• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
2
3<html>
4
5<head>
6<title>.dex &mdash; Dalvik Executable Format</title>
7<link rel=stylesheet href="dex-format.css">
8</head>
9
10<body>
11
12<h1 class="title"><code>.dex</code> &mdash; Dalvik Executable Format</h1>
13<p>Copyright &copy; 2007 The Android Open Source Project
14
15<p>This document describes the layout and contents of <code>.dex</code>
16files, which are used to hold a set of class definitions and their associated
17adjunct data.</p>
18
19<h1>Guide To Types</h1>
20
21<table class="guide">
22<thead>
23<tr>
24  <th>Name</th>
25  <th>Description</th>
26</tr>
27</thead>
28<tbody>
29<tr>
30  <td>byte</td>
31  <td>8-bit signed int</td>
32</tr>
33<tr>
34  <td>ubyte</td>
35  <td>8-bit unsigned int</td>
36</tr>
37<tr>
38  <td>short</td>
39  <td>16-bit signed int, little-endian</td>
40</tr>
41<tr>
42  <td>ushort</td>
43  <td>16-bit unsigned int, little-endian</td>
44</tr>
45<tr>
46  <td>int</td>
47  <td>32-bit signed int, little-endian</td>
48</tr>
49<tr>
50  <td>uint</td>
51  <td>32-bit unsigned int, little-endian</td>
52</tr>
53<tr>
54  <td>long</td>
55  <td>64-bit signed int, little-endian</td>
56</tr>
57<tr>
58  <td>ulong</td>
59  <td>64-bit unsigned int, little-endian</td>
60</tr>
61<tr>
62  <td>sleb128</td>
63  <td>signed LEB128, variable-length (see below)</td>
64</tr>
65<tr>
66  <td>uleb128</td>
67  <td>unsigned LEB128, variable-length (see below)</td>
68</tr>
69<tr>
70  <td>uleb128p1</td>
71  <td>unsigned LEB128 plus <code>1</code>, variable-length (see below)</td>
72</tr>
73</tbody>
74</table>
75
76<h3>LEB128</h3>
77
78<p>LEB128 ("<b>L</b>ittle-<b>E</b>ndian <b>B</b>ase <b>128</b>") is a
79variable-length encoding for
80arbitrary signed or unsigned integer quantities. The format was
81borrowed from the <a href="http://dwarfstd.org/Dwarf3Std.php">DWARF3</a>
82specification. In a <code>.dex</code> file, LEB128 is only ever used to
83encode 32-bit quantities.</p>
84
85<p>Each LEB128 encoded value consists of one to five
86bytes, which together represent a single 32-bit value. Each
87byte has its most significant bit set except for the final byte in the
88sequence, which has its most significant bit clear. The remaining
89seven bits of each byte are payload, with the least significant seven
90bits of the quantity in the first byte, the next seven in the second
91byte and so on. In the case of a signed LEB128 (<code>sleb128</code>),
92the most significant payload bit of the final byte in the sequence is
93sign-extended to produce the final value. In the unsigned case
94(<code>uleb128</code>), any bits not explicitly represented are
95interpreted as <code>0</code>.
96
97<table class="leb128Bits">
98<thead>
99<tr><th colspan="16">Bitwise diagram of a two-byte LEB128 value</th></tr>
100<tr>
101  <th colspan="8">First byte</td>
102  <th colspan="8">Second byte</td>
103</tr>
104</thead>
105<tbody>
106<tr>
107  <td class="start1"><code>1</code></td>
108  <td>bit<sub>6</sub></td>
109  <td>bit<sub>5</sub></td>
110  <td>bit<sub>4</sub></td>
111  <td>bit<sub>3</sub></td>
112  <td>bit<sub>2</sub></td>
113  <td>bit<sub>1</sub></td>
114  <td>bit<sub>0</sub></td>
115  <td class="start2"><code>0</code></td>
116  <td>bit<sub>13</sub></td>
117  <td>bit<sub>12</sub></td>
118  <td>bit<sub>11</sub></td>
119  <td>bit<sub>10</sub></td>
120  <td>bit<sub>9</sub></td>
121  <td>bit<sub>8</sub></td>
122  <td class="end2">bit<sub>7</sub></td>
123</tr>
124</tbody>
125</table>
126
127<p>The variant <code>uleb128p1</code> is used to represent a signed
128value, where the representation is of the value <i>plus one</i> encoded
129as a <code>uleb128</code>. This makes the encoding of <code>-1</code>
130(alternatively thought of as the unsigned value <code>0xffffffff</code>)
131&mdash; but no other negative number &mdash; a single byte, and is
132useful in exactly those cases where the represented number must either
133be non-negative or <code>-1</code> (or <code>0xffffffff</code>),
134and where no other negative values are allowed (or where large unsigned
135values are unlikely to be needed).</p>
136
137<p>Here are some examples of the formats:</p>
138
139<table class="leb128">
140<thead>
141<tr>
142  <th>Encoded Sequence</th>
143  <th>As <code>sleb128</code></th>
144  <th>As <code>uleb128</code></th>
145  <th>As <code>uleb128p1</code></th>
146</tr>
147</thead>
148<tbody>
149  <tr><td>00</td><td>0</td><td>0</td><td>-1</td></tr>
150  <tr><td>01</td><td>1</td><td>1</td><td>0</td></tr>
151  <tr><td>7f</td><td>-1</td><td>127</td><td>126</td></tr>
152  <tr><td>80 7f</td><td>-128</td><td>16256</td><td>16255</td></tr>
153</tbody>
154</table>
155
156<h1>Overall File Layout</h1>
157
158<table class="format">
159<thead>
160<tr>
161  <th>Name</th>
162  <th>Format</th>
163  <th>Description</th>
164</tr>
165</thead>
166<tbody>
167<tr>
168  <td>header</td>
169  <td>header_item</td>
170  <td>the header</td>
171</tr>
172<tr>
173  <td>string_ids</td>
174  <td>string_id_item[]</td>
175  <td>string identifiers list. These are identifiers for all the strings
176    used by this file, either for internal naming (e.g., type descriptors)
177    or as constant objects referred to by code. This list must be sorted
178    by string contents, using UTF-16 code point values (not in a
179    locale-sensitive manner).
180  </td>
181</tr>
182<tr>
183  <td>type_ids</td>
184  <td>type_id_item[]</td>
185  <td>type identifiers list. These are identifiers for all types (classes,
186    arrays, or primitive types) referred to by this file, whether defined
187    in the file or not. This list must be sorted by <code>string_id</code>
188    index.
189  </td>
190</tr>
191<tr>
192  <td>proto_ids</td>
193  <td>proto_id_item[]</td>
194  <td>method prototype identifiers list. These are identifiers for all
195    prototypes referred to by this file. This list must be sorted in
196    return-type (by <code>type_id</code> index) major order, and then
197    by arguments (also by <code>type_id</code> index).
198  </td>
199</tr>
200<tr>
201  <td>field_ids</td>
202  <td>field_id_item[]</td>
203  <td>field identifiers list. These are identifiers for all fields
204    referred to by this file, whether defined in the file or not. This
205    list must be sorted, where the defining type (by <code>type_id</code>
206    index) is the major order, field name (by <code>string_id</code> index)
207    is the intermediate order, and type (by <code>type_id</code> index)
208    is the minor order.
209  </td>
210</tr>
211<tr>
212  <td>method_ids</td>
213  <td>method_id_item[]</td>
214  <td>method identifiers list. These are identifiers for all methods
215    referred to by this file, whether defined in the file or not. This
216    list must be sorted, where the defining type (by <code>type_id</code>
217    index) is the major order, method name (by <code>string_id</code>
218    index) is the intermediate order, and method
219    prototype (by <code>proto_id</code> index) is the minor order.
220  </td>
221</tr>
222<tr>
223  <td>class_defs</td>
224  <td>class_def_item[]</td>
225  <td>class definitions list. The classes must be ordered such that a given
226    class's superclass and implemented interfaces appear in the
227    list earlier than the referring class.
228  </td>
229</tr>
230<tr>
231  <td>data</td>
232  <td>ubyte[]</td>
233  <td>data area, containing all the support data for the tables listed above.
234    Different items have different alignment requirements, and
235    padding bytes are inserted before each item if necessary to achieve
236    proper alignment.
237  </td>
238</tr>
239<tr>
240  <td>link_data</td>
241  <td>ubyte[]</td>
242  <td>data used in statically linked files. The format of the data in
243    this section is left unspecified by this document;
244    this section is empty in unlinked files, and runtime implementations
245    may use it as they see fit.
