diff options
Diffstat (limited to 'lib')
| -rw-r--r-- | lib/pure/parseutils.nim | 536 | ||||
| -rw-r--r-- | lib/system.nim | 14 | ||||
| -rw-r--r-- | lib/system/jssys.nim | 21 | ||||
| -rw-r--r-- | lib/system/strmantle.nim | 24 |
4 files changed, 481 insertions, 114 deletions
diff --git a/lib/pure/parseutils.nim b/lib/pure/parseutils.nim index 1e3c2f72c..2bb23c626 100644 --- a/lib/pure/parseutils.nim +++ b/lib/pure/parseutils.nim @@ -50,6 +50,8 @@ include "system/inclrtl" +template toOa(s: string): openArray[char] = openArray[char](s) + const Whitespace = {' ', '\t', '\v', '\r', '\l', '\f'} IdentChars = {'a'..'z', 'A'..'Z', '0'..'9', '_'} @@ -59,8 +61,7 @@ const proc toLower(c: char): char {.inline.} = result = if c in {'A'..'Z'}: chr(ord(c)-ord('A')+ord('a')) else: c -proc parseBin*[T: SomeInteger](s: string, number: var T, start = 0, - maxLen = 0): int {.noSideEffect.} = +proc parseBin*[T: SomeInteger](s: openArray[char], number: var T, maxLen = 0): int {.noSideEffect.} = ## Parses a binary number and stores its value in ``number``. ## ## Returns the number of the parsed characters or 0 in case of an error. @@ -89,7 +90,7 @@ proc parseBin*[T: SomeInteger](s: string, number: var T, start = 0, var num64: int64 doAssert parseBin("0100111001101001111011010100111001101001", num64) == 40 doAssert num64 == 336784608873 - var i = start + var i = 0 var output = T(0) var foundDigit = false let last = min(s.len, if maxLen == 0: s.len else: i + maxLen) @@ -104,10 +105,9 @@ proc parseBin*[T: SomeInteger](s: string, number: var T, start = 0, inc(i) if foundDigit: number = output - result = i - start + result = i -proc parseOct*[T: SomeInteger](s: string, number: var T, start = 0, - maxLen = 0): int {.noSideEffect.} = +proc parseOct*[T: SomeInteger](s: openArray[char], number: var T, maxLen = 0): int {.noSideEffect.} = ## Parses an octal number and stores its value in ``number``. ## ## Returns the number of the parsed characters or 0 in case of an error. @@ -136,7 +136,7 @@ proc parseOct*[T: SomeInteger](s: string, number: var T, start = 0, var num64: int64 doAssert parseOct("2346475523464755", num64) == 16 doAssert num64 == 86216859871725 - var i = start + var i = 0 var output = T(0) var foundDigit = false let last = min(s.len, if maxLen == 0: s.len else: i + maxLen) @@ -151,10 +151,9 @@ proc parseOct*[T: SomeInteger](s: string, number: var T, start = 0, inc(i) if foundDigit: number = output - result = i - start + result = i -proc parseHex*[T: SomeInteger](s: string, number: var T, start = 0, - maxLen = 0): int {.noSideEffect.} = +proc parseHex*[T: SomeInteger](s: openArray[char], number: var T, maxLen = 0): int {.noSideEffect.} = ## Parses a hexadecimal number and stores its value in ``number``. ## ## Returns the number of the parsed characters or 0 in case of an error. @@ -184,7 +183,7 @@ proc parseHex*[T: SomeInteger](s: string, number: var T, start = 0, var num64: int64 doAssert parseHex("4E69ED4E69ED", num64) == 12 doAssert num64 == 86216859871725 - var i = start + var i = 0 var output = T(0) var foundDigit = false let last = min(s.len, if maxLen == 0: s.len else: i + maxLen) @@ -206,9 +205,9 @@ proc parseHex*[T: SomeInteger](s: string, number: var T, start = 0, inc(i) if foundDigit: number = output - result = i - start + result = i -proc parseIdent*(s: string, ident: var string, start = 0): int = +proc parseIdent*(s: openArray[char], ident: var string): int = ## Parses an identifier and stores it in ``ident``. Returns ## the number of the parsed characters or 0 in case of an error. ## If error, the value of `ident` is not changed. @@ -220,14 +219,14 @@ proc parseIdent*(s: string, ident: var string, start = 0): int = doAssert res == "ello" doAssert parseIdent("Hello World", res, 6) == 5 doAssert res == "World" - var i = start + var i = 0 if i < s.len and s[i] in IdentStartChars: inc(i) while i < s.len and s[i] in IdentChars: inc(i) - ident = substr(s, start, i-1) - result = i-start + ident = substr(s.toOpenArray(0, i-1)) + result = i -proc parseIdent*(s: string, start = 0): string = +proc parseIdent*(s: openArray[char]): string = ## Parses an identifier and returns it or an empty string in ## case of an error. runnableExamples: @@ -236,13 +235,13 @@ proc parseIdent*(s: string, start = 0): string = doAssert parseIdent("Hello World", 5) == "" doAssert parseIdent("Hello World", 6) == "World" result = "" - var i = start + var i = 0 if i < s.len and s[i] in IdentStartChars: inc(i) while i < s.len and s[i] in IdentChars: inc(i) - result = substr(s, start, i-1) + result = substr(s.toOpenArray(0, i - 1)) -proc parseChar*(s: string, c: var char, start = 0): int = +proc parseChar*(s: openArray[char], c: var char): int = ## Parses a single character, stores it in `c` and returns 1. ## In case of error (if start >= s.len) it returns 0 ## and the value of `c` is unchanged. @@ -252,11 +251,11 @@ proc parseChar*(s: string, c: var char, start = 0): int = doAssert c == '\0' doAssert "nim".parseChar(c, 0) == 1 doAssert c == 'n' - if start < s.len: - c = s[start] + if s.len > 0: + c = s[0] result = 1 -proc skipWhitespace*(s: string, start = 0): int {.inline.