DexFile.h revision 4b0750e8df91220690bb417f45d7ae8b7851b220 
1/*
2 * Copyright (C) 2008 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 *      http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16/*
17 * Access .dex (Dalvik Executable Format) files.  The code here assumes that
18 * the DEX file has been rewritten (byte-swapped, word-aligned) and that
19 * the contents can be directly accessed as a collection of C arrays.  Please
20 * see docs/dalvik/dex-format.html for a detailed description.
21 *
22 * The structure and field names were chosen to match those in the DEX spec.
23 *
24 * It's generally assumed that the DEX file will be stored in shared memory,
25 * obviating the need to copy code and constant pool entries into newly
26 * allocated storage.  Maintaining local pointers to items in the shared area
27 * is valid and encouraged.
28 *
29 * All memory-mapped structures are 32-bit aligned unless otherwise noted.
30 */
31#ifndef _LIBDEX_DEXFILE
32#define _LIBDEX_DEXFILE
33
34#include "vm/Common.h"      // basic type defs, e.g. u1/u2/u4/u8, and LOG
35#include "libdex/SysUtil.h"
36
37/*
38 * gcc-style inline management -- ensures we have a copy of all functions
39 * in the library, so code that links against us will work whether or not
40 * it was built with optimizations enabled.
41 */
42#ifndef _DEX_GEN_INLINES             /* only defined by DexInlines.c */
43# define DEX_INLINE extern __inline__
44#else
45# define DEX_INLINE
46#endif
47
48/* DEX file magic number */
49#define DEX_MAGIC       "dex\n"
50/* version, encoded in 4 bytes of ASCII */
51#define DEX_MAGIC_VERS  "035\0"
52
53/* same, but for optimized DEX header */
54#define DEX_OPT_MAGIC   "dey\n"
55#define DEX_OPT_MAGIC_VERS  "036\0"
56
57#define DEX_DEP_MAGIC   "deps"
58
59/*
60 * 160-bit SHA-1 digest.
61 */
62enum { kSHA1DigestLen = 20,
63       kSHA1DigestOutputLen = kSHA1DigestLen*2 +1 };
64
65/* general constants */
66enum {
67    kDexEndianConstant = 0x12345678,    /* the endianness indicator */
68    kDexNoIndex = 0xffffffff,           /* not a valid index value */
69};
70
71/*
72 * access flags and masks; the "standard" ones are all <= 0x4000
73 *
74 * Note: There are related declarations in vm/oo/Object.h in the ClassFlags
75 * enum.
76 */
77enum {
78    ACC_PUBLIC       = 0x00000001,       // class, field, method, ic
79    ACC_PRIVATE      = 0x00000002,       // field, method, ic
80    ACC_PROTECTED    = 0x00000004,       // field, method, ic
81    ACC_STATIC       = 0x00000008,       // field, method, ic
82    ACC_FINAL        = 0x00000010,       // class, field, method, ic
83    ACC_SYNCHRONIZED = 0x00000020,       // method (only allowed on natives)
84    ACC_SUPER        = 0x00000020,       // class (not used in Dalvik)
85    ACC_VOLATILE     = 0x00000040,       // field
86    ACC_BRIDGE       = 0x00000040,       // method (1.5)
87    ACC_TRANSIENT    = 0x00000080,       // field
88    ACC_VARARGS      = 0x00000080,       // method (1.5)
89    ACC_NATIVE       = 0x00000100,       // method
90    ACC_INTERFACE    = 0x00000200,       // class, ic
91    ACC_ABSTRACT     = 0x00000400,       // class, method, ic
92    ACC_STRICT       = 0x00000800,       // method
93    ACC_SYNTHETIC    = 0x00001000,       // field, method, ic
94    ACC_ANNOTATION   = 0x00002000,       // class, ic (1.5)
95    ACC_ENUM         = 0x00004000,       // class, field, ic (1.5)
96    ACC_CONSTRUCTOR  = 0x00010000,       // method (Dalvik only)
97    ACC_DECLARED_SYNCHRONIZED =
98                       0x00020000,       // method (Dalvik only)
99    ACC_CLASS_MASK =
100        (ACC_PUBLIC | ACC_FINAL | ACC_INTERFACE | ACC_ABSTRACT
101                | ACC_SYNTHETIC | ACC_ANNOTATION | ACC_ENUM),
102    ACC_INNER_CLASS_MASK =
103        (ACC_CLASS_MASK | ACC_PRIVATE | ACC_PROTECTED | ACC_STATIC),
104    ACC_FIELD_MASK =
105        (ACC_PUBLIC | ACC_PRIVATE | ACC_PROTECTED | ACC_STATIC | ACC_FINAL
106                | ACC_VOLATILE | ACC_TRANSIENT | ACC_SYNTHETIC | ACC_ENUM),
107    ACC_METHOD_MASK =
108        (ACC_PUBLIC | ACC_PRIVATE | ACC_PROTECTED | ACC_STATIC | ACC_FINAL
109                | ACC_SYNCHRONIZED | ACC_BRIDGE | ACC_VARARGS | ACC_NATIVE
110                | ACC_ABSTRACT | ACC_STRICT | ACC_SYNTHETIC | ACC_CONSTRUCTOR
111                | ACC_DECLARED_SYNCHRONIZED),
112};
113
114/* annotation constants */
115enum {
116    kDexVisibilityBuild         = 0x00,     /* annotation visibility */
117    kDexVisibilityRuntime       = 0x01,