246  </td>
247</tr>
248</tbody>
249</table>
250
251<h1>Bitfield, String, and Constant Definitions</h1>
252
253<h2><code>DEX_FILE_MAGIC</code></h2>
254<h4>embedded in <code>header_item</code></h4>
255
256<p>The constant array/string <code>DEX_FILE_MAGIC</code> is the list of
257bytes that must appear at the beginning of a <code>.dex</code> file
258in order for it to be recognized as such. The value intentionally
259contains a newline (<code>"\n"</code> or <code>0x0a</code>) and a
260null byte (<code>"\0"</code> or <code>0x00</code>) in order to help
261in the detection of certain forms of corruption. The value also
262encodes a format version number as three decimal digits, which is
263expected to increase monotonically over time as the format evolves.</p>
264
265<pre>
266ubyte[8] DEX_FILE_MAGIC = { 0x64 0x65 0x78 0x0a 0x30 0x33 0x35 0x00 }
267                        = "dex\n035\0"
268</pre>
269
270<p><b>Note:</b> At least a couple earlier versions of the format have
271been used in widely-available public software releases. For example,
272version <code>009</code> was used for the M3 releases of the
273Android platform (November-December 2007),
274and version <code>013</code> was used for the M5 releases of the Android
275platform (February-March 2008). In several respects, these earlier versions
276of the format differ significantly from the version described in this
277document.</p>
278
279<h2><code>ENDIAN_CONSTANT</code> and <code>REVERSE_ENDIAN_CONSTANT</code></h2>
280<h4>embedded in <code>header_item</code></h4>
281
282<p>The constant <code>ENDIAN_CONSTANT</code> is used to indicate the
283endianness of the file in which it is found. Although the standard
284<code>.dex</code> format is little-endian, implementations may choose
285to perform byte-swapping. Should an implementation come across a
286header whose <code>endian_tag</code> is <code>REVERSE_ENDIAN_CONSTANT</code>
287instead of <code>ENDIAN_CONSTANT</code>, it would know that the file
288has been byte-swapped from the expected form.</p>
289
290<pre>
291uint ENDIAN_CONSTANT = 0x12345678;
292uint REVERSE_ENDIAN_CONSTANT = 0x78563412;
293</pre>
294
295<h2><code>NO_INDEX</code></h2>
296<h4>embedded in <code>class_def_item</code> and
297<code>debug_info_item</code></h4>
298
299<p>The constant <code>NO_INDEX</code> is used to indicate that
300an index value is absent.</p>
301
302<p><b>Note:</b> This value isn't defined to be
303<code>0</code>, because that is in fact typically a valid index.</p>
304
305<p><b>Also Note:</b> The chosen value for <code>NO_INDEX</code> is
306representable as a single byte in the <code>uleb128p1</code> encoding.</p>
307
308<pre>
309uint NO_INDEX = 0xffffffff;    // == -1 if treated as a signed int
310</pre>
311
312<h2><code>access_flags</code> Definitions</h2>
313<h4>embedded in <code>class_def_item</code>,
314<code>field_item</code>, <code>method_item</code>, and
315<code>InnerClass</code></h4>
316
317<p>Bitfields of these flags are used to indicate the accessibility and
318overall properties of classes and class members.</p>
319
320<table class="accessFlags">
321<thead>
322<tr>
323  <th>Name</th>
324  <th>Value</th>
325  <th>For Classes (and <code>InnerClass</code> annotations)</th>
326  <th>For Fields</th>
327  <th>For Methods</th>
328</tr>
329</thead>
330<tbody>
331<tr>
332  <td>ACC_PUBLIC</td>
333  <td>0x1</td>
334  <td><code>public</code>: visible everywhere</td>
335  <td><code>public</code>: visible everywhere</td>
336  <td><code>public</code>: visible everywhere</td>
337</tr>
338<tr>
339  <td>ACC_PRIVATE</td>
340  <td>0x2</td>
341  <td><super>*</super>
342    <code>private</code>: only visible to defining class
343  </td>
344  <td><code>private</code>: only visible to defining class</td>
345  <td><code>private</code>: only visible to defining class</td>
346</tr>
347<tr>
348  <td>ACC_PROTECTED</td>
349  <td>0x4</td>
350  <td><super>*</super>
351    <code>protected</code>: visible to package and subclasses
352  </td>
353  <td><code>protected</code>: visible to package and subclasses</td>
354  <td><code>protected</code>: visible to package and subclasses</td>
355</tr>
356<tr>
357  <td>ACC_STATIC</td>
358  <td>0x8</td>
359  <td><super>*</super>
360    <code>static</code>: is not constructed with an outer
361    <code>this</code> reference</td>
362  <td><code>static</code>: global to defining class</td>
363  <td><code>static</code>: does not take a <code>this</code> argument</td>
364</tr>
365<tr>
366  <td>ACC_FINAL</td>
367  <td>0x10</td>
368  <td><code>final</code>: not subclassable</td>
369  <td><code>final</code>: immutable after construction</td>
370  <td><code>final</code>: not overridable</td>
371</tr>
372<tr>
373  <td>ACC_SYNCHRONIZED</td>
374  <td>0x20</td>
375  <td>&nbsp;</td>
376  <td>&nbsp;</td>
377  <td><code>synchronized</code>: associated lock automatically acquired
378    around call to this method. <b>Note:</b> This is only valid to set when
379    <code>ACC_NATIVE</code> is also set.</td>
380</tr>
381<tr>
382  <td>ACC_VOLATILE</td>
383  <td>0x40</td>
384  <td>&nbsp;</td>
385  <td><code>volatile</code>: special access rules to help with thread
386    safety</td>
387  <td>&nbsp;</td>
388</tr>
389<tr>
390  <td>ACC_BRIDGE</td>
391  <td>0x40</td>
392  <td>&nbsp;</td>
393  <td>&nbsp;</td>
394  <td>bridge method, added automatically by compiler as a type-safe
395    bridge</td>
396</tr>
397<tr>
398  <td>ACC_TRANSIENT</td>
399  <td>0x80</td>
400  <td>&nbsp;</td>
401  <td><code>transient</code>: not to be saved by default serialization</td>
402  <td>&nbsp;</td>
403</tr>
404<tr>
405  <td>ACC_VARARGS</td>
406  <td>0x80</td>
407  <td>&nbsp;</td>
408  <td>&nbsp;</td>
409  <td>last argument should be treated as a "rest" argument by compiler</td>
410</tr>
411<tr>
412  <td>ACC_NATIVE</td>
413  <td>0x100</td>
414  <td>&nbsp;</td>
415  <td>&nbsp;</td>
416  <td><code>native</code>: implemented in native code</td>
417</tr>
418<tr>
419  <td>ACC_INTERFACE</td>
420  <td>0x200</td>
421  <td><code>interface</code>: multiply-implementable abstract class</td>
422  <td>&nbsp;</td>
423  <td>&nbsp;</td>
424</tr>
425<tr>
426  <td>ACC_ABSTRACT</td>
427  <td>0x400</td>
428  <td><code>abstract</code>: not directly instantiable</td>
429  <td>&nbsp;</td>
430  <td><code>abstract</code>: unimplemented by this class</td>
431</tr>
432<tr>
433  <td>ACC_STRICT</td>
434  <td>0x800</td>
435  <td>&nbsp;</td>
436  <td>&nbsp;</td>
437  <td><code>strictfp</code>: strict rules for floating-point arithmetic</td>
438</tr>
439<tr>
440  <td>ACC_SYNTHETIC</td>
441  <td>0x1000</td>
442  <td>not directly defined in source code</td>
443  <td>not directly defined in source code</td>
444  <td>not directly defined in source code</td>
445</tr>
446<tr>
447  <td>ACC_ANNOTATION</td>
448  <td>0x2000</td>
449  <td>declared as an annotation class</td>
450  <td>&nbsp;</td>
451  <td>&nbsp;</td>
452</tr>
453<tr>
454  <td>ACC_ENUM</td>
455  <td>0x4000</td>
456  <td>declared as an enumerated type</td>
457  <td>declared as an enumerated value</td>
458  <td>&nbsp;</td>
459</tr>
460<tr>
461  <td><i>(unused)</i></td>
462  <td>0x8000</td>
463  <td>&nbsp;</td>
464  <td>&nbsp;</td>
465  <td>&nbsp;</td>
466</tr>
467<tr>
468  <td>ACC_CONSTRUCTOR</td>
469  <td>0x10000</td>
470  <td>&nbsp;</td>
471  <td>&nbsp;</td>
472  <td>constructor method (class or instance initializer)</td>
473</tr>
474<tr>
475  <td>ACC_DECLARED_<br/>SYNCHRONIZED</td>
476  <td>0x20000</td>
477  <td>&nbsp;</td>
478  <td>&nbsp;</td>
479  <td>declared <code>synchronized</code>. <b>Note:</b> This has no effect on
480    execution (other than in reflection of this flag, per se).
