nitlanguage / nit.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
examples : adds a little example using "threaded" annotation
[nit.git] / lib / standard / string.nit
1 # This file is part of NIT ( http://www.nitlanguage.org ).
2 #
3 # Copyright 2004-2008 Jean Privat <jean@pryen.org>
4 # Copyright 2006-2008 Floréal Morandat <morandat@lirmm.fr>
5 #
6 # This file is free software, which comes along with NIT. This software is
7 # distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY;
8 # without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
9 # PARTICULAR PURPOSE. You can modify it is you want, provided this header
10 # is kept unaltered, and a notification of the changes is added.
11 # You are allowed to redistribute it and sell it, alone or is a part of
12 # another product.
13
14 # Basic manipulations of strings of characters
15 module string
16
17 import math
18 import collection
19 intrude import collection::array
20
21 `{
22 #include <stdio.h>
23 #include <string.h>
24 `}
25
26 ###############################################################################
27 # String #
28 ###############################################################################
29
30 # High-level abstraction for all text representations
31 abstract class Text
32 super Comparable
33
34 redef type OTHER: Text
35
36 # Type of self (used for factorization of several methods, ex : substring_from, empty...)
37 type SELFTYPE: Text
38
39 # Gets a view on the chars of the Text object
40 #
41 # assert "hello".chars.to_a == ['h', 'e', 'l', 'l', 'o']
42 fun chars: SequenceRead[Char] is abstract
43
44 # Number of characters contained in self.
45 #
46 # assert "12345".length == 5
47 # assert "".length == 0
48 fun length: Int is abstract
49
50 # Create a substring.
51 #
52 # assert "abcd".substring(1, 2) == "bc"
53 # assert "abcd".substring(-1, 2) == "a"
54 # assert "abcd".substring(1, 0) == ""
55 # assert "abcd".substring(2, 5) == "cd"
56 #
57 # A `from` index < 0 will be replaced by 0.
58 # Unless a `count` value is > 0 at the same time.
59 # In this case, `from += count` and `count -= from`.
60 fun substring(from: Int, count: Int): SELFTYPE is abstract
61
62 # Iterates on the substrings of self if any
63 fun substrings: Iterator[Text] is abstract
64
65 # Is the current Text empty (== "")
66 #
67 # assert "".is_empty
68 # assert not "foo".is_empty
69 fun is_empty: Bool do return self.length == 0
70
71 # Returns an empty Text of the right type
72 #
73 # This method is used internally to get the right
74 # implementation of an empty string.
75 protected fun empty: SELFTYPE is abstract
76
77 # Gets the first char of the Text
78 #
79 # DEPRECATED : Use self.chars.first instead
80 fun first: Char do return self.chars[0]
81
82 # Access a character at `index` in the string.
83 #
84 # assert "abcd"[2] == 'c'
85 #
86 # DEPRECATED : Use self.chars.[] instead
87 fun [](index: Int): Char do return self.chars[index]
88
89 # Gets the index of the first occurence of 'c'
90 #
91 # Returns -1 if not found
92 #
93 # DEPRECATED : Use self.chars.index_of instead
94 fun index_of(c: Char): Int
95 do
96 return index_of_from(c, 0)
97 end
98
99 # Gets the last char of self
100 #
101 # DEPRECATED : Use self.chars.last instead
102 fun last: Char do return self.chars[length-1]
103
104 # Gets the index of the first occurence of ´c´ starting from ´pos´
105 #
106 # Returns -1 if not found
107 #
108 # DEPRECATED : Use self.chars.index_of_from instead
109 fun index_of_from(c: Char, pos: Int): Int
110 do
111 var iter = self.chars.iterator_from(pos)
112 while iter.is_ok do
113 if iter.item == c then return iter.index
114 iter.next
115 end
116 return -1
117 end
118
119 # Gets the last index of char ´c´
120 #
121 # Returns -1 if not found
122 #
123 # DEPRECATED : Use self.chars.last_index_of instead
124 fun last_index_of(c: Char): Int
125 do
126 return last_index_of_from(c, length - 1)
127 end
128
129 # Return a null terminated char *
130 fun to_cstring: NativeString do return flatten.to_cstring
131
132 # The index of the last occurrence of an element starting from pos (in reverse order).
133 #
134 # var s = "/etc/bin/test/test.nit"
135 # assert s.last_index_of_from('/', s.length-1) == 13
136 # assert s.last_index_of_from('/', 12) == 8
137 #
138 # Returns -1 if not found
139 #
140 # DEPRECATED : Use self.chars.last_index_of_from instead
141 fun last_index_of_from(item: Char, pos: Int): Int
142 do
143 var iter = self.chars.reverse_iterator_from(pos)
144 while iter.is_ok do
145 if iter.item == item then return iter.index
146 iter.next
147 end
148 return -1
149 end
150
151 # Gets an iterator on the chars of self
152 #
153 # DEPRECATED : Use self.chars.iterator instead
154 fun iterator: Iterator[Char]
155 do
156 return self.chars.iterator
157 end
158
159
160 # Gets an Array containing the chars of self
161 #
162 # DEPRECATED : Use self.chars.to_a instead
163 fun to_a: Array[Char] do return chars.to_a
164
165 # Create a substring from `self` beginning at the `from` position
166 #
167 # assert "abcd".substring_from(1) == "bcd"
168 # assert "abcd".substring_from(-1) == "abcd"
169 # assert "abcd".substring_from(2) == "cd"
170 #
171 # As with substring, a `from` index < 0 will be replaced by 0
172 fun substring_from(from: Int): SELFTYPE
173 do
174 if from >= self.length then return empty
175 if from < 0 then from = 0
176 return substring(from, length - from)
177 end
178
179 # Does self have a substring `str` starting from position `pos`?
180 #
181 # assert "abcd".has_substring("bc",1) == true
182 # assert "abcd".has_substring("bc",2) == false
183 #
184 # Returns true iff all characters of `str` are presents
185 # at the expected index in `self.`
186 # The first character of `str` being at `pos`, the second
187 # character being at `pos+1` and so on...
188 #
189 # This means that all characters of `str` need to be inside `self`.
190 #
191 # assert "abcd".has_substring("xab", -1) == false
192 # assert "abcd".has_substring("cdx", 2) == false
193 #
194 # And that the empty string is always a valid substring.
195 #
196 # assert "abcd".has_substring("", 2) == true
197 # assert "abcd".has_substring("", 200) == true
198 fun has_substring(str: String, pos: Int): Bool
199 do
200 if str.is_empty then return true
201 if pos < 0 or pos + str.length > length then return false
202 var myiter = self.chars.iterator_from(pos)
203 var itsiter = str.chars.iterator
204 while myiter.is_ok and itsiter.is_ok do
205 if myiter.item != itsiter.item then return false
206 myiter.next
207 itsiter.next
208 end
209 if itsiter.is_ok then return false
210 return true
211 end
212
213 # Is this string prefixed by `prefix`?
214 #
215 # assert "abcd".has_prefix("ab") == true
216 # assert "abcbc".has_prefix("bc") == false
217 # assert "ab".has_prefix("abcd") == false
218 fun has_prefix(prefix: String): Bool do return has_substring(prefix,0)
219
220 # Is this string suffixed by `suffix`?
221 #
222 # assert "abcd".has_suffix("abc") == false
223 # assert "abcd".has_suffix("bcd") == true
224 fun has_suffix(suffix: String): Bool do return has_substring(suffix, length - suffix.length)
225
226 # If `self` contains only digits, return the corresponding integer
227 #
228 # assert "123".to_i == 123
229 # assert "-1".to_i == -1
230 fun to_i: Int
231 do
232 # Shortcut
233 return to_s.to_cstring.atoi
234 end
235
236 # If `self` contains a float, return the corresponding float
237 #
238 # assert "123".to_f == 123.0
239 # assert "-1".to_f == -1.0
240 # assert "-1.2e-3".to_f == -0.0012
241 fun to_f: Float
242 do
243 # Shortcut
244 return to_s.to_cstring.atof
245 end
246
247 # If `self` contains only digits and alpha <= 'f', return the corresponding integer.
