Mercurial > projects > ldc
annotate dmd/module.c @ 598:13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
name as the source file.
| author | Christian Kamm <kamm incasoftware de> |
|---|---|
| date | Sun, 14 Sep 2008 22:49:19 +0200 |
| parents | 8cc0c46064b1 |
| children | a30fc28e8f23 |
| rev | line source |
|---|---|
| 336 | 1 |
| 2 // Compiler implementation of the D programming language | |
| 3 // Copyright (c) 1999-2007 by Digital Mars | |
| 4 // All Rights Reserved | |
| 5 // written by Walter Bright | |
| 6 // http://www.digitalmars.com | |
| 7 // License for redistribution is by either the Artistic License | |
| 8 // in artistic.txt, or the GNU General Public License in gnu.txt. | |
| 9 // See the included readme.txt for details. | |
| 10 | |
| 11 #include <stdio.h> | |
| 12 #include <stdlib.h> | |
| 13 #include <assert.h> | |
| 14 | |
| 15 #if _MSC_VER || __MINGW32__ | |
| 16 #include <malloc.h> | |
| 17 #endif | |
| 18 | |
| 19 #if IN_GCC | |
| 20 #include "gdc_alloca.h" | |
| 21 #endif | |
| 22 | |
| 23 #include "mem.h" | |
| 24 | |
| 25 #include "mars.h" | |
| 26 #include "module.h" | |
| 27 #include "parse.h" | |
| 28 #include "scope.h" | |
| 29 #include "identifier.h" | |
| 30 #include "id.h" | |
| 31 #include "import.h" | |
| 32 #include "dsymbol.h" | |
| 33 #include "hdrgen.h" | |
| 34 #include "lexer.h" | |
| 35 | |
| 36 #define MARS 1 | |
| 37 #include "html.h" | |
| 38 | |
| 39 #ifdef IN_GCC | |
| 40 #include "d-dmd-gcc.h" | |
| 41 #endif | |
| 42 | |
| 43 ClassDeclaration *Module::moduleinfo; | |
| 44 | |
| 45 Module *Module::rootModule; | |
| 46 DsymbolTable *Module::modules; | |
| 47 Array Module::amodules; | |
| 48 | |
| 49 Array Module::deferred; // deferred Dsymbol's needing semantic() run on them | |
| 50 unsigned Module::dprogress; | |
| 51 | |
| 52 void Module::init() | |
| 53 { | |
| 54 modules = new DsymbolTable(); | |
| 55 } | |
| 56 | |
| 57 Module::Module(char *filename, Identifier *ident, int doDocComment, int doHdrGen) | |
| 58 : Package(ident) | |
| 59 { | |
| 60 FileName *srcfilename; | |
| 61 | |
| 62 // printf("Module::Module(filename = '%s', ident = '%s')\n", filename, ident->toChars()); | |
| 63 this->arg = filename; | |
| 64 md = NULL; | |
| 65 errors = 0; | |
| 66 numlines = 0; | |
| 67 members = NULL; | |
| 68 isHtml = 0; | |
| 69 isDocFile = 0; | |
| 70 needmoduleinfo = 0; | |
| 71 #ifdef IN_GCC | |
| 72 strictlyneedmoduleinfo = 0; | |
| 73 #endif | |
| 74 insearch = 0; | |
| 75 searchCacheIdent = NULL; | |
| 76 searchCacheSymbol = NULL; | |
| 77 searchCacheFlags = 0; | |
| 78 semanticstarted = 0; | |
| 79 semanticdone = 0; | |
| 80 decldefs = NULL; | |
| 81 vmoduleinfo = NULL; | |
| 82 massert = NULL; | |
| 83 marray = NULL; | |
| 84 sictor = NULL; | |
| 85 sctor = NULL; | |
| 86 sdtor = NULL; | |
| 87 stest = NULL; | |
| 88 sfilename = NULL; | |
| 89 root = 0; | |
| 90 importedFrom = NULL; | |
| 91 srcfile = NULL; | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
92 objfile = NULL; |
| 336 | 93 docfile = NULL; |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
94 hdrfile = NULL; |
| 336 | 95 |
| 96 debuglevel = 0; | |
| 97 debugids = NULL; | |
| 98 debugidsNot = NULL; | |
| 99 versionlevel = 0; | |
| 100 versionids = NULL; | |
| 101 versionidsNot = NULL; | |
| 102 | |
| 103 macrotable = NULL; | |
| 104 escapetable = NULL; | |
| 105 doppelganger = 0; | |
| 106 cov = NULL; | |
| 107 covb = NULL; | |
| 108 | |
| 109 srcfilename = FileName::defaultExt(filename, global.mars_ext); | |
| 110 if (!srcfilename->equalsExt(global.mars_ext) && | |
|
510
6aee82889553
Merged DMD 1.034, array operations are not yet implemented ;)
Tomas Lindquist Olsen <tomas.l.olsen@gmail.com>
parents:
486
diff
changeset
|
111 !srcfilename->equalsExt(global.hdr_ext) && |
| 336 | 112 !srcfilename->equalsExt("dd")) |
| 113 { | |
| 114 if (srcfilename->equalsExt("html") || | |
| 115 srcfilename->equalsExt("htm") || | |
| 116 srcfilename->equalsExt("xhtml")) | |
| 117 isHtml = 1; | |
| 118 else | |
| 119 { error("source file name '%s' must have .%s extension", srcfilename->toChars(), global.mars_ext); | |
| 120 fatal(); | |
| 121 } | |
| 122 } | |
| 123 srcfile = new File(srcfilename); | |
| 124 | |
|
486
a34078905d01
Added pragma(llvmdc, "string") for misc per-module compiler configuration, currently "string" can only be "verbose" which forces -vv for module it appears in.