481  </td>
482</tr>
483</tbody>
484</table>
485
486<p><super>*</super> Only allowed on for <code>InnerClass</code> annotations,
487and must not ever be on in a <code>class_def_item</code>.</p>
488
489<h2>MUTF-8 (Modified UTF-8) Encoding</h2>
490
491<p>As a concession to easier legacy support, the <code>.dex</code> format
492encodes its string data in a de facto standard modified UTF-8 form, hereafter
493referred to as MUTF-8. This form is identical to standard UTF-8, except:</p>
494
495<ul>
496  <li>Only the one-, two-, and three-byte encodings are used.</li>
497  <li>Code points in the range <code>U+10000</code> &hellip;
498    <code>U+10ffff</code> are encoded as a surrogate pair, each of
499    which is represented as a three-byte encoded value.</li>
500  <li>The code point <code>U+0000</code> is encoded in two-byte form.</li>
501  <li>A plain null byte (value <code>0</code>) indicates the end of
502    a string, as is the standard C language interpretation.</li>
503</ul>
504
505<p>The first two items above can be summarized as: MUTF-8
506is an encoding format for UTF-16, instead of being a more direct
507encoding format for Unicode characters.</p>
508
509<p>The final two items above make it simultaneously possible to include
510the code point <code>U+0000</code> in a string <i>and</i> still manipulate
511it as a C-style null-terminated string.</p>
512
513<p>However, the special encoding of <code>U+0000</code> means that, unlike
514normal UTF-8, the result of calling the standard C function
515<code>strcmp()</code> on a pair of MUTF-8 strings does not always
516indicate the properly signed result of comparison of <i>unequal</i> strings.
517When ordering (not just equality) is a concern, the most straightforward
518way to compare MUTF-8 strings is to decode them character by character,
519and compare the decoded values. (However, more clever implementations are
520also possible.)</p>
521
522<p>Please refer to <a href="http://unicode.org">The Unicode
523Standard</a> for further information about character encoding.
524MUTF-8 is actually closer to the (relatively less well-known) encoding
525<a href="http://www.unicode.org/reports/tr26/">CESU-8</a> than to UTF-8
526per se.</p>
527
528<h2><code>encoded_value</code> Encoding</h2>
529<h4>embedded in <code>annotation_element</code> and
530<code>encoded_array_item</code></h4>
531
532<p>An <code>encoded_value</code> is an encoded piece of (nearly)
533arbitrary hierarchically structured data. The encoding is meant to
534be both compact and straightforward to parse.</p>
535
536<table class="format">
537<thead>
538<tr>
539  <th>Name</th>
540  <th>Format</th>
541  <th>Description</th>
542</tr>
543</thead>
544<tbody>
545<tr>
546  <td>(value_arg &lt;&lt; 5) | value_type</td>
547  <td>ubyte</td>
548  <td>byte indicating the type of the immediately subsequent
549    <code>value</code> along
550    with an optional clarifying argument in the high-order three bits.
551    See below for the various <code>value</code> definitions.
552    In most cases, <code>value_arg</code> encodes the length of
553    the immediately-subsequent <code>value</code> in bytes, as
554    <code>(size - 1)</code>, e.g., <code>0</code> means that
555    the value requires one byte, and <code>7</code> means it requires
556    eight bytes; however, there are exceptions as noted below.
557  </td>
558</tr>
559<tr>
560  <td>value</td>
561  <td>ubyte[]</td>
562  <td>bytes representing the value, variable in length and interpreted
563    differently for different <code>value_type</code> bytes, though
564    always little-endian. See the various value definitions below for
565    details.
566  </td>
567</tr>
568</tbody>
569</table>
570
571<h3>Value Formats</h3>
572
573<table class="encodedValue">
574<thead>
575<tr>
576  <th>Type Name</th>
577  <th><code>value_type</code></th>
578  <th><code>value_arg</code> Format</th>
579  <th><code>value</code> Format</th>
580  <th>Description</th>
581</tr>
582</thead>
583<tbody>
584<tr>
585  <td>VALUE_BYTE</td>
586  <td>0x00</td>
587  <td><i>(none; must be <code>0</code>)</i></td>
588  <td>ubyte[1]</td>
589  <td>signed one-byte integer value</td>
590</tr>
591<tr>
592  <td>VALUE_SHORT</td>
593  <td>0x02</td>
594  <td>size - 1 (0&hellip;1)</td>
595  <td>ubyte[size]</td>
596  <td>signed two-byte integer value, sign-extended</td>
597</tr>
598<tr>
599  <td>VALUE_CHAR</td>
600  <td>0x03</td>
601  <td>size - 1 (0&hellip;1)</td>
602  <td>ubyte[size]</td>
603  <td>unsigned two-byte integer value, zero-extended</td>
604</tr>
605<tr>
606  <td>VALUE_INT</td>
607  <td>0x04</td>
608  <td>size - 1 (0&hellip;3)</td>
609  <td>ubyte[size]</td>
610  <td>signed four-byte integer value, sign-extended</td>
611</tr>
612<tr>
613  <td>VALUE_LONG</td>
614  <td>0x06</td>
615  <td>size - 1 (0&hellip;7)</td>
616  <td>ubyte[size]</td>
617  <td>signed eight-byte integer value, sign-extended</td>
618</tr>
619<tr>
620  <td>VALUE_FLOAT</td>
621  <td>0x10</td>
622  <td>size - 1 (0&hellip;3)</td>
623  <td>ubyte[size]</td>
624  <td>four-byte bit pattern, zero-extended <i>to the right</i>, and
625    interpreted as an IEEE754 32-bit floating point value
626  </td>
627</tr>
628<tr>
629  <td>VALUE_DOUBLE</td>
630  <td>0x11</td>
631  <td>size - 1 (0&hellip;7)</td>
632  <td>ubyte[size]</td>
633  <td>eight-byte bit pattern, zero-extended <i>to the right</i>, and
634    interpreted as an IEEE754 64-bit floating point value
635  </td>
636</tr>
637<tr>
638  <td>VALUE_STRING</td>
639  <td>0x17</td>
640  <td>size - 1 (0&hellip;3)</td>
641  <td>ubyte[size]</td>
642  <td>unsigned (zero-extended) four-byte integer value,
643    interpreted as an index into
644    the <code>string_ids</code> section and representing a string value
645  </td>
646</tr>
647<tr>
648  <td>VALUE_TYPE</td>
649  <td>0x18</td>
650  <td>size - 1 (0&hellip;3)</td>
651  <td>ubyte[size]</td>
652  <td>unsigned (zero-extended) four-byte integer value,
653    interpreted as an index into
654    the <code>type_ids</code> section and representing a reflective
655    type/class value
656  </td>
657</tr>
658<tr>
659  <td>VALUE_FIELD</td>
660  <td>0x19</td>
661  <td>size - 1 (0&hellip;3)</td>
662  <td>ubyte[size]</td>
663  <td>unsigned (zero-extended) four-byte integer value,
664    interpreted as an index into
665    the <code>field_ids</code> section and representing a reflective
666    field value
667  </td>
668</tr>
669<tr>
670  <td>VALUE_METHOD</td>
671  <td>0x1a</td>
672  <td>size - 1 (0&hellip;3)</td>
673  <td>ubyte[size]</td>
674  <td>unsigned (zero-extended) four-byte integer value,
675    interpreted as an index into
676    the <code>method_ids</code> section and representing a reflective
677    method value
678  </td>
679</tr>
680<tr>
681  <td>VALUE_ENUM</td>
682  <td>0x1b</td>
683  <td>size - 1 (0&hellip;3)</td>
684  <td>ubyte[size]</td>
685  <td>unsigned (zero-extended) four-byte integer value,
686    interpreted as an index into
687    the <code>field_ids</code> section and representing the value of
688    an enumerated type constant
689  </td>
690</tr>
691<tr>
692  <td>VALUE_ARRAY</td>
693  <td>0x1c</td>
694  <td><i>(none; must be <code>0</code>)</i></td>
695  <td>encoded_array</td>
696  <td>an array of values, in the format specified by
697    "<code>encoded_array</code> Format" below. The size
698    of the <code>value</code> is implicit in the encoding.