248 #
249 # assert "ff".to_hex == 255
250 fun to_hex: Int do return a_to(16)
251
252 # If `self` contains only digits and letters, return the corresponding integer in a given base
253 #
254 # assert "120".a_to(3) == 15
255 fun a_to(base: Int) : Int
256 do
257 var i = 0
258 var neg = false
259
260 for j in [0..length[ do
261 var c = chars[j]
262 var v = c.to_i
263 if v > base then
264 if neg then
265 return -i
266 else
267 return i
268 end
269 else if v < 0 then
270 neg = true
271 else
272 i = i * base + v
273 end
274 end
275 if neg then
276 return -i
277 else
278 return i
279 end
280 end
281
282 # Returns `true` if the string contains only Numeric values (and one "," or one "." character)
283 #
284 # assert "123".is_numeric == true
285 # assert "1.2".is_numeric == true
286 # assert "1,2".is_numeric == true
287 # assert "1..2".is_numeric == false
288 fun is_numeric: Bool
289 do
290 var has_point_or_comma = false
291 for i in [0..length[ do
292 var c = chars[i]
293 if not c.is_numeric then
294 if (c == '.' or c == ',') and not has_point_or_comma then
295 has_point_or_comma = true
296 else
297 return false
298 end
299 end
300 end
301 return true
302 end
303
304 # Returns `true` if the string contains only Hex chars
305 #
306 # assert "048bf".is_hex == true
307 # assert "ABCDEF".is_hex == true
308 # assert "0G".is_hex == false
309 fun is_hex: Bool
310 do
311 for i in [0..length[ do
312 var c = chars[i]
313 if not (c >= 'a' and c <= 'f') and
314 not (c >= 'A' and c <= 'F') and
315 not (c >= '0' and c <= '9') then return false
316 end
317 return true
318 end
319
320 # Are all letters in `self` upper-case ?
321 #
322 # assert "HELLO WORLD".is_upper == true
323 # assert "%$&%!".is_upper == true
324 # assert "hello world".is_upper == false
325 # assert "Hello World".is_upper == false
326 fun is_upper: Bool
327 do
328 for i in [0..length[ do
329 var char = chars[i]
330 if char.is_lower then return false
331 end
332 return true
333 end
334
335 # Are all letters in `self` lower-case ?
336 #
337 # assert "hello world".is_lower == true
338 # assert "%$&%!".is_lower == true
339 # assert "Hello World".is_lower == false
340 fun is_lower: Bool
341 do
342 for i in [0..length[ do
343 var char = chars[i]
344 if char.is_upper then return false
345 end
346 return true
347 end
348
349 # Removes the whitespaces at the beginning of self
350 #
351 # assert " \n\thello \n\t".l_trim == "hello \n\t"
352 #
353 # `Char::is_whitespace` determines what is a whitespace.
354 fun l_trim: SELFTYPE
355 do
356 var iter = self.chars.iterator
357 while iter.is_ok do
358 if not iter.item.is_whitespace then break
359 iter.next
360 end
361 if iter.index == length then return self.empty
362 return self.substring_from(iter.index)
363 end
364
365 # Removes the whitespaces at the end of self
366 #
367 # assert " \n\thello \n\t".r_trim == " \n\thello"
368 #
369 # `Char::is_whitespace` determines what is a whitespace.
370 fun r_trim: SELFTYPE
371 do
372 var iter = self.chars.reverse_iterator
373 while iter.is_ok do
374 if not iter.item.is_whitespace then break
375 iter.next
376 end
377 if iter.index < 0 then return self.empty
378 return self.substring(0, iter.index + 1)
379 end
380
381 # Trims trailing and preceding white spaces
382 #
383 # assert " Hello World ! ".trim == "Hello World !"
384 # assert "\na\nb\tc\t".trim == "a\nb\tc"
385 #
386 # `Char::is_whitespace` determines what is a whitespace.
387 fun trim: SELFTYPE do return (self.l_trim).r_trim
388
389 # Is the string non-empty but only made of whitespaces?
390 #
391 # assert " \n\t ".is_whitespace == true
392 # assert " hello ".is_whitespace == false
393 # assert "".is_whitespace == false
394 #
395 # `Char::is_whitespace` determines what is a whitespace.
396 fun is_whitespace: Bool
397 do
398 if is_empty then return false
399 for c in self.chars do
400 if not c.is_whitespace then return false
401 end
402 return true
403 end
404
405 # Returns `self` removed from its last line terminator (if any).
406 #
407 # assert "Hello\n".chomp == "Hello"
408 # assert "Hello".chomp == "Hello"
409 #
410 # assert "\n".chomp == ""
411 # assert "".chomp == ""
412 #
413 # Line terminators are `"\n"`, `"\r\n"` and `"\r"`.
414 # A single line terminator, the last one, is removed.
415 #
416 # assert "\r\n".chomp == ""
417 # assert "\r\n\n".chomp == "\r\n"
418 # assert "\r\n\r\n".chomp == "\r\n"
419 # assert "\r\n\r".chomp == "\r\n"
420 #
421 # Note: unlike with most IO methods like `Reader::read_line`,
422 # a single `\r` is considered here to be a line terminator and will be removed.
423 fun chomp: SELFTYPE
424 do
425 var len = length
426 if len == 0 then return self
427 var l = self.chars.last
428 if l == '\r' then
429 return substring(0, len-1)
430 else if l != '\n' then
431 return self
432 else if len > 1 and self.chars[len-2] == '\r' then
433 return substring(0, len-2)
434 else
435 return substring(0, len-1)
436 end
437 end
438
439 # Justify a self in a space of `length`
440 #
441 # `left` is the space ratio on the left side.
442 # * 0.0 for left-justified (no space at the left)
443 # * 1.0 for right-justified (all spaces at the left)
444 # * 0.5 for centered (half the spaces at the left)
445 #
446 # Examples
447 #
448 # assert "hello".justify(10, 0.0) == "hello "
449 # assert "hello".justify(10, 1.0) == " hello"
450 # assert "hello".justify(10, 0.5) == " hello "
451 #
452 # If `length` is not enough, `self` is returned as is.
453 #
454 # assert "hello".justify(2, 0.0) == "hello"
455 #
456 # REQUIRE: `left >= 0.0 and left <= 1.0`
457 # ENSURE: `self.length <= length implies result.length == length`
458 # ENSURE: `self.length >= length implies result == self`
459 fun justify(length: Int, left: Float): String
460 do
461 var diff = length - self.length
462 if diff <= 0 then return to_s
463 assert left >= 0.0 and left <= 1.0
464 var before = (diff.to_f * left).to_i
465 return " " * before + self + " " * (diff-before)
466 end
467
468 # Mangle a string to be a unique string only made of alphanumeric characters and underscores.
469 #
470 # This method is injective (two different inputs never produce the same
471 # output) and the returned string always respect the following rules:
472 #
473 # * Contains only US-ASCII letters, digits and underscores.
474 # * Never starts with a digit.
475 # * Never ends with an underscore.
476 # * Never contains two contiguous underscores.
477 #
478 # assert "42_is/The answer!".to_cmangle == "_52d2_is_47dThe_32danswer_33d"
479 # assert "__".to_cmangle == "_95d_95d"
480 # assert "__d".to_cmangle == "_95d_d"
481 # assert "_d_".to_cmangle == "_d_95d"
482 # assert "_42".to_cmangle == "_95d42"
483 # assert "foo".to_cmangle == "foo"
484 # assert "".to_cmangle == ""
485 fun to_cmangle: String
486 do
487 if is_empty then return ""
488 var res = new FlatBuffer
489 var underscore = false
490 var start = 0
491 var c = chars[0]
492
493 if c >= '0' and c <= '9' then
494 res.add('_')
495 res.append(c.ascii.to_s)
496 res.add('d')
497 start = 1
498 end
499 for i in [start..length[ do
500 c = chars[i]
501 if (c >= 'a' and c <= 'z') or (c >='A' and c <= 'Z') then
502 res.add(c)
503 underscore = false
504 continue
505 end
506 if underscore then
507 res.append('_'.ascii.to_s)
508 res.add('d')
509 end
510 if c >= '0' and c <= '9' then
511 res.add(c)
512 underscore = false
513 else if c == '_' then
514 res.add(c)
515 underscore = true
516 else
517 res.add('_')
518 res.append(c.ascii.to_s)
519 res.add('d')
520 underscore = false
521 end
522 end
523 if underscore then
524 res.append('_'.ascii.to_s)
525 res.add('d')
526 end
527 return res.to_s
528 end
529
530 # Escape " \ ' and non printable characters using the rules of literal C strings and characters
531 #
532 # assert "abAB12<>&".escape_to_c == "abAB12<>&"
533 # assert "\n\"'\\".escape_to_c == "\\n\\\"\\'\\\\"
534 #
535 # Most non-printable characters (bellow ASCII 32) are escaped to an octal form `\nnn`.
536 # Three digits are always used to avoid following digits to be interpreted as an element
537 # of the octal sequence.
538 #
539 # assert "{0.ascii}{1.ascii}{8.ascii}{31.ascii}{32.ascii}".escape_to_c == "\\000\\001\\010\\037 "
540 #
541 # The exceptions are the common `\t` and `\n`.