Tomas Lindquist Olsen <tomas.l.olsen@gmail.com>
parents:
336
diff
changeset
|
125 // LLVMDC |
|
a34078905d01
Added pragma(llvmdc, "string") for misc per-module compiler configuration, currently "string" can only be "verbose" which forces -vv for module it appears in.
Tomas Lindquist Olsen <tomas.l.olsen@gmail.com>
parents:
336
diff
changeset
|
126 llvmForceLogging = false; |
| 336 | 127 } |
| 128 | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
129 File* Module::buildFilePath(char* forcename, char* path, char* ext) |
| 336 | 130 { |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
131 char *argobj; |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
132 if (forcename) |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
133 argobj = forcename; |
| 336 | 134 else |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
135 { |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
136 if (global.params.preservePaths) |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
137 argobj = (char*)this->arg; |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
138 else |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
139 argobj = FileName::name((char*)this->arg); |
| 336 | 140 |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
141 if (global.params.fqnNames) |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
142 if(md) |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
143 argobj = FileName::replaceName(argobj, md->toChars()); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
144 else |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
145 argobj = FileName::replaceName(argobj, toChars()); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
146 |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
147 int clen = strlen(argobj); |
|
595
bbfb66c623e7
Fix output files being .d.bc. They're now .bc again.
Christian Kamm <kamm incasoftware de>
parents:
580
diff
changeset
|
148 char* tmp = (char *)alloca(clen + 1); |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
149 memcpy(tmp, argobj, clen); |
|
595
bbfb66c623e7
Fix output files being .d.bc. They're now .bc again.
Christian Kamm <kamm incasoftware de>
parents:
580
diff
changeset
|
150 tmp[clen] = 0; |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
151 argobj = tmp; |
| 336 | 152 } |
| 153 | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
154 if (!FileName::absolute(argobj)) |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
155 { |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
156 argobj = FileName::combine(path, argobj); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
157 } |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
158 |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
159 FileName::ensurePathExists(FileName::path(argobj)); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
160 |
|
597
8cc0c46064b1
Fix path handling, hopefully.
Christian Kamm <kamm incasoftware de>
parents:
595
diff
changeset
|
161 if (forcename) |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
162 return new File(argobj); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
163 else |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
164 return new File(FileName::forceExt(argobj, ext)); |
| 336 | 165 } |
| 166 | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
167 void Module::buildTargetFiles() |
| 336 | 168 { |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
169 if(objfile && docfile && hdrfile) |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
170 return; |
| 336 | 171 |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
172 objfile = Module::buildFilePath(global.params.objname, global.params.objdir, global.bc_ext); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
173 docfile = Module::buildFilePath(global.params.docname, global.params.docdir, global.doc_ext); |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
174 hdrfile = Module::buildFilePath(global.params.hdrname, global.params.hdrdir, global.hdr_ext); |