699  </td>
700</tr>
701<tr>
702  <td>VALUE_ANNOTATION</td>
703  <td>0x1d</td>
704  <td><i>(none; must be <code>0</code>)</i></td>
705  <td>encoded_annotation</td>
706  <td>a sub-annotation, in the format specified by
707    "<code>encoded_annotation</code> Format" below. The size
708    of the <code>value</code> is implicit in the encoding.
709  </td>
710</tr>
711<tr>
712  <td>VALUE_NULL</td>
713  <td>0x1e</td>
714  <td><i>(none; must be <code>0</code>)</i></td>
715  <td><i>(none)</i></td>
716  <td><code>null</code> reference value</td>
717</tr>
718<tr>
719  <td>VALUE_BOOLEAN</td>
720  <td>0x1f</td>
721  <td>boolean (0&hellip;1)</td>
722  <td><i>(none)</i></td>
723  <td>one-bit value; <code>0</code> for <code>false</code> and
724    <code>1</code> for <code>true</code>. The bit is represented in the
725    <code>value_arg</code>.
726  </td>
727</tr>
728</tbody>
729</table>
730
731<h3><code>encoded_array</code> Format</h3>
732
733<table class="format">
734<thead>
735<tr>
736  <th>Name</th>
737  <th>Format</th>
738  <th>Description</th>
739</tr>
740</thead>
741<tbody>
742<tr>
743  <td>size</td>
744  <td>uleb128</td>
745  <td>number of elements in the array</td>
746</tr>
747<tr>
748  <td>values</td>
749  <td>encoded_value[size]</td>
750  <td>a series of <code>size</code> <code>encoded_value</code> byte
751    sequences in the format specified by this section, concatenated
752    sequentially.
753  </td>
754</tr>
755</tbody>
756</table>
757
758<h3><code>encoded_annotation</code> Format</h3>
759
760<table class="format">
761<thead>
762<tr>
763  <th>Name</th>
764  <th>Format</th>
765  <th>Description</th>
766</tr>
767</thead>
768<tbody>
769<tr>
770  <td>type_idx</td>
771  <td>uleb128</td>
772  <td>type of the annotation. This must be a class (not array or primitive)
773    type.
774  </td>
775</tr>
776<tr>
777  <td>size</td>
778  <td>uleb128</td>
779  <td>number of name-value mappings in this annotation</td>
780</tr>
781<tr>
782  <td>elements</td>
783  <td>annotation_element[size]</td>
784  <td>elements of the annotataion, represented directly in-line (not as
785    offsets). Elements must be sorted in increasing order by
786    <code>string_id</code> index.
787  </td>
788</tr>
789</tbody>
790</table>
791
792<h3><code>annotation_element</code> Format</h3>
793
794<table class="format">
795<thead>
796<tr>
797  <th>Name</th>
798  <th>Format</th>
799  <th>Description</th>
800</tr>
801</thead>
802<tbody>
803<tr>
804  <td>name_idx</td>
805  <td>uleb128</td>
806  <td>element name, represented as an index into the
807    <code>string_ids</code> section. The string must conform to the
808    syntax for <i>MemberName</i>, defined above.
809  </td>
810</tr>
811<tr>
812  <td>value</td>
813  <td>encoded_value</td>
814  <td>element value</td>
815</tr>
816</tbody>
817</table>
818
819<h2>String Syntax</h2>
820
821<p>There are several kinds of item in a <code>.dex</code> file which
822ultimately refer to a string. The following BNF-style definitions
823indicate the acceptable syntax for these strings.</p>
824
825<h3><i>SimpleName</i></h3>
826
827<p>A <i>SimpleName</i> is the basis for the syntax of the names of other
828things. The <code>.dex</code> format allows a fair amount of latitude
829here (much more than most common source languages). In brief, a simple
830name may consist of any low-ASCII alphabetic character or digit, a few
831specific low-ASCII symbols, and most non-ASCII code points that are not
832control, space, or special characters. Note that surrogate code points
833(in the range <code>U+d800</code> &hellip; <code>U+dfff</code>) are not
834considered valid name characters, per se, but Unicode supplemental
835characters <i>are</i> valid (which are represented by the final
836alternative of the rule for <i>SimpleNameChar</i>), and they should be
837represented in a file as pairs of surrogate code points in the MUTF-8
838encoding.</p>
839
840<table class="bnf">
841  <tr><td colspan="2" class="def"><i>SimpleName</i> &rarr;</td></tr>
842  <tr>
843    <td/>
844    <td><i>SimpleNameChar</i> (<i>SimpleNameChar</i>)*</td>
845  </tr>
846
847  <tr><td colspan="2" class="def"><i>SimpleNameChar</i> &rarr;</td></tr>
848  <tr>
849    <td/>
850    <td><code>'A'</code> &hellip; <code>'Z'</code></td>
851  </tr>
852  <tr>
853    <td class="bar">|</td>
854    <td><code>'a'</code> &hellip; <code>'z'</code></td>
855  </tr>
856  <tr>
857    <td class="bar">|</td>
858    <td><code>'0'</code> &hellip; <code>'9'</code></td>
859  </tr>
860  <tr>
861    <td class="bar">|</td>
862    <td><code>'$'</code></td>
863  </tr>
864  <tr>
865    <td class="bar">|</td>
866    <td><code>'-'</code></td>
867  </tr>
868  <tr>
869    <td class="bar">|</td>
870    <td><code>'_'</code></td>
871  </tr>
872  <tr>
873    <td class="bar">|</td>
874    <td><code>U+00a1</code> &hellip; <code>U+1fff</code></td>
875  </tr>
876  <tr>
877    <td class="bar">|</td>
878    <td><code>U+2010</code> &hellip; <code>U+2027</code></td>
879  </tr>
880  <tr>
881    <td class="bar">|</td>
882    <td><code>U+2030</code> &hellip; <code>U+d7ff</code></td>
883  </tr>
884  <tr>
885    <td class="bar">|</td>
886    <td><code>U+e000</code> &hellip; <code>U+ffef</code></td>
887  </tr>
888  <tr>
889    <td class="bar">|</td>
890    <td><code>U+10000</code> &hellip; <code>U+10ffff</code></td>
891  </tr>
892</table>
893
894<h3><i>MemberName</i></h3>
895<h4>used by <code>field_id_item</code> and <code>method_id_item</code></h4>
896
897<p>A <i>MemberName</i> is the name of a member of a class, members being
898fields, methods, and inner classes.</p>
899
900<table class="bnf">
901  <tr><td colspan="2" class="def"><i>MemberName</i> &rarr;</td></tr>
902  <tr>
903    <td/>
904    <td><i>SimpleName</i></td>
905  </tr>
906  <tr>
907    <td class="bar">|</td>
908    <td><code>'&lt;'</code> <i>SimpleName</i> <code>'&gt;'</code></td>
909  </tr>
910</table>
911
912<h3><i>FullClassName</i></h3>
913
914<p>A <i>FullClassName</i> is a fully-qualified class name, including an
915optional package specifier followed by a required name.</p>
916
917<table class="bnf">
918  <tr><td colspan="2" class="def"><i>FullClassName</i> &rarr;</td></tr>
919  <tr>
920    <td/>
921    <td><i>OptionalPackagePrefix</i> <i>SimpleName</i></td>
922  </tr>
923
924  <tr><td colspan="2" class="def"><i>OptionalPackagePrefix</i> &rarr;</td></tr>
925  <tr>
926    <td/>
927    <td>(<i>SimpleName</i> <code>'/'</code>)*</td>
928  </tr>
929</table>
930
931<h3><i>TypeDescriptor</i></h3>
932<h4>used by <code>type_id_item</code></h4>
933
934<p>A <i>TypeDescriptor</i> is the representation of any type, including
935primitives, classes, arrays, and <code>void</code>. See below for
936the meaning of the various versions.</p>
937
938<table class="bnf">
939  <tr><td colspan="2" class="def"><i>TypeDescriptor</i> &rarr;</td></tr>
940  <tr>
