cwstring.pp 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380
  1. {
  2. This file is part of the Free Pascal run time library.
  3. Copyright (c) 2005 by Florian Klaempfl,
  4. member of the Free Pascal development team.
  5. libc based wide string support
  6. See the file COPYING.FPC, included in this distribution,
  7. for details about the copyright.
  8. This program is distributed in the hope that it will be useful,
  9. but WITHOUT ANY WARRANTY; without even the implied warranty of
  10. MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  11. **********************************************************************}
  12. {$mode objfpc}
  13. unit cwstring;
  14. interface
  15. procedure SetCWidestringManager;
  16. implementation
  17. {$linklib c}
  18. {$ifndef linux} // Linux (and maybe glibc platforms in general), have iconv in glibc.
  19. {$linklib iconv}
  20. {$define useiconv}
  21. {$endif linux}
  22. Uses
  23. BaseUnix,
  24. ctypes,
  25. unix,
  26. unixtype,
  27. initc;
  28. Const
  29. {$ifndef useiconv}
  30. libiconvname='c'; // is in libc under Linux.
  31. {$else}
  32. libiconvname='iconv';
  33. {$endif}
  34. { Case-mapping "arrays" }
  35. var
  36. AnsiUpperChars: AnsiString; // 1..255
  37. AnsiLowerChars: AnsiString; // 1..255
  38. WideUpperChars: WideString; // 1..65535
  39. WideLowerChars: WideString; // 1..65535
  40. { the following declarations are from the libc unit for linux so they
  41. might be very linux centric
  42. maybe this needs to be splitted in an os depend way later }
  43. function towlower(__wc:wint_t):wint_t;cdecl;external libiconvname name 'towlower';
  44. function towupper(__wc:wint_t):wint_t;cdecl;external libiconvname name 'towupper';
  45. function wcscoll (__s1:pwchar_t; __s2:pwchar_t):cint;cdecl;external libiconvname name 'wcscoll';
  46. function strcoll (__s1:pchar; __s2:pchar):cint;cdecl;external libiconvname name 'strcoll';
  47. const
  48. {$ifdef linux}
  49. __LC_CTYPE = 0;
  50. _NL_CTYPE_CLASS = (__LC_CTYPE shl 16);
  51. _NL_CTYPE_CODESET_NAME = (_NL_CTYPE_CLASS)+14;
  52. CODESET = _NL_CTYPE_CODESET_NAME;
  53. {$else linux}
  54. {$ifdef darwin}
  55. CODESET = 0;
  56. {$else darwin}
  57. {$ifdef FreeBSD} // actually FreeBSD5. internationalisation is afaik not default on 4.
  58. __LC_CTYPE = 0;
  59. _NL_CTYPE_CLASS = (__LC_CTYPE shl 16);
  60. _NL_CTYPE_CODESET_NAME = (_NL_CTYPE_CLASS)+14;
  61. CODESET = _NL_CTYPE_CODESET_NAME;
  62. {$else freebsd}
  63. {$ifdef solaris}
  64. CODESET=49;
  65. {$else}
  66. {$error lookup the value of CODESET in /usr/include/langinfo.h for your OS }
  67. // and while doing it, check if iconv is in libc, and if the symbols are prefixed with iconv_ or libiconv_
  68. {$endif solaris}
  69. {$endif FreeBSD}
  70. {$endif darwin}
  71. {$endif linux}
  72. { unicode encoding name }
  73. {$ifdef FPC_LITTLE_ENDIAN}
  74. unicode_encoding = 'UNICODELITTLE';
  75. {$else FPC_LITTLE_ENDIAN}
  76. unicode_encoding = 'UNICODEBIG';
  77. {$endif FPC_LITTLE_ENDIAN}
  78. type
  79. piconv_t = ^iconv_t;
  80. iconv_t = pointer;
  81. nl_item = cint;
  82. function nl_langinfo(__item:nl_item):pchar;cdecl;external libiconvname name 'nl_langinfo';
  83. {$ifndef bsd}
  84. function iconv_open(__tocode:pchar; __fromcode:pchar):iconv_t;cdecl;external libiconvname name 'iconv_open';
  85. function iconv(__cd:iconv_t; __inbuf:ppchar; __inbytesleft:psize_t; __outbuf:ppchar; __outbytesleft:psize_t):size_t;cdecl;external libiconvname name 'iconv';
  86. function iconv_close(__cd:iconv_t):cint;cdecl;external libiconvname name 'iconv_close';
  87. {$else}