} = +proc skipWhitespace*(s: openArray[char]): int {.inline.} = ## Skips the whitespace starting at ``s[start]``. Returns the number of ## skipped characters. runnableExamples: @@ -265,9 +264,9 @@ proc skipWhitespace*(s: string, start = 0): int {.inline.} = doAssert skipWhitespace("Hello World", 5) == 1 doAssert skipWhitespace("Hello World", 5) == 2 result = 0 - while start+result < s.len and s[start+result] in Whitespace: inc(result) + while result < s.len and s[result] in Whitespace: inc(result) -proc skip*(s, token: string, start = 0): int {.inline.} = +proc skip*(s, token: openArray[char]): int {.inline.} = ## Skips the `token` starting at ``s[start]``. Returns the length of `token` ## or 0 if there was no `token` at ``s[start]``. runnableExamples: @@ -277,22 +276,22 @@ proc skip*(s, token: string, start = 0): int {.inline.} = doAssert skip("CAPlow", "CAP", 0) == 3 doAssert skip("CAPlow", "cap", 0) == 0 result = 0 - while start+result < s.len and result < token.len and - s[result+start] == token[result]: + while result < s.len and result < token.len and + s[result] == token[result]: inc(result) if result != token.len: result = 0 -proc skipIgnoreCase*(s, token: string, start = 0): int = +proc skipIgnoreCase*(s, token: openArray[char]): int = ## Same as `skip` but case is ignored for token matching. runnableExamples: doAssert skipIgnoreCase("CAPlow", "CAP", 0) == 3 doAssert skipIgnoreCase("CAPlow", "cap", 0) == 3 result = 0 - while start+result < s.len and result < token.len and - toLower(s[result+start]) == toLower(token[result]): inc(result) + while result < s.len and result < token.len and + toLower(s[result]) == toLower(token[result]): inc(result) if result != token.len: result = 0 -proc skipUntil*(s: string, until: set[char], start = 0): int {.inline.} = +proc skipUntil*(s: openArray[char], until: set[char]): int {.inline.} = ## Skips all characters until one char from the set `until` is found ## or the end is reached. ## Returns number of characters skipped. @@ -301,9 +300,9 @@ proc skipUntil*(s: string, until: set[char], start = 0): int {.inline.} = doAssert skipUntil("Hello World", {'W'}, 0) == 6 doAssert skipUntil("Hello World", {'W', 'd'}, 0) == 6 result = 0 - while start+result < s.len and s[result+start] notin until: inc(result) + while result < s.len and s[result] notin until: inc(result) -proc skipUntil*(s: string, until: char, start = 0): int {.inline.} = +proc skipUntil*(s: openArray[char], until: char): int {.inline.} = ## Skips all characters until the char `until` is found ## or the end is reached. ## Returns number of characters skipped. @@ -313,9 +312,9 @@ proc skipUntil*(s: string, until: char, start = 0): int {.inline.} = doAssert skipUntil("Hello World", 'W', 0) == 6 doAssert skipUntil("Hello World", 'w', 0) == 11 result = 0 - while start+result < s.len and s[result+start] != until: inc(result) + while result < s.len and s[result] != until: inc(result) -proc skipWhile*(s: string, toSkip: set[char], start = 0): int {.inline.} = +proc skipWhile*(s: openArray[char], toSkip: set[char]): int {.inline.} = ## Skips all characters while one char from the set `toSkip` is found. ## Returns number of characters skipped. runnableExamples: @@ -323,14 +322,13 @@ proc skipWhile*(s: string, toSkip: set[char], start = 0): int {.inline.} = doAssert skipWhile("Hello World", {'e'}) == 0 doAssert skipWhile("Hello World", {'W', 'o', 'r'}, 6) == 3 result = 0 - while start+result < s.len and s[result+start] in toSkip: inc(result) + while result < s.len and s[result] in toSkip: inc(result) -proc fastSubstr(s: string; token: var string; start, length: int) = +proc fastSubstr(s: openArray[char]; token: var string; length: int) = token.setLen length - for i in 0 ..< length: token[i] = s[i+start] + for i in 0 ..< length: token[i] = s[i] -proc parseUntil*(s: string, token: var string, until: set[char], - start = 0): int {.inline.} = +proc parseUntil*(s: openArray[char], token: var string, until: set[char]): int {.inline.} = ## Parses a token and stores it in ``token``. Returns ## the number of the parsed characters or 0 in case of an error. A token ## consists of the characters notin `until`. @@ -342,14 +340,13 @@ proc parseUntil*(s: string, token: var string, until: set[char], doAssert myToken == "Hello " doAssert parseUntil("Hello World", myToken, {'W', 'r'}, 3) == 3 doAssert myToken == "lo " - var i = start + var i = 0 while i < s.len and s[i] notin until: inc(i) - result = i-start - fastSubstr(s, token, start, result) + result = i + fastSubstr(s, token, result) #token = substr(s, start, i-1) -proc parseUntil*(s: string, token: var string, until: char, - start = 0): int {.inline.} = +proc parseUntil*(s: openArray[char], token: var string, until: char): int {.inline.} = ## Parses a token and stores it in ``token``. Returns ## the number of the parsed characters or 0 in case of an error. A token ## consists of any character that is not the `until` character. @@ -361,14 +358,13 @@ proc parseUntil*(s: string, token: var string, until: char, doAssert myToken == "Hell" doAssert parseUntil("Hello World", myToken, 'o', 2) == 2 doAssert myToken == "ll" - var i = start + var i = 0 while i < s.len and s[i] != until: inc(i) - result = i-start - fastSubstr(s, token, start, result) + result = i + fastSubstr(s, token, result) #token = substr(s, start, i-1) -proc parseUntil*(s: string, token: var string, until: string, - start = 0): int {.inline.