Add missing exa-prefixed units, from_string(), to_string(), update example

README.md

@@ -1,4 +1,4 @@
-## Data units converter
+## The data size units converter
 
 Example:
 

cmd/dataunit.v

@@ -33,48 +33,6 @@ import os
 import flag
 import dataunits
 
-const units = {
-	'bit':    dataunits.bit
-	'nibble': dataunits.nibble
-	'bytes':  dataunits.bytes
-	'kb':     dataunits.kb
-	'mb':     dataunits.mb
-	'gb':     dataunits.gb
-	'tb':     dataunits.tb
-	'pb':     dataunits.pb
-	'zb':     dataunits.zb
-	'yb':     dataunits.yb
-	'kib':    dataunits.kib
-	'mib':    dataunits.mib
-	'gib':    dataunits.gib
-	'tib':    dataunits.tib
-	'pib':    dataunits.pib
-	'zib':    dataunits.zib
-	'yib':    dataunits.yib
-	'kbit':   dataunits.kbit
-	'mbit':   dataunits.mbit
-	'gbit':   dataunits.gbit
-	'tbit':   dataunits.tbit
-	'pbit':   dataunits.pbit
-	'zbit':   dataunits.zbit
-	'ybit':   dataunits.ybit
-	'kibit':  dataunits.kibit
-	'mibit':  dataunits.mibit
-	'gibit':  dataunits.gibit
-	'tibit':  dataunits.tibit
-	'pibit':  dataunits.pibit
-	'zibit':  dataunits.zibit
-	'yibit':  dataunits.yibit
-}
-
-fn units_str() string {
-	mut str := []string{}
-	for key, _ in units {
-		str << key
-	}
-	return str.join(', ')
-}
-
 @[name: 'dataunit']
 struct FlagConfig {
 	help bool
@@ -96,7 +54,7 @@ fn main() {
 		flags.value = no_matches[0].f64()
 	}
 	if flags.help {
-		println('convert the value between *from* and *to* units.')
+		println('convert the value between data size units.')
 		println('usage: dataunit -f <unit> -t <unit> <value>')
 		println('options:')
 		println('  -help      print this help message and exit')
@@ -112,15 +70,15 @@ fn main() {
 		eprintln('no value passed, see -help for info')
 		exit(2)
 	}
-	unit_from := units[flags.from.to_lower()] or {
+	src := dataunits.from_string(flags.from) or {
-		eprintln('invalid unit ${flags.from}, valid ones: ${units_str()}')
+		eprintln('invalid source unit: ${err}')
 		exit(1)
 	}
-	unit_to := units[flags.to.to_lower()] or {
+	dst := dataunits.from_string(flags.to) or {
-		eprintln('invalid unit ${flags.to}, valid ones: ${units_str()}')
+		eprintln('invalid destination unit: ${err}')
 		exit(1)
 	}
-	result := '${dataunits.convert(flags.value, unit_from, unit_to):.20f}'
+	result := '${dataunits.convert(flags.value, src, dst):.20f}'
 	splitted := result.split('.')
 	if splitted[1].contains_only('0') {
 		println(splitted[0])

src/dataunits.v

@@ -25,9 +25,11 @@
 
 module dataunits
 
+import maps
+
 pub type DataSize = f64
 
-pub fn (d DataSize) bits() f64 {
+pub fn (d DataSize) bit() f64 {
 	return f64(d)
 }
 
@@ -59,6 +61,10 @@ pub fn (d DataSize) pb() f64 {
 	return f64(d / pb)
 }
 
+pub fn (d DataSize) eb() f64 {
+	return f64(d / eb)
+}
+
 pub fn (d DataSize) zb() f64 {
 	return f64(d / zb)
 }
@@ -87,6 +93,10 @@ pub fn (d DataSize) pib() f64 {
 	return f64(d / pib)
 }
 
+pub fn (d DataSize) eib() f64 {
+	return f64(d / eib)
+}
+
 pub fn (d DataSize) zib() f64 {
 	return f64(d / zib)
 }
@@ -115,6 +125,10 @@ pub fn (d DataSize) pbit() f64 {
 	return f64(d / pbit)
 }
 
+pub fn (d DataSize) ebit() f64 {
+	return f64(d / ebit)
+}
+
 pub fn (d DataSize) zbit() f64 {
 	return f64(d / zbit)
 }
@@ -143,6 +157,10 @@ pub fn (d DataSize) pibit() f64 {
 	return f64(d / pibit)
 }
 
