X-Git-Url: http://nitlanguage.org diff --git a/lib/standard/string.nit b/lib/standard/string.nit index bf59f0b..8db6a4e 100644 --- a/lib/standard/string.nit +++ b/lib/standard/string.nit @@ -350,12 +350,12 @@ abstract class Text # # assert " \n\thello \n\t".l_trim == "hello \n\t" # - # A whitespace is defined as any character which ascii value is less than or equal to 32 + # `Char::is_whitespace` determines what is a whitespace. fun l_trim: SELFTYPE do var iter = self.chars.iterator while iter.is_ok do - if iter.item.ascii > 32 then break + if not iter.item.is_whitespace then break iter.next end if iter.index == length then return self.empty @@ -366,12 +366,12 @@ abstract class Text # # assert " \n\thello \n\t".r_trim == " \n\thello" # - # A whitespace is defined as any character which ascii value is less than or equal to 32 + # `Char::is_whitespace` determines what is a whitespace. fun r_trim: SELFTYPE do var iter = self.chars.reverse_iterator while iter.is_ok do - if iter.item.ascii > 32 then break + if not iter.item.is_whitespace then break iter.next end if iter.index < 0 then return self.empty @@ -379,23 +379,61 @@ abstract class Text end # Trims trailing and preceding white spaces - # A whitespace is defined as any character which ascii value is less than or equal to 32 # # assert " Hello World ! ".trim == "Hello World !" # assert "\na\nb\tc\t".trim == "a\nb\tc" + # + # `Char::is_whitespace` determines what is a whitespace. fun trim: SELFTYPE do return (self.l_trim).r_trim - # Returns `self` removed from its last `\n` (if any). + # Is the string non-empty but only made of whitespaces? + # + # assert " \n\t ".is_whitespace == true + # assert " hello ".is_whitespace == false + # assert "".is_whitespace == false + # + # `Char::is_whitespace` determines what is a whitespace. + fun is_whitespace: Bool + do + if is_empty then return false + for c in self.chars do + if not c.is_whitespace then return false + end + return true + end + + # Returns `self` removed from its last line terminator (if any). # # assert "Hello\n".chomp == "Hello" # assert "Hello".chomp == "Hello" - # assert "\n\n\n".chomp == "\n\n" # - # This method is mainly used to remove the LINE_FEED character from lines of text. + # assert "\n".chomp == "" + # assert "".chomp == "" + # + # Line terminators are `"\n"`, `"\r\n"` and `"\r"`. + # A single line terminator, the last one, is removed. + # + # assert "\r\n".chomp == "" + # assert "\r\n\n".chomp == "\r\n" + # assert "\r\n\r\n".chomp == "\r\n" + # assert "\r\n\r".chomp == "\r\n" + # + # Note: unlike with most IO methods like `Reader::read_line`, + # a single `\r` is considered here to be a line terminator and will be removed. fun chomp: SELFTYPE do - if self.chars.last != '\n' then return self - return substring(0, length-1) + var len = length + if len == 0 then return self + var l = self.chars.last + if l == '\r' then + return substring(0, len-1) + else if l != '\n' then + return self + else if len > 1 and self.chars[len-2] == '\r' then + return substring(0, len-2) + else + return substring(0, len-1) + end end # Justify a self in a space of `length` @@ -405,6 +443,8 @@ abstract class Text # * 1.0 for right-justified (all spaces at the left) # * 0.5 for centered (half the spaces at the left) # + # Examples + # # assert "hello".justify(10, 0.0) == "hello " # assert "hello".justify(10, 1.0) == " hello" # assert "hello".justify(10, 0.5) == " hello " @@ -413,25 +453,51 @@ abstract class Text # # assert "hello".justify(2, 0.0) == "hello" # - # REQUIRE: left >= 0.0 and left <= 1.0 + # REQUIRE: `left >= 0.0 and left <= 1.0` # ENSURE: `self.length <= length implies result.length == length` - # ENSURE: `self.length >= length implies result == self - fun justify(length: Int, left: Float): SELFTYPE + # ENSURE: `self.length >= length implies result == self` + fun justify(length: Int, left: Float): String do var diff = length - self.length - if diff <= 0 then return self + if diff <= 0 then return to_s assert left >= 0.0 and left <= 1.0 var before = (diff.to_f * left).to_i return " " * before + self + " " * (diff-before) end - # Mangle a string to be a unique string only made of alphanumeric characters + # Mangle a string to be a unique string only made of alphanumeric characters and underscores. + # + # This method is injective (two different inputs never produce the same + # output) and the returned string always respect the following rules: + # + # * Contains only US-ASCII letters, digits and underscores. + # * Never starts with a digit. + # * Never ends with an underscore. + # * Never contains two contiguous underscores. + # + # assert "42_is/The answer!".to_cmangle == "_52d2_is_47dThe_32danswer_33d" + # assert "__".to_cmangle == "_95d_95d" + # assert "__d".to_cmangle == "_95d_d" + # assert "_d_".to_cmangle == "_d_95d" + # assert "_42".to_cmangle == "_95d42" + # assert "foo".to_cmangle == "foo" + # assert "".to_cmangle == "" fun to_cmangle: String do + if is_empty then return "" var res = new FlatBuffer var underscore = false - for i in [0..length[ do - var c = chars[i] + var start = 0 + var c = chars[0] + + if c >= '0' and c <= '9' then + res.add('_') + res.append(c.ascii.to_s) + res.add('d') + start = 1 + end + for i in [start..length[ do + c = chars[i] if (c >= 'a' and c <= 'z') or (c >='A' and c <= 'Z') then res.add(c) underscore = false @@ -454,6 +520,10 @@ abstract class Text underscore = false end end + if underscore then + res.append('_'.ascii.to_s) + res.add('d') + end return res.to_s end @@ -461,6 +531,14 @@ abstract class Text # # assert "abAB12<>&".escape_to_c == "abAB12<>&" # assert "\n\"'\\".escape_to_c == "\\n\\\"\\'\\\\" + # + # Most non-printable characters (bellow ASCII 32) are escaped to an octal form `\nnn`. + # Three digits are always used to avoid following digits to be interpreted as an element + # of the octal sequence. + # + # assert "{0.ascii}{1.ascii}{8.ascii}{31.ascii}{32.ascii}".escape_to_c == "\\000\\001\\010\\037 " + # + # The exceptions are the common `\t` and `\n`. fun escape_to_c: String do var b = new FlatBuffer @@ -468,8 +546,10 @@ abstract class Text var c = chars[i] if c == '\n' then b.append("\\n") + else if c == '\t' then + b.append("\\t") else if c == '\0' then - b.append("\\0") + b.append("\\000") else if c == '"' then b.append("\\\"") else if c == '\'' then @@ -477,7 +557,17 @@ abstract class Text else if c == '\\' then b.append("\\\\") else if c.ascii < 32 then - b.append("\\{c.ascii.to_base(8, false)}") + b.add('\\') + var oct = c.ascii.to_base(8, false) + # Force 3 octal digits since it is the + # maximum allowed in the C specification + if oct.length == 1 then + b.add('0') + b.add('0') + else if oct.length == 2 then + b.add('0') + end + b.append(oct) else b.add(c) end @@ -658,7 +748,7 @@ abstract class Text # assert "a&b-<>\"x\"/'".html_escape == "a&b-<>"x"/'" # # SEE: - fun html_escape: SELFTYPE + fun html_escape: String do var buf = new FlatBuffer @@ -774,6 +864,23 @@ abstract class FlatText # Real items, used as cache for to_cstring is called private var real_items: nullable NativeString = null + # Returns a char* starting at position `index_from` + # + # WARNING: If you choose to use this service, be careful of the following. + # + # Strings and NativeString are *ideally* always allocated through a Garbage Collector. + # Since the GC tracks the use of the pointer for the beginning of the char*, it may be + # deallocated at any moment, rendering the pointer returned by this function invalid. + # Any access to freed memory may very likely cause undefined behaviour or a crash. + # (Failure to do so will most certainly result in long and painful debugging hours) + # + # The only safe use of this pointer is if it is ephemeral (e.g. read in a C function + # then immediately return). + # + # As always, do not