X-Git-Url: http://nitlanguage.org diff --git a/lib/standard/string.nit b/lib/standard/string.nit index 489654e..aa757b7 100644 --- a/lib/standard/string.nit +++ b/lib/standard/string.nit @@ -60,7 +60,7 @@ abstract class Text fun substring(from: Int, count: Int): SELFTYPE is abstract # Iterates on the substrings of self if any - fun substrings: Iterator[Text] is abstract + fun substrings: Iterator[FlatText] is abstract # Is the current Text empty (== "") # @@ -345,7 +345,7 @@ abstract class Text end return true end - + # Removes the whitespaces at the beginning of self # # assert " \n\thello \n\t".l_trim == "hello \n\t" @@ -531,6 +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 @@ -538,8 +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 @@ -547,7 +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 @@ -829,6 +849,60 @@ abstract class Text return hash_cache.as(not null) end + # Gives the formatted string back as a Nit string with `args` in place + # + # assert "This %1 is a %2.".format("String", "formatted String") == "This String is a formatted String." + # assert "\\%1 This string".format("String") == "\\%1 This string" + fun format(args: Object...): String do + var s = new Array[Text] + var curr_st = 0 + var i = 0 + while i < length do + # Skip escaped characters + if self[i] == '\\' then + i += 1 + # In case of format + else if self[i] == '%' then + var fmt_st = i + i += 1 + var ciph_st = i + while i < length and self[i].is_numeric do + i += 1 + end + i -= 1 + var fmt_end = i + var ciph_len = fmt_end - ciph_st + 1 + s.push substring(curr_st, fmt_st - curr_st) + s.push args[substring(ciph_st, ciph_len).to_i - 1].to_s + curr_st = i + 1 + end + i += 1 + end + s.push substring(curr_st, length - curr_st) + return s.to_s + end + + # Copies `n` bytes from `self` at `src_offset` into `dest` starting at `dest_offset` + # + # Basically a high-level synonym of NativeString::copy_to + # + # REQUIRE: `n` must be large enough to contain `len` bytes + # + # var ns = new NativeString(8) + # "Text is String".copy_to_native(ns, 8, 2, 0) + # assert ns.to_s_with_length(8) == "xt is St" + # + fun copy_to_native(dest: NativeString, n, src_offset, dest_offset: Int) do + var mypos = src_offset + var itspos = dest_offset + while n > 0 do + dest[itspos] = self.chars[mypos] + itspos += 1 + mypos += 1 + n -= 1 + end + end + end # All kinds of array-based text representations. @@ -844,6 +918,23 @@ 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 @@ -856,6 +947,10 @@ abstract class FlatText end redef fun flatten do return self + + redef fun copy_to_native(dest, n, src_offset, dest_offset) do + items.copy_to(dest, n, src_offset, dest_offset) + end end # Abstract class for the SequenceRead compatible @@ -916,7 +1011,7 @@ abstract class String # assert "helloworld".insert_at(" ", 5) == "hello world" fun insert_at(s: String, pos: Int): SELFTYPE is abstract - redef fun substrings: Iterator[String] is abstract + redef fun substrings is abstract # Returns a reversed version of self # @@ -935,41 +1030,61 @@ abstract class String # assert "Hello World!".to_lower == "hello world!" fun to_lower : SELFTYPE is abstract - # Takes a camel case `self` and converts it to snake case + # Takes a camel case `self` and converts it to snake case # # assert "randomMethodId".to_snake_case == "random_method_id" # - # If `self` is upper, it is returned unchanged + # The rules are the following: # - # assert "RANDOM_METHOD_ID".to_snake_case == "RANDOM_METHOD_ID" + # An uppercase is always converted to a lowercase # - # If the identifier is prefixed by an underscore, the underscore is ignored + # assert "HELLO_WORLD".to_snake_case == "hello_world" + # + # An uppercase that follows a lowercase is prefixed with an underscore + # + # assert "HelloTheWORLD".to_snake_case == "hello_the_world" # - # assert "_privateField".to_snake_case == "_private_field" + # An uppercase that follows an uppercase and is followed by a lowercase, is prefixed with an underscore + # + # assert "HelloTHEWorld".to_snake_case == "hello_the_world" + # + # All other characters are kept as is; `self` does not need to be a proper CamelCased string. + # + # assert "=-_H3ll0Th3W0rld_-=".to_snake_case == "=-_h3ll0th3w0rld_-=" fun to_snake_case: SELFTYPE do - if self.is_upper then return self + if self.is_lower then return self var new_str = new FlatBuffer.with_capacity(self.length) - var is_first_char = true + var prev_is_lower = false + var prev_is_upper = false for i in [0..length[ do var char = chars[i] - if is_first_char then - new_str.add(char.to_lower) - is_first_char = false + if char.is_lower then + new_str.add(char) + prev_is_lower = true + prev_is_upper = false else if char.is_upper then - new_str.add('_') + if prev_is_lower then + new_str.add('_') + else if prev_is_upper and i+1 < length and chars[i+1].is_lower then + new_str.add('_') + end new_str.add(char.to_lower) + prev_is_lower = false + prev_is_upper = true else new_str.add(char) + prev_is_lower = false + prev_is_upper = false end end - + return new_str.to_s end - # Takes a snake case `self` and converts it to camel case + # Takes a snake case `self` and converts it to camel case # # assert "random_method_id".to_camel_case == "randomMethodId" # @@ -1069,7 +1184,7 @@ 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 @@ -1099,6 +1214,8 @@ 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 @@ -1180,10 +1297,14 @@ 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 @@ -1539,10 +1660,12 @@ class FlatBuffer super FlatText super Buffer - 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` @@ -1627,6 +1750,20 @@ 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 @@ -1651,7 +1788,6 @@ 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 @@ -1695,11 +1831,10 @@ 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 @@ -1895,9 +2030,7 @@ 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 @@ -1937,6 +2070,10 @@ redef class Int # assert 1.to_s == "1" # assert (-123).to_s == "-123" redef fun to_s do + # 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' @@ -2083,6 +2220,12 @@ redef class Collection[E] # Concatenate elements. redef fun to_s do + return plain_to_s + end + + # Concatenate element without separators + fun plain_to_s: String + do var s = new FlatBuffer for e in self do if e != null then s.append(e.to_s) return s.to_s @@ -2118,7 +2261,7 @@ end redef class Array[E] # Fast implementation - redef fun to_s + redef fun plain_to_s do var l = length if l == 0 then return "" @@ -2164,6 +2307,48 @@ 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`. @@ -2207,6 +2392,12 @@ 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