941    <td/>
942    <td><code>'V'</code></td>
943  </tr>
944  <tr>
945    <td class="bar">|</td>
946    <td><i>FieldTypeDescriptor</i></td>
947  </tr>
948
949  <tr><td colspan="2" class="def"><i>FieldTypeDescriptor</i> &rarr;</td></tr>
950  <tr>
951    <td/>
952    <td><i>NonArrayFieldTypeDescriptor</i></td>
953  </tr>
954  <tr>
955    <td class="bar">|</td>
956    <td>(<code>'['</code> * 1&hellip;255)
957      <i>NonArrayFieldTypeDescriptor</i></td>
958  </tr>
959
960  <tr>
961    <td colspan="2" class="def"><i>NonArrayFieldTypeDescriptor</i>&rarr;</td>
962  </tr>
963  <tr>
964    <td/>
965    <td><code>'Z'</code></td>
966  </tr>
967  <tr>
968    <td class="bar">|</td>
969    <td><code>'B'</code></td>
970  </tr>
971  <tr>
972    <td class="bar">|</td>
973    <td><code>'S'</code></td>
974  </tr>
975  <tr>
976    <td class="bar">|</td>
977    <td><code>'C'</code></td>
978  </tr>
979  <tr>
980    <td class="bar">|</td>
981    <td><code>'I'</code></td>
982  </tr>
983  <tr>
984    <td class="bar">|</td>
985    <td><code>'J'</code></td>
986  </tr>
987  <tr>
988    <td class="bar">|</td>
989    <td><code>'F'</code></td>
990  </tr>
991  <tr>
992    <td class="bar">|</td>
993    <td><code>'D'</code></td>
994  </tr>
995  <tr>
996    <td class="bar">|</td>
997    <td><code>'L'</code> <i>FullClassName</i> <code>';'</code></td>
998  </tr>
999</table>
1000
1001<h3><i>ShortyDescriptor</i></h3>
1002<h4>used by <code>proto_id_item</code></h4>
1003
1004<p>A <i>ShortyDescriptor</i> is the short form representation of a method
1005prototype, including return and parameter types, except that there is
1006no distinction between various reference (class or array) types. Instead,
1007all reference types are represented by a single <code>'L'</code> character.</p>
1008
1009<table class="bnf">
1010  <tr><td colspan="2" class="def"><i>ShortyDescriptor</i> &rarr;</td></tr>
1011  <tr>
1012    <td/>
1013    <td><i>ShortyReturnType</i> (<i>ShortyFieldType</i>)*</td>
1014  </tr>
1015
1016  <tr><td colspan="2" class="def"><i>ShortyReturnType</i> &rarr;</td></tr>
1017  <tr>
1018    <td/>
1019    <td><code>'V'</code></td>
1020  </tr>
1021  <tr>
1022    <td class="bar">|</td>
1023    <td><i>ShortyFieldType</i></td>
1024  </tr>
1025
1026  <tr><td colspan="2" class="def"><i>ShortyFieldType</i> &rarr;</td></tr>
1027  <tr>
1028    <td/>
1029    <td><code>'Z'</code></td>
1030  </tr>
1031  <tr>
1032    <td class="bar">|</td>
1033    <td><code>'B'</code></td>
1034  </tr>
1035  <tr>
1036    <td class="bar">|</td>
1037    <td><code>'S'</code></td>
1038  </tr>
1039  <tr>
1040    <td class="bar">|</td>
1041    <td><code>'C'</code></td>
1042  </tr>
1043  <tr>
1044    <td class="bar">|</td>
1045    <td><code>'I'</code></td>
1046  </tr>
1047  <tr>
1048    <td class="bar">|</td>
1049    <td><code>'J'</code></td>
1050  </tr>
1051  <tr>
1052    <td class="bar">|</td>
1053    <td><code>'F'</code></td>
1054  </tr>
1055  <tr>
1056    <td class="bar">|</td>
1057    <td><code>'D'</code></td>
1058  </tr>
1059  <tr>
1060    <td class="bar">|</td>
1061    <td><code>'L'</code></td>
1062  </tr>
1063</table>
1064
1065<h2><i>TypeDescriptor</i> Semantics</h2>
1066
1067<p>This is the meaning of each of the variants of <i>TypeDescriptor</i>.</p>
1068
1069<table class="descriptor">
1070<thead>
1071<tr>
1072  <th>Syntax</th>
1073  <th>Meaning</th>
1074</tr>
1075</thead>
1076<tbody>
1077<tr>
1078  <td>V</td>
1079  <td><code>void</code>; only valid for return types</td>
1080</tr>
1081<tr>
1082  <td>Z</td>
1083  <td><code>boolean</code></td>
1084</tr>
1085<tr>
1086  <td>B</td>
1087  <td><code>byte</code></td>
1088</tr>
1089<tr>
1090  <td>S</td>
1091  <td><code>short</code></td>
1092</tr>
1093<tr>
1094  <td>C</td>
1095  <td><code>char</code></td>
1096</tr>
1097<tr>
1098  <td>I</td>
1099  <td><code>int</code></td>
1100</tr>
1101<tr>
1102  <td>J</td>
1103  <td><code>long</code></td>
1104</tr>
1105<tr>
1106  <td>F</td>
1107  <td><code>float</code></td>
1108</tr>
1109<tr>
1110  <td>D</td>
1111  <td><code>double</code></td>
1112</tr>
1113<tr>
1114  <td>L<i>fully/qualified/Name</i>;</td>
1115  <td>the class <code><i>fully.qualified.Name</i></code></td>
1116</tr>
1117<tr>
1118  <td>[<i>descriptor</i></td>
1119  <td>array of <code><i>descriptor</i></code>, usable recursively for
1120    arrays-of-arrays, though it is invalid to have more than 255
1121    dimensions.
1122  </td>
1123</tr>
1124</tbody>
1125</table>
1126
1127<h1>Items and Related Structures</h1>
1128
1129<p>This section includes definitions for each of the top-level items that
1130may appear in a <code>.dex</code> file.
1131
1132<h2><code>header_item</code></h2>
1133<h4>appears in the <code>header</code> section</h4>
1134<h4>alignment: 4 bytes</h4>
1135
1136<table class="format">
1137<thead>
1138<tr>
1139  <th>Name</th>
1140  <th>Format</th>
1141  <th>Description</th>
1142</tr>
1143</thead>
1144<tbody>
1145<tr>
1146  <td>magic</td>
1147  <td>ubyte[8] = DEX_FILE_MAGIC</td>
1148  <td>magic value. See discussion above under "<code>DEX_FILE_MAGIC</code>"
1149    for more details.
1150  </td>
1151</tr>
1152<tr>
1153  <td>checksum</td>
1154  <td>uint</td>
1155  <td>adler32 checksum of the rest of the file (everything but
1156    <code>magic</code> and this field); used to detect file corruption
1157  </td>
1158</tr>
1159<tr>
1160  <td>signature</td>
1161  <td>ubyte[20]</td>
1162  <td>SHA-1 signature (hash) of the rest of the file (everything but
1163    <code>magic</code>, <code>checksum</code>, and this field); used
1164    to uniquely identify files
1165  </td>
1166</tr>
1167<tr>
1168  <td>file_size</td>
1169  <td>uint</td>
1170  <td>size of the entire file (including the header), in bytes
1171</tr>
1172<tr>
1173  <td>header_size</td>
1174  <td>uint = 0x70</td>
1175  <td>size of the header (this entire section), in bytes. This allows for at
1176    least a limited amount of backwards/forwards compatibility without
1177    invalidating the format.
1178  </td>
1179</tr>
1180<tr>
1181  <td>endian_tag</td>
1182  <td>uint = ENDIAN_CONSTANT</td>
1183  <td>endianness tag. See discussion above under "<code>ENDIAN_CONSTANT</code>
1184    and <code>REVERSE_ENDIAN_CONSTANT</code>" for more details.
1185  </td>
1186</tr>
1187<tr>
1188  <td>link_size</td>
1189  <td>uint</td>
1190  <td>size of the link section, or <code>0</code> if this file isn't
1191    statically linked</td>
1192</tr>
1193<tr>
1194  <td>link_off</td>
1195  <td>uint</td>
1196  <td>offset from the start of the file to the link section, or
1197    <code>0</code> if <code>link_size == 0</code>. The offset, if non-zero,
1198    should be to an offset into the <code>link_data</code> section. The
1199    format of the data pointed at is left unspecified by this document;
1200    this header field (and the previous) are left as hooks for use by
1201    runtime implementations.