118    kDexVisibilitySystem        = 0x02,
119
120    kDexAnnotationByte          = 0x00,
121    kDexAnnotationShort         = 0x02,
122    kDexAnnotationChar          = 0x03,
123    kDexAnnotationInt           = 0x04,
124    kDexAnnotationLong          = 0x06,
125    kDexAnnotationFloat         = 0x10,
126    kDexAnnotationDouble        = 0x11,
127    kDexAnnotationString        = 0x17,
128    kDexAnnotationType          = 0x18,
129    kDexAnnotationField         = 0x19,
130    kDexAnnotationMethod        = 0x1a,
131    kDexAnnotationEnum          = 0x1b,
132    kDexAnnotationArray         = 0x1c,
133    kDexAnnotationAnnotation    = 0x1d,
134    kDexAnnotationNull          = 0x1e,
135    kDexAnnotationBoolean       = 0x1f,
136
137    kDexAnnotationValueTypeMask = 0x1f,     /* low 5 bits */
138    kDexAnnotationValueArgShift = 5,
139};
140
141/* map item type codes */
142enum {
143    kDexTypeHeaderItem               = 0x0000,
144    kDexTypeStringIdItem             = 0x0001,
145    kDexTypeTypeIdItem               = 0x0002,
146    kDexTypeProtoIdItem              = 0x0003,
147    kDexTypeFieldIdItem              = 0x0004,
148    kDexTypeMethodIdItem             = 0x0005,
149    kDexTypeClassDefItem             = 0x0006,
150    kDexTypeMapList                  = 0x1000,
151    kDexTypeTypeList                 = 0x1001,
152    kDexTypeAnnotationSetRefList     = 0x1002,
153    kDexTypeAnnotationSetItem        = 0x1003,
154    kDexTypeClassDataItem            = 0x2000,
155    kDexTypeCodeItem                 = 0x2001,
156    kDexTypeStringDataItem           = 0x2002,
157    kDexTypeDebugInfoItem            = 0x2003,
158    kDexTypeAnnotationItem           = 0x2004,
159    kDexTypeEncodedArrayItem         = 0x2005,
160    kDexTypeAnnotationsDirectoryItem = 0x2006,
161};
162
163/* auxillary data section chunk codes */
164enum {
165    kDexChunkClassLookup            = 0x434c4b50,   /* CLKP */
166    kDexChunkRegisterMaps           = 0x524d4150,   /* RMAP */
167
168    kDexChunkReducingIndexMap       = 0x5249584d,   /* RIXM */
169    kDexChunkExpandingIndexMap      = 0x4549584d,   /* EIXM */
170
171    kDexChunkEnd                    = 0x41454e44,   /* AEND */
172};
173
174/* debug info opcodes and constants */
175enum {
176    DBG_END_SEQUENCE         = 0x00,
177    DBG_ADVANCE_PC           = 0x01,
178    DBG_ADVANCE_LINE         = 0x02,
179    DBG_START_LOCAL          = 0x03,
180    DBG_START_LOCAL_EXTENDED = 0x04,
181    DBG_END_LOCAL            = 0x05,
182    DBG_RESTART_LOCAL        = 0x06,
183    DBG_SET_PROLOGUE_END     = 0x07,
184    DBG_SET_EPILOGUE_BEGIN   = 0x08,
185    DBG_SET_FILE             = 0x09,
186    DBG_FIRST_SPECIAL        = 0x0a,
187    DBG_LINE_BASE            = -4,
188    DBG_LINE_RANGE           = 15,
189};
190
191/*
192 * Direct-mapped "header_item" struct.
193 */
194typedef struct DexHeader {
195    u1  magic[8];           /* includes version number */
196    u4  checksum;           /* adler32 checksum */
197    u1  signature[kSHA1DigestLen]; /* SHA-1 hash */
198    u4  fileSize;           /* length of entire file */
199    u4  headerSize;         /* offset to start of next section */
200    u4  endianTag;
201    u4  linkSize;
202    u4  linkOff;
203    u4  mapOff;
204    u4  stringIdsSize;
205    u4  stringIdsOff;
206    u4  typeIdsSize;
207    u4  typeIdsOff;
208    u4  protoIdsSize;
209    u4  protoIdsOff;
210    u4  fieldIdsSize;
211    u4  fieldIdsOff;
212    u4  methodIdsSize;
213    u4  methodIdsOff;
214    u4  classDefsSize;
215    u4  classDefsOff;
216    u4  dataSize;
217    u4  dataOff;
218} DexHeader;
219
220/*
221 * Direct-mapped "map_item".
222 */
223typedef struct DexMapItem {
224    u2  type;              /* type code (see kDexType* above) */
225    u2  unused;
226    u4  size;              /* count of items of the indicated type */
227    u4  offset;            /* file offset to the start of data */
228} DexMapItem;
229
230/*
231 * Direct-mapped "map_list".
232 */
233typedef struct DexMapList {
234    u4  size;               /* #of entries in list */
235    DexMapItem list[1];     /* entries */
236} DexMapList;
237
238/*
239 * Direct-mapped "string_id_item".
240 */
241typedef struct DexStringId {
242    u4  stringDataOff;      /* file offset to string_data_item */
243} DexStringId;
244
245/*
246 * Direct-mapped "type_id_item".
247 */
248typedef struct DexTypeId {
249    u4  descriptorIdx;      /* index into stringIds list for type descriptor */
250} DexTypeId;
251
252/*
253 * Direct-mapped "field_id_item".