|
598
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
175 |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
176 // safety check: never allow obj, doc or hdr file to have the source file's name |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
177 if(stricmp(FileName::name(objfile->name->str), FileName::name((char*)this->arg)) == 0 || |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
178 stricmp(FileName::name(docfile->name->str), FileName::name((char*)this->arg)) == 0 || |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
179 stricmp(FileName::name(hdrfile->name->str), FileName::name((char*)this->arg)) == 0) |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
180 { |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
181 error("Object-, ddoc-, and header- output files with the same name as the source file are forbidden"); |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
182 fatal(); |
|
13ff06605226
To prevert source-overwriting in the future, forbit output files with the same
Christian Kamm <kamm incasoftware de>
parents:
597
diff
changeset
|
183 } |
| 336 | 184 } |
| 185 | |
| 186 void Module::deleteObjFile() | |
| 187 { | |
| 188 if (global.params.obj) | |
| 189 objfile->remove(); | |
| 190 //if (global.params.llvmBC) | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
191 //bcfile->remove(); |
| 336 | 192 if (docfile) |
| 193 docfile->remove(); | |
| 194 } | |
| 195 | |
| 196 Module::~Module() | |
| 197 { | |
| 198 } | |
| 199 | |
| 200 const char *Module::kind() | |
| 201 { | |
| 202 return "module"; | |
| 203 } | |
| 204 | |
| 205 Module *Module::load(Loc loc, Array *packages, Identifier *ident) | |
| 206 { Module *m; | |
| 207 char *filename; | |
| 208 | |
| 209 //printf("Module::load(ident = '%s')\n", ident->toChars()); | |
| 210 | |
| 211 // Build module filename by turning: | |
| 212 // foo.bar.baz | |
| 213 // into: | |
| 214 // foo\bar\baz | |
| 215 filename = ident->toChars(); | |
| 216 if (packages && packages->dim) | |
| 217 { | |
| 218 OutBuffer buf; | |
| 219 int i; | |
| 220 | |
| 221 for (i = 0; i < packages->dim; i++) | |
| 222 { Identifier *pid = (Identifier *)packages->data[i]; | |
| 223 | |
| 224 buf.writestring(pid->toChars()); | |
| 225 #if _WIN32 | |
| 226 buf.writeByte('\\'); | |
| 227 #else | |
| 228 buf.writeByte('/'); | |
| 229 #endif | |
| 230 } | |
| 231 buf.writestring(filename); | |
| 232 buf.writeByte(0); | |
| 233 filename = (char *)buf.extractData(); | |
| 234 } | |
| 235 | |
| 236 m = new Module(filename, ident, 0, 0); | |
| 237 m->loc = loc; | |
| 238 | |
| 239 /* Search along global.path for .di file, then .d file. | |
| 240 */ | |
| 241 char *result = NULL; | |
| 242 FileName *fdi = FileName::forceExt(filename, global.hdr_ext); | |
| 243 FileName *fd = FileName::forceExt(filename, global.mars_ext); | |
| 244 char *sdi = fdi->toChars(); | |
| 245 char *sd = fd->toChars(); | |
| 246 | |
| 247 if (FileName::exists(sdi)) | |
| 248 result = sdi; | |
| 249 else if (FileName::exists(sd)) | |
| 250 result = sd; | |
| 251 else if (FileName::absolute(filename)) | |
| 252 ; | |
| 253 else if (!global.path) | |
| 254 ; | |
| 255 else | |
| 256 { | |
| 257 for (size_t i = 0; i < global.path->dim; i++) | |
| 258 { | |
| 259 char *p = (char *)global.path->data[i]; | |
| 260 char *n = FileName::combine(p, sdi); | |
| 261 if (FileName::exists(n)) | |
| 262 { result = n; | |
| 263 break; | |
| 264 } | |
| 265 mem.free(n); | |
| 266 n = FileName::combine(p, sd); | |
| 267 if (FileName::exists(n)) | |
| 268 { result = n; | |
| 269 break; | |
| 270 } | |
| 271 mem.free(n); | |
| 272 } | |
| 273 } | |
| 274 if (result) | |
| 275 m->srcfile = new File(result); | |
| 276 | |
| 277 if (global.params.verbose) | |
| 278 { | |
| 279 printf("import "); | |
| 280 if (packages) | |
| 281 { | |
| 282 for (size_t i = 0; i < packages->dim; i++) | |
| 283 { Identifier *pid = (Identifier *)packages->data[i]; | |
| 284 printf("%s.", pid->toChars()); | |
| 285 } | |