} = +proc parseUntil*(s: openArray[char], token: var string, until: string): int {.inline.} = ## Parses a token and stores it in ``token``. Returns ## the number of the parsed characters or 0 in case of an error. A token ## consists of any character that comes before the `until` token. @@ -382,7 +378,7 @@ proc parseUntil*(s: string, token: var string, until: string, if until.len == 0: token.setLen(0) return 0 - var i = start + var i = 0 while i < s.len: if until.len > 0 and s[i] == until[0]: var u = 1 @@ -390,12 +386,11 @@ proc parseUntil*(s: string, token: var string, until: string, inc u if u >= until.len: break inc(i) - result = i-start - fastSubstr(s, token, start, result) + result = i + fastSubstr(s, token, result) #token = substr(s, start, i-1) -proc parseWhile*(s: string, token: var string, validChars: set[char], - start = 0): int {.inline.} = +proc parseWhile*(s: openArray[char], token: var string, validChars: set[char]): int {.inline.} = ## Parses a token and stores it in ``token``. Returns ## the number of the parsed characters or 0 in case of an error. A token ## consists of the characters in `validChars`. @@ -405,22 +400,22 @@ proc parseWhile*(s: string, token: var string, validChars: set[char], doAssert myToken.len() == 0 doAssert parseWhile("Hello World", myToken, {'W', 'o', 'r'}, 6) == 3 doAssert myToken == "Wor" - var i = start + var i = 0 while i < s.len and s[i] in validChars: inc(i) - result = i-start - fastSubstr(s, token, start, result) + result = i + fastSubstr(s, token, result) #token = substr(s, start, i-1) -proc captureBetween*(s: string, first: char, second = '\0', start = 0): string = +proc captureBetween*(s: openArray[char], first: char, second = '\0'): string = ## Finds the first occurrence of ``first``, then returns everything from there ## up to ``second`` (if ``second`` is '\0', then ``first`` is used). runnableExamples: doAssert captureBetween("Hello World", 'e') == "llo World" doAssert captureBetween("Hello World", 'e', 'r') == "llo Wo" - doAssert captureBetween("Hello World", 'l', start = 6) == "d" - var i = skipUntil(s, first, start)+1+start + doAssert captureBetween("Hello World".toOpenArray(6, "Hello World".high), 'l') == "d" + var i = skipUntil(s, first) + 1 result = "" - discard s.parseUntil(result, if second == '\0': first else: second, i) + discard parseUntil(s.toOpenArray(i, s.high), result, if second == '\0': first else: second) proc integerOutOfRangeError() {.noinline.} = raise newException(ValueError, "Parsed integer outside of valid range") @@ -429,10 +424,10 @@ proc integerOutOfRangeError() {.noinline.} = when defined(js): {.push overflowChecks: off.} -proc rawParseInt(s: string, b: var BiggestInt, start = 0): int = +proc rawParseInt(s: openArray[char], b: var BiggestInt): int = var sign: BiggestInt = -1 - i = start + i = 0 if i < s.len: if s[i] == '+': inc(i) elif s[i] == '-': @@ -452,30 +447,30 @@ proc rawParseInt(s: string, b: var BiggestInt, start = 0): int = integerOutOfRangeError() else: b = b * sign - result = i - start + result = i when defined(js): {.pop.} # overflowChecks: off -proc parseBiggestInt*(s: string, number: var BiggestInt, start = 0): int {. +proc parseBiggestInt*(s: openArray[char], number: var BiggestInt): int {. rtl, extern: "npuParseBiggestInt", noSideEffect, raises: [ValueError].} = - ## Parses an integer starting at `start` and stores the value into `number`. + ## Parses an integer and stores the value into `number`. ## Result is the number of processed chars or 0 if there is no integer. ## `ValueError` is raised if the parsed integer is out of the valid range. runnableExamples: var res: BiggestInt - doAssert parseBiggestInt("9223372036854775807", res, 0) == 19 + doAssert parseBiggestInt("9223372036854775807", res) == 19 doAssert res == 9223372036854775807 var res = BiggestInt(0) # use 'res' for exception safety (don't write to 'number' in case of an # overflow exception): - result = rawParseInt(s, res, start) + result = rawParseInt(s, res) if result != 0: number = res -proc parseInt*(s: string, number: var int, start = 0): int {. +proc parseInt*(s: openArray[char], number: var int): int {. rtl, extern: "npuParseInt", noSideEffect, raises: [ValueError].} = - ## Parses an integer starting at `start` and stores the value into `number`. + ## Parses an integer and stores the value into `number`. ## Result is the number of processed chars or 0 if there is no integer. ## `ValueError` is raised if the parsed integer is out of the valid range. runnableExamples: @@ -485,14 +480,14 @@ proc parseInt*(s: string, number: var int, start = 0): int {. doAssert parseInt("2019", res, 2) == 2 doAssert res == 19 var res = BiggestInt(0) - result = parseBiggestInt(s, res, start) + result = parseBiggestInt(s, res) when sizeof(int) <= 4: if res < low(int) or res > high(int): integerOutOfRangeError() if result != 0: number = int(res) -proc parseSaturatedNatural*(s: string, b: var int, start = 0): int {. +proc parseSaturatedNatural*(s: openArray[char], b: var int): int {. raises: [].