modify the content of the String in C code, if this is what you want + # copy locally the char* as Nit Strings are immutable. + private fun fast_cstring: NativeString is abstract + redef var length: Int = 0 redef fun output @@ -824,7 +931,7 @@ end abstract class String super Text - redef type SELFTYPE: String + redef type SELFTYPE: String is fixed redef fun to_s do return self @@ -999,7 +1106,7 @@ class FlatString # Indes in _items of the last item of the string private var index_to: Int is noinit - redef var chars: SequenceRead[Char] = new FlatStringCharView(self) + redef var chars: SequenceRead[Char] = new FlatStringCharView(self) is lazy redef fun [](index) do @@ -1029,6 +1136,8 @@ class FlatString return native.to_s_with_length(self.length) end + redef fun fast_cstring do return items.fast_cstring(index_from) + redef fun substring(from, count) do assert count >= 0 @@ -1039,15 +1148,19 @@ class FlatString from = 0 end - var realFrom = index_from + from + var new_from = index_from + from - if (realFrom + count) > index_to then return new FlatString.with_infos(items, index_to - realFrom + 1, realFrom, index_to) + if (new_from + count) > index_to then + var new_len = index_to - new_from + 1 + if new_len <= 0 then return empty + return new FlatString.with_infos(items, new_len, new_from, index_to) + end - if count == 0 then return empty + if count <= 0 then return empty - var to = realFrom + count - 1 + var to = new_from + count - 1 - return new FlatString.with_infos(items, to - realFrom + 1, realFrom, to) + return new FlatString.with_infos(items, to - new_from + 1, new_from, to) end redef fun empty do return "".as(FlatString) @@ -1106,10 +1219,14 @@ class FlatString # String Specific Methods # ################################################## - private init with_infos(items: NativeString, len: Int, from: Int, to: Int) + # Low-level creation of a new string with given data. + # + # `items` will be used as is, without copy, to retrieve the characters of the string. + # Aliasing issues is the responsibility of the caller. + private init with_infos(items: NativeString, length: Int, from: Int, to: Int) do self.items = items - length = len + self.length = length index_from = from index_to = to end @@ -1342,7 +1459,7 @@ end abstract class Buffer super Text - redef type SELFTYPE: Buffer + redef type SELFTYPE: Buffer is fixed # Specific implementations MUST set this to `true` in order to invalidate caches protected var is_dirty = true @@ -1465,12 +1582,12 @@ class FlatBuffer super FlatText super Buffer - redef type SELFTYPE: FlatBuffer - - redef var chars: Sequence[Char] = new FlatBufferCharView(self) + redef var chars: Sequence[Char] = new FlatBufferCharView(self) is lazy private var capacity: Int = 0 + redef fun fast_cstring do return items.fast_cstring(0) + redef fun substrings do return new FlatSubstringsIter(self) # Re-copies the `NativeString` into a new one and sets it as the new `Buffer` @@ -1555,6 +1672,20 @@ class FlatBuffer # Create a new empty string. init do end + # Low-level creation a new buffer with given data. + # + # `items` will be used as is, without copy, to store the characters of the buffer. + # Aliasing issues is the responsibility of the caller. + # + # If `items` is shared, `written` should be set to true after the creation + # so that a modification will do a copy-on-write. + private init with_infos(items: NativeString, capacity, length: Int) + do + self.items = items + self.length = length + self.capacity = capacity + end + # Create a new string copied from `s`. init from(s: Text) do @@ -1579,7 +1710,6 @@ class FlatBuffer init with_capacity(cap: Int) do assert cap >= 0 - # _items = new NativeString.calloc(cap) items = new NativeString(cap+1) capacity = cap length = 0 @@ -1623,11 +1753,10 @@ class FlatBuffer if from < 0 then from = 0 if count > length then count = length if from < count then - var r = new