542 fun escape_to_c: String
543 do
544 var b = new FlatBuffer
545 for i in [0..length[ do
546 var c = chars[i]
547 if c == '\n' then
548 b.append("\\n")
549 else if c == '\t' then
550 b.append("\\t")
551 else if c == '\0' then
552 b.append("\\000")
553 else if c == '"' then
554 b.append("\\\"")
555 else if c == '\'' then
556 b.append("\\\'")
557 else if c == '\\' then
558 b.append("\\\\")
559 else if c.ascii < 32 then
560 b.add('\\')
561 var oct = c.ascii.to_base(8, false)
562 # Force 3 octal digits since it is the
563 # maximum allowed in the C specification
564 if oct.length == 1 then
565 b.add('0')
566 b.add('0')
567 else if oct.length == 2 then
568 b.add('0')
569 end
570 b.append(oct)
571 else
572 b.add(c)
573 end
574 end
575 return b.to_s
576 end
577
578 # Escape additionnal characters
579 # The result might no be legal in C but be used in other languages
580 #
581 # assert "ab|\{\}".escape_more_to_c("|\{\}") == "ab\\|\\\{\\\}"
582 fun escape_more_to_c(chars: String): String
583 do
584 var b = new FlatBuffer
585 for c in escape_to_c.chars do
586 if chars.chars.has(c) then
587 b.add('\\')
588 end
589 b.add(c)
590 end
591 return b.to_s
592 end
593
594 # Escape to C plus braces
595 #
596 # assert "\n\"'\\\{\}".escape_to_nit == "\\n\\\"\\'\\\\\\\{\\\}"
597 fun escape_to_nit: String do return escape_more_to_c("\{\}")
598
599 # Escape to POSIX Shell (sh).
600 #
601 # Abort if the text contains a null byte.
602 #
603 # assert "\n\"'\\\{\}0".escape_to_sh == "'\n\"'\\''\\\{\}0'"
604 fun escape_to_sh: String do
605 var b = new FlatBuffer
606 b.chars.add '\''
607 for i in [0..length[ do
608 var c = chars[i]
609 if c == '\'' then
610 b.append("'\\''")
611 else
612 assert without_null_byte: c != '\0'
613 b.add(c)
614 end
615 end
616 b.chars.add '\''
617 return b.to_s
618 end
619
620 # Escape to include in a Makefile
621 #
622 # Unfortunately, some characters are not escapable in Makefile.
623 # These characters are `;`, `|`, `\`, and the non-printable ones.
624 # They will be rendered as `"?{hex}"`.
625 fun escape_to_mk: String do
626 var b = new FlatBuffer
627 for i in [0..length[ do
628 var c = chars[i]
629 if c == '$' then
630 b.append("$$")
631 else if c == ':' or c == ' ' or c == '#' then
632 b.add('\\')
633 b.add(c)
634 else if c.ascii < 32 or c == ';' or c == '|' or c == '\\' or c == '=' then
635 b.append("?{c.ascii.to_base(16, false)}")
636 else
637 b.add(c)
638 end
639 end
640 return b.to_s
641 end
642
643 # Return a string where Nit escape sequences are transformed.
644 #
645 # var s = "\\n"
646 # assert s.length == 2
647 # var u = s.unescape_nit
648 # assert u.length == 1
649 # assert u.chars[0].ascii == 10 # (the ASCII value of the "new line" character)
650 fun unescape_nit: String
651 do
652 var res = new FlatBuffer.with_capacity(self.length)
653 var was_slash = false
654 for i in [0..length[ do
655 var c = chars[i]
656 if not was_slash then
657 if c == '\\' then
658 was_slash = true
659 else
660 res.add(c)
661 end
662 continue
663 end
664 was_slash = false
665 if c == 'n' then
666 res.add('\n')
667 else if c == 'r' then
668 res.add('\r')
669 else if c == 't' then
670 res.add('\t')
671 else if c == '0' then
672 res.add('\0')
673 else
674 res.add(c)
675 end
676 end
677 return res.to_s
678 end
679
680 # Encode `self` to percent (or URL) encoding
681 #
682 # assert "aBc09-._~".to_percent_encoding == "aBc09-._~"
683 # assert "%()< >".to_percent_encoding == "%25%28%29%3c%20%3e"
684 # assert ".com/post?e=asdf&f=123".to_percent_encoding == ".com%2fpost%3fe%3dasdf%26f%3d123"
685 fun to_percent_encoding: String
686 do
687 var buf = new FlatBuffer
688
689 for i in [0..length[ do
690 var c = chars[i]
691 if (c >= '0' and c <= '9') or
692 (c >= 'a' and c <= 'z') or
693 (c >= 'A' and c <= 'Z') or
694 c == '-' or c == '.' or
695 c == '_' or c == '~'
696 then
697 buf.add c
698 else buf.append "%{c.ascii.to_hex}"
699 end
700
701 return buf.to_s
702 end
703
704 # Decode `self` from percent (or URL) encoding to a clear string
705 #
706 # Replace invalid use of '%' with '?'.
707 #
708 # assert "aBc09-._~".from_percent_encoding == "aBc09-._~"
709 # assert "%25%28%29%3c%20%3e".from_percent_encoding == "%()< >"
710 # assert ".com%2fpost%3fe%3dasdf%26f%3d123".from_percent_encoding == ".com/post?e=asdf&f=123"
711 # assert "%25%28%29%3C%20%3E".from_percent_encoding == "%()< >"
712 # assert "incomplete %".from_percent_encoding == "incomplete ?"
713 # assert "invalid % usage".from_percent_encoding == "invalid ? usage"
714 fun from_percent_encoding: String
715 do
716 var buf = new FlatBuffer
717
718 var i = 0
719 while i < length do
720 var c = chars[i]
721 if c == '%' then
722 if i + 2 >= length then
723 # What follows % has been cut off
724 buf.add '?'
725 else
726 i += 1
727 var hex_s = substring(i, 2)
728 if hex_s.is_hex then
729 var hex_i = hex_s.to_hex
730 buf.add hex_i.ascii
731 i += 1
732 else
733 # What follows a % is not Hex
734 buf.add '?'
735 i -= 1
736 end
737 end
738 else buf.add c
739
740 i += 1
741 end
742
743 return buf.to_s
744 end
745
746 # Escape the characters `<`, `>`, `&`, `"`, `'` and `/` as HTML/XML entity references.
747 #
748 # assert "a&b-<>\"x\"/'".html_escape == "a&amp;b-&lt;&gt;&#34;x&#34;&#47;&#39;"
749 #
750 # SEE: <https://www.owasp.org/index.php/XSS_%28Cross_Site_Scripting%29_Prevention_Cheat_Sheet#RULE_.231_-_HTML_Escape_Before_Inserting_Untrusted_Data_into_HTML_Element_Content>
751 fun html_escape: String
752 do
753 var buf = new FlatBuffer
754
755 for i in [0..length[ do
756 var c = chars[i]
757 if c == '&' then
758 buf.append "&amp;"
759 else if c == '<' then
760 buf.append "&lt;"
761 else if c == '>' then
762 buf.append "&gt;"
763 else if c == '"' then
764 buf.append "&#34;"
765 else if c == '\'' then
766 buf.append "&#39;"
767 else if c == '/' then
768 buf.append "&#47;"
769 else buf.add c
770 end
771
772 return buf.to_s
773 end
774
775 # Equality of text
776 # Two pieces of text are equals if thez have the same characters in the same order.
777 #
778 # assert "hello" == "hello"
779 # assert "hello" != "HELLO"
780 # assert "hello" == "hel"+"lo"
781 #
782 # Things that are not Text are not equal.
783 #
784 # assert "9" != '9'
785 # assert "9" != ['9']
786 # assert "9" != 9
787 #
788 # assert "9".chars.first == '9' # equality of Char
789 # assert "9".chars == ['9'] # equality of Sequence
790 # assert "9".to_i == 9 # equality of Int
791 redef fun ==(o)
792 do
793 if o == null then return false
794 if not o isa Text then return false
795 if self.is_same_instance(o) then return true
796 if self.length != o.length then return false
797 return self.chars == o.chars
798 end
799
800 # Lexicographical comparaison
801 #
802 # assert "abc" < "xy"
803 # assert "ABC" < "abc"
804 redef fun <(other)
805 do
806 var self_chars = self.chars.iterator
807 var other_chars = other.chars.iterator
808
809 while self_chars.is_ok and other_chars.is_ok do
810 if self_chars.item < other_chars.item then return true
811 if self_chars.item > other_chars.item then return false
812 self_chars.next
813 other_chars.next
814 end
815
816 if self_chars.is_ok then
817 return false
818 else
819 return true
820 end
821 end
822
823 # Escape string used in labels for graphviz
824 #
825 # assert ">><<".escape_to_dot == "\\>\\>\\<\\<"
826 fun escape_to_dot: String
827 do
828 return escape_more_to_c("|\{\}<>")
829 end
830
831 # Flat representation of self
832 fun flatten: FlatText is abstract
833
834 private var hash_cache: nullable Int = null
835
836 redef fun hash
837 do
838 if hash_cache == null then
839 # djb2 hash algorithm
840 var h = 5381
841
842 for i in [0..length[ do
843 var char = chars[i]
844 h = h.lshift(5) + h + char.ascii
845 end
846
847 hash_cache = h
848 end
849 return hash_cache.as(not null)
850 end
851
852 end
853
854 # All kinds of array-based text representations.
855 abstract class FlatText
856 super Text
857
858 # Underlying C-String (`char*`)
859 #
860 # Warning : Might be void in some subclasses, be sure to check
861 # if set before using it.