| 286 } | |
| 287 printf("%s\t(%s)\n", ident->toChars(), m->srcfile->toChars()); | |
| 288 } | |
| 289 | |
| 290 m->read(loc); | |
| 291 m->parse(); | |
| 292 | |
| 293 #ifdef IN_GCC | |
| 294 d_gcc_magic_module(m); | |
| 295 #endif | |
| 296 | |
| 297 return m; | |
| 298 } | |
| 299 | |
| 300 void Module::read(Loc loc) | |
| 301 { | |
| 302 //printf("Module::read('%s') file '%s'\n", toChars(), srcfile->toChars()); | |
| 303 if (srcfile->read()) | |
| 304 { error(loc, "cannot read file '%s'", srcfile->toChars()); | |
| 305 fatal(); | |
| 306 } | |
| 307 } | |
| 308 | |
| 309 inline unsigned readwordLE(unsigned short *p) | |
| 310 { | |
| 311 #if __I86__ | |
| 312 return *p; | |
| 313 #else | |
| 314 return (((unsigned char *)p)[1] << 8) | ((unsigned char *)p)[0]; | |
| 315 #endif | |
| 316 } | |
| 317 | |
| 318 inline unsigned readwordBE(unsigned short *p) | |
| 319 { | |
| 320 return (((unsigned char *)p)[0] << 8) | ((unsigned char *)p)[1]; | |
| 321 } | |
| 322 | |
| 323 inline unsigned readlongLE(unsigned *p) | |
| 324 { | |
| 325 #if __I86__ | |
| 326 return *p; | |
| 327 #else | |
| 328 return ((unsigned char *)p)[0] | | |
| 329 (((unsigned char *)p)[1] << 8) | | |
| 330 (((unsigned char *)p)[2] << 16) | | |
| 331 (((unsigned char *)p)[3] << 24); | |
| 332 #endif | |
| 333 } | |
| 334 | |
| 335 inline unsigned readlongBE(unsigned *p) | |
| 336 { | |
| 337 return ((unsigned char *)p)[3] | | |
| 338 (((unsigned char *)p)[2] << 8) | | |
| 339 (((unsigned char *)p)[1] << 16) | | |
| 340 (((unsigned char *)p)[0] << 24); | |
| 341 } | |
| 342 | |
| 343 #if IN_GCC | |
| 344 void Module::parse(bool dump_source) | |
| 345 #else | |
| 346 void Module::parse() | |
| 347 #endif | |
| 348 { char *srcname; | |
| 349 unsigned char *buf; | |
| 350 unsigned buflen; | |
| 351 unsigned le; | |
| 352 unsigned bom; | |
| 353 | |
| 354 //printf("Module::parse()\n"); | |
| 355 | |
| 356 srcname = srcfile->name->toChars(); | |
| 357 //printf("Module::parse(srcname = '%s')\n", srcname); | |
| 358 | |
| 359 buf = srcfile->buffer; | |
| 360 buflen = srcfile->len; | |
| 361 | |
| 362 if (buflen >= 2) | |
| 363 { | |
| 364 /* Convert all non-UTF-8 formats to UTF-8. | |
| 365 * BOM : http://www.unicode.org/faq/utf_bom.html | |
| 366 * 00 00 FE FF UTF-32BE, big-endian | |
| 367 * FF FE 00 00 UTF-32LE, little-endian | |
| 368 * FE FF UTF-16BE, big-endian | |
| 369 * FF FE UTF-16LE, little-endian | |
| 370 * EF BB BF UTF-8 | |
| 371 */ | |
| 372 | |
| 373 bom = 1; // assume there's a BOM | |
| 374 if (buf[0] == 0xFF && buf[1] == 0xFE) | |
| 375 { | |
| 376 if (buflen >= 4 && buf[2] == 0 && buf[3] == 0) | |
| 377 { // UTF-32LE | |
| 378 le = 1; | |
| 379 | |
| 380 Lutf32: | |
| 381 OutBuffer dbuf; | |
| 382 unsigned *pu = (unsigned *)(buf); | |
| 383 unsigned *pumax = &pu[buflen / 4]; | |
| 384 | |
| 385 if (buflen & 3) | |
| 386 { error("odd length of UTF-32 char source %u", buflen); | |
| 387 fatal(); | |
| 388 } | |
| 389 | |
| 390 dbuf.reserve(buflen / 4); | |
| 391 for (pu += bom; pu < pumax; pu++) | |
| 392 { unsigned u; | |
| 393 | |
| 394 u = le ? readlongLE(pu) : readlongBE(pu); | |
| 395 if (u & ~0x7F) | |
| 396 { | |
| 397 if (u > 0x10FFFF) | |
| 398 { error("UTF-32 value %08x greater than 0x10FFFF", u); | |
| 399 fatal(); | |
| 400 } | |
| 401 dbuf.writeUTF8(u); | |
| 402 } | |
| 403 else | |
| 404 dbuf.writeByte(u); | |
| 405 } | |
| 406 dbuf.writeByte(0); // add 0 as sentinel for scanner | |
| 407 buflen = dbuf.offset - 1; // don't include sentinel in count | |
| 408 buf = (unsigned char *) dbuf.extractData(); | |
| 409 } | |
| 410 else | |
| 411 { // UTF-16LE (X86) | |
| 412 // Convert it to UTF-8 | |
| 413 le = 1; | |
| 414 | |
| 415 Lutf16: | |
| 416 OutBuffer dbuf; | |
| 417 unsigned short *pu = (unsigned short *)(buf); | |
| 418 unsigned short *pumax = &pu[buflen / 2]; | |
| 419 | |
| 420 if (buflen & 1) | |
| 421 { error("odd length of UTF-16 char source %u", buflen); | |
| 422 fatal(); | |
| 423 } | |
| 424 | |
| 425 dbuf.reserve(buflen / 2); | |
