0); - + auto matches = [&](StringView str, StringView filename = "", bool print_filename = false, bool is_binary = false) { + size_t last_printed_char_pos { 0 }; if (is_binary && binary_mode == BinaryFileMode::Skip) - return 1; + return false; - auto matched = matches(line_view, "stdin", false, is_binary); - did_match_something = did_match_something || matched; - if (matched && is_binary && binary_mode == BinaryFileMode::Binary) - return 0; - } - } else { - if (recursive) { - add_directory(".", {}, add_directory); + auto result = re.match(str, PosixFlags::Global); + if (result.success ^ invert_match) { + if (is_binary && binary_mode == BinaryFileMode::Binary) { + outln("binary file \x1B[34m{}\x1B[0m matches", filename); + } else { + if ((result.matches.size() || invert_match) && print_filename) { + out("\x1B[34m{}:\x1B[0m", filename); + } - } else { - bool print_filename { files.size() > 1 }; - for (auto& filename : files) { - if (!handle_file(filename, print_filename)) + for (auto& match : result.matches) { + + out("{}\x1B[32m{}\x1B[0m", + StringView(&str[last_printed_char_pos], match.global_offset - last_printed_char_pos), + match.view.to_string()); + last_printed_char_pos = match.global_offset + match.view.length(); + } + outln("{}", StringView(&str[last_printed_char_pos], str.length() - last_printed_char_pos)); + } + + return true; + } + + return false; + }; + + auto handle_file = [&matches, binary_mode](StringView filename, bool print_filename) -> bool { + auto file = Core::File::construct(filename); + if (!file->open(Core::OpenMode::ReadOnly)) { + warnln("Failed to open {}: {}", filename, file->error_string()); + return false; + } + + while (file->can_read_line()) { + auto line = file->read_line(); + auto is_binary = memchr(line.characters(), 0, line.length()) != nullptr; + + if (matches(line, filename, print_filename, is_binary) && is_binary && binary_mode == BinaryFileMode::Binary) + return true; + } + return true; + }; + + auto add_directory = [&handle_file](String base, Optional recursive, auto handle_directory) -> void { + Core::DirIterator it(recursive.value_or(base), Core::DirIterator::Flags::SkipDots); + while (it.has_next()) { + auto path = it.next_full_path(); + if (!Core::File::is_directory(path)) { + auto key = path.substring_view(base.length() + 1, path.length() - base.length() - 1); + handle_file(key, true); + } else { + handle_directory(base, path, handle_directory); + } + } + }; + + bool did_match_something = false; + if (!files.size() && !recursive) { + char* line = nullptr; + size_t line_len = 0; + ssize_t nread = 0; + ScopeGuard free_line = [line] { free(line); }; + while ((nread = getline(&line, &line_len, stdin)) != -1) { + VERIFY(nread > 0); + if (line[nread - 1] == '\n') + --nread; + StringView line_view(line, nread); + bool is_binary = line_view.contains(0); + + if (is_binary && binary_mode == BinaryFileMode::Skip) return 1; + + auto matched = matches(line_view, "stdin", false, is_binary); + did_match_something = did_match_something || matched; + if (matched && is_binary && binary_mode == BinaryFileMode::Binary) + return 0; + } + } else { + if (recursive) { + add_directory(".", {}, add_directory); + + } else { + bool print_filename { files.size() > 1 }; + for (auto& filename : files) { + if (!handle_file(filename, print_filename)) + return 1; + } } } - } - return did_match_something ? 0 : 1; + return did_match_something ? 0 : 1; + }; + + if (use_ere) + return grep_logic(Regex(pattern, options)); + + return grep_logic(Regex(pattern, options)); }