862 private var items: NativeString is noinit
863
864 # Real items, used as cache for to_cstring is called
865 private var real_items: nullable NativeString = null
866
867 redef var length: Int = 0
868
869 redef fun output
870 do
871 var i = 0
872 while i < length do
873 items[i].output
874 i += 1
875 end
876 end
877
878 redef fun flatten do return self
879 end
880
881 # Abstract class for the SequenceRead compatible
882 # views on String and Buffer objects
883 private abstract class StringCharView
884 super SequenceRead[Char]
885
886 type SELFTYPE: Text
887
888 var target: SELFTYPE
889
890 redef fun is_empty do return target.is_empty
891
892 redef fun length do return target.length
893
894 redef fun iterator: IndexedIterator[Char] do return self.iterator_from(0)
895
896 redef fun reverse_iterator do return self.reverse_iterator_from(self.length - 1)
897 end
898
899 # View on Buffer objects, extends Sequence
900 # for mutation operations
901 private abstract class BufferCharView
902 super StringCharView
903 super Sequence[Char]
904
905 redef type SELFTYPE: Buffer
906
907 end
908
909 # A `String` holds and manipulates an arbitrary sequence of characters.
910 #
911 # String objects may be created using literals.
912 #
913 # assert "Hello World!" isa String
914 abstract class String
915 super Text
916
917 redef type SELFTYPE: String is fixed
918
919 redef fun to_s do return self
920
921 # Concatenates `o` to `self`
922 #
923 # assert "hello" + "world" == "helloworld"
924 # assert "" + "hello" + "" == "hello"
925 fun +(o: Text): SELFTYPE is abstract
926
927 # Concatenates self `i` times
928 #
929 # assert "abc" * 4 == "abcabcabcabc"
930 # assert "abc" * 1 == "abc"
931 # assert "abc" * 0 == ""
932 fun *(i: Int): SELFTYPE is abstract
933
934 # Insert `s` at `pos`.
935 #
936 # assert "helloworld".insert_at(" ", 5) == "hello world"
937 fun insert_at(s: String, pos: Int): SELFTYPE is abstract
938
939 redef fun substrings: Iterator[String] is abstract
940
941 # Returns a reversed version of self
942 #
943 # assert "hello".reversed == "olleh"
944 # assert "bob".reversed == "bob"
945 # assert "".reversed == ""
946 fun reversed: SELFTYPE is abstract
947
948 # A upper case version of `self`
949 #
950 # assert "Hello World!".to_upper == "HELLO WORLD!"
951 fun to_upper: SELFTYPE is abstract
952
953 # A lower case version of `self`
954 #
955 # assert "Hello World!".to_lower == "hello world!"
956 fun to_lower : SELFTYPE is abstract
957
958 # Takes a camel case `self` and converts it to snake case
959 #
960 # assert "randomMethodId".to_snake_case == "random_method_id"
961 #
962 # If `self` is upper, it is returned unchanged
963 #
964 # assert "RANDOM_METHOD_ID".to_snake_case == "RANDOM_METHOD_ID"
965 #
966 # If the identifier is prefixed by an underscore, the underscore is ignored
967 #
968 # assert "_privateField".to_snake_case == "_private_field"
969 fun to_snake_case: SELFTYPE
970 do
971 if self.is_upper then return self
972
973 var new_str = new FlatBuffer.with_capacity(self.length)
974 var is_first_char = true
975
976 for i in [0..length[ do
977 var char = chars[i]
978 if is_first_char then
979 new_str.add(char.to_lower)
980 is_first_char = false
981 else if char.is_upper then
982 new_str.add('_')
983 new_str.add(char.to_lower)
984 else
985 new_str.add(char)
986 end
987 end
988
989 return new_str.to_s
990 end
991
992 # Takes a snake case `self` and converts it to camel case
993 #
994 # assert "random_method_id".to_camel_case == "randomMethodId"
995 #
996 # If the identifier is prefixed by an underscore, the underscore is ignored
997 #
998 # assert "_private_field".to_camel_case == "_privateField"
999 #
1000 # If `self` is upper, it is returned unchanged
1001 #
1002 # assert "RANDOM_ID".to_camel_case == "RANDOM_ID"
1003 #
1004 # If there are several consecutive underscores, they are considered as a single one
1005 #
1006 # assert "random__method_id".to_camel_case == "randomMethodId"
1007 fun to_camel_case: SELFTYPE
1008 do
1009 if self.is_upper then return self
1010
1011 var new_str = new FlatBuffer
1012 var is_first_char = true
1013 var follows_us = false
1014
1015 for i in [0..length[ do
1016 var char = chars[i]
1017 if is_first_char then
1018 new_str.add(char)
1019 is_first_char = false
1020 else if char == '_' then
1021 follows_us = true
1022 else if follows_us then
1023 new_str.add(char.to_upper)
1024 follows_us = false
1025 else
1026 new_str.add(char)
1027 end
1028 end
1029
1030 return new_str.to_s
1031 end
1032
1033 # Returns a capitalized `self`
1034 #
1035 # Letters that follow a letter are lowercased
1036 # Letters that follow a non-letter are upcased.
1037 #
1038 # SEE : `Char::is_letter` for the definition of letter.
1039 #
1040 # assert "jAVASCRIPT".capitalized == "Javascript"
1041 # assert "i am root".capitalized == "I Am Root"
1042 # assert "ab_c -ab0c ab\nc".capitalized == "Ab_C -Ab0C Ab\nC"
1043 fun capitalized: SELFTYPE do
1044 if length == 0 then return self
1045
1046 var buf = new FlatBuffer.with_capacity(length)
1047
1048 var curr = chars[0].to_upper
1049 var prev = curr
1050 buf[0] = curr
1051
1052 for i in [1 .. length[ do
1053 prev = curr
1054 curr = self[i]
1055 if prev.is_letter then
1056 buf[i] = curr.to_lower
1057 else
1058 buf[i] = curr.to_upper
1059 end
1060 end
1061
1062 return buf.to_s
1063 end
1064 end
1065
1066 private class FlatSubstringsIter
1067 super Iterator[FlatText]
1068
1069 var tgt: nullable FlatText
1070
1071 redef fun item do
1072 assert is_ok
1073 return tgt.as(not null)
1074 end
1075
1076 redef fun is_ok do return tgt != null
1077
1078 redef fun next do tgt = null
1079 end
1080
1081 # Immutable strings of characters.
1082 class FlatString
1083 super FlatText
1084 super String
1085
1086 # Index in _items of the start of the string
1087 private var index_from: Int is noinit
1088
1089 # Indes in _items of the last item of the string
1090 private var index_to: Int is noinit
1091
1092 redef var chars: SequenceRead[Char] = new FlatStringCharView(self) is lazy
1093
1094 redef fun [](index)
1095 do
1096 # Check that the index (+ index_from) is not larger than indexTo
1097 # In other terms, if the index is valid
1098 assert index >= 0
1099 assert (index + index_from) <= index_to
1100 return items[index + index_from]
1101 end
1102
1103 ################################################
1104 # AbstractString specific methods #
1105 ################################################
1106
1107 redef fun reversed
1108 do
1109 var native = new NativeString(self.length + 1)
1110 var length = self.length
1111 var items = self.items
1112 var pos = 0
1113 var ipos = length-1
1114 while pos < length do
1115 native[pos] = items[ipos]
1116 pos += 1
1117 ipos -= 1
1118 end
1119 return native.to_s_with_length(self.length)
1120 end
1121
1122 redef fun substring(from, count)
1123 do
1124 assert count >= 0
1125
1126 if from < 0 then
1127 count += from
1128 if count < 0 then count = 0
1129 from = 0
1130 end
1131
1132 var new_from = index_from + from
1133
1134 if (new_from + count) > index_to then
1135 var new_len = index_to - new_from + 1
1136 if new_len <= 0 then return empty
1137 return new FlatString.with_infos(items, new_len, new_from, index_to)
1138 end
1139
1140 if count <= 0 then return empty
1141
1142 var to = new_from + count - 1
1143
1144 return new FlatString.with_infos(items, to - new_from + 1, new_from, to)
1145 end
1146
1147 redef fun empty do return "".as(FlatString)
1148
1149 redef fun to_upper
1150 do
1151 var outstr = new NativeString(self.length + 1)
1152 var out_index = 0
1153
1154 var myitems = self.items
1155 var index_from = self.index_from
1156 var max = self.index_to
1157
1158 while index_from <= max do
1159 outstr[out_index] = myitems[index_from].to_upper
1160 out_index += 1
1161 index_from += 1
1162 end
1163
1164 outstr[self.length] = '\0'
1165
1166 return outstr.to_s_with_length(self.length)
1167 end
1168
1169 redef fun to_lower
1170 do
1171 var outstr = new NativeString(self.length + 1)
1172 var out_index = 0
1173
1174 var myitems = self.items
1175 var index_from = self.index_from
1176 var max = self.index_to
1177
1178 while index_from <= max do
1179 outstr[out_index] = myitems[index_from].to_lower
1180 out_index += 1
1181 index_from += 1
1182 end
1183
1184 outstr[self.length] = '\0'
1185
1186 return outstr.to_s_with_length(self.length)
1187 end
1188
1189 redef fun output
1190 do
1191 var i = self.index_from
1192 var imax = self.index_to
1193 while i <= imax do
1194 items[i].output
1195 i += 1
1196 end
1197 end
1198
1199 ##################################################
1200 # String Specific Methods #
1201 ##################################################
1202
1203 # Low-level creation of a new string with given data.