} = ## Parses a natural number into ``b``. This cannot raise an overflow ## error. ``high(int)`` is returned for an overflow. @@ -502,7 +497,7 @@ proc parseSaturatedNatural*(s: string, b: var int, start = 0): int {. var res = 0 discard parseSaturatedNatural("848", res) doAssert res == 848 - var i = start + var i = 0 if i < s.len and s[i] == '+': inc(i) if i < s.len and s[i] in {'0'..'9'}: b = 0 @@ -514,13 +509,13 @@ proc parseSaturatedNatural*(s: string, b: var int, start = 0): int {. b = high(int) inc(i) while i < s.len and s[i] == '_': inc(i) # underscores are allowed and ignored - result = i - start + result = i -proc rawParseUInt(s: string, b: var BiggestUInt, start = 0): int = +proc rawParseUInt(s: openArray[char], b: var BiggestUInt): int = var res = 0.BiggestUInt prev = 0.BiggestUInt - i = start + i = 0 if i < s.len - 1 and s[i] == '-' and s[i + 1] in {'0'..'9'}: integerOutOfRangeError() if i < s.len and s[i] == '+': inc(i) # Allow @@ -534,11 +529,11 @@ proc rawParseUInt(s: string, b: var BiggestUInt, start = 0): int = inc(i) while i < s.len and s[i] == '_': inc(i) # underscores are allowed and ignored b = res - result = i - start + result = i -proc parseBiggestUInt*(s: string, number: var BiggestUInt, start = 0): int {. +proc parseBiggestUInt*(s: openArray[char], number: var BiggestUInt): int {. rtl, extern: "npuParseBiggestUInt", noSideEffect, raises: [ValueError].} = - ## Parses an unsigned integer starting at `start` and stores the value + ## Parses an unsigned integer and stores the value ## into `number`. ## `ValueError` is raised if the parsed integer is out of the valid range. runnableExamples: @@ -550,13 +545,13 @@ proc parseBiggestUInt*(s: string, number: var BiggestUInt, start = 0): int {. var res = BiggestUInt(0) # use 'res' for exception safety (don't write to 'number' in case of an # overflow exception): - result = rawParseUInt(s, res, start) + result = rawParseUInt(s, res) if result != 0: number = res -proc parseUInt*(s: string, number: var uint, start = 0): int {. +proc parseUInt*(s: openArray[char], number: var uint): int {. rtl, extern: "npuParseUInt", noSideEffect, raises: [ValueError].} = - ## Parses an unsigned integer starting at `start` and stores the value + ## Parses an unsigned integer and stores the value ## into `number`. ## `ValueError` is raised if the parsed integer is out of the valid range. runnableExamples: @@ -566,22 +561,22 @@ proc parseUInt*(s: string, number: var uint, start = 0): int {. doAssert parseUInt("3450", res, 2) == 2 doAssert res == 50 var res = BiggestUInt(0) - result = parseBiggestUInt(s, res, start) + result = parseBiggestUInt(s, res) when sizeof(BiggestUInt) > sizeof(uint) and sizeof(uint) <= 4: if res > 0xFFFF_FFFF'u64: integerOutOfRangeError() if result != 0: number = uint(res) -proc parseBiggestFloat*(s: string, number: var BiggestFloat, start = 0): int {. +proc parseBiggestFloat*(s: openArray[char], number: var BiggestFloat): int {. magic: "ParseBiggestFloat", importc: "nimParseBiggestFloat", noSideEffect.} - ## Parses a float starting at `start` and stores the value into `number`. + ## Parses a float and stores the value into `number`. ## Result is the number of processed chars or 0 if a parsing error ## occurred. -proc parseFloat*(s: string, number: var float, start = 0): int {. +proc parseFloat*(s: openArray[char], number: var float): int {. rtl, extern: "npuParseFloat", noSideEffect.} = - ## Parses a float starting at `start` and stores the value into `number`. + ## Parses a float and stores the value into `number`. ## Result is the number of processed chars or 0 if there occurred a parsing ## error. runnableExamples: @@ -593,7 +588,7 @@ proc parseFloat*(s: string, number: var float, start = 0): int {. doAssert parseFloat("32.57", res, 3) == 2 doAssert res == 57.00 var bf = BiggestFloat(0.0) - result = parseBiggestFloat(s, bf, start) + result = parseBiggestFloat(s, bf) if result != 0: number = bf @@ -606,7 +601,7 @@ type ikVar, ## ``var`` part of the interpolated string ikExpr ## ``expr`` part of the interpolated string -iterator interpolatedFragments*(s: string): tuple[kind: InterpolatedKind, +iterator interpolatedFragments*(s: openArray[char]): tuple[kind: InterpolatedKind, value: string] = ## Tokenizes the string `s` into substrings for interpolation purposes. ## @@ -641,7 +636,7 @@ iterator interpolatedFragments*(s: string): tuple[kind: InterpolatedKind, else: discard inc j raise newException(ValueError, - "Expected closing '}': " & substr(s, i, s.high)) + "Expected closing '}': " & substr(s.toOpenArray(i, s.high))) inc i, 2 # skip ${ kind = ikExpr elif j+1 < s.len and s[j+1] in IdentStartChars: @@ -655,15 +650,370 @@ iterator interpolatedFragments*(s: string): tuple[kind: InterpolatedKind, kind = ikDollar else: raise newException(ValueError, - "Unable to parse a variable name at " & substr(s, i, s.high)) + "Unable to parse a variable name at " & substr(s.toOpenArray(i, s.high))) else: while j < s.len and s[j] != '$': inc j kind = ikStr if j > i: # do not copy the trailing } for ikExpr: - yield (kind, substr(s, i, j-1-ord(kind == ikExpr))) + yield (kind, substr(s.toOpenArray(i, j-1-ord(kind == ikExpr)))) else: break i = j {.pop.