  88. function iconv_open(__tocode:pchar; __fromcode:pchar):iconv_t;cdecl;external libiconvname name 'libiconv_open';
  89. function iconv(__cd:iconv_t; __inbuf:ppchar; __inbytesleft:psize_t; __outbuf:ppchar; __outbytesleft:psize_t):size_t;cdecl;external libiconvname name 'libiconv';
  90. function iconv_close(__cd:iconv_t):cint;cdecl;external libiconvname name 'libiconv_close';
  91. {$endif}
  92. var
  93. iconv_ansi2ucs4,
  94. iconv_ucs42ansi,
  95. iconv_ansi2wide,
  96. iconv_wide2ansi : iconv_t;
  97. lock_ansi2ucs4 : integer = -1;
  98. lock_ucs42ansi : integer = -1;
  99. lock_ansi2wide : integer = -1;
  100. lock_wide2ansi : integer = -1;
  101. iconv_lock : TRTLcriticalsection;
  102. {
  103. procedure lockiconv(var lockcount: integer);
  104. begin
  105. while interlockedincrement(lockcount) <> 0 do begin
  106. interlockeddecrement(lockcount);
  107. sleep(0);
  108. end;
  109. end;
  110. procedure unlockiconv(var lockcount: integer);
  111. begin
  112. interlockeddecrement(lockcount);
  113. end;
  114. }
  115. procedure Wide2AnsiMove(source:pwidechar;var dest:ansistring;len:SizeInt);
  116. var
  117. outlength,
  118. outoffset,
  119. srclen,
  120. outleft : size_t;
  121. srcpos : pwidechar;
  122. destpos: pchar;
  123. mynil : pchar;
  124. my0 : size_t;
  125. begin
  126. mynil:=nil;
  127. my0:=0;
  128. { rought estimation }
  129. setlength(dest,len*3);
  130. outlength:=len*3;
  131. srclen:=len*2;
  132. srcpos:=source;
  133. destpos:=pchar(dest);
  134. outleft:=outlength;
  135. entercriticalsection(iconv_lock);
  136. while iconv(iconv_wide2ansi,ppchar(@srcpos),@srclen,@destpos,@outleft)=size_t(-1) do
  137. begin
  138. case fpgetCerrno of
  139. ESysEILSEQ:
  140. begin
  141. { skip and set to '?' }
  142. inc(srcpos);
  143. dec(srclen,2);
  144. destpos^:='?';
  145. inc(destpos);
  146. dec(outleft);
  147. { reset }
  148. iconv(iconv_wide2ansi,@mynil,@my0,@mynil,@my0);
  149. end;
  150. ESysE2BIG:
  151. begin
  152. outoffset:=destpos-pchar(dest);
  153. { extend }
  154. setlength(dest,outlength+len*3);
  155. inc(outleft,len*3);
  156. inc(outlength,len*3);
  157. { string could have been moved }
  158. destpos:=pchar(dest)+outoffset;
  159. end;
  160. else
  161. leavecriticalsection(iconv_lock);
  162. runerror(231);
  163. end;
  164. end;
  165. leavecriticalsection(iconv_lock);
  166. // truncate string
  167. setlength(dest,length(dest)-outleft);
  168. end;
  169. procedure Ansi2WideMove(source:pchar;var dest:widestring;len:SizeInt);
  170. var
  171. outlength,
  172. outoffset,
  173. outleft : size_t;
  174. srcpos,
  175. destpos: pchar;
  176. mynil : pchar;
  177. my0 : size_t;
  178. begin
  179. mynil:=nil;
  180. my0:=0;
  181. // extra space
  182. outlength:=len+1;
  183. setlength(dest,outlength);
  184. outlength:=len+1;
  185. srcpos:=source;
  186. destpos:=pchar(dest);
  187. outleft:=outlength*2;
  188. entercriticalsection(iconv_lock);
  189. while iconv(iconv_ansi2wide,@srcpos,psize(@len),@destpos,@outleft)=size_t(-1) do
  190. begin
  191. case fpgetCerrno of
  192. ESysEILSEQ:
  193. begin
  194. { skip and set to '?' }
  195. inc(srcpos);
  196. pwidechar(destpos)^:='?';
  197. inc(destpos,2);
  198. dec(outleft,2);
  199. { reset }
  200. iconv(iconv_wide2ansi,@mynil,@my0,@mynil,@my0);
  201. end;
  202. ESysE2BIG:
  203. begin
  204. outoffset:=destpos-pchar(dest);
  205. { extend }
  206. setlength(dest,outlength+len);
  207. inc(outleft,len*2);
  208. inc(outlength,len);
  209. { string could have been moved }
  210. destpos:=pchar(dest)+outoffset;
  211. end;
  212. else
  213. leavecriticalsection(iconv_lock);
  214. runerror(231);
  215. end;
  216. end;
