2003-12-26 Guilhem Lavaux <guilhem@kaffe.org>
[official-gcc.git] / gcc / java / jcf-io.c
blob101be8ee1d3a389c54dd76f321ebd610c5a8adbc
1 /* Utility routines for finding and reading Java(TM) .class files.
2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2002, 2003
3 Free Software Foundation, Inc.
5 This file is part of GCC.
7 GCC is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
10 any later version.
12 GCC is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with GCC; see the file COPYING. If not, write to
19 the Free Software Foundation, 59 Temple Place - Suite 330,
20 Boston, MA 02111-1307, USA.
22 Java and all Java-based marks are trademarks or registered trademarks
23 of Sun Microsystems, Inc. in the United States and other countries.
24 The Free Software Foundation is independent of Sun Microsystems, Inc. */
26 /* Written by Per Bothner <bothner@cygnus.com>, February 1996. */
28 #include "config.h"
29 #include "system.h"
30 #include "coretypes.h"
31 #include "tm.h"
33 #include "jcf.h"
34 #include "tree.h"
35 #include "toplev.h"
36 #include "java-tree.h"
37 #include "hashtab.h"
38 #if JCF_USE_SCANDIR
39 #include <dirent.h>
40 #include <fnmatch.h>
41 #endif
43 #include "zlib.h"
45 /* DOS brain-damage */
46 #ifndef O_BINARY
47 #define O_BINARY 0 /* MS-DOS brain-damage */
48 #endif
50 int
51 jcf_unexpected_eof (JCF *jcf, int count ATTRIBUTE_UNUSED)
53 if (jcf->filename)
54 fprintf (stderr, "Premature end of .class file %s.\n", jcf->filename);
55 else
56 fprintf (stderr, "Premature end of .class file <stdin>.\n");
57 exit (-1);
60 void
61 jcf_trim_old_input (JCF *jcf)
63 int count = jcf->read_ptr - jcf->buffer;
64 if (count > 0)
66 memmove (jcf->buffer, jcf->read_ptr, jcf->read_end - jcf->read_ptr);
67 jcf->read_ptr -= count;
68 jcf->read_end -= count;
72 int
73 jcf_filbuf_from_stdio (JCF *jcf, int count)
75 FILE *file = (FILE*) (jcf->read_state);
76 if (count > jcf->buffer_end - jcf->read_ptr)
78 JCF_u4 old_read_ptr = jcf->read_ptr - jcf->buffer;
79 JCF_u4 old_read_end = jcf->read_end - jcf->buffer;
80 JCF_u4 old_size = jcf->buffer_end - jcf->buffer;
81 JCF_u4 new_size = (old_size == 0 ? 2000 : 2 * old_size) + count;
82 unsigned char *new_buffer = jcf->buffer == NULL ? ALLOC (new_size)
83 : REALLOC (jcf->buffer, new_size);
84 jcf->buffer = new_buffer;
85 jcf->buffer_end = new_buffer + new_size;
86 jcf->read_ptr = new_buffer + old_read_ptr;
87 jcf->read_end = new_buffer + old_read_end;
89 count -= jcf->read_end - jcf->read_ptr;
90 if (count <= 0)
91 return 0;
92 if ((int) fread (jcf->read_end, 1, count, file) != count)
93 jcf_unexpected_eof (jcf, count);
94 jcf->read_end += count;
95 return 0;
98 #include "zipfile.h"
100 struct ZipFile *SeenZipFiles = NULL;
102 /* Open a zip file with the given name, and cache directory and file
103 descriptor. If the file is missing, treat it as an empty archive.
104 Return NULL if the .zip file is malformed.
107 ZipFile *
108 opendir_in_zip (const char *zipfile, int is_system)
110 struct ZipFile* zipf;
111 char magic [4];
112 int fd;
113 for (zipf = SeenZipFiles; zipf != NULL; zipf = zipf->next)
115 if (strcmp (zipf->name, zipfile) == 0)
116 return zipf;
119 zipf = ALLOC (sizeof (struct ZipFile) + strlen (zipfile) + 1);
120 zipf->next = SeenZipFiles;
121 zipf->name = (char*)(zipf+1);
122 strcpy (zipf->name, zipfile);
123 SeenZipFiles = zipf;
124 fd = open (zipfile, O_RDONLY | O_BINARY);
125 zipf->fd = fd;
126 if (fd < 0)
128 /* A missing zip file is not considered an error.