} + + +proc parseBin*[T: SomeInteger](s: string, number: var T, start = 0, + maxLen = 0): int {.noSideEffect.} = + ## Parses a binary number and stores its value in ``number``. + ## + ## Returns the number of the parsed characters or 0 in case of an error. + ## If error, the value of ``number`` is not changed. + ## + ## If ``maxLen == 0``, the parsing continues until the first non-bin character + ## or to the end of the string. Otherwise, no more than ``maxLen`` characters + ## are parsed starting from the ``start`` position. + ## + ## It does not check for overflow. If the value represented by the string is + ## too big to fit into ``number``, only the value of last fitting characters + ## will be stored in ``number`` without producing an error. + runnableExamples: + var num: int + doAssert parseBin("0100_1110_0110_1001_1110_1101", num) == 29 + doAssert num == 5138925 + doAssert parseBin("3", num) == 0 + var num8: int8 + doAssert parseBin("0b_0100_1110_0110_1001_1110_1101", num8) == 32 + doAssert num8 == 0b1110_1101'i8 + doAssert parseBin("0b_0100_1110_0110_1001_1110_1101", num8, 3, 9) == 9 + doAssert num8 == 0b0100_1110'i8 + var num8u: uint8 + doAssert parseBin("0b_0100_1110_0110_1001_1110_1101", num8u) == 32 + doAssert num8u == 237 + var num64: int64 + doAssert parseBin("0100111001101001111011010100111001101001", num64) == 40 + doAssert num64 == 336784608873 + parseBin(s.toOpenArray(start, s.high), number, maxLen) + +proc parseOct*[T: SomeInteger](s: string, number: var T, start = 0, + maxLen = 0): int {.noSideEffect.} = + ## Parses an octal number and stores its value in ``number``. + ## + ## Returns the number of the parsed characters or 0 in case of an error. + ## If error, the value of ``number`` is not changed. + ## + ## If ``maxLen == 0``, the parsing continues until the first non-oct character + ## or to the end of the string. Otherwise, no more than ``maxLen`` characters + ## are parsed starting from the ``start`` position. + ## + ## It does not check for overflow. If the value represented by the string is + ## too big to fit into ``number``, only the value of last fitting characters + ## will be stored in ``number`` without producing an error. + runnableExamples: + var num: int + doAssert parseOct("0o23464755", num) == 10 + doAssert num == 5138925 + doAssert parseOct("8", num) == 0 + var num8: int8 + doAssert parseOct("0o_1464_755", num8) == 11 + doAssert num8 == -19 + doAssert parseOct("0o_1464_755", num8, 3, 3) == 3 + doAssert num8 == 102 + var num8u: uint8 + doAssert parseOct("1464755", num8u) == 7 + doAssert num8u == 237 + var num64: int64 + doAssert parseOct("2346475523464755", num64) == 16 + doAssert num64 == 86216859871725 + parseOct(s.toOpenArray(start, s.high), number, maxLen) + +proc parseHex*[T: SomeInteger](s: string, number: var T, start = 0, + maxLen = 0): int {.noSideEffect.} = + ## Parses a hexadecimal number and stores its value in ``number``. + ## + ## Returns the number of the parsed characters or 0 in case of an error. + ## If error, the value of ``number`` is not changed. + ## + ## If ``maxLen == 0``, the parsing continues until the first non-hex character + ## or to the end of the string. Otherwise, no more than ``maxLen`` characters + ## are parsed starting from the ``start`` position. + ## + ## It does not check for overflow. If the value represented by the string is + ## too big to fit into ``number``, only the value of last fitting characters + ## will be stored in ``number`` without producing an error. + runnableExamples: + var num: int + doAssert parseHex("4E_69_ED", num) == 8 + doAssert num == 5138925 + doAssert parseHex("X", num) == 0 + doAssert parseHex("#ABC", num) == 4 + var num8: int8 + doAssert parseHex("0x_4E_69_ED", num8) == 11 + doAssert num8 == 0xED'i8 + doAssert parseHex("0x_4E_69_ED", num8, 3, 2) == 2 + doAssert num8 == 0x4E'i8 + var num8u: uint8 + doAssert parseHex("0x_4E_69_ED", num8u) == 11 + doAssert num8u == 237 + var num64: int64 + doAssert parseHex("4E69ED4E69ED", num64) == 12 + doAssert num64 == 86216859871725 + parseHex(s.toOpenArray(start, s.high), number, maxLen) + +proc parseIdent*(s: string, ident: var string, start = 0): int = + ## Parses an identifier and stores it in ``ident``. Returns + ## the number of the parsed characters or 0 in case of an error. + ## If error, the value of `ident` is not changed. + runnableExamples: + var res: string + doAssert parseIdent("Hello World", res, 0) == 5 + doAssert res == "Hello" + doAssert parseIdent("Hello World", res, 1) == 4 + doAssert res == "ello" + doAssert parseIdent("Hello World", res, 6) == 5 + doAssert res == "World" + parseIdent(s.toOpenArray(start, s.high), ident) + +proc parseIdent*(s: string, start = 0): string = + ## Parses an identifier and returns it or an empty string in + ## case of an error. + runnableExamples: + doAssert parseIdent("Hello World", 0) == "Hello" + doAssert parseIdent("Hello World", 1) == "ello" + doAssert parseIdent("Hello World", 5) == "" + doAssert parseIdent("Hello World", 6) == "World" + parseIdent(s.toOpenArray(start, s.high)) + +proc parseChar*(s: string, c: var char, start = 0): int = + ## Parses a single character, stores it in `c` and returns 1. + ## In case of error (if start >= s.len) it returns 0 + ## and the value of `c` is unchanged. + runnableExamples: + var c: char + doAssert "nim".parseChar(c, 3) == 0 + doAssert c == '\0' + doAssert "nim".parseChar(c, 0) == 1 + doAssert c == 'n' + parseChar(s.toOpenArray(start, s.high), c) + +proc skipWhitespace*(s: string, start = 0): int {.inline.