254 */
255typedef struct DexFieldId {
256    u2  classIdx;           /* index into typeIds list for defining class */
257    u2  typeIdx;            /* index into typeIds for field type */
258    u4  nameIdx;            /* index into stringIds for field name */
259} DexFieldId;
260
261/*
262 * Direct-mapped "method_id_item".
263 */
264typedef struct DexMethodId {
265    u2  classIdx;           /* index into typeIds list for defining class */
266    u2  protoIdx;           /* index into protoIds for method prototype */
267    u4  nameIdx;            /* index into stringIds for method name */
268} DexMethodId;
269
270/*
271 * Direct-mapped "proto_id_item".
272 */
273typedef struct DexProtoId {
274    u4  shortyIdx;          /* index into stringIds for shorty descriptor */
275    u4  returnTypeIdx;      /* index into typeIds list for return type */
276    u4  parametersOff;      /* file offset to type_list for parameter types */
277} DexProtoId;
278
279/*
280 * Direct-mapped "class_def_item".
281 */
282typedef struct DexClassDef {
283    u4  classIdx;           /* index into typeIds for this class */
284    u4  accessFlags;
285    u4  superclassIdx;      /* index into typeIds for superclass */
286    u4  interfacesOff;      /* file offset to DexTypeList */
287    u4  sourceFileIdx;      /* index into stringIds for source file name */
288    u4  annotationsOff;     /* file offset to annotations_directory_item */
289    u4  classDataOff;       /* file offset to class_data_item */
290    u4  staticValuesOff;    /* file offset to DexEncodedArray */
291} DexClassDef;
292
293/*
294 * Direct-mapped "type_item".
295 */
296typedef struct DexTypeItem {
297    u2  typeIdx;            /* index into typeIds */
298} DexTypeItem;
299
300/*
301 * Direct-mapped "type_list".
302 */
303typedef struct DexTypeList {
304    u4  size;               /* #of entries in list */
305    DexTypeItem list[1];    /* entries */
306} DexTypeList;
307
308/*
309 * Direct-mapped "code_item".
310 *
311 * The "catches" table is used when throwing an exception,
312 * "debugInfo" is used when displaying an exception stack trace or
313 * debugging. An offset of zero indicates that there are no entries.
314 */
315typedef struct DexCode {
316    u2  registersSize;
317    u2  insSize;
318    u2  outsSize;
319    u2  triesSize;
320    u4  debugInfoOff;       /* file offset to debug info stream */
321    u4  insnsSize;          /* size of the insns array, in u2 units */
322    u2  insns[1];
323    /* followed by optional u2 padding */
324    /* followed by try_item[triesSize] */
325    /* followed by uleb128 handlersSize */
326    /* followed by catch_handler_item[handlersSize] */
327} DexCode;
328
329/*
330 * Direct-mapped "try_item".
331 */
332typedef struct DexTry {
333    u4  startAddr;          /* start address, in 16-bit code units */
334    u2  insnCount;          /* instruction count, in 16-bit code units */
335    u2  handlerOff;         /* offset in encoded handler data to handlers */
336} DexTry;
337
338/*
339 * Link table.  Currently undefined.
340 */
341typedef struct DexLink {
342    u1  bleargh;
343} DexLink;
344
345
346/*
347 * Direct-mapped "annotations_directory_item".
348 */
349typedef struct DexAnnotationsDirectoryItem {
350    u4  classAnnotationsOff;  /* offset to DexAnnotationSetItem */
351    u4  fieldsSize;           /* count of DexFieldAnnotationsItem */
352    u4  methodsSize;          /* count of DexMethodAnnotationsItem */
353    u4  parametersSize;       /* count of DexParameterAnnotationsItem */
354    /* followed by DexFieldAnnotationsItem[fieldsSize] */
355    /* followed by DexMethodAnnotationsItem[methodsSize] */
356    /* followed by DexParameterAnnotationsItem[parametersSize] */
357} DexAnnotationsDirectoryItem;
358
359/*
360 * Direct-mapped "field_annotations_item".
361 */
362typedef struct DexFieldAnnotationsItem {
363    u4  fieldIdx;
364    u4  annotationsOff;             /* offset to DexAnnotationSetItem */
365} DexFieldAnnotationsItem;
366
367/*
368 * Direct-mapped "method_annotations_item".
369 */
370typedef struct DexMethodAnnotationsItem {
371    u4  methodIdx;
372    u4  annotationsOff;             /* offset to DexAnnotationSetItem */
373} DexMethodAnnotationsItem;
374
375/*
376 * Direct-mapped "parameter_annotations_item".
377 */
378typedef struct DexParameterAnnotationsItem {
379    u4  methodIdx;
380    u4  annotationsOff;             /* offset to DexAnotationSetRefList */
381} DexParameterAnnotationsItem;
382
383/*
384 * Direct-mapped "annotation_set_ref_item".
385 */
386typedef struct DexAnnotationSetRefItem {
387    u4  annotationsOff;             /* offset to DexAnnotationSetItem */
388} DexAnnotationSetRefItem;
389
390/*
391 * Direct-mapped "annotation_set_ref_list".
392 */
393typedef struct DexAnnotationSetRefList {
394    u4  size;
395    DexAnnotationSetRefItem list[1];
396} DexAnnotationSetRefList;
397
398/*
399 * Direct-mapped "anotation_set_item".
400 */
401typedef struct DexAnnotationSetItem {
402    u4  size;
403    u4  entries[1];                 /* offset to DexAnnotationItem */
404} DexAnnotationSetItem;
405
406/*
407 * Direct-mapped "annotation_item".