| 426 for (pu += bom; pu < pumax; pu++) | |
| 427 { unsigned u; | |
| 428 | |
| 429 u = le ? readwordLE(pu) : readwordBE(pu); | |
| 430 if (u & ~0x7F) | |
| 431 { if (u >= 0xD800 && u <= 0xDBFF) | |
| 432 { unsigned u2; | |
| 433 | |
| 434 if (++pu > pumax) | |
| 435 { error("surrogate UTF-16 high value %04x at EOF", u); | |
| 436 fatal(); | |
| 437 } | |
| 438 u2 = le ? readwordLE(pu) : readwordBE(pu); | |
| 439 if (u2 < 0xDC00 || u2 > 0xDFFF) | |
| 440 { error("surrogate UTF-16 low value %04x out of range", u2); | |
| 441 fatal(); | |
| 442 } | |
| 443 u = (u - 0xD7C0) << 10; | |
| 444 u |= (u2 - 0xDC00); | |
| 445 } | |
| 446 else if (u >= 0xDC00 && u <= 0xDFFF) | |
| 447 { error("unpaired surrogate UTF-16 value %04x", u); | |
| 448 fatal(); | |
| 449 } | |
| 450 else if (u == 0xFFFE || u == 0xFFFF) | |
| 451 { error("illegal UTF-16 value %04x", u); | |
| 452 fatal(); | |
| 453 } | |
| 454 dbuf.writeUTF8(u); | |
| 455 } | |
| 456 else | |
| 457 dbuf.writeByte(u); | |
| 458 } | |
| 459 dbuf.writeByte(0); // add 0 as sentinel for scanner | |
| 460 buflen = dbuf.offset - 1; // don't include sentinel in count | |
| 461 buf = (unsigned char *) dbuf.extractData(); | |
| 462 } | |
| 463 } | |
| 464 else if (buf[0] == 0xFE && buf[1] == 0xFF) | |
| 465 { // UTF-16BE | |
| 466 le = 0; | |
| 467 goto Lutf16; | |
| 468 } | |
| 469 else if (buflen >= 4 && buf[0] == 0 && buf[1] == 0 && buf[2] == 0xFE && buf[3] == 0xFF) | |
| 470 { // UTF-32BE | |
| 471 le = 0; | |
| 472 goto Lutf32; | |
| 473 } | |
| 474 else if (buflen >= 3 && buf[0] == 0xEF && buf[1] == 0xBB && buf[2] == 0xBF) | |
| 475 { // UTF-8 | |
| 476 | |
| 477 buf += 3; | |
| 478 buflen -= 3; | |
| 479 } | |
| 480 else | |
| 481 { | |
| 482 /* There is no BOM. Make use of Arcane Jill's insight that | |
| 483 * the first char of D source must be ASCII to | |
| 484 * figure out the encoding. | |
| 485 */ | |
| 486 | |
| 487 bom = 0; | |
| 488 if (buflen >= 4) | |
| 489 { if (buf[1] == 0 && buf[2] == 0 && buf[3] == 0) | |
| 490 { // UTF-32LE | |
| 491 le = 1; | |
| 492 goto Lutf32; | |
| 493 } | |
| 494 else if (buf[0] == 0 && buf[1] == 0 && buf[2] == 0) | |
| 495 { // UTF-32BE | |
| 496 le = 0; | |
| 497 goto Lutf32; | |
| 498 } | |
| 499 } | |
| 500 if (buflen >= 2) | |
| 501 { | |
| 502 if (buf[1] == 0) | |
| 503 { // UTF-16LE | |
| 504 le = 1; | |
| 505 goto Lutf16; | |
| 506 } | |
| 507 else if (buf[0] == 0) | |
| 508 { // UTF-16BE | |
| 509 le = 0; | |
| 510 goto Lutf16; | |
| 511 } | |
| 512 } | |
| 513 | |
| 514 // It's UTF-8 | |
| 515 if (buf[0] >= 0x80) | |
| 516 { error("source file must start with BOM or ASCII character, not \\x%02X", buf[0]); | |
| 517 fatal(); | |
| 518 } | |
| 519 } | |
| 520 } | |
| 521 | |
| 522 #ifdef IN_GCC | |
| 523 // dump utf-8 encoded source | |
| 524 if (dump_source) | |
| 525 { // %% srcname could contain a path ... | |
| 526 d_gcc_dump_source(srcname, "utf-8", buf, buflen); | |
| 527 } | |
| 528 #endif | |
| 529 | |
| 530 /* If it starts with the string "Ddoc", then it's a documentation | |
| 531 * source file. | |
| 532 */ | |
| 533 if (buflen >= 4 && memcmp(buf, "Ddoc", 4) == 0) | |
| 534 { | |
| 535 comment = buf + 4; | |
| 536 isDocFile = 1; | |
| 537 return; | |
| 538 } | |
| 539 if (isHtml) | |
| 540 { | |
| 541 OutBuffer *dbuf = new OutBuffer(); | |
| 542 Html h(srcname, buf, buflen); | |
| 543 h.extractCode(dbuf); | |
| 544 buf = dbuf->data; | |
| 545 buflen = dbuf->offset; | |
| 546 #ifdef IN_GCC | |
| 547 // dump extracted source | |
| 548 if (dump_source) | |
| 549 d_gcc_dump_source(srcname, "d.utf-8", buf, buflen); | |
| 550 #endif | |
| 551 } | |
| 552 Parser p(this, buf, buflen, docfile != NULL); | |
| 553 p.nextToken(); | |
| 554 members = p.parseModule(); | |
| 555 md = p.md; | |
| 556 numlines = p.loc.linnum; | |
| 557 | |
| 558 DsymbolTable *dst; | |
| 559 | |
| 560 if (md) | |
| 561 { this->ident = md->id; | |
| 562 dst = Package::resolve(md->packages, &this->parent, NULL); | |
| 563 } | |
| 564 else | |