+pub fn (d DataSize) eibit() f64 {
+	return f64(d / eibit)
+}
+
 pub fn (d DataSize) zibit() f64 {
 	return f64(d / zibit)
 }
@@ -160,7 +178,8 @@ pub const mb = kb * 1000
 pub const gb = mb * 1000
 pub const tb = gb * 1000
 pub const pb = tb * 1000
-pub const zb = tb * 1000
+pub const eb = pb * 1000
+pub const zb = eb * 1000
 pub const yb = zb * 1000
 
 pub const kib = bytes * 1024
@@ -168,7 +187,8 @@ pub const mib = kib * 1024
 pub const gib = mib * 1024
 pub const tib = gib * 1024
 pub const pib = tib * 1024
-pub const zib = tib * 1024
+pub const eib = pib * 1024
+pub const zib = eib * 1024
 pub const yib = zib * 1024
 
 pub const kbit = bit * 1000
@@ -176,7 +196,8 @@ pub const mbit = kbit * 1000
 pub const gbit = mbit * 1000
 pub const tbit = gbit * 1000
 pub const pbit = tbit * 1000
-pub const zbit = pbit * 1000
+pub const ebit = pbit * 1000
+pub const zbit = ebit * 1000
 pub const ybit = zbit * 1000
 
 pub const kibit = bit * 1024
@@ -184,9 +205,69 @@ pub const mibit = kibit * 1024
 pub const gibit = mibit * 1024
 pub const tibit = gibit * 1024
 pub const pibit = tibit * 1024
-pub const zibit = tibit * 1024
+pub const eibit = pibit * 1024
+pub const zibit = eibit * 1024
 pub const yibit = zibit * 1024
 
+const units = {
+	'bit':    bit
+	'nibble': nibble
+	'bytes':  bytes
+	'kB':     kb
+	'MB':     mb
+	'GB':     gb
+	'TB':     tb
+	'PB':     pb
+	'EB':     eb
+	'ZB':     zb
+	'YB':     yb
+	'KiB':    kib
+	'MiB':    mib
+	'GiB':    gib
+	'TiB':    tib
+	'PiB':    pib
+	'EiB':    eib
+	'ZiB':    zib
+	'YiB':    yib
+	'kbit':   kbit
+	'Mbit':   mbit
+	'Gbit':   gbit
+	'Tbit':   tbit
+	'Pbit':   pbit
+	'Ebit':   ebit
+	'Zbit':   zbit
+	'Ybit':   ybit
+	'Kibit':  kibit
+	'Mibit':  mibit
+	'Gibit':  gibit
+	'Tibit':  tibit
+	'Pibit':  pibit
+	'Eibit':  eibit
+	'Zibit':  zibit
+	'Yibit':  yibit
+}
+
+pub const prefixes = {
+	// Metric size (1000^N)
+	'kilo':  'k'
+	'mega':  'M'
+	'giga':  'G'
+	'tera':  'T'
+	'peta':  'P'
+	'exa':   'E'
+	'zetta': 'Z'
+	'yotta': 'Y'
+	// Binary size (1024^N)
+	'kibi':  'Ki'
+	'mebi':  'Mi'
+	'gibi':  'Gi'
+	'tebi':  'Ti'
+	'pebi':  'Pi'
+	'exbi':  'Ei'
+	'zebi':  'Zi'
+	'yobi':  'Yi'
+}
+
 // convert returns the value converted between the *from* and *to* units.
 // Example:
 // ```
@@ -198,3 +279,94 @@ pub fn convert(value f64, from DataSize, to DataSize) f64 {
 	}
 	return f64(value * from / to)
 }
+
+// from_string parses input and returns the actual DataSize.
+// Note: Case insensitivity makes unit abbreviations such as `Mb` (megabit) and `MB` (megabyte)
+// ambiguous. Use `bit` suffix for bit units. The `b` suffix will be accepted as byte unit.
+// Example:
+// ```
+// assert dataunits.from_string('GiB')! == dataunits.gib
+// assert dataunits.from_string('M')! == dataunits.mib
+// assert dataunits.from_string('M', bits: true, metric: true)! == dataunits.mbit
+// assert dataunits.from_string('ZeTtAbYtEs', ci: true)! == dataunits.zb
+// ```
+pub fn from_string(input string, params ParseParams) !DataSize {
+	if !input.is_pure_ascii() {
+		return error('${input} non-ASCII characters is not allowed in data size unit')
+	}
+	unit := parse_unit_str(input, params)
+	if params.ci {
+		for key, value in units {
+			if key.to_lower_ascii() == unit {
+				return value
+			}
+		}
+	}
+	return units[unit] or { error('${input} is not a valid data size unit') }
+}
+
+fn parse_unit_str(input string, params ParseParams) string {
+	mut unit := ''
+	match true {
+		input.to_lower_ascii() in ['byte', 'bytes'] {
+			return 'bytes'
+		}
+		input.to_lower_ascii() in ['bit', 'bits'] {
+			return 'bit'
+		}
+		input.len == 1 {
+			if params.metric {
+				unit = input
+			} else {
+				unit = input.to_upper_ascii() + 'i'
+			}
+			if params.bits {
+				unit += 'bit'
+			} else {
+				unit += 'B'
+			}
+			return unit
+		}
+		input.len == 2 && input[1] == u8(`b`) && params.ci == false {
+			if input[0] != u8(`k`) {
+				return input[..1] + 'bit'
+			}
+			return input[..1].to_upper_ascii() + 'bit'
+		}
+		else {
+			unit = input
+		}
+	}
+	if params.ci {
+		unit = unit.to_lower_ascii()
+	}
+	if unit.len == 5 && unit.ends_with('ibit') {
+		// prevent Gibit --> Git transform
+		return unit
+	}
+	unit = unit.replace_each(maps.flat_map[string, string, string](prefixes, |k, v| [
+		k,
+		v,
+	]))
+	unit = unit.replace_each(['bytes', 'B', 'byte', 'B']).replace_once('bits', 'bit')
+	return unit
+}
+
+@[params]
+pub struct ParseParams {
+pub:
+	ci     bool // if true parse string in case insensitive mode
+	bits   bool // if true interpret single letter abbreviations as bit, otherwise as byte
+	metric bool // if ture apply single letter as metric prefix (power of ten), otherwise as binary
+}
+
+// to_string returns a string representation of data size unit in short form
+// e.g. kB, Mbit, GiB, etc.
+pub fn to_string(input DataSize) !string {
+	for key, value in units {
+		if value == input {
+			return key
+		}
+	}
+	return error('invalid input data size unit')
+}