408 *
409 * NOTE: this structure is byte-aligned.
410 */
411typedef struct DexAnnotationItem {
412    u1  visibility;
413    u1  annotation[1];              /* data in encoded_annotation format */
414} DexAnnotationItem;
415
416/*
417 * Direct-mapped "encoded_array".
418 *
419 * NOTE: this structure is byte-aligned.
420 */
421typedef struct DexEncodedArray {
422    u1  array[1];                   /* data in encoded_array format */
423} DexEncodedArray;
424
425/*
426 * Lookup table for classes.  It provides a mapping from class name to
427 * class definition.  Used by dexFindClass().
428 *
429 * We calculate this at DEX optimization time and embed it in the file so we
430 * don't need the same hash table in every VM.  This is slightly slower than
431 * a hash table with direct pointers to the items, but because it's shared
432 * there's less of a penalty for using a fairly sparse table.
433 */
434typedef struct DexClassLookup {
435    int     size;                       // total size, including "size"
436    int     numEntries;                 // size of table[]; always power of 2
437    struct {
438        u4      classDescriptorHash;    // class descriptor hash code
439        int     classDescriptorOffset;  // in bytes, from start of DEX
440        int     classDefOffset;         // in bytes, from start of DEX
441    } table[1];
442} DexClassLookup;
443
444/*
445 * Map constant pool indices from one form to another.  Some or all of these
446 * may be NULL.
447 *
448 * The map values are 16-bit unsigned values.  If the values we map to
449 * require a larger range, we omit the mapping for that category (which
450 * requires that the lookup code recognize that the data will not be
451 * there for all DEX files in all categories.)
452 */
453typedef struct DexIndexMap {
454    const u2* classMap;         /* map, either expanding or reducing */
455    u4  classFullCount;         /* same as typeIdsSize */
456    u4  classReducedCount;      /* post-reduction count */
457    const u2* methodMap;
458    u4  methodFullCount;
459    u4  methodReducedCount;
460    const u2* fieldMap;
461    u4  fieldFullCount;
462    u4  fieldReducedCount;
463    const u2* stringMap;
464    u4  stringFullCount;
465    u4  stringReducedCount;
466} DexIndexMap;
467
468/*
469 * Header added by DEX optimization pass.  Values are always written in
470 * local byte and structure padding.  The first field (magic + version)
471 * is guaranteed to be present and directly readable for all expected
472 * compiler configurations; the rest is version-dependent.
473 *
474 * Try to keep this simple and fixed-size.
475 */
476typedef struct DexOptHeader {
477    u1  magic[8];           /* includes version number */
478
479    u4  dexOffset;          /* file offset of DEX header */
480    u4  dexLength;
481    u4  depsOffset;         /* offset of optimized DEX dependency table */
482    u4  depsLength;
483    u4  auxOffset;          /* file offset of pre-calc auxillary data */
484    u4  auxLength;
485
486    u4  flags;              /* some info flags */
487    u4  checksum;           /* adler32 checksum covering deps/aux */
488
489    /* pad for 64-bit alignment if necessary */
490} DexOptHeader;
491
492#define DEX_FLAG_VERIFIED           (1)     /* tried to verify all classes */
493#define DEX_OPT_FLAG_BIG            (1<<1)  /* swapped to big-endian */
494#define DEX_OPT_FLAG_FIELDS         (1<<2)  /* field access optimized */
495#define DEX_OPT_FLAG_INVOCATIONS    (1<<3)  /* method calls optimized */
496
497#define DEX_INTERFACE_CACHE_SIZE    128     /* must be power of 2 */
498
499/*
500 * Structure representing a DEX file.
501 *
502 * Code should regard DexFile as opaque, using the API calls provided here
503 * to access specific structures.
504 */
505typedef struct DexFile {
506    /* directly-mapped "opt" header */
507    const DexOptHeader* pOptHeader;
508
509    /* pointers to directly-mapped structs and arrays in base DEX */
510    const DexHeader*    pHeader;
511    const DexStringId*  pStringIds;
512    const DexTypeId*    pTypeIds;
513    const DexFieldId*   pFieldIds;
514    const DexMethodId*  pMethodIds;
515    const DexProtoId*   pProtoIds;
516    const DexClassDef*  pClassDefs;
517    const DexLink*      pLinkData;
518
519    /*
520     * These are mapped out of the "auxillary" section, and may not be
521     * included in the file.
522     */
523    const DexClassLookup* pClassLookup;
524    DexIndexMap         indexMap;
525    const void*         pRegisterMapPool;       // RegisterMapClassPool
526
527    /* points to start of DEX file data */
528    const u1*           baseAddr;
529
530    /* track memory overhead for auxillary structures */
531    int                 overhead;
532
533    /* additional app-specific data structures associated with the DEX */
534    //void*               auxData;
535} DexFile;
536
537/*
538 * Utility function -- rounds up to the nearest power of 2.
539 */
540u4 dexRoundUpPower2(u4 val);
541
542/*
543 * Parse an optimized or unoptimized .dex file sitting in memory.
544 *
545 * On success, return a newly-allocated DexFile.
546 */
547DexFile* dexFileParse(const u1* data, size_t length, int flags);
548
549/* bit values for "flags" argument to dexFileParse */
550enum {
551    kDexParseDefault            = 0,
552    kDexParseVerifyChecksum     = 1,
553    kDexParseContinueOnError    = (1 << 1),
554};
555
556/*
557 * Fix the byte ordering of all fields in the DEX file, and do
558 * structural verification. This is only required for code that opens
559 * "raw" DEX files, such as the DEX optimizer.