1204 #
1205 # `items` will be used as is, without copy, to retrieve the characters of the string.
1206 # Aliasing issues is the responsibility of the caller.
1207 private init with_infos(items: NativeString, length: Int, from: Int, to: Int)
1208 do
1209 self.items = items
1210 self.length = length
1211 index_from = from
1212 index_to = to
1213 end
1214
1215 redef fun to_cstring: NativeString
1216 do
1217 if real_items != null then
1218 return real_items.as(not null)
1219 else
1220 var newItems = new NativeString(length + 1)
1221 self.items.copy_to(newItems, length, index_from, 0)
1222 newItems[length] = '\0'
1223 self.real_items = newItems
1224 return newItems
1225 end
1226 end
1227
1228 redef fun ==(other)
1229 do
1230 if not other isa FlatString then return super
1231
1232 if self.object_id == other.object_id then return true
1233
1234 var my_length = length
1235
1236 if other.length != my_length then return false
1237
1238 var my_index = index_from
1239 var its_index = other.index_from
1240
1241 var last_iteration = my_index + my_length
1242
1243 var itsitems = other.items
1244 var myitems = self.items
1245
1246 while my_index < last_iteration do
1247 if myitems[my_index] != itsitems[its_index] then return false
1248 my_index += 1
1249 its_index += 1
1250 end
1251
1252 return true
1253 end
1254
1255 redef fun <(other)
1256 do
1257 if not other isa FlatString then return super
1258
1259 if self.object_id == other.object_id then return false
1260
1261 var my_curr_char : Char
1262 var its_curr_char : Char
1263
1264 var curr_id_self = self.index_from
1265 var curr_id_other = other.index_from
1266
1267 var my_items = self.items
1268 var its_items = other.items
1269
1270 var my_length = self.length
1271 var its_length = other.length
1272
1273 var max_iterations = curr_id_self + my_length
1274
1275 while curr_id_self < max_iterations do
1276 my_curr_char = my_items[curr_id_self]
1277 its_curr_char = its_items[curr_id_other]
1278
1279 if my_curr_char != its_curr_char then
1280 if my_curr_char < its_curr_char then return true
1281 return false
1282 end
1283
1284 curr_id_self += 1
1285 curr_id_other += 1
1286 end
1287
1288 return my_length < its_length
1289 end
1290
1291 redef fun +(s)
1292 do
1293 var my_length = self.length
1294 var its_length = s.length
1295
1296 var total_length = my_length + its_length
1297
1298 var target_string = new NativeString(my_length + its_length + 1)
1299
1300 self.items.copy_to(target_string, my_length, index_from, 0)
1301 if s isa FlatString then
1302 s.items.copy_to(target_string, its_length, s.index_from, my_length)
1303 else if s isa FlatBuffer then
1304 s.items.copy_to(target_string, its_length, 0, my_length)
1305 else
1306 var curr_pos = my_length
1307 for i in [0..s.length[ do
1308 var c = s.chars[i]
1309 target_string[curr_pos] = c
1310 curr_pos += 1
1311 end
1312 end
1313
1314 target_string[total_length] = '\0'
1315
1316 return target_string.to_s_with_length(total_length)
1317 end
1318
1319 redef fun *(i)
1320 do
1321 assert i >= 0
1322
1323 var my_length = self.length
1324
1325 var final_length = my_length * i
1326
1327 var my_items = self.items
1328
1329 var target_string = new NativeString(final_length + 1)
1330
1331 target_string[final_length] = '\0'
1332
1333 var current_last = 0
1334
1335 for iteration in [1 .. i] do
1336 my_items.copy_to(target_string, my_length, 0, current_last)
1337 current_last += my_length
1338 end
1339
1340 return target_string.to_s_with_length(final_length)
1341 end
1342
1343 redef fun hash
1344 do
1345 if hash_cache == null then
1346 # djb2 hash algorithm
1347 var h = 5381
1348 var i = index_from
1349
1350 var myitems = items
1351
1352 while i <= index_to do
1353 h = h.lshift(5) + h + myitems[i].ascii
1354 i += 1
1355 end
1356
1357 hash_cache = h
1358 end
1359
1360 return hash_cache.as(not null)
1361 end
1362
1363 redef fun substrings do return new FlatSubstringsIter(self)
1364 end
1365
1366 private class FlatStringReverseIterator
1367 super IndexedIterator[Char]
1368
1369 var target: FlatString
1370
1371 var target_items: NativeString
1372
1373 var curr_pos: Int
1374
1375 init with_pos(tgt: FlatString, pos: Int)
1376 do
1377 target = tgt
1378 target_items = tgt.items
1379 curr_pos = pos + tgt.index_from
1380 end
1381
1382 redef fun is_ok do return curr_pos >= target.index_from
1383
1384 redef fun item do return target_items[curr_pos]
1385
1386 redef fun next do curr_pos -= 1
1387
1388 redef fun index do return curr_pos - target.index_from
1389
1390 end
1391
1392 private class FlatStringIterator
1393 super IndexedIterator[Char]
1394
1395 var target: FlatString
1396
1397 var target_items: NativeString
1398
1399 var curr_pos: Int
1400
1401 init with_pos(tgt: FlatString, pos: Int)
1402 do
1403 target = tgt
1404 target_items = tgt.items
1405 curr_pos = pos + target.index_from
1406 end
1407
1408 redef fun is_ok do return curr_pos <= target.index_to
1409
1410 redef fun item do return target_items[curr_pos]
1411
1412 redef fun next do curr_pos += 1
1413
1414 redef fun index do return curr_pos - target.index_from
1415
1416 end
1417
1418 private class FlatStringCharView
1419 super StringCharView
1420
1421 redef type SELFTYPE: FlatString
1422
1423 redef fun [](index)
1424 do
1425 # Check that the index (+ index_from) is not larger than indexTo
1426 # In other terms, if the index is valid
1427 assert index >= 0
1428 var target = self.target
1429 assert (index + target.index_from) <= target.index_to
1430 return target.items[index + target.index_from]
1431 end
1432
1433 redef fun iterator_from(start) do return new FlatStringIterator.with_pos(target, start)
1434
1435 redef fun reverse_iterator_from(start) do return new FlatStringReverseIterator.with_pos(target, start)
1436
1437 end
1438
1439 # A mutable sequence of characters.
1440 abstract class Buffer
1441 super Text
1442
1443 redef type SELFTYPE: Buffer is fixed
1444
1445 # Specific implementations MUST set this to `true` in order to invalidate caches
1446 protected var is_dirty = true
1447
1448 # Copy-On-Write flag
1449 #
1450 # If the `Buffer` was to_s'd, the next in-place altering
1451 # operation will cause the current `Buffer` to be re-allocated.
1452 #
1453 # The flag will then be set at `false`.
1454 protected var written = false
1455
1456 # Modifies the char contained at pos `index`
1457 #
1458 # DEPRECATED : Use self.chars.[]= instead
1459 fun []=(index: Int, item: Char) is abstract
1460
1461 # Adds a char `c` at the end of self
1462 #
1463 # DEPRECATED : Use self.chars.add instead
1464 fun add(c: Char) is abstract
1465
1466 # Clears the buffer
1467 #
1468 # var b = new FlatBuffer
1469 # b.append "hello"
1470 # assert not b.is_empty
1471 # b.clear
1472 # assert b.is_empty
1473 fun clear is abstract
1474
1475 # Enlarges the subsequent array containing the chars of self
1476 fun enlarge(cap: Int) is abstract
1477
1478 # Adds the content of text `s` at the end of self
1479 #
1480 # var b = new FlatBuffer
1481 # b.append "hello"
1482 # b.append "world"
1483 # assert b == "helloworld"
1484 fun append(s: Text) is abstract
1485
1486 # `self` is appended in such a way that `self` is repeated `r` times
1487 #
1488 # var b = new FlatBuffer
1489 # b.append "hello"
1490 # b.times 3
1491 # assert b == "hellohellohello"
1492 fun times(r: Int) is abstract
1493
1494 # Reverses itself in-place
1495 #
1496 # var b = new FlatBuffer
1497 # b.append("hello")
1498 # b.reverse
1499 # assert b == "olleh"
1500 fun reverse is abstract
1501
1502 # Changes each lower-case char in `self` by its upper-case variant
1503 #
1504 # var b = new FlatBuffer
1505 # b.append("Hello World!")
1506 # b.upper
1507 # assert b == "HELLO WORLD!"
1508 fun upper is abstract
1509
1510 # Changes each upper-case char in `self` by its lower-case variant
1511 #
1512 # var b = new FlatBuffer
1513 # b.append("Hello World!")