} = + ## Skips the whitespace starting at ``s[start]``. Returns the number of + ## skipped characters. + runnableExamples: + doAssert skipWhitespace("Hello World", 0) == 0 + doAssert skipWhitespace(" Hello World", 0) == 1 + doAssert skipWhitespace("Hello World", 5) == 1 + doAssert skipWhitespace("Hello World", 5) == 2 + skipWhitespace(s.toOpenArray(start, s.high)) + +proc skip*(s, token: string, start = 0): int {.inline.} = + ## Skips the `token` starting at ``s[start]``. Returns the length of `token` + ## or 0 if there was no `token` at ``s[start]``. + runnableExamples: + doAssert skip("2019-01-22", "2019", 0) == 4 + doAssert skip("2019-01-22", "19", 0) == 0 + doAssert skip("2019-01-22", "19", 2) == 2 + doAssert skip("CAPlow", "CAP", 0) == 3 + doAssert skip("CAPlow", "cap", 0) == 0 + skip(s.toOpenArray(start, s.high), token) + +proc skipIgnoreCase*(s, token: string, start = 0): int = + ## Same as `skip` but case is ignored for token matching. + runnableExamples: + doAssert skipIgnoreCase("CAPlow", "CAP", 0) == 3 + doAssert skipIgnoreCase("CAPlow", "cap", 0) == 3 + skipIgnoreCase(s.toOpenArray(start, s.high), token) + +proc skipUntil*(s: string, until: set[char], start = 0): int {.inline.} = + ## Skips all characters until one char from the set `until` is found + ## or the end is reached. + ## Returns number of characters skipped. + runnableExamples: + doAssert skipUntil("Hello World", {'W', 'e'}, 0) == 1 + doAssert skipUntil("Hello World", {'W'}, 0) == 6 + doAssert skipUntil("Hello World", {'W', 'd'}, 0) == 6 + skipUntil(s.toOpenArray(start, s.high), until) + +proc skipUntil*(s: string, until: char, start = 0): int {.inline.} = + ## Skips all characters until the char `until` is found + ## or the end is reached. + ## Returns number of characters skipped. + runnableExamples: + doAssert skipUntil("Hello World", 'o', 0) == 4 + doAssert skipUntil("Hello World", 'o', 4) == 0 + doAssert skipUntil("Hello World", 'W', 0) == 6 + doAssert skipUntil("Hello World", 'w', 0) == 11 + skipUntil(s.toOpenArray(start, s.high), until) + +proc skipWhile*(s: string, toSkip: set[char], start = 0): int {.inline.} = + ## Skips all characters while one char from the set `toSkip` is found. + ## Returns number of characters skipped. + runnableExamples: + doAssert skipWhile("Hello World", {'H', 'e'}) == 2 + doAssert skipWhile("Hello World", {'e'}) == 0 + doAssert skipWhile("Hello World", {'W', 'o', 'r'}, 6) == 3 + skipWhile(s.toOpenArray(start, s.high), toSkip) + +proc parseUntil*(s: string, token: var string, until: set[char], + start = 0): int {.inline.} = + ## Parses a token and stores it in ``token``. Returns + ## the number of the parsed characters or 0 in case of an error. A token + ## consists of the characters notin `until`. + runnableExamples: + var myToken: string + doAssert parseUntil("Hello World", myToken, {'W', 'o', 'r'}) == 4 + doAssert myToken == "Hell" + doAssert parseUntil("Hello World", myToken, {'W', 'r'}) == 6 + doAssert myToken == "Hello " + doAssert parseUntil("Hello World", myToken, {'W', 'r'}, 3) == 3 + doAssert myToken == "lo " + parseUntil(s.toOpenArray(start, s.high), token, until) + +proc parseUntil*(s: string, token: var string, until: char, + start = 0): int {.inline.} = + ## Parses a token and stores it in ``token``. Returns + ## the number of the parsed characters or 0 in case of an error. A token + ## consists of any character that is not the `until` character. + runnableExamples: + var myToken: string + doAssert parseUntil("Hello World", myToken, 'W') == 6 + doAssert myToken == "Hello " + doAssert parseUntil("Hello World", myToken, 'o') == 4 + doAssert myToken == "Hell" + doAssert parseUntil("Hello World", myToken, 'o', 2) == 2 + doAssert myToken == "ll" + parseUntil(s.toOpenArray(start, s.high), token, until) + +proc parseUntil*(s: string, token: var string, until: string, + start = 0): int {.inline.} = + ## Parses a token and stores it in ``token``. Returns + ## the number of the parsed characters or 0 in case of an error. A token + ## consists of any character that comes before the `until` token. + runnableExamples: + var myToken: string + doAssert parseUntil("Hello World", myToken, "Wor") == 6 + doAssert myToken == "Hello " + doAssert parseUntil("Hello World", myToken, "Wor", 2) == 4 + doAssert myToken == "llo " + parseUntil(s.toOpenArray(start, s.high), token, until) + +proc parseWhile*(s: string, token: var string, validChars: set[char], + start = 0): int {.inline.