560 *
561 * Return 0 on success.
562 */
563int dexSwapAndVerify(u1* addr, int len);
564
565/*
566 * Detect the file type of the given memory buffer via magic number.
567 * Call dexSwapAndVerify() on an unoptimized DEX file, do nothing
568 * but return successfully on an optimized DEX file, and report an
569 * error for all other cases.
570 *
571 * Return 0 on success.
572 */
573int dexSwapAndVerifyIfNecessary(u1* addr, int len);
574
575/*
576 * Compute DEX checksum.
577 */
578u4 dexComputeChecksum(const DexHeader* pHeader);
579
580/*
581 * Free a DexFile structure, along with any associated structures.
582 */
583void dexFileFree(DexFile* pDexFile);
584
585/*
586 * Create class lookup table.
587 */
588DexClassLookup* dexCreateClassLookup(DexFile* pDexFile);
589
590/*
591 * Find a class definition by descriptor.
592 */
593const DexClassDef* dexFindClass(const DexFile* pFile, const char* descriptor);
594
595/*
596 * Set up the basic raw data pointers of a DexFile. This function isn't
597 * meant for general use.
598 */
599void dexFileSetupBasicPointers(DexFile* pDexFile, const u1* data);
600
601/* return the DexMapList of the file, if any */
602DEX_INLINE const DexMapList* dexGetMap(const DexFile* pDexFile) {
603    u4 mapOff = pDexFile->pHeader->mapOff;
604
605    if (mapOff == 0) {
606        return NULL;
607    } else {
608        return (const DexMapList*) (pDexFile->baseAddr + mapOff);
609    }
610}
611
612/* return the const char* string data referred to by the given string_id */
613DEX_INLINE const char* dexGetStringData(const DexFile* pDexFile,
614        const DexStringId* pStringId) {
615    const u1* ptr = pDexFile->baseAddr + pStringId->stringDataOff;
616
617    // Skip the uleb128 length.
618    while (*(ptr++) > 0x7f) /* empty */ ;
619
620    return (const char*) ptr;
621}
622/* return the StringId with the specified index */
623DEX_INLINE const DexStringId* dexGetStringId(const DexFile* pDexFile, u4 idx) {
624    assert(idx < pDexFile->pHeader->stringIdsSize);
625    return &pDexFile->pStringIds[idx];
626}
627/* return the UTF-8 encoded string with the specified string_id index */
628DEX_INLINE const char* dexStringById(const DexFile* pDexFile, u4 idx) {
629    const DexStringId* pStringId = dexGetStringId(pDexFile, idx);
630    return dexGetStringData(pDexFile, pStringId);
631}
632
633/* Return the UTF-8 encoded string with the specified string_id index,
634 * also filling in the UTF-16 size (number of 16-bit code points).*/
635const char* dexStringAndSizeById(const DexFile* pDexFile, u4 idx,
636        u4* utf16Size);
637
638/* return the TypeId with the specified index */
639DEX_INLINE const DexTypeId* dexGetTypeId(const DexFile* pDexFile, u4 idx) {
640    assert(idx < pDexFile->pHeader->typeIdsSize);
641    return &pDexFile->pTypeIds[idx];
642}
643
644/*
645 * Get the descriptor string associated with a given type index.
646 * The caller should not free() the returned string.
647 */
648DEX_INLINE const char* dexStringByTypeIdx(const DexFile* pDexFile, u4 idx) {
649    const DexTypeId* typeId = dexGetTypeId(pDexFile, idx);
650    return dexStringById(pDexFile, typeId->descriptorIdx);
651}
652
653/* return the MethodId with the specified index */
654DEX_INLINE const DexMethodId* dexGetMethodId(const DexFile* pDexFile, u4 idx) {
655    assert(idx < pDexFile->pHeader->methodIdsSize);
656    return &pDexFile->pMethodIds[idx];
657}
658
659/* return the FieldId with the specified index */
660DEX_INLINE const DexFieldId* dexGetFieldId(const DexFile* pDexFile, u4 idx) {
661    assert(idx < pDexFile->pHeader->fieldIdsSize);
662    return &pDexFile->pFieldIds[idx];
663}
664
665/* return the ProtoId with the specified index */
666DEX_INLINE const DexProtoId* dexGetProtoId(const DexFile* pDexFile, u4 idx) {
667    assert(idx < pDexFile->pHeader->protoIdsSize);
668    return &pDexFile->pProtoIds[idx];
669}
670
671/*
672 * Get the parameter list from a ProtoId. The returns NULL if the ProtoId
673 * does not have a parameter list.