FlatBuffer.with_capacity(count - from) - while from < count do - r.chars.push(items[from]) - from += 1 - end + var len = count - from + var r_items = new NativeString(len) + items.copy_to(r_items, len, from, 0) + var r = new FlatBuffer.with_infos(r_items, len, len) return r else return new FlatBuffer @@ -1823,9 +1952,7 @@ end redef class Int # Wrapper of strerror C function - private fun strerror_ext: NativeString is extern `{ - return strerror(recv); - `} + private fun strerror_ext: NativeString is extern "strerror" # Returns a string describing error number fun strerror: String do return strerror_ext.to_s @@ -1854,15 +1981,26 @@ redef class Int end end + # C function to calculate the length of the `NativeString` to receive `self` + private fun int_to_s_len: Int is extern "native_int_length_str" + # C function to convert an nit Int to a NativeString (char*) - private fun native_int_to_s: NativeString is extern "native_int_to_s" + private fun native_int_to_s(nstr: NativeString, strlen: Int) is extern "native_int_to_s" # return displayable int in base 10 and signed # # assert 1.to_s == "1" # assert (-123).to_s == "-123" redef fun to_s do - return native_int_to_s.to_s + # Fast case for common numbers + if self == 0 then return "0" + if self == 1 then return "1" + + var nslen = int_to_s_len + var ns = new NativeString(nslen + 1) + ns[nslen] = '\0' + native_int_to_s(ns, nslen + 1) + return ns.to_s_with_length(nslen) end # return displayable int in hexadecimal @@ -1954,23 +2092,6 @@ redef class Float return p1 + "." + p2 end - - # `self` representation with `nb` digits after the '.'. - # - # assert 12.345.to_precision_native(1) == "12.3" - # assert 12.345.to_precision_native(2) == "12.35" - # assert 12.345.to_precision_native(3) == "12.345" - # assert 12.345.to_precision_native(4) == "12.3450" - fun to_precision_native(nb: Int): String import NativeString.to_s `{ - int size; - char *str; - - size = snprintf(NULL, 0, "%.*f", (int)nb, recv); - str = malloc(size + 1); - sprintf(str, "%.*f", (int)nb, recv ); - - return NativeString_to_s( str ); - `} end redef class Char @@ -2102,6 +2223,48 @@ redef class Array[E] end end +redef class NativeArray[E] + # Join all the elements using `to_s` + # + # REQUIRE: `self isa NativeArray[String]` + # REQUIRE: all elements are initialized + fun native_to_s: String + do + assert self isa NativeArray[String] + var l = length + var na = self + var i = 0 + var sl = 0 + var mypos = 0 + while i < l do + sl += na[i].length + i += 1 + mypos += 1 + end + var ns = new NativeString(sl + 1) + ns[sl] = '\0' + i = 0 + var off = 0 + while i < mypos do + var tmp = na[i] + var tpl = tmp.length + if tmp isa FlatString then + tmp.items.copy_to(ns, tpl, tmp.index_from, off) + off += tpl + else + for j in tmp.substrings do + var s = j.as(FlatString) + var slen = s.length + s.items.copy_to(ns, slen, s.index_from, off) + off += slen + end + end + i += 1 + end + return ns.to_s_with_length(sl) + end +end + redef class Map[K,V] # Concatenate couple of 'key value'. # key and value are separated by `couple_sep`. @@ -2121,7 +2284,7 @@ redef class Map[K,V] var i = iterator var k = i.key var e = i.item - s.append("{k}{couple_sep}{e or else ""}") + s.append("{k or else ""}{couple_sep}{e or else ""}") # Concat other items i.next @@ -2129,7 +2292,7 @@ redef class Map[K,V] s.append(sep) k = i.key e = i.item - s.append("{k}{couple_sep}{e or else ""}") + s.append("{k or else ""}{couple_sep}{e or else ""}") i.next end return s.to_s @@ -2145,6 +2308,12 @@ extern class NativeString `{ char* `} # Creates a new NativeString with a capacity of `length` new(length: Int) is intern + # Returns a char* starting at `index`. + # + # WARNING: Unsafe for extern code, use only for temporary + # pointer manipulation purposes (e.g. write to file or such) + fun fast_cstring(index: Int): NativeString is intern + # Get char at `index`. fun [](index: Int): Char is intern