129 We may want to re-consider that. FIXME. */
130 zipf->count = 0;
131 zipf->dir_size = 0;
132 zipf->central_directory = NULL;
134 else
136 jcf_dependency_add_file (zipfile, is_system);
137 if (read (fd, magic, 4) != 4 || GET_u4 (magic) != (JCF_u4)ZIPMAGIC)
138 return NULL;
139 lseek (fd, 0L, SEEK_SET);
140 if (read_zip_archive (zipf) != 0)
141 return NULL;
143 return zipf;
146 /* Returns:
147 0: OK - zipmember found.
148 -1: Not found.
149 -2: Malformed archive.
153 open_in_zip (JCF *jcf, const char *zipfile, const char *zipmember,
154 int is_system)
156 ZipDirectory *zipd;
157 int i, len;
158 ZipFile *zipf = opendir_in_zip (zipfile, is_system);
160 if (zipf == NULL)
161 return -2;
163 if (!zipmember)
164 return 0;
166 len = strlen (zipmember);
168 zipd = (struct ZipDirectory*) zipf->central_directory;
169 for (i = 0; i < zipf->count; i++, zipd = ZIPDIR_NEXT (zipd))
171 if (len == zipd->filename_length &&
172 strncmp (ZIPDIR_FILENAME (zipd), zipmember, len) == 0)
174 JCF_ZERO (jcf);
176 jcf->filename = xstrdup (zipfile);
177 jcf->classname = xstrdup (zipmember);
178 return read_zip_member(jcf, zipd, zipf);
181 return -1;
184 /* Read data from zip archive member. */
187 read_zip_member (JCF *jcf, ZipDirectory *zipd, ZipFile *zipf)
189 jcf->filbuf = jcf_unexpected_eof;
190 jcf->zipd = (void *)zipd;
192 if (zipd->compression_method == Z_NO_COMPRESSION)
194 jcf->buffer = ALLOC (zipd->size);
195 jcf->buffer_end = jcf->buffer + zipd->size;
196 jcf->read_ptr = jcf->buffer;
197 jcf->read_end = jcf->buffer_end;
198 if (lseek (zipf->fd, zipd->filestart, 0) < 0
199 || read (zipf->fd, jcf->buffer, zipd->size) != (long) zipd->size)
200 return -2;
202 else
204 char *buffer;
205 z_stream d_stream; /* decompression stream */
206 d_stream.zalloc = (alloc_func) 0;
207 d_stream.zfree = (free_func) 0;
208 d_stream.opaque = (voidpf) 0;
210 jcf->buffer = ALLOC (zipd->uncompressed_size);
211 d_stream.next_out = jcf->buffer;
212 d_stream.avail_out = zipd->uncompressed_size;
213 jcf->buffer_end = jcf->buffer + zipd->uncompressed_size;
214 jcf->read_ptr = jcf->buffer;
215 jcf->read_end = jcf->buffer_end;
216 buffer = ALLOC (zipd->size);
217 d_stream.next_in = buffer;
218 d_stream.avail_in = zipd->size;
219 if (lseek (zipf->fd, zipd->filestart, 0) < 0
220 || read (zipf->fd, buffer, zipd->size) != (long) zipd->size)
221 return -2;
222 /* Handle NO_HEADER using undocumented zlib feature.