674 */
675DEX_INLINE const DexTypeList* dexGetProtoParameters(
676    const DexFile *pDexFile, const DexProtoId* pProtoId) {
677    if (pProtoId->parametersOff == 0) {
678        return NULL;
679    }
680    return (const DexTypeList*)
681        (pDexFile->baseAddr + pProtoId->parametersOff);
682}
683
684/* return the ClassDef with the specified index */
685DEX_INLINE const DexClassDef* dexGetClassDef(const DexFile* pDexFile, u4 idx) {
686    assert(idx < pDexFile->pHeader->classDefsSize);
687    return &pDexFile->pClassDefs[idx];
688}
689
690/* given a ClassDef pointer, recover its index */
691DEX_INLINE u4 dexGetIndexForClassDef(const DexFile* pDexFile,
692    const DexClassDef* pClassDef)
693{
694    assert(pClassDef >= pDexFile->pClassDefs &&
695           pClassDef < pDexFile->pClassDefs + pDexFile->pHeader->classDefsSize);
696    return pClassDef - pDexFile->pClassDefs;
697}
698
699/* get the interface list for a DexClass */
700DEX_INLINE const DexTypeList* dexGetInterfacesList(const DexFile* pDexFile,
701    const DexClassDef* pClassDef)
702{
703    if (pClassDef->interfacesOff == 0)
704        return NULL;
705    return (const DexTypeList*)
706        (pDexFile->baseAddr + pClassDef->interfacesOff);
707}
708/* return the Nth entry in a DexTypeList. */
709DEX_INLINE const DexTypeItem* dexGetTypeItem(const DexTypeList* pList,
710    u4 idx)
711{
712    assert(idx < pList->size);
713    return &pList->list[idx];
714}
715/* return the type_idx for the Nth entry in a TypeList */
716DEX_INLINE u4 dexTypeListGetIdx(const DexTypeList* pList, u4 idx) {
717    const DexTypeItem* pItem = dexGetTypeItem(pList, idx);
718    return pItem->typeIdx;
719}
720
721/* get the static values list for a DexClass */
722DEX_INLINE const DexEncodedArray* dexGetStaticValuesList(
723    const DexFile* pDexFile, const DexClassDef* pClassDef)
724{
725    if (pClassDef->staticValuesOff == 0)
726        return NULL;
727    return (const DexEncodedArray*)
728        (pDexFile->baseAddr + pClassDef->staticValuesOff);
729}
730
731/* get the annotations directory item for a DexClass */
732DEX_INLINE const DexAnnotationsDirectoryItem* dexGetAnnotationsDirectoryItem(
733    const DexFile* pDexFile, const DexClassDef* pClassDef)
734{
735    if (pClassDef->annotationsOff == 0)
736        return NULL;
737    return (const DexAnnotationsDirectoryItem*)
738        (pDexFile->baseAddr + pClassDef->annotationsOff);
739}
740
741/* get the source file string */
742DEX_INLINE const char* dexGetSourceFile(
743    const DexFile* pDexFile, const DexClassDef* pClassDef)
744{
745    if (pClassDef->sourceFileIdx == 0xffffffff)
746        return NULL;
747    return dexStringById(pDexFile, pClassDef->sourceFileIdx);
748}
749
750/* get the size, in bytes, of a DexCode */
751size_t dexGetDexCodeSize(const DexCode* pCode);
752
753/* Get the list of "tries" for the given DexCode. */
754DEX_INLINE const DexTry* dexGetTries(const DexCode* pCode) {
755    const u2* insnsEnd = &pCode->insns[pCode->insnsSize];
756
757    // Round to four bytes.
758    if ((((u4) insnsEnd) & 3) != 0) {
759        insnsEnd++;
760    }
761
762    return (const DexTry*) insnsEnd;
763}
764
765/* Get the base of the encoded data for the given DexCode. */
766DEX_INLINE const u1* dexGetCatchHandlerData(const DexCode* pCode) {
767    const DexTry* pTries = dexGetTries(pCode);
768    return (const u1*) &pTries[pCode->triesSize];
769}
770
771/* get a pointer to the start of the debugging data */
772DEX_INLINE const u1* dexGetDebugInfoStream(const DexFile* pDexFile,
773    const DexCode* pCode)
774{
775    if (pCode->debugInfoOff == 0) {
776        return NULL;
777    } else {
778        return pDexFile->baseAddr + pCode->debugInfoOff;
779    }
780}
781
782/*
783 * Callback for "new position table entry".
784 * Returning non-0 causes the decoder to stop early.
785 */
786typedef int (*DexDebugNewPositionCb)(void *cnxt, u4 address, u4 lineNum);
787
788/*
789 * Callback for "new locals table entry". "signature" is an empty string
790 * if no signature is available for an entry.
791 */
792typedef void (*DexDebugNewLocalCb)(void *cnxt, u2 reg, u4 startAddress,
793        u4 endAddress, const char *name, const char *descriptor,
794        const char *signature);
795
796/*
797 * Decode debug info for method.
798 *
799 * posCb is called in ascending address order.
800 * localCb is called in order of ascending end address.