1202  </td>
1203</tr>
1204<tr>
1205  <td>map_off</td>
1206  <td>uint</td>
1207  <td>offset from the start of the file to the map item, or
1208    <code>0</code> if this file has no map. The offset, if non-zero,
1209    should be to an offset into the <code>data</code> section,
1210    and the data should be in the format specified by "<code>map_list</code>"
1211    below.
1212  </td>
1213</tr>
1214<tr>
1215  <td>string_ids_size</td>
1216  <td>uint</td>
1217  <td>count of strings in the string identifiers list</td>
1218</tr>
1219<tr>
1220  <td>string_ids_off</td>
1221  <td>uint</td>
1222  <td>offset from the start of the file to the string identifiers list, or
1223    <code>0</code> if <code>string_ids_size == 0</code> (admittedly a
1224    strange edge case). The offset, if non-zero,
1225    should be to the start of the <code>string_ids</code> section.
1226  </td>
1227</tr>
1228<tr>
1229  <td>type_ids_size</td>
1230  <td>uint</td>
1231  <td>count of elements in the type identifiers list</td>
1232</tr>
1233<tr>
1234  <td>type_ids_off</td>
1235  <td>uint</td>
1236  <td>offset from the start of the file to the type identifiers list, or
1237    <code>0</code> if <code>type_ids_size == 0</code> (admittedly a
1238    strange edge case). The offset, if non-zero,
1239    should be to the start of the <code>type_ids</code>
1240    section.
1241  </td>
1242</tr>
1243<tr>
1244  <td>proto_ids_size</td>
1245  <td>uint</td>
1246  <td>count of elements in the prototype identifiers list</td>
1247</tr>
1248<tr>
1249  <td>proto_ids_off</td>
1250  <td>uint</td>
1251  <td>offset from the start of the file to the prototype identifiers list, or
1252    <code>0</code> if <code>proto_ids_size == 0</code> (admittedly a
1253    strange edge case). The offset, if non-zero,
1254    should be to the start of the <code>proto_ids</code>
1255    section.
1256  </td>
1257</tr>
1258<tr>
1259  <td>field_ids_size</td>
1260  <td>uint</td>
1261  <td>count of elements in the field identifiers list</td>
1262</tr>
1263<tr>
1264  <td>field_ids_off</td>
1265  <td>uint</td>
1266  <td>offset from the start of the file to the field identifiers list, or
1267    <code>0</code> if <code>field_ids_size == 0</code>. The offset, if
1268    non-zero, should be to the start of the <code>field_ids</code>
1269    section.</td>
1270</td>
1271</tr>
1272<tr>
1273  <td>method_ids_size</td>
1274  <td>uint</td>
1275  <td>count of elements in the method identifiers list</td>
1276</tr>
1277<tr>
1278  <td>method_ids_off</td>
1279  <td>uint</td>
1280  <td>offset from the start of the file to the method identifiers list, or
1281    <code>0</code> if <code>method_ids_size == 0</code>. The offset, if
1282    non-zero, should be to the start of the <code>method_ids</code>
1283    section.</td>
1284</tr>
1285<tr>
1286  <td>class_defs_size</td>
1287  <td>uint</td>
1288  <td>count of elements in the class definitions list</td>
1289</tr>
1290<tr>
1291  <td>class_defs_off</td>
1292  <td>uint</td>
1293  <td>offset from the start of the file to the class definitions list, or
1294    <code>0</code> if <code>class_defs_size == 0</code> (admittedly a
1295    strange edge case). The offset, if non-zero,
1296    should be to the start of the <code>class_defs</code> section.
1297  </td>
1298</tr>
1299<tr>
1300  <td>data_size</td>
1301  <td>uint</td>
1302  <td>Size of <code>data</code> section in bytes. Must be an even
1303    multiple of sizeof(uint).</td>
1304</tr>
1305<tr>
1306  <td>data_off</td>
1307  <td>uint</td>
1308  <td>offset from the start of the file to the start of the
1309   <code>data</code> section.
1310  </td>
1311</tr>
1312</tbody>
1313</table>
1314
1315<h2><code>map_list</code></h2>
1316<h4>appears in the <code>data</code> section</h4>
1317<h4>referenced from <code>header_item</code></h4>
1318<h4>alignment: 4 bytes</h4>
1319
1320<p>This is a list of the entire contents of a file, in order. It
1321contains some redundancy with respect to the <code>header_item</code>
1322but is intended to be an easy form to use to iterate over an entire
1323file. A given type may appear at most once in a map, but there is no
1324restriction on what order types may appear in, other than the
1325restrictions implied by the rest of the format (e.g., a
1326<code>header</code> section must appear first, followed by a
1327<code>string_ids</code> section, etc.). Additionally, the map entries must
1328be ordered by initial offset and must not overlap.</p>
1329
1330<table class="format">
1331<thead>
1332<tr>
1333  <th>Name</th>
1334  <th>Format</th>
1335  <th>Description</th>
1336</tr>
1337</thead>
1338<tbody>
1339<tr>
1340  <td>size</td>
1341  <td>uint</td>
1342  <td>size of the list, in entries</td>
1343</tr>
1344<tr>
1345  <td>list</td>
1346  <td>map_item[size]</td>
1347  <td>elements of the list</td>
1348</tr>
1349</tbody>
1350</table>
1351
1352<h3><code>map_item</code> Format</h3>
1353
1354<table class="format">
1355<thead>
1356<tr>
1357  <th>Name</th>
1358  <th>Format</th>
1359  <th>Description</th>
1360</tr>
1361</thead>
1362<tbody>
1363<tr>
1364  <td>type</td>
1365  <td>ushort</td>
1366  <td>type of the items; see table below</td>
1367</tr>
1368<tr>
1369  <td>unused</td>
1370  <td>ushort</td>
1371  <td><i>(unused)</i></td>
1372</tr>
1373<tr>
1374  <td>size</td>
1375  <td>uint</td>
1376  <td>count of the number of items to be found at the indicated offset</td>
1377</tr>
1378<tr>
1379  <td>offset</td>
1380  <td>uint</td>
1381  <td>offset from the start of the file to the items in question</td>
1382</tr>
1383</tbody>
1384</table>
1385
1386
1387<h3>Type Codes</h3>
1388
1389<table class="typeCodes">
1390<thead>
1391<tr>
1392  <th>Item Type</th>
1393  <th>Constant</th>
1394  <th>Value</th>
1395  <th>Item Size In Bytes</th>
1396</tr>
1397</thead>
1398<tbody>
1399<tr>
1400  <td>header_item</td>
1401  <td>TYPE_HEADER_ITEM</td>
1402  <td>0x0000</td>
1403  <td>0x70</td>
1404</tr>
1405<tr>
1406  <td>string_id_item</td>
1407  <td>TYPE_STRING_ID_ITEM</td>
1408  <td>0x0001</td>
1409  <td>0x04</td>
1410</tr>
1411<tr>
1412  <td>type_id_item</td>
1413  <td>TYPE_TYPE_ID_ITEM</td>
1414  <td>0x0002</td>
1415  <td>0x04</td>
1416</tr>
1417<tr>
1418  <td>proto_id_item</td>
1419  <td>TYPE_PROTO_ID_ITEM</td>
1420  <td>0x0003</td>
1421  <td>0x0c</td>
1422</tr>
1423<tr>
1424  <td>field_id_item</td>
1425  <td>TYPE_FIELD_ID_ITEM</td>
1426  <td>0x0004</td>
1427  <td>0x08</td>
1428</tr>
1429<tr>
1430  <td>method_id_item</td>
1431  <td>TYPE_METHOD_ID_ITEM</td>
1432  <td>0x0005</td>
1433  <td>0x08</td>
1434</tr>
1435<tr>
1436  <td>class_def_item</td>
1437  <td>TYPE_CLASS_DEF_ITEM</td>
1438  <td>0x0006</td>
1439  <td>0x20</td>
1440</tr>
1441<tr>
1442  <td>map_list</td>
1443  <td>TYPE_MAP_LIST</td>
1444  <td>0x1000</td>
1445  <td>4 + (item.size * 12)</td>
1446</tr>
1447<tr>
1448  <td>type_list</td>
1449  <td>TYPE_TYPE_LIST</td>
1450  <td>0x1001</td>
1451  <td>4 + (item.size * 2)</td>
1452</tr>
1453<tr>
1454  <td>annotation_set_ref_list</td>
1455  <td>TYPE_ANNOTATION_SET_REF_LIST</td>
1456  <td>0x1002</td>
1457  <td>4 + (item.size * 4)</td>
1458</tr>
1459<tr>
1460  <td>annotation_set_item</td>
1461  <td>TYPE_ANNOTATION_SET_ITEM</td>
