///|
priv struct Counts {
mut lines : Int64
mut words : Int64
mut chars : Int64
mut bytes : Int64
}
///|
fn Counts::new() -> Counts {
{ lines: 0, words: 0, chars: 0, bytes: 0 }
}
///|
fn Counts::add(self : Counts, other : Counts) -> Unit {
self.lines += other.lines
self.words += other.words
self.chars += other.chars
self.bytes += other.bytes
}
///|
fn is_space_byte(b : Byte) -> Bool {
b is (b' ' | b'\t' | b'\n' | b'\r' | 0x0B | 0x0C)
}
///|
fn is_continuation(b : Byte) -> Bool {
(b.to_int() & 0xC0) == 0x80
}
///|
/// Count valid UTF-8 encoded characters. Bytes that are not part of a valid
/// sequence are skipped without being counted, like GNU wc -m.
fn count_utf8_chars(data : Bytes) -> Int64 {
let mut count : Int64 = 0
let mut i = 0
while i < data.length() {
let b0 = data[i].to_int()
if b0 < 0x80 {
count += 1
i += 1
continue
}
// (sequence length, valid range for the second byte)
let (len, lo1, hi1) = if b0 >= 0xC2 && b0 <= 0xDF {
(2, 0x80, 0xBF)
} else if b0 == 0xE0 {
(3, 0xA0, 0xBF)
} else if (b0 >= 0xE1 && b0 <= 0xEC) || b0 == 0xEE || b0 == 0xEF {
(3, 0x80, 0xBF)
} else if b0 == 0xED {
(3, 0x80, 0x9F)
} else if b0 == 0xF0 {
(4, 0x90, 0xBF)
} else if b0 >= 0xF1 && b0 <= 0xF3 {
(4, 0x80, 0xBF)
} else if b0 == 0xF4 {
(4, 0x80, 0x8F)
} else {
(0, 0, 0)
}
if len == 0 || i + len > data.length() {
i += 1
continue
}
let second = data[i + 1].to_int()
if second < lo1 || second > hi1 {
i += 1
continue
}
let mut ok = true
for k in 2.. Counts {
let counts = Counts::new()
counts.bytes = data.length().to_int64()
counts.chars = count_utf8_chars(data)
let mut in_word = false
for b in data {
if b is b'\n' {
counts.lines += 1
}
if is_space_byte(b) {
in_word = false
} else if !in_word {
in_word = true
counts.words += 1
}
}
counts
}
///|
priv struct Selection {
lines : Bool
words : Bool
chars : Bool
bytes : Bool
}
///|
fn render_counts(
counts : Counts,
selection : Selection,
name : String?,
) -> String {
let parts : Array[String] = []
if selection.lines {
parts.push(counts.lines.to_string())
}
if selection.words {
parts.push(counts.words.to_string())
}
if selection.chars {
parts.push(counts.chars.to_string())
}
if selection.bytes {
parts.push(counts.bytes.to_string())
}
if name is Some(n) {
parts.push(n)
}
parts.join(" ")
}
///|
fn wc_command() -> @argparse.Command {
Command(
"wc",
about="Count lines, words, and bytes for each input.",
flags=[
FlagArg("lines", short='l', about="Print newline counts."),
FlagArg("words", short='w', about="Print word counts."),
FlagArg("bytes", short='c', about="Print byte counts."),
FlagArg("chars", short='m', about="Print character (UTF-8) counts."),
],
positionals=[
PositionArg(
"files",
about="[file...] ('-' or no file reads stdin)",
num_args=ValueRange(lower=0),
),
],
disable_help_subcommand=true,
)
}
///|
async fn read_source(path : String) -> Bytes {
if path == "-" {
@stdio.stdin.read_all().binary()
} else {
@fs.read_file_to_bytes(path)
}
}
///|
async fn main {
let args = @env.args()[1:]
let command = wc_command()
let matches = command.parse(argv=args, env=Map([])) catch {
err => {
@stdio.stderr.write("\{err}\n")
@sys.exit(2)
return
}
}
let lines = matches.flags.get_or_default("lines", false)
let words = matches.flags.get_or_default("words", false)
let chars = matches.flags.get_or_default("chars", false)
let bytes = matches.flags.get_or_default("bytes", false)
let selection : Selection = if lines || words || chars || bytes {
{ lines, words, chars, bytes }
} else {
{ lines: true, words: true, chars: false, bytes: true }
}
let files = matches.values.get("files").unwrap_or([])
let total = Counts::new()
let mut failed = false
if files.is_empty() {
let counts = count(@stdio.stdin.read_all().binary())
@stdio.stdout.write(render_counts(counts, selection, None) + "\n")
} else {
for path in files {
let data = read_source(path) catch {
err => {
@stdio.stderr.write("wc: \{err}\n")
failed = true
continue
}
}
let counts = count(data)
total.add(counts)
@stdio.stdout.write(render_counts(counts, selection, Some(path)) + "\n")
}
if files.length() > 1 {
@stdio.stdout.write(render_counts(total, selection, Some("total")) + "\n")
}
}
if failed {
@sys.exit(1)
}
}