801 */
802void dexDecodeDebugInfo(
803            const DexFile* pDexFile,
804            const DexCode* pDexCode,
805            const char* classDescriptor,
806            u4 protoIdx,
807            u4 accessFlags,
808            DexDebugNewPositionCb posCb, DexDebugNewLocalCb localCb,
809            void* cnxt);
810
811/* DexClassDef convenience - get class descriptor */
812DEX_INLINE const char* dexGetClassDescriptor(const DexFile* pDexFile,
813    const DexClassDef* pClassDef)
814{
815    return dexStringByTypeIdx(pDexFile, pClassDef->classIdx);
816}
817
818/* DexClassDef convenience - get superclass descriptor */
819DEX_INLINE const char* dexGetSuperClassDescriptor(const DexFile* pDexFile,
820    const DexClassDef* pClassDef)
821{
822    if (pClassDef->superclassIdx == 0)
823        return NULL;
824    return dexStringByTypeIdx(pDexFile, pClassDef->superclassIdx);
825}
826
827/* DexClassDef convenience - get class_data_item pointer */
828DEX_INLINE const u1* dexGetClassData(const DexFile* pDexFile,
829    const DexClassDef* pClassDef)
830{
831    if (pClassDef->classDataOff == 0)
832        return NULL;
833    return (const u1*) (pDexFile->baseAddr + pClassDef->classDataOff);
834}
835
836/* Get an annotation set at a particular offset. */
837DEX_INLINE const DexAnnotationSetItem* dexGetAnnotationSetItem(
838    const DexFile* pDexFile, u4 offset)
839{
840    return (const DexAnnotationSetItem*) (pDexFile->baseAddr + offset);
841}
842/* get the class' annotation set */
843DEX_INLINE const DexAnnotationSetItem* dexGetClassAnnotationSet(
844    const DexFile* pDexFile, const DexAnnotationsDirectoryItem* pAnnoDir)
845{
846    if (pAnnoDir->classAnnotationsOff == 0)
847        return NULL;
848    return dexGetAnnotationSetItem(pDexFile, pAnnoDir->classAnnotationsOff);
849}
850
851/* get the class' field annotation list */
852DEX_INLINE const DexFieldAnnotationsItem* dexGetFieldAnnotations(
853    const DexFile* pDexFile, const DexAnnotationsDirectoryItem* pAnnoDir)
854{
855    if (pAnnoDir->fieldsSize == 0)
856        return NULL;
857
858    // Skip past the header to the start of the field annotations.
859    return (const DexFieldAnnotationsItem*) &pAnnoDir[1];
860}
861
862/* get field annotation list size */
863DEX_INLINE int dexGetFieldAnnotationsSize(const DexFile* pDexFile,
864    const DexAnnotationsDirectoryItem* pAnnoDir)
865{
866    return pAnnoDir->fieldsSize;
867}
868
869/* return a pointer to the field's annotation set */
870DEX_INLINE const DexAnnotationSetItem* dexGetFieldAnnotationSetItem(
871    const DexFile* pDexFile, const DexFieldAnnotationsItem* pItem)
872{
873    return dexGetAnnotationSetItem(pDexFile, pItem->annotationsOff);
874}
875
876/* get the class' method annotation list */
877DEX_INLINE const DexMethodAnnotationsItem* dexGetMethodAnnotations(
878    const DexFile* pDexFile, const DexAnnotationsDirectoryItem* pAnnoDir)
879{
880    if (pAnnoDir->methodsSize == 0)
881        return NULL;
882
883    /*
884     * Skip past the header and field annotations to the start of the
885     * method annotations.
886     */
887    const u1* addr = (const u1*) &pAnnoDir[1];
888    addr += pAnnoDir->fieldsSize * sizeof (DexFieldAnnotationsItem);
889    return (const DexMethodAnnotationsItem*) addr;
890}
891
892/* get method annotation list size */
893DEX_INLINE int dexGetMethodAnnotationsSize(const DexFile* pDexFile,
894    const DexAnnotationsDirectoryItem* pAnnoDir)
895{
896    return pAnnoDir->methodsSize;
897}
898
899/* return a pointer to the method's annotation set */
900DEX_INLINE const DexAnnotationSetItem* dexGetMethodAnnotationSetItem(
901    const DexFile* pDexFile, const DexMethodAnnotationsItem* pItem)
902{
903    return dexGetAnnotationSetItem(pDexFile, pItem->annotationsOff);
904}
905
906/* get the class' parameter annotation list */
907DEX_INLINE const DexParameterAnnotationsItem* dexGetParameterAnnotations(
908    const DexFile* pDexFile, const DexAnnotationsDirectoryItem* pAnnoDir)
909{
910    if (pAnnoDir->parametersSize == 0)
911        return NULL;
912
913    /*
914     * Skip past the header, field annotations, and method annotations
915     * to the start of the parameter annotations.
916     */
917    const u1* addr = (const u1*) &pAnnoDir[1];
918    addr += pAnnoDir->fieldsSize * sizeof (DexFieldAnnotationsItem);
919    addr += pAnnoDir->methodsSize * sizeof (DexMethodAnnotationsItem);
920    return (const DexParameterAnnotationsItem*) addr;
921}
922
923/* get method annotation list size */
924DEX_INLINE int dexGetParameterAnnotationsSize(const DexFile* pDexFile,
925    const DexAnnotationsDirectoryItem* pAnnoDir)
926{
927    return pAnnoDir->parametersSize;
928}
929
930/* return the parameter annotation ref list */
931DEX_INLINE const DexAnnotationSetRefList* dexGetParameterAnnotationSetRefList(
932    const DexFile* pDexFile, const DexParameterAnnotationsItem* pItem)
933{
934    return (const DexAnnotationSetRefList*)
935        (pDexFile->baseAddr + pItem->annotationsOff);
936}
937
938/* get method annotation list size */
939DEX_INLINE int dexGetParameterAnnotationSetRefSize(const DexFile* pDexFile,
940    const DexParameterAnnotationsItem* pItem)
941{
942    if (pItem->annotationsOff == 0)
943        return 0;
944    return dexGetParameterAnnotationSetRefList(pDexFile, pItem)->size;
945}
946
947/* return the Nth entry from an annotation set ref list */
948DEX_INLINE const DexAnnotationSetRefItem* dexGetParameterAnnotationSetRef(
949    const DexAnnotationSetRefList* pList, u4 idx)
950{
951    assert(idx < pList->size);
952    return &pList->list[idx];
953}
954
955/* given a DexAnnotationSetRefItem, return the DexAnnotationSetItem */
956DEX_INLINE const DexAnnotationSetItem* dexGetSetRefItemItem(
957    const DexFile* pDexFile, const DexAnnotationSetRefItem* pItem)
958{
959    return dexGetAnnotationSetItem(pDexFile, pItem->annotationsOff);
960}
961
962/* return the Nth annotation offset from a DexAnnotationSetItem */
963DEX_INLINE u4 dexGetAnnotationOff(
964    const DexAnnotationSetItem* pAnnoSet, u4 idx)
965{
966    assert(idx < pAnnoSet->size);
967    return pAnnoSet->entries[idx];
968}
969
970/* return the Nth annotation item from a DexAnnotationSetItem */
971DEX_INLINE const DexAnnotationItem* dexGetAnnotationItem(
972    const DexFile* pDexFile, const DexAnnotationSetItem* pAnnoSet, u4 idx)
973{
974    return (const DexAnnotationItem*)
975        (pDexFile->baseAddr + dexGetAnnotationOff(pAnnoSet, idx));
976}
977
978
979/*
980 * ===========================================================================
981 *      Utility Functions
982 * ===========================================================================
983 */
984
985/*
986 * Retrieve the next UTF-16 character from a UTF-8 string.