223 This is a very common hack. */
224 inflateInit2 (&d_stream, -MAX_WBITS);
225 inflate (&d_stream, Z_NO_FLUSH);
226 inflateEnd (&d_stream);
227 FREE (buffer);
230 return 0;
233 const char *
234 open_class (const char *filename, JCF *jcf, int fd, const char *dep_name)
236 if (jcf)
238 struct stat stat_buf;
239 if (fstat (fd, &stat_buf) != 0
240 || ! S_ISREG (stat_buf.st_mode))
242 perror ("Could not figure length of .class file");
243 return NULL;
245 if (dep_name != NULL)
246 jcf_dependency_add_file (dep_name, 0);
247 JCF_ZERO (jcf);
248 jcf->buffer = ALLOC (stat_buf.st_size);
249 jcf->buffer_end = jcf->buffer + stat_buf.st_size;
250 jcf->read_ptr = jcf->buffer;
251 jcf->read_end = jcf->buffer_end;
252 jcf->read_state = NULL;
253 jcf->filename = filename;
254 if (read (fd, jcf->buffer, stat_buf.st_size) != stat_buf.st_size)
256 perror ("Failed to read .class file");
257 return NULL;
259 close (fd);
260 jcf->filbuf = jcf_unexpected_eof;
262 else
263 close (fd);
264 return filename;
268 const char *
269 find_classfile (char *filename, JCF *jcf, const char *dep_name)
271 int fd = open (filename, O_RDONLY | O_BINARY);
272 if (fd < 0)
273 return NULL;
274 return open_class (filename, jcf, fd, dep_name);
277 #if JCF_USE_SCANDIR
279 /* A comparison function (as for qsort) that compares KEY (a char *
280 giving the basename of a file) with the name stored in ENTRY (a
281 dirent **). */
283 static int
284 compare_path (const void *key, const void *entry)
286 return strcmp ((const char *) key,
287 (*((const struct dirent **) entry))->d_name);
290 /* Returns nonzero if ENTRY names a .java or .class file. */
292 static int
293 java_or_class_file (const struct dirent *entry)
295 const char *base = lbasename (entry->d_name);
296 return (fnmatch ("*.java", base, 0) == 0 ||
297 fnmatch ("*.class", base, 0) == 0);
300 /* Information about the files present in a particular directory. */
301 typedef struct memoized_dirlist_entry
303 /* The name of the directory. */
304 const char *dir;
305 /* The number of .java and .class files present, or -1 if we could
306 not, for some reason, obtain the list. */
307 int num_files;
308 /* The .java and .class files in the directory, in alphabetical
309 order. */
310 struct dirent **files;
311 } memoized_dirlist_entry;
313 /* Returns true if ENTRY (a memoized_dirlist_entry *) corresponds to
314 the directory given by KEY (a char *) giving the directory
315 name. */
317 static int
318 memoized_dirlist_lookup_eq (const void *entry, const void *key)
320 return strcmp ((const char *) key,
321 ((const memoized_dirlist_entry *) entry)->dir) == 0;
324 /* A hash table mapping directory names to the lists of .java and
325 .class files in that directory. */
327 static htab_t memoized_dirlists;
329 #endif
331 /* Like stat, but avoids actually making the stat system call if we
332 know that it cannot succeed. FILENAME and BUF are as for stat. */
334 static int
335 caching_stat (char *filename, struct stat *buf)
337 #if JCF_USE_SCANDIR
338 char *sep;
339 char origsep = 0;
340 char *base;
341 memoized_dirlist_entry *dent;
342 void **slot;
344 /* If the hashtable has not already been created, create it now. */
345 if (!memoized_dirlists)
346 memoized_dirlists = htab_create (37,
347 htab_hash_string,
348 memoized_dirlist_lookup_eq,
349 NULL);
351 /* Get the name of the directory. */
352 sep = strrchr (filename, DIR_SEPARATOR);
353 #ifdef DIR_SEPARATOR_2
354 if (! sep)
355 sep = strrchr (filename, DIR_SEPARATOR_2);
356 #endif
357 if (sep)
359 origsep = *sep;
360 *sep = '\0';
361 base = sep + 1;
363 else
364 base = filename;
366 /* Obtain the entry for this directory from the hash table. */
367 slot = htab_find_slot (memoized_dirlists, filename, INSERT);
368 if (!*slot)
370 /* We have not already scanned this directory; scan it now. */
371 dent = ((memoized_dirlist_entry *)
372 ALLOC (sizeof (memoized_dirlist_entry)));
373 dent->dir = xstrdup (filename);
374 /* Unfortunately, scandir is not fully standardized. In
375 particular, the type of the function pointer passed as the