} = + ## Parses a token and stores it in ``token``. Returns + ## the number of the parsed characters or 0 in case of an error. A token + ## consists of the characters in `validChars`. + runnableExamples: + var myToken: string + doAssert parseWhile("Hello World", myToken, {'W', 'o', 'r'}, 0) == 0 + doAssert myToken.len() == 0 + doAssert parseWhile("Hello World", myToken, {'W', 'o', 'r'}, 6) == 3 + doAssert myToken == "Wor" + parseWhile(s.toOpenArray(start, s.high), token, validChars) + +proc captureBetween*(s: string, first: char, second = '\0', start = 0): string = + ## Finds the first occurrence of ``first``, then returns everything from there + ## up to ``second`` (if ``second`` is '\0', then ``first`` is used). + runnableExamples: + doAssert captureBetween("Hello World", 'e') == "llo World" + doAssert captureBetween("Hello World", 'e', 'r') == "llo Wo" + doAssert captureBetween("Hello World", 'l', start = 6) == "d" + captureBetween(s.toOpenArray(start, s.high), first, second) + +proc parseBiggestInt*(s: string, number: var BiggestInt, start = 0): int {.noSideEffect, raises: [ValueError].} = + ## Parses an integer starting at `start` and stores the value into `number`. + ## Result is the number of processed chars or 0 if there is no integer. + ## `ValueError` is raised if the parsed integer is out of the valid range. + runnableExamples: + var res: BiggestInt + doAssert parseBiggestInt("9223372036854775807", res, 0) == 19 + doAssert res == 9223372036854775807 + parseBiggestInt(s.toOpenArray(start, s.high), number) + +proc parseInt*(s: string, number: var int, start = 0): int {.noSideEffect, raises: [ValueError].} = + ## Parses an integer starting at `start` and stores the value into `number`. + ## Result is the number of processed chars or 0 if there is no integer. + ## `ValueError` is raised if the parsed integer is out of the valid range. + runnableExamples: + var res: int + doAssert parseInt("2019", res, 0) == 4 + doAssert res == 2019 + doAssert parseInt("2019", res, 2) == 2 + doAssert res == 19 + parseInt(s.toOpenArray(start, s.high), number) + + +proc parseSaturatedNatural*(s: string, b: var int, start = 0): int {. + raises: [].} = + ## Parses a natural number into ``b``. This cannot raise an overflow + ## error. ``high(int)`` is returned for an overflow. + ## The number of processed character is returned. + ## This is usually what you really want to use instead of `parseInt`:idx:. + runnableExamples: + var res = 0 + discard parseSaturatedNatural("848", res) + doAssert res == 848 + parseSaturatedNatural(s.toOpenArray(start, s.high), b) + + +proc parseBiggestUInt*(s: string, number: var BiggestUInt, start = 0): int {.noSideEffect, raises: [ValueError].} = + ## Parses an unsigned integer starting at `start` and stores the value + ## into `number`. + ## `ValueError` is raised if the parsed integer is out of the valid range. + runnableExamples: + var res: BiggestUInt + doAssert parseBiggestUInt("12", res, 0) == 2 + doAssert res == 12 + doAssert parseBiggestUInt("1111111111111111111", res, 0) == 19 + doAssert res == 1111111111111111111'u64 + parseBiggestUInt(s.toOpenArray(start, s.high), number) + +proc parseUInt*(s: string, number: var uint, start = 0): int {.noSideEffect, raises: [ValueError].} = + ## Parses an unsigned integer starting at `start` and stores the value + ## into `number`. + ## `ValueError` is raised if the parsed integer is out of the valid range. + runnableExamples: + var res: uint + doAssert parseUInt("3450", res) == 4 + doAssert res == 3450 + doAssert parseUInt("3450", res, 2) == 2 + doAssert res == 50 + parseUInt(s.toOpenArray(start, s.high), number) + +proc parseBiggestFloat*(s: string, number: var BiggestFloat, start = 0): int {.noSideEffect.} = + ## Parses a float starting at `start` and stores the value into `number`. + ## Result is the number of processed chars or 0 if a parsing error + ## occurred. + parseFloat(s.toOpenArray(start, s.high), number) + +proc parseFloat*(s: string, number: var float, start = 0): int {.noSideEffect.} = + ## Parses a float starting at `start` and stores the value into `number`. + ## Result is the number of processed chars or 0 if there occurred a parsing + ## error. + runnableExamples: + var res: float + doAssert parseFloat("32", res, 0) == 2 + doAssert res == 32.0 + doAssert parseFloat("32.57", res, 0) == 5 + doAssert res == 32.57 + doAssert parseFloat("32.57", res, 3) == 2 + doAssert res == 57.00 + parseFloat(s.toOpenArray(start, s.high), number) + +iterator interpolatedFragments*(s: string): tuple[kind: InterpolatedKind, + value: string] = + ## Tokenizes the string `s` into substrings for interpolation purposes. + ## + runnableExamples: + var outp: seq[tuple[kind: InterpolatedKind, value: string]] + for k, v in interpolatedFragments(" $this is ${an example} $$"): + outp.add (k, v) + doAssert outp == @[(ikStr, " "), + (ikVar, "this"), + (ikStr, " is "), + (ikExpr, "an example"), + (ikStr, " "), + (ikDollar, "$")] + for x in s.toOa.interpolatedFragments: + yield x + diff --git a/lib/system.nim b/lib/system.nim index c0e95a5d2..c28479a44 100644 --- a/lib/system.nim +++ b/lib/system.nim @@ -2580,7 +2580,19 @@ template once*(body: untyped): untyped = {.pop.