1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-14 08:54:58 +00:00

cut: Add -c,--characters option

This commit is contained in:
Kyle Lanmon 2024-01-26 13:29:22 -06:00 committed by Andrew Kaster
parent f926d96c74
commit e09245badd

View file

@ -7,6 +7,7 @@
#include <AK/ByteString.h> #include <AK/ByteString.h>
#include <AK/QuickSort.h> #include <AK/QuickSort.h>
#include <AK/StdLibExtras.h> #include <AK/StdLibExtras.h>
#include <AK/String.h>
#include <AK/Vector.h> #include <AK/Vector.h>
#include <LibCore/ArgsParser.h> #include <LibCore/ArgsParser.h>
#include <LibCore/File.h> #include <LibCore/File.h>
@ -31,6 +32,11 @@ struct Range {
m_from = min(m_from, other.m_from); m_from = min(m_from, other.m_from);
m_to = max(m_to, other.m_to); m_to = max(m_to, other.m_to);
} }
bool contains(size_t x) const
{
return m_from <= x && m_to >= x;
}
}; };
static bool expand_list(ByteString& list, Vector<Range>& ranges) static bool expand_list(ByteString& list, Vector<Range>& ranges)
@ -134,6 +140,22 @@ static void process_line_bytes(StringView line, Vector<Range> const& ranges)
outln(); outln();
} }
static void process_line_characters(StringView line, Vector<Range> const& ranges)
{
for (auto const& range : ranges) {
if (range.m_from >= line.length())
continue;
auto s = String::from_utf8(line).release_value_but_fixme_should_propagate_errors();
size_t i = 1;
for (auto c : s.code_points()) {
if (range.contains(i++))
out("{}", String::from_code_point(c));
}
}
outln();
}
static void process_line_fields(StringView line, Vector<Range> const& ranges, char delimiter, bool only_print_delimited_lines) static void process_line_fields(StringView line, Vector<Range> const& ranges, char delimiter, bool only_print_delimited_lines)
{ {
auto string_split = ByteString(line).split(delimiter, SplitBehavior::KeepEmpty); auto string_split = ByteString(line).split(delimiter, SplitBehavior::KeepEmpty);
@ -157,6 +179,7 @@ static void process_line_fields(StringView line, Vector<Range> const& ranges, ch
ErrorOr<int> serenity_main(Main::Arguments arguments) ErrorOr<int> serenity_main(Main::Arguments arguments)
{ {
ByteString byte_list = ""; ByteString byte_list = "";
ByteString character_list = "";
ByteString fields_list = ""; ByteString fields_list = "";
ByteString delimiter = "\t"; ByteString delimiter = "\t";
bool only_print_delimited_lines = false; bool only_print_delimited_lines = false;
@ -166,24 +189,26 @@ ErrorOr<int> serenity_main(Main::Arguments arguments)
Core::ArgsParser args_parser; Core::ArgsParser args_parser;
args_parser.add_positional_argument(files, "file(s) to cut", "file", Core::ArgsParser::Required::No); args_parser.add_positional_argument(files, "file(s) to cut", "file", Core::ArgsParser::Required::No);
args_parser.add_option(byte_list, "select only these bytes", "bytes", 'b', "list"); args_parser.add_option(byte_list, "select only these bytes", "bytes", 'b', "list");
args_parser.add_option(character_list, "select only these characters", "characters", 'c', "list");
args_parser.add_option(fields_list, "select only these fields", "fields", 'f', "list"); args_parser.add_option(fields_list, "select only these fields", "fields", 'f', "list");
args_parser.add_option(delimiter, "set a custom delimiter", "delimiter", 'd', "delimiter"); args_parser.add_option(delimiter, "set a custom delimiter", "delimiter", 'd', "delimiter");
args_parser.add_option(only_print_delimited_lines, "suppress lines which don't contain any field delimiter characters", "only-delimited", 's'); args_parser.add_option(only_print_delimited_lines, "suppress lines which don't contain any field delimiter characters", "only-delimited", 's');
args_parser.parse(arguments); args_parser.parse(arguments);
bool selected_bytes = (byte_list != ""); bool const selected_bytes = (byte_list != "");
bool selected_fields = (fields_list != ""); bool const selected_characters = (character_list != "");
bool const selected_fields = (fields_list != "");
int selected_options_count = (selected_bytes ? 1 : 0) + (selected_fields ? 1 : 0); int const selected_options_count = (selected_bytes ? 1 : 0) + (selected_characters ? 1 : 0) + (selected_fields ? 1 : 0);
if (selected_options_count == 0) { if (selected_options_count == 0) {
warnln("cut: you must specify a list of bytes, or fields"); warnln("cut: you must specify a list of bytes, characters, or fields");
args_parser.print_usage(stderr, arguments.strings[0]); args_parser.print_usage(stderr, arguments.strings[0]);
return 1; return 1;
} }
if (selected_options_count > 1) { if (selected_options_count > 1) {
warnln("cut: you must specify only one of bytes, or fields"); warnln("cut: you must specify only one of bytes, characters, or fields");
args_parser.print_usage(stderr, arguments.strings[0]); args_parser.print_usage(stderr, arguments.strings[0]);
return 1; return 1;
} }
@ -199,6 +224,8 @@ ErrorOr<int> serenity_main(Main::Arguments arguments)
if (selected_bytes) { if (selected_bytes) {
ranges_list = byte_list; ranges_list = byte_list;
} else if (selected_characters) {
ranges_list = character_list;
} else if (selected_fields) { } else if (selected_fields) {
ranges_list = fields_list; ranges_list = fields_list;
} else { } else {
@ -252,6 +279,8 @@ ErrorOr<int> serenity_main(Main::Arguments arguments)
if (selected_bytes) { if (selected_bytes) {
process_line_bytes(line, disjoint_ranges); process_line_bytes(line, disjoint_ranges);
} else if (selected_characters) {
process_line_characters(line, disjoint_ranges);
} else if (selected_fields) { } else if (selected_fields) {
process_line_fields(line, disjoint_ranges, delimiter[0], only_print_delimited_lines); process_line_fields(line, disjoint_ranges, delimiter[0], only_print_delimited_lines);
} else { } else {