1514 # b.lower
1515 # assert b == "hello world!"
1516 fun lower is abstract
1517
1518 # Capitalizes each word in `self`
1519 #
1520 # Letters that follow a letter are lowercased
1521 # Letters that follow a non-letter are upcased.
1522 #
1523 # SEE: `Char::is_letter` for the definition of a letter.
1524 #
1525 # var b = new FlatBuffer.from("jAVAsCriPt")
1526 # b.capitalize
1527 # assert b == "Javascript"
1528 # b = new FlatBuffer.from("i am root")
1529 # b.capitalize
1530 # assert b == "I Am Root"
1531 # b = new FlatBuffer.from("ab_c -ab0c ab\nc")
1532 # b.capitalize
1533 # assert b == "Ab_C -Ab0C Ab\nC"
1534 fun capitalize do
1535 if length == 0 then return
1536 var c = self[0].to_upper
1537 self[0] = c
1538 var prev = c
1539 for i in [1 .. length[ do
1540 prev = c
1541 c = self[i]
1542 if prev.is_letter then
1543 self[i] = c.to_lower
1544 else
1545 self[i] = c.to_upper
1546 end
1547 end
1548 end
1549
1550 redef fun hash
1551 do
1552 if is_dirty then hash_cache = null
1553 return super
1554 end
1555
1556 # In Buffers, the internal sequence of character is mutable
1557 # Thus, `chars` can be used to modify the buffer.
1558 redef fun chars: Sequence[Char] is abstract
1559 end
1560
1561 # Mutable strings of characters.
1562 class FlatBuffer
1563 super FlatText
1564 super Buffer
1565
1566 redef var chars: Sequence[Char] = new FlatBufferCharView(self) is lazy
1567
1568 private var capacity: Int = 0
1569
1570 redef fun substrings do return new FlatSubstringsIter(self)
1571
1572 # Re-copies the `NativeString` into a new one and sets it as the new `Buffer`
1573 #
1574 # This happens when an operation modifies the current `Buffer` and
1575 # the Copy-On-Write flag `written` is set at true.
1576 private fun reset do
1577 var nns = new NativeString(capacity)
1578 items.copy_to(nns, length, 0, 0)
1579 items = nns
1580 written = false
1581 end
1582
1583 redef fun [](index)
1584 do
1585 assert index >= 0
1586 assert index < length
1587 return items[index]
1588 end
1589
1590 redef fun []=(index, item)
1591 do
1592 is_dirty = true
1593 if index == length then
1594 add(item)
1595 return
1596 end
1597 if written then reset
1598 assert index >= 0 and index < length
1599 items[index] = item
1600 end
1601
1602 redef fun add(c)
1603 do
1604 is_dirty = true
1605 if capacity <= length then enlarge(length + 5)
1606 items[length] = c
1607 length += 1
1608 end
1609
1610 redef fun clear do
1611 is_dirty = true
1612 if written then reset
1613 length = 0
1614 end
1615
1616 redef fun empty do return new FlatBuffer
1617
1618 redef fun enlarge(cap)
1619 do
1620 var c = capacity
1621 if cap <= c then return
1622 while c <= cap do c = c * 2 + 2
1623 # The COW flag can be set at false here, since
1624 # it does a copy of the current `Buffer`
1625 written = false
1626 var a = new NativeString(c+1)
1627 if length > 0 then items.copy_to(a, length, 0, 0)
1628 items = a
1629 capacity = c
1630 end
1631
1632 redef fun to_s: String
1633 do
1634 written = true
1635 if length == 0 then items = new NativeString(1)
1636 return new FlatString.with_infos(items, length, 0, length - 1)
1637 end
1638
1639 redef fun to_cstring
1640 do
1641 if is_dirty then
1642 var new_native = new NativeString(length + 1)
1643 new_native[length] = '\0'
1644 if length > 0 then items.copy_to(new_native, length, 0, 0)
1645 real_items = new_native
1646 is_dirty = false
1647 end
1648 return real_items.as(not null)
1649 end
1650
1651 # Create a new empty string.
1652 init do end
1653
1654 # Low-level creation a new buffer with given data.
1655 #
1656 # `items` will be used as is, without copy, to store the characters of the buffer.
1657 # Aliasing issues is the responsibility of the caller.
1658 #
1659 # If `items` is shared, `written` should be set to true after the creation
1660 # so that a modification will do a copy-on-write.
1661 private init with_infos(items: NativeString, capacity, length: Int)
1662 do
1663 self.items = items
1664 self.length = length
1665 self.capacity = capacity
1666 end
1667
1668 # Create a new string copied from `s`.
1669 init from(s: Text)
1670 do
1671 capacity = s.length + 1
1672 length = s.length
1673 items = new NativeString(capacity)
1674 if s isa FlatString then
1675 s.items.copy_to(items, length, s.index_from, 0)
1676 else if s isa FlatBuffer then
1677 s.items.copy_to(items, length, 0, 0)
1678 else
1679 var curr_pos = 0
1680 for i in [0..s.length[ do
1681 var c = s.chars[i]
1682 items[curr_pos] = c
1683 curr_pos += 1
1684 end
1685 end
1686 end
1687
1688 # Create a new empty string with a given capacity.
1689 init with_capacity(cap: Int)
1690 do
1691 assert cap >= 0
1692 items = new NativeString(cap+1)
1693 capacity = cap
1694 length = 0
1695 end
1696
1697 redef fun append(s)
1698 do
1699 if s.is_empty then return
1700 is_dirty = true
1701 var sl = s.length
1702 if capacity < length + sl then enlarge(length + sl)
1703 if s isa FlatString then
1704 s.items.copy_to(items, sl, s.index_from, length)
1705 else if s isa FlatBuffer then
1706 s.items.copy_to(items, sl, 0, length)
1707 else
1708 var curr_pos = self.length
1709 for i in [0..s.length[ do
1710 var c = s.chars[i]
1711 items[curr_pos] = c
1712 curr_pos += 1
1713 end
1714 end
1715 length += sl
1716 end
1717
1718 # Copies the content of self in `dest`
1719 fun copy(start: Int, len: Int, dest: Buffer, new_start: Int)
1720 do
1721 var self_chars = self.chars
1722 var dest_chars = dest.chars
1723 for i in [0..len-1] do
1724 dest_chars[new_start+i] = self_chars[start+i]
1725 end
1726 end
1727
1728 redef fun substring(from, count)
1729 do
1730 assert count >= 0
1731 count += from
1732 if from < 0 then from = 0
1733 if count > length then count = length
1734 if from < count then
1735 var len = count - from
1736 var r_items = new NativeString(len)
1737 items.copy_to(r_items, len, from, 0)
1738 var r = new FlatBuffer.with_infos(r_items, len, len)
1739 return r
1740 else
1741 return new FlatBuffer
1742 end
1743 end
1744
1745 redef fun reverse
1746 do
1747 written = false
1748 var ns = new NativeString(capacity)
1749 var si = length - 1
1750 var ni = 0
1751 var it = items
1752 while si >= 0 do
1753 ns[ni] = it[si]
1754 ni += 1
1755 si -= 1
1756 end
1757 items = ns
1758 end
1759
1760 redef fun times(repeats)
1761 do
1762 var x = new FlatString.with_infos(items, length, 0, length - 1)
1763 for i in [1..repeats[ do
1764 append(x)
1765 end
1766 end
1767
1768 redef fun upper
1769 do
1770 if written then reset
1771 var it = items
1772 var id = length - 1
1773 while id >= 0 do
1774 it[id] = it[id].to_upper
1775 id -= 1
1776 end
1777 end
1778
1779 redef fun lower
1780 do
1781 if written then reset
1782 var it = items
1783 var id = length - 1
1784 while id >= 0 do
1785 it[id] = it[id].to_lower
1786 id -= 1
1787 end
1788 end
1789 end
1790
1791 private class FlatBufferReverseIterator
1792 super IndexedIterator[Char]
1793
1794 var target: FlatBuffer
1795
1796 var target_items: NativeString
1797
1798 var curr_pos: Int
1799
1800 init with_pos(tgt: FlatBuffer, pos: Int)
1801 do
1802 target = tgt
1803 if tgt.length > 0 then target_items = tgt.items
1804 curr_pos = pos
1805 end
1806
1807 redef fun index do return curr_pos
1808
1809 redef fun is_ok do return curr_pos >= 0
1810
1811 redef fun item do return target_items[curr_pos]
1812
1813 redef fun next do curr_pos -= 1
1814
1815 end
1816
1817 private class FlatBufferCharView
1818 super BufferCharView
1819
1820 redef type SELFTYPE: FlatBuffer
1821
1822 redef fun [](index) do return target.items[index]
1823
1824 redef fun []=(index, item)
1825 do
1826 assert index >= 0 and index <= length
1827 if index == length then
1828 add(item)
1829 return
1830 end
1831 target.items[index] = item
1832 end
1833
1834 redef fun push(c)
1835 do
1836 target.add(c)
1837 end
1838
1839 redef fun add(c)
1840 do
1841 target.add(c)
1842 end
1843
1844 fun enlarge(cap: Int)
1845 do
1846 target.enlarge(cap)
1847 end
1848
1849 redef fun append(s)
1850 do
1851 var s_length = s.length
1852 if target.capacity < s.length then enlarge(s_length + target.length)
1853 end
1854
1855 redef fun iterator_from(pos) do return new FlatBufferIterator.with_pos(target, pos)
1856
1857 redef fun reverse_iterator_from(pos) do return new FlatBufferReverseIterator.with_pos(target, pos)
1858
1859 end
1860
1861 private class FlatBufferIterator
1862 super IndexedIterator[Char]
1863
1864 var target: FlatBuffer
1865
1866 var target_items: NativeString
1867
1868 var curr_pos: Int
1869
1870 init with_pos(tgt: FlatBuffer, pos: Int)
1871 do
1872 target = tgt
1873 if tgt.length > 0 then target_items = tgt.items
1874 curr_pos = pos
1875 end
1876
1877 redef fun index do return curr_pos
1878
1879 redef fun is_ok do return curr_pos < target.length
1880
1881 redef fun item do return target_items[curr_pos]
1882
1883 redef fun next do curr_pos += 1
1884
1885 end
1886
1887 ###############################################################################
1888 # Refinement #
1889 ###############################################################################
1890
1891 redef class Object
1892 # User readable representation of `self`.