| 565 { | |
| 566 dst = modules; | |
| 567 | |
| 568 /* Check to see if module name is a valid identifier | |
| 569 */ | |
| 570 if (!Lexer::isValidIdentifier(this->ident->toChars())) | |
| 571 error("has non-identifier characters in filename, use module declaration instead"); | |
| 572 } | |
| 573 | |
| 574 // Update global list of modules | |
| 575 if (!dst->insert(this)) | |
| 576 { | |
| 577 if (md) | |
| 578 error(loc, "is in multiple packages %s", md->toChars()); | |
| 579 else | |
| 580 error(loc, "is in multiple defined"); | |
| 581 } | |
| 582 else | |
| 583 { | |
| 584 amodules.push(this); | |
| 585 } | |
| 586 } | |
| 587 | |
| 588 void Module::semantic() | |
| 589 { int i; | |
| 590 | |
| 591 if (semanticstarted) | |
| 592 return; | |
| 593 | |
| 594 //printf("+Module::semantic(this = %p, '%s'): parent = %p\n", this, toChars(), parent); | |
| 595 semanticstarted = 1; | |
| 596 | |
| 597 // Note that modules get their own scope, from scratch. | |
| 598 // This is so regardless of where in the syntax a module | |
| 599 // gets imported, it is unaffected by context. | |
| 600 Scope *sc = Scope::createGlobal(this); // create root scope | |
| 601 | |
| 602 //printf("Module = %p, linkage = %d\n", sc->scopesym, sc->linkage); | |
| 603 | |
| 604 // Add import of "object" if this module isn't "object" | |
| 605 if (ident != Id::object) | |
| 606 { | |
| 607 Import *im = new Import(0, NULL, Id::object, NULL, 0); | |
| 608 members->shift(im); | |
| 609 } | |
| 610 | |
| 611 // Add all symbols into module's symbol table | |
| 612 symtab = new DsymbolTable(); | |
| 613 for (i = 0; i < members->dim; i++) | |
| 614 { Dsymbol *s; | |
| 615 | |
| 616 s = (Dsymbol *)members->data[i]; | |
| 617 s->addMember(NULL, sc->scopesym, 1); | |
| 618 } | |
| 619 | |
| 620 // Pass 1 semantic routines: do public side of the definition | |
| 621 for (i = 0; i < members->dim; i++) | |
| 622 { Dsymbol *s; | |
| 623 | |
| 624 s = (Dsymbol *)members->data[i]; | |
| 625 //printf("\tModule('%s'): '%s'.semantic()\n", toChars(), s->toChars()); | |
| 626 s->semantic(sc); | |
| 627 runDeferredSemantic(); | |
| 628 } | |
| 629 | |
| 630 sc = sc->pop(); | |
| 631 sc->pop(); | |
| 632 semanticdone = semanticstarted; | |
| 633 //printf("-Module::semantic(this = %p, '%s'): parent = %p\n", this, toChars(), parent); | |
| 634 } | |
| 635 | |
| 636 void Module::semantic2() | |
| 637 { int i; | |
| 638 | |
| 639 if (deferred.dim) | |
| 640 { | |
| 641 for (int i = 0; i < deferred.dim; i++) | |
| 642 { | |
| 643 Dsymbol *sd = (Dsymbol *)deferred.data[i]; | |
| 644 | |
| 645 sd->error("unable to resolve forward reference in definition"); | |
| 646 } | |
| 647 return; | |
| 648 } | |
| 649 //printf("Module::semantic2('%s'): parent = %p\n", toChars(), parent); | |
| 650 if (semanticstarted >= 2) | |
| 651 return; | |
| 652 assert(semanticstarted == 1); | |
| 653 semanticstarted = 2; | |
| 654 | |
| 655 // Note that modules get their own scope, from scratch. | |
| 656 // This is so regardless of where in the syntax a module | |
| 657 // gets imported, it is unaffected by context. | |
| 658 Scope *sc = Scope::createGlobal(this); // create root scope | |
| 659 //printf("Module = %p\n", sc.scopesym); | |
| 660 | |
| 661 // Pass 2 semantic routines: do initializers and function bodies | |
| 662 for (i = 0; i < members->dim; i++) | |
| 663 { Dsymbol *s; | |
| 664 | |
| 665 s = (Dsymbol *)members->data[i]; | |
| 666 s->semantic2(sc); | |
| 667 } | |
| 668 | |
| 669 sc = sc->pop(); | |
| 670 sc->pop(); | |
| 671 semanticdone = semanticstarted; | |
| 672 //printf("-Module::semantic2('%s'): parent = %p\n", toChars(), parent); | |
| 673 } | |
| 674 | |
| 675 void Module::semantic3() | |
| 676 { int i; | |
| 677 | |
| 678 //printf("Module::semantic3('%s'): parent = %p\n", toChars(), parent); | |
| 679 if (semanticstarted >= 3) | |
| 680 return; | |
| 681 assert(semanticstarted == 2); | |
| 682 semanticstarted = 3; | |
| 683 | |
| 684 // Note that modules get their own scope, from scratch. | |