  217. leavecriticalsection(iconv_lock);
  218. // truncate string
  219. setlength(dest,length(dest)-outleft div 2);
  220. end;
  221. function LowerWideString(const s : WideString) : WideString;
  222. var
  223. i : SizeInt;
  224. begin
  225. SetLength(result,length(s));
  226. for i:=1 to length(s) do
  227. result[i]:=WideChar(towlower(wint_t(s[i])));
  228. end;
  229. function UpperWideString(const s : WideString) : WideString;
  230. var
  231. i : SizeInt;
  232. begin
  233. SetLength(result,length(s));
  234. for i:=1 to length(s) do
  235. result[i]:=WideChar(towupper(wint_t(s[i])));
  236. end;
  237. procedure Ansi2UCS4Move(source:pchar;var dest:UCS4String;len:SizeInt);
  238. var
  239. outlength,
  240. outoffset,
  241. outleft : size_t;
  242. srcpos,
  243. destpos: pchar;
  244. mynil : pchar;
  245. my0 : size_t;
  246. begin
  247. mynil:=nil;
  248. my0:=0;
  249. // extra space
  250. outlength:=len+1;
  251. setlength(dest,outlength);
  252. outlength:=len+1;
  253. srcpos:=source;
  254. destpos:=pchar(dest);
  255. outleft:=outlength*4;
  256. entercriticalsection(iconv_lock);
  257. while iconv(iconv_ansi2ucs4,@srcpos,psize(@len),@destpos,@outleft)=size_t(-1) do
  258. begin
  259. case fpgetCerrno of
  260. ESysE2BIG:
  261. begin
  262. outoffset:=destpos-pchar(dest);
  263. { extend }
  264. setlength(dest,outlength+len);
  265. inc(outleft,len*4);
  266. inc(outlength,len);
  267. { string could have been moved }
  268. destpos:=pchar(dest)+outoffset;
  269. end;
  270. else
  271. leavecriticalsection(iconv_lock);
  272. runerror(231);
  273. end;
  274. end;
  275. leavecriticalsection(iconv_lock);
  276. // truncate string
  277. setlength(dest,length(dest)-outleft div 4);
  278. end;
  279. function CompareWideString(const s1, s2 : WideString) : PtrInt;
  280. var
  281. hs1,hs2 : UCS4String;
  282. begin
  283. hs1:=WideStringToUCS4String(s1);
  284. hs2:=WideStringToUCS4String(s2);
  285. result:=wcscoll(pwchar_t(hs1),pwchar_t(hs2));
  286. end;
  287. function CompareTextWideString(const s1, s2 : WideString): PtrInt;
  288. begin
  289. result:=CompareWideString(UpperWideString(s1),UpperWideString(s2));
  290. end;
  291. function StrCompAnsi(s1,s2 : PChar): PtrInt;
  292. begin
  293. result:=strcoll(s1,s2);
  294. end;
  295. Procedure SetCWideStringManager;
  296. Var
  297. CWideStringManager : TWideStringManager;
  298. begin
  299. CWideStringManager:=widestringmanager;
  300. With CWideStringManager do
  301. begin
  302. Wide2AnsiMoveProc:=@Wide2AnsiMove;
  303. Ansi2WideMoveProc:=@Ansi2WideMove;
  304. UpperWideStringProc:=@UpperWideString;
  305. LowerWideStringProc:=@LowerWideString;
  306. CompareWideStringProc:=@CompareWideString;
  307. CompareTextWideStringProc:=@CompareTextWideString;
  308. {
  309. CharLengthPCharProc
  310. UpperAnsiStringProc
  311. LowerAnsiStringProc
  312. CompareStrAnsiStringProc
  313. CompareTextAnsiStringProc
  314. }
  315. StrCompAnsiStringProc:=@StrCompAnsi;
  316. {
  317. StrICompAnsiStringProc
  318. StrLCompAnsiStringProc
  319. StrLICompAnsiStringProc
  320. StrLowerAnsiStringProc
  321. StrUpperAnsiStringProc
  322. }
  323. end;
  324. SetWideStringManager(CWideStringManager);
  325. end;
  326. initialization
  327. SetCWideStringManager;
  328. { init conversion tables }
  329. iconv_wide2ansi:=iconv_open(nl_langinfo(CODESET),unicode_encoding);
  330. iconv_ansi2wide:=iconv_open(unicode_encoding,nl_langinfo(CODESET));
  331. iconv_ucs42ansi:=iconv_open(nl_langinfo(CODESET),'UCS4');
  332. iconv_ansi2ucs4:=iconv_open('UCS4',nl_langinfo(CODESET));
  333. finalization
  334. iconv_close(iconv_ansi2wide);
  335. end.