1893 fun to_s: String do return inspect
1894
1895 # The class name of the object in NativeString format.
1896 private fun native_class_name: NativeString is intern
1897
1898 # The class name of the object.
1899 #
1900 # assert 5.class_name == "Int"
1901 fun class_name: String do return native_class_name.to_s
1902
1903 # Developer readable representation of `self`.
1904 # Usually, it uses the form "<CLASSNAME:#OBJECTID bla bla bla>"
1905 fun inspect: String
1906 do
1907 return "<{inspect_head}>"
1908 end
1909
1910 # Return "CLASSNAME:#OBJECTID".
1911 # This function is mainly used with the redefinition of the inspect method
1912 protected fun inspect_head: String
1913 do
1914 return "{class_name}:#{object_id.to_hex}"
1915 end
1916 end
1917
1918 redef class Bool
1919 # assert true.to_s == "true"
1920 # assert false.to_s == "false"
1921 redef fun to_s
1922 do
1923 if self then
1924 return once "true"
1925 else
1926 return once "false"
1927 end
1928 end
1929 end
1930
1931 redef class Int
1932
1933 # Wrapper of strerror C function
1934 private fun strerror_ext: NativeString is extern `{
1935 return strerror(recv);
1936 `}
1937
1938 # Returns a string describing error number
1939 fun strerror: String do return strerror_ext.to_s
1940
1941 # Fill `s` with the digits in base `base` of `self` (and with the '-' sign if 'signed' and negative).
1942 # assume < to_c max const of char
1943 private fun fill_buffer(s: Buffer, base: Int, signed: Bool)
1944 do
1945 var n: Int
1946 # Sign
1947 if self < 0 then
1948 n = - self
1949 s.chars[0] = '-'
1950 else if self == 0 then
1951 s.chars[0] = '0'
1952 return
1953 else
1954 n = self
1955 end
1956 # Fill digits
1957 var pos = digit_count(base) - 1
1958 while pos >= 0 and n > 0 do
1959 s.chars[pos] = (n % base).to_c
1960 n = n / base # /
1961 pos -= 1
1962 end
1963 end
1964
1965 # C function to calculate the length of the `NativeString` to receive `self`
1966 private fun int_to_s_len: Int is extern "native_int_length_str"
1967
1968 # C function to convert an nit Int to a NativeString (char*)
1969 private fun native_int_to_s(nstr: NativeString, strlen: Int) is extern "native_int_to_s"
1970
1971 # return displayable int in base 10 and signed
1972 #
1973 # assert 1.to_s == "1"
1974 # assert (-123).to_s == "-123"
1975 redef fun to_s do
1976 # Fast case for common numbers
1977 if self == 0 then return "0"
1978 if self == 1 then return "1"
1979
1980 var nslen = int_to_s_len
1981 var ns = new NativeString(nslen + 1)
1982 ns[nslen] = '\0'
1983 native_int_to_s(ns, nslen + 1)
1984 return ns.to_s_with_length(nslen)
1985 end
1986
1987 # return displayable int in hexadecimal
1988 #
1989 # assert 1.to_hex == "1"
1990 # assert (-255).to_hex == "-ff"
1991 fun to_hex: String do return to_base(16,false)
1992
1993 # return displayable int in base base and signed
1994 fun to_base(base: Int, signed: Bool): String
1995 do
1996 var l = digit_count(base)
1997 var s = new FlatBuffer.from(" " * l)
1998 fill_buffer(s, base, signed)
1999 return s.to_s
2000 end
2001 end
2002
2003 redef class Float
2004 # Pretty representation of `self`, with decimals as needed from 1 to a maximum of 3
2005 #
2006 # assert 12.34.to_s == "12.34"
2007 # assert (-0120.030).to_s == "-120.03"
2008 #
2009 # see `to_precision` for a custom precision.
2010 redef fun to_s do
2011 var str = to_precision( 3 )
2012 if is_inf != 0 or is_nan then return str
2013 var len = str.length
2014 for i in [0..len-1] do
2015 var j = len-1-i
2016 var c = str.chars[j]
2017 if c == '0' then
2018 continue
2019 else if c == '.' then
2020 return str.substring( 0, j+2 )
2021 else
2022 return str.substring( 0, j+1 )
2023 end
2024 end
2025 return str
2026 end
2027
2028 # `String` representation of `self` with the given number of `decimals`
2029 #
2030 # assert 12.345.to_precision(0) == "12"
2031 # assert 12.345.to_precision(3) == "12.345"
2032 # assert (-12.345).to_precision(3) == "-12.345"
2033 # assert (-0.123).to_precision(3) == "-0.123"
2034 # assert 0.999.to_precision(2) == "1.00"
2035 # assert 0.999.to_precision(4) == "0.9990"
2036 fun to_precision(decimals: Int): String
2037 do
2038 if is_nan then return "nan"
2039
2040 var isinf = self.is_inf
2041 if isinf == 1 then
2042 return "inf"
2043 else if isinf == -1 then
2044 return "-inf"
2045 end
2046
2047 if decimals == 0 then return self.to_i.to_s
2048 var f = self
2049 for i in [0..decimals[ do f = f * 10.0
2050 if self > 0.0 then
2051 f = f + 0.5
2052 else
2053 f = f - 0.5
2054 end
2055 var i = f.to_i
2056 if i == 0 then return "0." + "0"*decimals
2057
2058 # Prepare both parts of the float, before and after the "."
2059 var s = i.abs.to_s
2060 var sl = s.length
2061 var p1
2062 var p2
2063 if sl > decimals then
2064 # Has something before the "."
2065 p1 = s.substring(0, sl-decimals)
2066 p2 = s.substring(sl-decimals, decimals)
2067 else
2068 p1 = "0"
2069 p2 = "0"*(decimals-sl) + s
2070 end
2071
2072 if i < 0 then p1 = "-" + p1
2073
2074 return p1 + "." + p2
2075 end
2076 end
2077
2078 redef class Char
2079 # assert 'x'.to_s == "x"
2080 redef fun to_s
2081 do
2082 var s = new FlatBuffer.with_capacity(1)
2083 s.chars[0] = self
2084 return s.to_s
2085 end
2086
2087 # Returns true if the char is a numerical digit
2088 #
2089 # assert '0'.is_numeric
2090 # assert '9'.is_numeric
2091 # assert not 'a'.is_numeric
2092 # assert not '?'.is_numeric
2093 fun is_numeric: Bool
2094 do
2095 return self >= '0' and self <= '9'
2096 end
2097
2098 # Returns true if the char is an alpha digit
2099 #
2100 # assert 'a'.is_alpha
2101 # assert 'Z'.is_alpha
2102 # assert not '0'.is_alpha
2103 # assert not '?'.is_alpha
2104 fun is_alpha: Bool
2105 do
2106 return (self >= 'a' and self <= 'z') or (self >= 'A' and self <= 'Z')
2107 end
2108
2109 # Returns true if the char is an alpha or a numeric digit
2110 #
2111 # assert 'a'.is_alphanumeric
2112 # assert 'Z'.is_alphanumeric
2113 # assert '0'.is_alphanumeric
2114 # assert '9'.is_alphanumeric
2115 # assert not '?'.is_alphanumeric
2116 fun is_alphanumeric: Bool
2117 do
2118 return self.is_numeric or self.is_alpha
2119 end
2120 end
2121
2122 redef class Collection[E]
2123 # Concatenate elements.
2124 redef fun to_s
2125 do
2126 var s = new FlatBuffer
2127 for e in self do if e != null then s.append(e.to_s)
2128 return s.to_s
2129 end
2130
2131 # Concatenate and separate each elements with `sep`.