| 685 // This is so regardless of where in the syntax a module | |
| 686 // gets imported, it is unaffected by context. | |
| 687 Scope *sc = Scope::createGlobal(this); // create root scope | |
| 688 //printf("Module = %p\n", sc.scopesym); | |
| 689 | |
| 690 // Pass 3 semantic routines: do initializers and function bodies | |
| 691 for (i = 0; i < members->dim; i++) | |
| 692 { Dsymbol *s; | |
| 693 | |
| 694 s = (Dsymbol *)members->data[i]; | |
| 695 //printf("Module %s: %s.semantic3()\n", toChars(), s->toChars()); | |
| 696 s->semantic3(sc); | |
| 697 } | |
| 698 | |
| 699 sc = sc->pop(); | |
| 700 sc->pop(); | |
| 701 semanticdone = semanticstarted; | |
| 702 } | |
| 703 | |
| 704 void Module::inlineScan() | |
| 705 { int i; | |
| 706 | |
| 707 if (semanticstarted >= 4) | |
| 708 return; | |
| 709 assert(semanticstarted == 3); | |
| 710 semanticstarted = 4; | |
| 711 | |
| 712 // Note that modules get their own scope, from scratch. | |
| 713 // This is so regardless of where in the syntax a module | |
| 714 // gets imported, it is unaffected by context. | |
| 715 //printf("Module = %p\n", sc.scopesym); | |
| 716 | |
| 717 for (i = 0; i < members->dim; i++) | |
| 718 { Dsymbol *s; | |
| 719 | |
| 720 s = (Dsymbol *)members->data[i]; | |
| 721 //if (global.params.verbose) | |
| 722 //printf("inline scan symbol %s\n", s->toChars()); | |
| 723 | |
| 724 s->inlineScan(); | |
| 725 } | |
| 726 semanticdone = semanticstarted; | |
| 727 } | |
| 728 | |
| 729 /**************************************************** | |
| 730 */ | |
| 731 | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
732 // is this used anywhere? |
|
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
733 /* |
| 336 | 734 void Module::gensymfile() |
| 735 { | |
| 736 OutBuffer buf; | |
| 737 HdrGenState hgs; | |
| 738 | |
| 739 //printf("Module::gensymfile()\n"); | |
| 740 | |
| 741 buf.printf("// Sym file generated from '%s'", srcfile->toChars()); | |
| 742 buf.writenl(); | |
| 743 | |
| 744 for (int i = 0; i < members->dim; i++) | |
| 745 { Dsymbol *s = (Dsymbol *)members->data[i]; | |
| 746 | |
| 747 s->toCBuffer(&buf, &hgs); | |
| 748 } | |
| 749 | |
| 750 // Transfer image to file | |
| 751 symfile->setbuffer(buf.data, buf.offset); | |
| 752 buf.data = NULL; | |
| 753 | |
| 754 symfile->writev(); | |
|
580
7824c21a58e3
Restructure path handling a bit. Fixes #66.
Christian Kamm <kamm incasoftware de>
parents:
550
diff
changeset
|
755 }*/ |
| 336 | 756 |
| 757 /********************************** | |
| 758 * Determine if we need to generate an instance of ModuleInfo | |
| 759 * for this Module. | |
| 760 */ | |
| 761 | |
| 762 int Module::needModuleInfo() | |
| 763 { | |
|
550
cbe08531430f
Removed unimplemented switches.
Tomas Lindquist Olsen <tomas.l.olsen@gmail.com>
parents:
510
diff
changeset
|
764 return needmoduleinfo; |
| 336 | 765 } |
| 766 | |
| 767 Dsymbol *Module::search(Loc loc, Identifier *ident, int flags) | |
| 768 { | |
| 769 /* Since modules can be circularly referenced, | |
| 770 * need to stop infinite recursive searches. | |
| 771 */ | |
| 772 | |
| 773 //printf("%s Module::search('%s', flags = %d) insearch = %d\n", toChars(), ident->toChars(), flags, insearch); | |
| 774 Dsymbol *s; | |
| 775 if (insearch) | |
| 776 s = NULL; | |
| 777 else if (searchCacheIdent == ident && searchCacheFlags == flags) | |
| 778 s = searchCacheSymbol; | |
| 779 else | |
| 780 { | |
| 781 insearch = 1; | |
| 782 s = ScopeDsymbol::search(loc, ident, flags); | |
| 783 insearch = 0; | |
| 784 | |
| 785 searchCacheIdent = ident; | |
| 786 searchCacheSymbol = s; | |
| 787 searchCacheFlags = flags; | |
| 788 } | |
| 789 return s; | |
| 790 } | |
| 791 | |
| 792 /******************************************* | |
| 793 * Can't run semantic on s now, try again later. | |
| 794 */ | |
| 795 | |
| 796 void Module::addDeferredSemantic(Dsymbol *s) | |
| 797 { | |
| 798 // Don't add it if it is already there | |
| 799 for (int i = 0; i < deferred.dim; i++) | |