376 third argument sometimes takes a "const struct dirent *"
377 parameter, and sometimes just a "struct dirent *". We cast
378 to (void *) so that either way it is quietly accepted. */
379 dent->num_files = scandir (filename, &dent->files,
380 (void *) java_or_class_file,
381 alphasort);
382 *slot = dent;
384 else
385 dent = *((memoized_dirlist_entry **) slot);
387 /* Put the separator back. */
388 if (sep)
389 *sep = origsep;
391 /* If the file is not in the list, there is no need to stat it; it
392 does not exist. */
393 if (dent->num_files != -1
394 && !bsearch (base, dent->files, dent->num_files,
395 sizeof (struct dirent *), compare_path))
396 return -1;
397 #endif
399 return stat (filename, buf);
402 /* Returns 1 if the CLASSNAME (really a char *) matches the name
403 stored in TABLE_ENTRY (also a char *). */
405 static int
406 memoized_class_lookup_eq (const void *table_entry, const void *classname)
408 return strcmp ((const char *)classname, (const char *)table_entry) == 0;
411 /* A hash table keeping track of class names that were not found
412 during class lookup. (There is no need to cache the values
413 associated with names that were found; they are saved in
414 IDENTIFIER_CLASS_VALUE.) */
415 static htab_t memoized_class_lookups;
417 /* Returns a freshly malloc'd string with the fully qualified pathname
418 of the .class file for the class CLASSNAME. CLASSNAME must be
419 allocated in permanent storage; this function may retain a pointer
420 to it. Returns NULL on failure. If JCF != NULL, it is suitably
421 initialized. SOURCE_OK is true if we should also look for .java
422 file. */
424 const char *
425 find_class (const char *classname, int classname_length, JCF *jcf,
426 int source_ok)
428 int fd;
429 int i, k, java = -1, class = -1;
430 struct stat java_buf, class_buf;
431 char *dep_file;
432 void *entry;
433 char *java_buffer;
434 int buflen;
435 char *buffer;
436 hashval_t hash;
438 /* Create the hash table, if it does not already exist. */
439 if (!memoized_class_lookups)
440 memoized_class_lookups = htab_create (37,
441 htab_hash_string,
442 memoized_class_lookup_eq,
443 NULL);
445 /* Loop for this class in the hashtable. If it is present, we've
446 already looked for this class and failed to find it. */
447 hash = htab_hash_string (classname);
448 if (htab_find_with_hash (memoized_class_lookups, classname, hash))
449 return NULL;
451 /* Allocate and zero out the buffer, since we don't explicitly put a
452 null pointer when we're copying it below. */
453 buflen = jcf_path_max_len () + classname_length + 10;
454 buffer = ALLOC (buflen);
455 memset (buffer, 0, buflen);
457 java_buffer = alloca (buflen);
459 jcf->java_source = 0;
461 for (entry = jcf_path_start (); entry != NULL; entry = jcf_path_next (entry))
463 const char *path_name = jcf_path_name (entry);
464 if (class != 0)
466 int dir_len;
468 strcpy (buffer, path_name);
469 i = strlen (buffer);
471 /* This is right because we know that `.zip' entries will have a
472 trailing slash. See jcf-path.c. */
473 dir_len = i - 1;
475 for (k = 0; k < classname_length; k++, i++)
477 char ch = classname[k];
478 buffer[i] = ch == '.' ? '/' : ch;
480 strcpy (buffer+i, ".class");
482 if (jcf_path_is_zipfile (entry))
484 int err_code;
485 JCF _jcf;
486 buffer[dir_len] = '\0';
487 SOURCE_FRONTEND_DEBUG
488 (("Trying [...%s]:%s",
489 &buffer[dir_len-(dir_len > 15 ? 15 : dir_len)],
490 buffer+dir_len+1));
491 if (jcf == NULL)
492 jcf = &_jcf;
493 err_code = open_in_zip (jcf, buffer, buffer+dir_len+1,
494 jcf_path_is_system (entry));
495 if (err_code == 0)
497 /* Should we check if .zip is out-of-date wrt .java? */
498 buffer[dir_len] = '(';
499 strcpy (buffer+i, ".class)");
500 if (jcf == &_jcf)
501 JCF_FINISH (jcf);
502 return buffer;
504 else
505 continue;
507 class = caching_stat(buffer, &class_buf);
510 if (source_ok)
512 /* Compute name of .java file. */
513 int l, m;
514 strcpy (java_buffer, path_name);
515 l = strlen (java_buffer);
516 for (m = 0; m < classname_length; ++m)
517 java_buffer[m + l] = (classname[m] == '.'