1462  <td>0x1003</td>
1463  <td>4 + (item.size * 4)</td>
1464</tr>
1465<tr>
1466  <td>class_data_item</td>
1467  <td>TYPE_CLASS_DATA_ITEM</td>
1468  <td>0x2000</td>
1469  <td><i>implicit; must parse</i></td>
1470</tr>
1471<tr>
1472  <td>code_item</td>
1473  <td>TYPE_CODE_ITEM</td>
1474  <td>0x2001</td>
1475  <td><i>implicit; must parse</i></td>
1476</tr>
1477<tr>
1478  <td>string_data_item</td>
1479  <td>TYPE_STRING_DATA_ITEM</td>
1480  <td>0x2002</td>
1481  <td><i>implicit; must parse</i></td>
1482</tr>
1483<tr>
1484  <td>debug_info_item</td>
1485  <td>TYPE_DEBUG_INFO_ITEM</td>
1486  <td>0x2003</td>
1487  <td><i>implicit; must parse</i></td>
1488</tr>
1489<tr>
1490  <td>annotation_item</td>
1491  <td>TYPE_ANNOTATION_ITEM</td>
1492  <td>0x2004</td>
1493  <td><i>implicit; must parse</i></td>
1494</tr>
1495<tr>
1496  <td>encoded_array_item</td>
1497  <td>TYPE_ENCODED_ARRAY_ITEM</td>
1498  <td>0x2005</td>
1499  <td><i>implicit; must parse</i></td>
1500</tr>
1501<tr>
1502  <td>annotations_directory_item</td>
1503  <td>TYPE_ANNOTATIONS_DIRECTORY_ITEM</td>
1504  <td>0x2006</td>
1505  <td><i>implicit; must parse</i></td>
1506</tr>
1507</tbody>
1508</table>
1509
1510
1511<h2><code>string_id_item</code></h2>
1512<h4>appears in the <code>string_ids</code> section</h4>
1513<h4>alignment: 4 bytes</h4>
1514
1515<table class="format">
1516<thead>
1517<tr>
1518  <th>Name</th>
1519  <th>Format</th>
1520  <th>Description</th>
1521</tr>
1522</thead>
1523<tbody>
1524<tr>
1525  <td>string_data_off</td>
1526  <td>uint</td>
1527  <td>offset from the start of the file to the string data for this
1528    item. The offset should be to a location
1529    in the <code>data</code> section, and the data should be in the
1530    format specified by "<code>string_data_item</code>" below.
1531    There is no alignment requirement for the offset.
1532  </td>
1533</tr>
1534</tbody>
1535</table>
1536
1537<h2><code>string_data_item</code></h2>
1538<h4>appears in the <code>data</code> section</h4>
1539<h4>alignment: none (byte-aligned)</h4>
1540
1541<table class="format">
1542<thead>
1543<tr>
1544  <th>Name</th>
1545  <th>Format</th>
1546  <th>Description</th>
1547</tr>
1548</thead>
1549<tbody>
1550<tr>
1551  <td>utf16_size</td>
1552  <td>uleb128</td>
1553  <td>size of this string, in UTF-16 code units (which is the "string
1554    length" in many systems). That is, this is the decoded length of
1555    the string. (The encoded length is implied by the position of
1556    the <code>0</code> byte.)</td>
1557</tr>
1558<tr>
1559  <td>data</td>
1560  <td>ubyte[]</td>
1561  <td>a series of MUTF-8 code units (a.k.a. octets, a.k.a. bytes)
1562    followed by a byte of value <code>0</code>. See
1563    "MUTF-8 (Modified UTF-8) Encoding" above for details and
1564    discussion about the data format.
1565    <p><b>Note:</b> It is acceptable to have a string which includes
1566    (the encoded form of) UTF-16 surrogate code units (that is,
1567    <code>U+d800</code> &hellip; <code>U+dfff</code>)
1568    either in isolation or out-of-order with respect to the usual
1569    encoding of Unicode into UTF-16. It is up to higher-level uses of
1570    strings to reject such invalid encodings, if appropriate.</p>
1571  </td>
1572</tr>
1573</tbody>
1574</table>
1575
1576<h2><code>type_id_item</code></h2>
1577<h4>appears in the <code>type_ids</code> section</h4>
1578<h4>alignment: 4 bytes</h4>
1579
1580<table class="format">
1581<thead>
1582<tr>
1583  <th>Name</th>
1584  <th>Format</th>
1585  <th>Description</th>
1586</tr>
1587</thead>
1588<tbody>
1589<tr>
1590  <td>descriptor_idx</td>
1591  <td>uint</td>
1592  <td>index into the <code>string_ids</code> list for the descriptor
1593    string of this type. The string must conform to the syntax for
1594    <i>TypeDescriptor</i>, defined above.
1595  </td>
1596</tr>
1597</tbody>
1598</table>
1599
1600<h2><code>proto_id_item</code></h2>
1601<h4>appears in the <code>proto_ids</code> section</h4>
1602<h4>alignment: 4 bytes</h4>
1603
1604<table class="format">
1605<thead>
1606<tr>
1607  <th>Name</th>
1608  <th>Format</th>
1609  <th>Description</th>
1610</tr>
1611</thead>
1612<tbody>
1613<tr>
1614  <td>shorty_idx</td>
1615  <td>uint</td>
1616  <td>index into the <code>string_ids</code> list for the short-form
1617    descriptor string of this prototype. The string must conform to the
1618    syntax for <i>ShortyDescriptor</i>, defined above, and must correspond
1619    to the return type and parameters of this item.
1620  </td>
1621</tr>
1622<tr>
1623  <td>return_type_idx</td>
1624  <td>uint</td>
1625  <td>index into the <code>type_ids</code> list for the return type
1626    of this prototype
1627  </td>
1628</tr>
1629<tr>
1630  <td>parameters_off</td>
1631  <td>uint</td>
1632  <td>offset from the start of the file to the list of parameter types
1633    for this prototype, or <code>0</code> if this prototype has no
1634    parameters. This offset, if non-zero, should be in the
1635    <code>data</code> section, and the data there should be in the
1636    format specified by <code>"type_list"</code> below. Additionally, there
1637    should be no reference to the type <code>void</code> in the list.
1638  </td>
1639</tr>
1640</tbody>
1641</table>
1642
1643<h2><code>field_id_item</code></h2>
1644<h4>appears in the <code>field_ids</code> section</h4>
1645<h4>alignment: 4 bytes</h4>
1646
1647<table class="format">
1648<thead>
1649<tr>
1650  <th>Name</th>
1651  <th>Format</th>
1652  <th>Description</th>
1653</tr>
1654</thead>
1655<tbody>
1656<tr>
1657  <td>class_idx</td>
1658  <td>ushort</td>
1659  <td>index into the <code>type_ids</code> list for the definer of this
1660    field. This must be a class type, and not an array or primitive type.
1661  </td>
1662</tr>
1663<tr>
1664  <td>type_idx</td>
1665  <td>ushort</td>
1666  <td>index into the <code>type_ids</code> list for the type of
1667    this field
1668  </td>
1669</tr>
1670<tr>
1671  <td>name_idx</td>
1672  <td>uint</td>
1673  <td>index into the <code>string_ids</code> list for the name of this
1674    field. The string must conform to the syntax for <i>MemberName</i>,
1675    defined above.
1676  </td>
1677</tr>
1678</tbody>
1679</table>
1680
1681<h2><code>method_id_item</code></h2>
1682<h4>appears in the <code>method_ids</code> section</h4>
1683<h4>alignment: 4 bytes</h4>
1684
1685<table class="format">
1686<thead>
1687<tr>
1688  <th>Name</th>
1689  <th>Format</th>
1690  <th>Description</th>
1691</tr>
1692</thead>
1693<tbody>
1694<tr>
1695  <td>class_idx</td>
1696  <td>ushort</td>
1697  <td>index into the <code>type_ids</code> list for the definer of this
1698    method. This must be a class or array type, and not a primitive type.
1699  </td>
1700</tr>
1701<tr>
1702  <td>proto_idx</td>
1703  <td>ushort</td>
1704  <td>index into the <code>proto_ids</code> list for the prototype of
1705    this method
1706  </td>
1707</tr>
1708<tr>
1709  <td>name_idx</td>
1710  <td>uint</td>
1711  <td>index into the <code>string_ids</code> list for the name of this
1712    method. The string must conform to the syntax for <i>MemberName</i>,
1713    defined above.