2132 #
2133 # assert [1, 2, 3].join(":") == "1:2:3"
2134 # assert [1..3].join(":") == "1:2:3"
2135 fun join(sep: Text): String
2136 do
2137 if is_empty then return ""
2138
2139 var s = new FlatBuffer # Result
2140
2141 # Concat first item
2142 var i = iterator
2143 var e = i.item
2144 if e != null then s.append(e.to_s)
2145
2146 # Concat other items
2147 i.next
2148 while i.is_ok do
2149 s.append(sep)
2150 e = i.item
2151 if e != null then s.append(e.to_s)
2152 i.next
2153 end
2154 return s.to_s
2155 end
2156 end
2157
2158 redef class Array[E]
2159
2160 # Fast implementation
2161 redef fun to_s
2162 do
2163 var l = length
2164 if l == 0 then return ""
2165 if l == 1 then if self[0] == null then return "" else return self[0].to_s
2166 var its = _items
2167 var na = new NativeArray[String](l)
2168 var i = 0
2169 var sl = 0
2170 var mypos = 0
2171 while i < l do
2172 var itsi = its[i]
2173 if itsi == null then
2174 i += 1
2175 continue
2176 end
2177 var tmp = itsi.to_s
2178 sl += tmp.length
2179 na[mypos] = tmp
2180 i += 1
2181 mypos += 1
2182 end
2183 var ns = new NativeString(sl + 1)
2184 ns[sl] = '\0'
2185 i = 0
2186 var off = 0
2187 while i < mypos do
2188 var tmp = na[i]
2189 var tpl = tmp.length
2190 if tmp isa FlatString then
2191 tmp.items.copy_to(ns, tpl, tmp.index_from, off)
2192 off += tpl
2193 else
2194 for j in tmp.substrings do
2195 var s = j.as(FlatString)
2196 var slen = s.length
2197 s.items.copy_to(ns, slen, s.index_from, off)
2198 off += slen
2199 end
2200 end
2201 i += 1
2202 end
2203 return ns.to_s_with_length(sl)
2204 end
2205 end
2206
2207 redef class NativeArray[E]
2208 # Join all the elements using `to_s`
2209 #
2210 # REQUIRE: `self isa NativeArray[String]`
2211 # REQUIRE: all elements are initialized
2212 fun native_to_s: String
2213 do
2214 assert self isa NativeArray[String]
2215 var l = length
2216 var na = self
2217 var i = 0
2218 var sl = 0
2219 var mypos = 0
2220 while i < l do
2221 sl += na[i].length
2222 i += 1
2223 mypos += 1
2224 end
2225 var ns = new NativeString(sl + 1)
2226 ns[sl] = '\0'
2227 i = 0
2228 var off = 0
2229 while i < mypos do
2230 var tmp = na[i]
2231 var tpl = tmp.length
2232 if tmp isa FlatString then
2233 tmp.items.copy_to(ns, tpl, tmp.index_from, off)
2234 off += tpl
2235 else
2236 for j in tmp.substrings do
2237 var s = j.as(FlatString)
2238 var slen = s.length
2239 s.items.copy_to(ns, slen, s.index_from, off)
2240 off += slen
2241 end
2242 end
2243 i += 1
2244 end
2245 return ns.to_s_with_length(sl)
2246 end
2247 end
2248
2249 redef class Map[K,V]
2250 # Concatenate couple of 'key value'.
2251 # key and value are separated by `couple_sep`.
2252 # each couple is separated each couple with `sep`.
2253 #
2254 # var m = new ArrayMap[Int, String]
2255 # m[1] = "one"
2256 # m[10] = "ten"
2257 # assert m.join("; ", "=") == "1=one; 10=ten"
2258 fun join(sep: String, couple_sep: String): String
2259 do
2260 if is_empty then return ""
2261
2262 var s = new FlatBuffer # Result
2263
2264 # Concat first item
2265 var i = iterator
2266 var k = i.key
2267 var e = i.item
2268 s.append("{k or else "<null>"}{couple_sep}{e or else "<null>"}")
2269
2270 # Concat other items
2271 i.next
2272 while i.is_ok do
2273 s.append(sep)
2274 k = i.key
2275 e = i.item
2276 s.append("{k or else "<null>"}{couple_sep}{e or else "<null>"}")
2277 i.next
2278 end
2279 return s.to_s
2280 end
2281 end
2282
2283 ###############################################################################
2284 # Native classes #
2285 ###############################################################################
2286
2287 # Native strings are simple C char *
2288 extern class NativeString `{ char* `}
2289 # Creates a new NativeString with a capacity of `length`
2290 new(length: Int) is intern
2291
2292 # Get char at `index`.
2293 fun [](index: Int): Char is intern
2294
2295 # Set char `item` at index.
2296 fun []=(index: Int, item: Char) is intern
2297
2298 # Copy `self` to `dest`.
2299 fun copy_to(dest: NativeString, length: Int, from: Int, to: Int) is intern
2300
2301 # Position of the first nul character.
2302 fun cstring_length: Int
2303 do
2304 var l = 0
2305 while self[l] != '\0' do l += 1
2306 return l
2307 end
2308
2309 # Parse `self` as an Int.
2310 fun atoi: Int is intern
2311
2312 # Parse `self` as a Float.
2313 fun atof: Float is extern "atof"
2314
2315 redef fun to_s
2316 do
2317 return to_s_with_length(cstring_length)
2318 end
2319
2320 # Returns `self` as a String of `length`.
2321 fun to_s_with_length(length: Int): FlatString
2322 do
2323 assert length >= 0
2324 var str = new FlatString.with_infos(self, length, 0, length - 1)
2325 return str
2326 end
2327
2328 # Returns `self` as a new String.
2329 fun to_s_with_copy: FlatString
2330 do
2331 var length = cstring_length
2332 var new_self = new NativeString(length + 1)
2333 copy_to(new_self, length, 0, 0)
2334 var str = new FlatString.with_infos(new_self, length, 0, length - 1)
2335 new_self[length] = '\0'
2336 str.real_items = new_self
2337 return str
2338 end
2339 end
2340
2341 redef class Sys
2342 private var args_cache: nullable Sequence[String]
2343
2344 # The arguments of the program as given by the OS
2345 fun program_args: Sequence[String]
2346 do
2347 if _args_cache == null then init_args
2348 return _args_cache.as(not null)
2349 end
2350
2351 # The name of the program as given by the OS
2352 fun program_name: String
2353 do
2354 return native_argv(0).to_s
2355 end
2356
2357 # Initialize `program_args` with the contents of `native_argc` and `native_argv`.
2358 private fun init_args
2359 do
2360 var argc = native_argc
2361 var args = new Array[String].with_capacity(0)
2362 var i = 1
2363 while i < argc do
2364 args[i-1] = native_argv(i).to_s
2365 i += 1
2366 end
2367 _args_cache = args
2368 end
2369
2370 # First argument of the main C function.
2371 private fun native_argc: Int is intern
2372
2373 # Second argument of the main C function.
2374 private fun native_argv(i: Int): NativeString is intern
2375 end
2376
2377 # Comparator that efficienlty use `to_s` to compare things
2378 #
2379 # The comparaison call `to_s` on object and use the result to order things.
2380 #
2381 # var a = [1, 2, 3, 10, 20]
2382 # (new CachedAlphaComparator).sort(a)
2383 # assert a == [1, 10, 2, 20, 3]
2384 #
2385 # Internally the result of `to_s` is cached in a HashMap to counter
2386 # uneficient implementation of `to_s`.
2387 #
2388 # Note: it caching is not usefull, see `alpha_comparator`
2389 class CachedAlphaComparator
2390 super Comparator
2391 redef type COMPARED: Object
2392
2393 private var cache = new HashMap[Object, String]
2394
2395 private fun do_to_s(a: Object): String do
2396 if cache.has_key(a) then return cache[a]
2397 var res = a.to_s
2398 cache[a] = res
2399 return res
2400 end
2401
2402 redef fun compare(a, b) do
2403 return do_to_s(a) <=> do_to_s(b)
2404 end
2405 end
2406
2407 # see `alpha_comparator`
2408 private class AlphaComparator
2409 super Comparator
2410 redef fun compare(a, b) do return a.to_s <=> b.to_s
2411 end
2412
2413 # Stateless comparator that naively use `to_s` to compare things.
2414 #
2415 # Note: the result of `to_s` is not cached, thus can be invoked a lot
2416 # on a single instace. See `CachedAlphaComparator` as an alternative.
2417 #
2418 # var a = [1, 2, 3, 10, 20]
2419 # alpha_comparator.sort(a)
2420 # assert a == [1, 10, 2, 20, 3]
2421 fun alpha_comparator: Comparator do return once new AlphaComparator
2422
2423 # The arguments of the program as given by the OS
2424 fun args: Sequence[String]
2425 do
2426 return sys.program_args
2427 end
Nit language project; an oo programing language and its tool suite.