518 ? DIR_SEPARATOR : classname[m]);
519 strcpy (java_buffer + m + l, ".java");
520 java = caching_stat (java_buffer, &java_buf);
521 if (java == 0)
522 break;
526 /* We preferably pick a class file if we have a chance. If the source
527 file is newer than the class file, we issue a warning and parse the
528 source file instead.
529 There should be a flag to allow people have the class file picked
530 up no matter what. FIXME. */
531 if (! java && ! class && java_buf.st_mtime > class_buf.st_mtime)
533 if (flag_newer)
534 warning ("source file for class `%s' is newer than its matching class file. Source file `%s' used instead", classname, java_buffer);
535 class = -1;
538 if (! java)
539 dep_file = java_buffer;
540 else
541 dep_file = buffer;
542 if (!class)
544 SOURCE_FRONTEND_DEBUG ((stderr, "[Class selected: %s]\n",
545 classname+classname_length-
546 (classname_length <= 30 ?
547 classname_length : 30)));
548 fd = JCF_OPEN_EXACT_CASE (buffer, O_RDONLY | O_BINARY);
549 if (fd >= 0)
550 goto found;
552 /* Give .java a try, if necessary */
553 if (!java)
555 strcpy (buffer, java_buffer);
556 SOURCE_FRONTEND_DEBUG ((stderr, "[Source selected: %s]\n",
557 classname+classname_length-
558 (classname_length <= 30 ?
559 classname_length : 30)));
560 fd = JCF_OPEN_EXACT_CASE (buffer, O_RDONLY);
561 if (fd >= 0)
563 jcf->java_source = 1;
564 goto found;
568 free (buffer);
570 /* Remember that this class could not be found so that we do not
571 have to look again. */
572 *htab_find_slot_with_hash (memoized_class_lookups, classname, hash, INSERT)
573 = (void *) classname;
575 return NULL;
576 found:
577 if (jcf->java_source)
579 JCF_ZERO (jcf); /* JCF_FINISH relies on this */
580 jcf->java_source = 1;
581 jcf->filename = xstrdup (buffer);
582 close (fd); /* We use STDIO for source file */
584 else
585 buffer = (char *) open_class (buffer, jcf, fd, dep_file);
586 jcf->classname = xstrdup (classname);
587 return buffer;
590 void
591 jcf_print_char (FILE *stream, int ch)
593 switch (ch)
595 case '\'':
596 case '\\':
597 case '\"':
598 fprintf (stream, "\\%c", ch);
599 break;
600 case '\n':
601 fprintf (stream, "\\n");
602 break;
603 case '\t':
604 fprintf (stream, "\\t");
605 break;
606 case '\r':
607 fprintf (stream, "\\r");
608 break;
609 default:
610 if (ch >= ' ' && ch < 127)
611 putc (ch, stream);
612 else if (ch < 256)
613 fprintf (stream, "\\%03x", ch);
614 else
615 fprintf (stream, "\\u%04x", ch);
619 /* Print UTF8 string at STR of length LENGTH bytes to STREAM. */
621 void
622 jcf_print_utf8 (FILE *stream, const unsigned char *str, int length)
624 const unsigned char * limit = str + length;
625 while (str < limit)
627 int ch = UTF8_GET (str, limit);
628 if (ch < 0)
630 fprintf (stream, "\\<invalid>");
631 return;
633 jcf_print_char (stream, ch);
637 /* Same as jcf_print_utf8, but print IN_CHAR as OUT_CHAR. */
639 void
640 jcf_print_utf8_replace (FILE *stream, const unsigned char *str, int length,
641 int in_char, int out_char)
643 const unsigned char *limit = str + length;
644 while (str < limit)
646 int ch = UTF8_GET (str, limit);
647 if (ch < 0)
649 fprintf (stream, "\\<invalid>");
650 return;
652 jcf_print_char (stream, ch == in_char ? out_char : ch);
656 /* Check that all the cross-references in the constant pool are
657 valid. Returns 0 on success.