| 800 { | |
| 801 Dsymbol *sd = (Dsymbol *)deferred.data[i]; | |
| 802 | |
| 803 if (sd == s) | |
| 804 return; | |
| 805 } | |
| 806 | |
| 807 //printf("Module::addDeferredSemantic('%s')\n", s->toChars()); | |
| 808 deferred.push(s); | |
| 809 } | |
| 810 | |
| 811 | |
| 812 /****************************************** | |
| 813 * Run semantic() on deferred symbols. | |
| 814 */ | |
| 815 | |
| 816 void Module::runDeferredSemantic() | |
| 817 { | |
| 818 size_t len; | |
| 819 | |
| 820 static int nested; | |
| 821 if (nested) | |
| 822 return; | |
| 823 //if (deferred.dim) printf("+Module::runDeferredSemantic('%s'), len = %d\n", toChars(), deferred.dim); | |
| 824 nested++; | |
| 825 | |
| 826 do | |
| 827 { | |
| 828 dprogress = 0; | |
| 829 len = deferred.dim; | |
| 830 if (!len) | |
| 831 break; | |
| 832 | |
| 833 Dsymbol **todo; | |
| 834 Dsymbol *tmp; | |
| 835 if (len == 1) | |
| 836 { | |
| 837 todo = &tmp; | |
| 838 } | |
| 839 else | |
| 840 { | |
| 841 todo = (Dsymbol **)alloca(len * sizeof(Dsymbol *)); | |
| 842 assert(todo); | |
| 843 } | |
| 844 memcpy(todo, deferred.data, len * sizeof(Dsymbol *)); | |
| 845 deferred.setDim(0); | |
| 846 | |
| 847 for (int i = 0; i < len; i++) | |
| 848 { | |
| 849 Dsymbol *s = todo[i]; | |
| 850 | |
| 851 s->semantic(NULL); | |
| 852 //printf("deferred: %s, parent = %s\n", s->toChars(), s->parent->toChars()); | |
| 853 } | |
| 854 //printf("\tdeferred.dim = %d, len = %d, dprogress = %d\n", deferred.dim, len, dprogress); | |
| 855 } while (deferred.dim < len || dprogress); // while making progress | |
| 856 nested--; | |
| 857 //printf("-Module::runDeferredSemantic('%s'), len = %d\n", toChars(), deferred.dim); | |
| 858 } | |
| 859 | |
| 860 /* =========================== ModuleDeclaration ===================== */ | |
| 861 | |
| 862 ModuleDeclaration::ModuleDeclaration(Array *packages, Identifier *id) | |
| 863 { | |
| 864 this->packages = packages; | |
| 865 this->id = id; | |
| 866 } | |
| 867 | |
| 868 char *ModuleDeclaration::toChars() | |
| 869 { | |
| 870 OutBuffer buf; | |
| 871 int i; | |
| 872 | |
| 873 if (packages && packages->dim) | |
| 874 { | |
| 875 for (i = 0; i < packages->dim; i++) | |
| 876 { Identifier *pid = (Identifier *)packages->data[i]; | |
| 877 | |
| 878 buf.writestring(pid->toChars()); | |
| 879 buf.writeByte('.'); | |
| 880 } | |
| 881 } | |
| 882 buf.writestring(id->toChars()); | |
| 883 buf.writeByte(0); | |
| 884 return (char *)buf.extractData(); | |
| 885 } | |
| 886 | |
| 887 /* =========================== Package ===================== */ | |
| 888 | |
| 889 Package::Package(Identifier *ident) | |
| 890 : ScopeDsymbol(ident) | |
| 891 { | |
| 892 } | |
| 893 | |
| 894 | |
| 895 const char *Package::kind() | |
| 896 { | |
| 897 return "package"; | |
| 898 } | |
| 899 | |
| 900 | |
| 901 DsymbolTable *Package::resolve(Array *packages, Dsymbol **pparent, Package **ppkg) | |
| 902 { | |
| 903 DsymbolTable *dst = Module::modules; | |
| 904 Dsymbol *parent = NULL; | |
| 905 | |
| 906 //printf("Package::resolve()\n"); | |
| 907 if (ppkg) | |
| 908 *ppkg = NULL; | |
| 909 | |
| 910 if (packages) | |
| 911 { int i; | |
| 912 | |
| 913 for (i = 0; i < packages->dim; i++) | |
| 914 { Identifier *pid = (Identifier *)packages->data[i]; | |
| 915 Dsymbol *p; | |
| 916 | |
| 917 p = dst->lookup(pid); | |
| 918 if (!p) | |
| 919 { | |
| 920 p = new Package(pid); | |
| 921 dst->insert(p); | |
| 922 p->parent = parent; | |
| 923 ((ScopeDsymbol *)p)->symtab = new DsymbolTable(); | |
| 924 } | |
| 925 else | |
| 926 { | |
| 927 assert(p->isPackage()); | |
| 928 if (p->isModule()) | |
| 929 { p->error("module and package have the same name"); | |
| 930 fatal(); | |
| 931 break; | |
| 932 } | |
| 933 } | |
| 934 parent = p; | |
| 935 dst = ((Package *)p)->symtab; | |
| 936 if (ppkg && !*ppkg) | |
| 937 *ppkg = (Package *)p; | |
| 938 } | |
| 939 if (pparent) | |
| 940 { | |
| 941 *pparent = parent; | |
| 942 } | |
| 943 } | |
| 944 return dst; | |
| 945 } |