} # warning[GcMem]: off, warning[Uninit]: off -proc substr*(s: string, first, last: int): string = +proc substr*(s: openArray[char]): string = + ## Copies a slice of `s` into a new string and returns this new + ## string. + runnableExamples: + let a = "abcdefgh" + assert a.substr(2, 5) == "cdef" + assert a.substr(2) == "cdefgh" + assert a.substr(5, 99) == "fgh" + result = newString(s.len) + for i, ch in s: + result[i] = ch + +proc substr*(s: string, first, last: int): string = # A bug with `magic: Slice` requires this to exist this way ## Copies a slice of `s` into a new string and returns this new ## string. ## diff --git a/lib/system/jssys.nim b/lib/system/jssys.nim index a31de0d86..9dfa80877 100644 --- a/lib/system/jssys.nim +++ b/lib/system/jssys.nim @@ -725,15 +725,20 @@ const IdentChars = {'a'..'z', 'A'..'Z', '0'..'9', '_'} -proc parseFloatNative(a: string): float = - let a2 = a.cstring +proc parseFloatNative(a: openarray[char]): float = + var str = "" + for x in a: + str.add x + + let cstr = cstring str + asm """ - `result` = Number(`a2`); + `result` = Number(`cstr`); """ -proc nimParseBiggestFloat(s: string, number: var BiggestFloat, start: int): int {.compilerproc.} = +proc nimParseBiggestFloat(s: openarray[char], number: var BiggestFloat): int {.compilerproc.} = var sign: bool - var i = start + var i = 0 if s[i] == '+': inc(i) elif s[i] == '-': sign = true @@ -743,14 +748,14 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, start: int): int if s[i+2] == 'N' or s[i+2] == 'n': if s[i+3] notin IdentChars: number = NaN - return i+3 - start + return i+3 return 0 if s[i] == 'I' or s[i] == 'i': if s[i+1] == 'N' or s[i+1] == 'n': if s[i+2] == 'F' or s[i+2] == 'f': if s[i+3] notin IdentChars: number = if sign: -Inf else: Inf - return i+3 - start + return i+3 return 0 var buf: string @@ -782,7 +787,7 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, start: int): int addInc() eatUnderscores() number = parseFloatNative(buf) - result = i - start + result = i # Workaround for IE, IE up to version 11 lacks 'Math.trunc'. We produce # 'Math.trunc' for Nim's ``div`` and ``mod`` operators: diff --git a/lib/system/strmantle.nim b/lib/system/strmantle.nim index feaac7817..cb67185ab 100644 --- a/lib/system/strmantle.nim +++ b/lib/system/strmantle.nim @@ -78,8 +78,8 @@ const when defined(nimHasInvariant): {.push staticBoundChecks: off.} -proc nimParseBiggestFloat(s: string, number: var BiggestFloat, - start = 0): int {.compilerproc.} = +proc nimParseBiggestFloat(s: openArray[char], number: var BiggestFloat, + ): int {.compilerproc.} = # This routine attempt to parse float that can parsed quickly. # i.e. whose integer part can fit inside a 53bits integer. # their real exponent must also be <= 22. If the float doesn't follow @@ -88,7 +88,7 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, # This avoid the problems of decimal character portability. # see: http://www.exploringbinary.com/fast-path-decimal-to-floating-point-conversion/ var - i = start + i = 0 sign = 1.0 kdigits, fdigits = 0 exponent = 0 @@ -111,7 +111,7 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, if s[i+2] == 'N' or s[i+2] == 'n': if i+3 >= s.len or s[i+3] notin IdentChars: number = NaN - return i+3 - start + return i+3 return 0 # Inf? @@ -120,7 +120,7 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, if s[i+2] == 'F' or s[i+2] == 'f': if i+3 >= s.len or s[i+3] notin IdentChars: number = Inf*sign - return i+3 - start + return i+3 return 0 if i < s.len and s[i] in {'0'..'9'}: @@ -154,8 +154,8 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, # if has no digits: return error if kdigits + fdigits <= 0 and - (i == start or # no char consumed (empty string). - (i == start + 1 and hasSign)): # or only '+' or '- + (i == 0 or # no char consumed (empty string). + (i == 1 and hasSign)): # or only '+' or '- return 0 if i+1 < s.len and s[i] in {'e', 'E'}: @@ -182,7 +182,7 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, number = 0.0*sign else: number = Inf*sign - return i - start + return i # if integer is representable in 53 bits: fast path # max fast path integer is 1<<53 - 1 or 8999999999999999 (16 digits) @@ -194,14 +194,14 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, number = sign * integer.float / powtens[absExponent] else: number = sign * integer.float * powtens[absExponent] - return i - start + return i # if exponent is greater try to fit extra exponent above 22 by multiplying # integer part is there is space left. let slop = 15 - kdigits - fdigits if absExponent <= 22 + slop and not expNegative: number = sign * integer.float * powtens[slop] * powtens[absExponent-slop] - return i - start + return i # if failed: slow path with strtod. var t: array[500, char] # flaviu says: 325 is the longest reasonable literal @@ -209,8 +209,8 @@ proc nimParseBiggestFloat(s: string, number: var BiggestFloat, let maxlen = t.high - "e+000".len # reserve enough space for exponent let endPos = i - result = endPos - start - i = start + result = endPos + i = 0 # re-parse without error checking, any error should be handled by the code above. if i < endPos and s[i] == '.': i.inc while i < endPos and s[i] in {'0'..'9','+','-'}: |