tests/convert_test.v

@@ -1,12 +0,0 @@
-import dataunits
-
-fn test_convert() {
-	assert (dataunits.nibble * 4).bytes() == 2
-	assert (dataunits.bit * 8).bits() == 8
-	assert (dataunits.bit * 8).bytes() == 1
-	assert (dataunits.gib * 10).mib() == 10240
-	assert (dataunits.gib * 5000).bytes() == i64(5368709120000)
-	assert (dataunits.mbit * 500).kb() == 62500
-	assert dataunits.convert(500, dataunits.mbit, dataunits.kb) == 62500
-	assert dataunits.DataSize(4000 * dataunits.gib).bytes() == f64(4294967296000)
-}

tests/dataunits_test.v

@@ -0,0 +1,31 @@
+import dataunits
+
+fn test_convert() {
+	assert (dataunits.nibble * 4).bytes() == 2
+	assert (dataunits.bit * 8).bit() == 8
+	assert (dataunits.bit * 8).bytes() == 1
+	assert (dataunits.gib * 10).mib() == 10240
+	assert (dataunits.gib * 5000).bytes() == i64(5368709120000)
+	assert (dataunits.mbit * 500).kb() == 62500
+	assert dataunits.convert(500, dataunits.mbit, dataunits.kb) == 62500
+	assert dataunits.DataSize(4000 * dataunits.gib).bytes() == f64(4294967296000)
+}
+
+fn test_from_string() {
+	assert dataunits.from_string('GiB')! == dataunits.gib
+	assert dataunits.from_string('M')! == dataunits.mib
+	assert dataunits.from_string('m', ci: true)! == dataunits.mib
+	assert dataunits.from_string('M', bits: true, metric: true)! == dataunits.mbit
+	assert dataunits.from_string('ZeTtAbYtEs', ci: true)! == dataunits.zb
+	assert dataunits.from_string('bytes')! == dataunits.bytes
+	assert dataunits.from_string('byte')! == dataunits.bytes
+	assert dataunits.from_string('megabytes')! == dataunits.mb
+	assert dataunits.from_string('megabyte')! == dataunits.mb
+	assert dataunits.from_string('exbibit')! == dataunits.eibit
+	assert dataunits.from_string('Mb')! == dataunits.mbit
+	assert dataunits.from_string('MB')! == dataunits.mb
+	assert dataunits.from_string('Mb', ci: true)! == dataunits.mb
+	assert dataunits.from_string('Gibit')! == dataunits.gibit
+	assert dataunits.from_string('gibit', ci: true)! == dataunits.gibit
+	assert dataunits.from_string('Gib', ci: true)! == dataunits.gib
+}