987 *
988 * Advances "*pUtf8Ptr" to the start of the next character.
989 *
990 * WARNING: If a string is corrupted by dropping a '\0' in the middle
991 * of a 3-byte sequence, you can end up overrunning the buffer with
992 * reads (and possibly with the writes if the length was computed and
993 * cached before the damage). For performance reasons, this function
994 * assumes that the string being parsed is known to be valid (e.g., by
995 * already being verified). Most strings we process here are coming
996 * out of dex files or other internal translations, so the only real
997 * risk comes from the JNI NewStringUTF call.
998 */
999DEX_INLINE u2 dexGetUtf16FromUtf8(const char** pUtf8Ptr)
1000{
1001    unsigned int one, two, three;
1002
1003    one = *(*pUtf8Ptr)++;
1004    if ((one & 0x80) != 0) {
1005        /* two- or three-byte encoding */
1006        two = *(*pUtf8Ptr)++;
1007        if ((one & 0x20) != 0) {
1008            /* three-byte encoding */
1009            three = *(*pUtf8Ptr)++;
1010            return ((one & 0x0f) << 12) |
1011                   ((two & 0x3f) << 6) |
1012                   (three & 0x3f);
1013        } else {
1014            /* two-byte encoding */
1015            return ((one & 0x1f) << 6) |
1016                   (two & 0x3f);
1017        }
1018    } else {
1019        /* one-byte encoding */
1020        return one;
1021    }
1022}
1023
1024/* Compare two '\0'-terminated modified UTF-8 strings, using Unicode
1025 * code point values for comparison. This treats different encodings
1026 * for the same code point as equivalent, except that only a real '\0'
1027 * byte is considered the string terminator. The return value is as
1028 * for strcmp(). */
1029int dexUtf8Cmp(const char* s1, const char* s2);
1030
1031
1032/* for dexIsValidMemberNameUtf8(), a bit vector indicating valid low ascii */
1033extern u4 DEX_MEMBER_VALID_LOW_ASCII[4];
1034
1035/* Helper for dexIsValidMemberUtf8(); do not call directly. */
1036bool dexIsValidMemberNameUtf8_0(const char** pUtf8Ptr);
1037
1038/* Return whether the pointed-at modified-UTF-8 encoded character is
1039 * valid as part of a member name, updating the pointer to point past
1040 * the consumed character. This will consume two encoded UTF-16 code
1041 * points if the character is encoded as a surrogate pair. Also, if
1042 * this function returns false, then the given pointer may only have
1043 * been partially advanced. */
1044DEX_INLINE bool dexIsValidMemberNameUtf8(const char** pUtf8Ptr) {
1045    u1 c = (u1) **pUtf8Ptr;
1046    if (c <= 0x7f) {
1047        // It's low-ascii, so check the table.
1048        u4 wordIdx = c >> 5;
1049        u4 bitIdx = c & 0x1f;
1050        (*pUtf8Ptr)++;
1051        return (DEX_MEMBER_VALID_LOW_ASCII[wordIdx] & (1 << bitIdx)) != 0;
1052    }
1053
1054    /*
1055     * It's a multibyte encoded character. Call a non-inline function
1056     * for the heavy lifting.
1057     */
1058    return dexIsValidMemberNameUtf8_0(pUtf8Ptr);
1059}
1060
1061/* Return whether the given string is a valid field or method name. */
1062bool dexIsValidMemberName(const char* s);
1063
1064/* Return whether the given string is a valid type descriptor. */
1065bool dexIsValidTypeDescriptor(const char* s);
1066
1067/* Return whether the given string is a valid reference descriptor. This
1068 * is true if dexIsValidTypeDescriptor() returns true and the descriptor
1069 * is for a class or array and not a primitive type. */
1070bool dexIsReferenceDescriptor(const char* s);
1071
1072/* Return whether the given string is a valid class descriptor. This
1073 * is true if dexIsValidTypeDescriptor() returns true and the descriptor
1074 * is for a class and not an array or primitive type. */
1075bool dexIsClassDescriptor(const char* s);
1076
1077/* Return whether the given string is a valid field type descriptor. This
1078 * is true if dexIsValidTypeDescriptor() returns true and the descriptor
1079 * is for anything but "void". */
1080bool dexIsFieldDescriptor(const char* s);
1081
1082#endif /*_LIBDEX_DEXFILE*/
1083