658 Otherwise, returns the index of the (first) invalid entry.
659 Only checks internal consistency, but does not check that
660 any classes, fields, or methods are valid.*/
663 verify_constant_pool (JCF *jcf)
665 int i, n;
666 for (i = 1; i < JPOOL_SIZE (jcf); i++)
668 switch (JPOOL_TAG (jcf, i))
670 case CONSTANT_NameAndType:
671 n = JPOOL_USHORT2 (jcf, i);
672 if (n <= 0 || n >= JPOOL_SIZE(jcf)
673 || JPOOL_TAG (jcf, n) != CONSTANT_Utf8)
674 return i;
675 /* ... fall through ... */
676 case CONSTANT_Class:
677 case CONSTANT_String:
678 n = JPOOL_USHORT1 (jcf, i);
679 if (n <= 0 || n >= JPOOL_SIZE(jcf)
680 || JPOOL_TAG (jcf, n) != CONSTANT_Utf8)
681 return i;
682 break;
683 case CONSTANT_Fieldref:
684 case CONSTANT_Methodref:
685 case CONSTANT_InterfaceMethodref:
686 n = JPOOL_USHORT1 (jcf, i);
687 if (n <= 0 || n >= JPOOL_SIZE(jcf)
688 || JPOOL_TAG (jcf, n) != CONSTANT_Class)
689 return i;
690 n = JPOOL_USHORT2 (jcf, i);
691 if (n <= 0 || n >= JPOOL_SIZE(jcf)
692 || JPOOL_TAG (jcf, n) != CONSTANT_NameAndType)
693 return i;
694 break;
695 case CONSTANT_Long:
696 case CONSTANT_Double:
697 i++;
698 break;
699 case CONSTANT_Float:
700 case CONSTANT_Integer:
701 case CONSTANT_Utf8:
702 case CONSTANT_Unicode:
703 break;
704 default:
705 return i;
708 return 0;
711 void
712 format_uint (char *buffer, uint64 value, int base)
714 #define WRITE_BUF_SIZE (4 + sizeof(uint64) * 8)
715 char buf[WRITE_BUF_SIZE];
716 char *buf_ptr = buf+WRITE_BUF_SIZE; /* End of buf. */
717 int chars_written;
718 int i;
720 /* Now do the actual conversion, placing the result at the *end* of buf. */
721 /* Note this code does not pretend to be optimized. */
722 do {
723 int digit = value % base;
724 static const char digit_chars[] = "0123456789abcdefghijklmnopqrstuvwxyz";
725 *--buf_ptr = digit_chars[digit];
726 value /= base;
727 } while (value != 0);
729 chars_written = buf+WRITE_BUF_SIZE - buf_ptr;
730 for (i = 0; i < chars_written; i++)
731 buffer[i] = *buf_ptr++;
732 buffer[i] = 0;
735 void
736 format_int (char *buffer, jlong value, int base)
738 uint64 abs_value;
739 if (value < 0)
741 abs_value = -(uint64)value;
742 *buffer++ = '-';
744 else
745 abs_value = (uint64) value;
746 format_uint (buffer, abs_value, base);