1714  </td>
1715</tr>
1716</tbody>
1717</table>
1718
1719<h2><code>class_def_item</code></h2>
1720<h4>appears in the <code>class_defs</code> section</h4>
1721<h4>alignment: 4 bytes</h4>
1722
1723<table class="format">
1724<thead>
1725<tr>
1726  <th>Name</th>
1727  <th>Format</th>
1728  <th>Description</th>
1729</tr>
1730</thead>
1731<tbody>
1732<tr>
1733  <td>class_idx</td>
1734  <td>uint</td>
1735  <td>index into the <code>type_ids</code> list for this class.
1736    This must be a class type, and not an array or primitive type.
1737  </td>
1738</tr>
1739<tr>
1740  <td>access_flags</td>
1741  <td>uint</td>
1742  <td>access flags for the class (<code>public</code>, <code>final</code>,
1743    etc.). See "<code>access_flags</code> Definitions" for details.
1744  </td>
1745</tr>
1746<tr>
1747  <td>superclass_idx</td>
1748  <td>uint</td>
1749  <td>index into the <code>type_ids</code> list for the superclass, or
1750    the constant value <code>NO_INDEX</code> if this class has no
1751    superclass (i.e., it is a root class such as <code>Object</code>).
1752    If present, this must be a class type, and not an array or primitive type.
1753  </td>
1754</tr>
1755<tr>
1756  <td>interfaces_off</td>
1757  <td>uint</td>
1758  <td>offset from the start of the file to the list of interfaces, or
1759    <code>0</code> if there are none. This offset
1760    should be in the <code>data</code> section, and the data
1761    there should be in the format specified by
1762    "<code>type_list</code>" below. Each of the elements of the list
1763    must be a class type (not an array or primitive type), and there
1764    must not be any duplicates.
1765  </td>
1766</tr>
1767<tr>
1768  <td>source_file_idx</td>
1769  <td>uint</td>
1770  <td>index into the <code>string_ids</code> list for the name of the
1771    file containing the original source for (at least most of) this class,
1772    or the special value <code>NO_INDEX</code> to represent a lack of
1773    this information. The <code>debug_info_item</code> of any given method
1774    may override this source file, but the expectation is that most classes
1775    will only come from one source file.
1776  </td>
1777</tr>
1778<tr>
1779  <td>annotations_off</td>
1780  <td>uint</td>
1781  <td>offset from the start of the file to the annotations structure
1782    for this class, or <code>0</code> if there are no annotations on
1783    this class. This offset, if non-zero, should be in the
1784    <code>data</code> section, and the data there should be in
1785    the format specified by "<code>annotations_directory_item</code>" below,
1786    with all items referring to this class as the definer.
1787  </td>
1788</tr>
1789<tr>
1790  <td>class_data_off</td>
1791  <td>uint</td>
1792  <td>offset from the start of the file to the associated
1793    class data for this item, or <code>0</code> if there is no class
1794    data for this class. (This may be the case, for example, if this class
1795    is a marker interface.) The offset, if non-zero, should be in the
1796    <code>data</code> section, and the data there should be in the
1797    format specified by "<code>class_data_item</code>" below, with all
1798    items referring to this class as the definer.
1799  </td>
1800</tr>
1801<tr>
1802  <td>static_values_off</td>
1803  <td>uint</td>
1804  <td>offset from the start of the file to the list of initial
1805    values for <code>static</code> fields, or <code>0</code> if there
1806    are none (and all <code>static</code> fields are to be initialized with
1807    <code>0</code> or <code>null</code>). This offset should be in the
1808    <code>data</code> section, and the data there should be in the
1809    format specified by "<code>encoded_array_item</code>" below. The size
1810    of the array must be no larger than the number of <code>static</code>
1811    fields declared by this class, and the elements correspond to the
1812    <code>static</code> fields in the same order as declared in the
1813    corresponding <code>field_list</code>. The type of each array
1814    element must match the declared type of its corresponding field.
1815    If there are fewer elements in the array than there are
1816    <code>static</code> fields, then the leftover fields are initialized
1817    with a type-appropriate <code>0</code> or <code>null</code>.
1818  </td>
1819</tr>
1820</tbody>
1821</table>
1822
1823<h2><code>class_data_item</code></h2>
1824<h4>referenced from <code>class_def_item</code></h4>
1825<h4>appears in the <code>data</code> section</h4>
1826<h4>alignment: none (byte-aligned)</h4>
1827
1828<table class="format">
1829<thead>
1830<tr>
1831  <th>Name</th>
1832  <th>Format</th>
1833  <th>Description</th>
1834</tr>
1835</thead>
1836<tbody>
1837<tr>
1838  <td>static_fields_size</td>
1839  <td>uleb128</td>
1840  <td>the number of static fields defined in this item</td>
1841</tr>
1842<tr>
1843  <td>instance_fields_size</td>
1844  <td>uleb128</td>
1845  <td>the number of instance fields defined in this item</td>
1846</tr>
1847<tr>
1848  <td>direct_methods_size</td>
1849  <td>uleb128</td>
1850  <td>the number of direct methods defined in this item</td>
1851</tr>
1852<tr>
1853  <td>virtual_methods_size</td>
1854  <td>uleb128</td>
1855  <td>the number of virtual methods defined in this item</td>
1856</tr>
1857<tr>
1858  <td>static_fields</td>
1859  <td>encoded_field[static_fields_size]</td>
1860  <td>the defined static fields, represented as a sequence of
1861    encoded elements. The fields must be sorted by
1862    <code>field_idx</code> in increasing order.
1863  </td>
1864</tr>
1865<tr>
1866  <td>instance_fields</td>
1867  <td>encoded_field[instance_fields_size]</td>
1868  <td>the defined instance fields, represented as a sequence of
1869    encoded elements. The fields must be sorted by
1870    <code>field_idx</code> in increasing order.
1871  </td>
1872</tr>
1873<tr>
1874  <td>direct_methods</td>
1875  <td>encoded_method[direct_methods_size]</td>
1876  <td>the defined direct (any of <code>static</code>, <code>private</code>,
1877    or constructor) methods, represented as a sequence of
1878    encoded elements. The methods must be sorted by
1879    <code>method_idx</code> in increasing order.
1880  </td>
1881</tr>
1882<tr>
1883  <td>virtual_methods</td>
1884  <td>encoded_method[virtual_methods_size]</td>
1885  <td>the defined virtual (none of <code>static</code>, <code>private</code>,
1886    or constructor) methods, represented as a sequence of
1887    encoded elements. This list should <i>not</i> include inherited
1888    methods unless overridden by the class that this item represents. The
1889    methods must be sorted by <code>method_idx</code> in increasing order.
1890  </td>
1891</tr>
1892</tbody>
1893</table>
1894
1895<p><b>Note:</b> All elements' <code>field_id</code>s and
1896<code>method_id</code>s must refer to the same defining class.</p>
1897
1898<h3><code>encoded_field</code> Format</h3>
1899
1900<table class="format">
1901<thead>
1902<tr>
1903  <th>Name</th>
1904  <th>Format</th>
1905  <th>Description</th>
1906</tr>
1907</thead>
1908<tbody>
1909<tr>
1910  <td>field_idx_diff</td>
1911  <td>uleb128</td>
1912  <td>index into the <code>field_ids</code> list for the identity of this
1913    field (includes the name and descriptor), represented as a difference
1914    from the index of previous element in the list. The index of the
1915    first element in a list is represented directly.
1916  </td>
1917</tr>
1918<tr>
1919  <td>access_flags</td>
1920  <td>uleb128</td>
1921  <td>access flags for the field (<code>public</code>, <code>final</code>,
1922    etc.). See "<code>access_flags</code> Definitions" for details.
1923  </td>
1924</tr>
1925</tbody>
1926</table>
1927
1928<h3><code>encoded_method</code> Format</h3>
1929
1930<table class="format">
1931<thead>
1932<tr>
1933  <th>Name</th>
1934  <th>Format</th>
1935  <th>Description</th>
1936</tr>
1937</thead>
1938<tbody>
1939<tr>
1940  <td>method_idx_diff</td>
1941  <td>uleb128</td>
1942  <td>index into the <code>method_ids</code> list for the identity of this
1943    method (includes the name and descriptor), represented as a difference
1944    from the index of previous element in the list. The index of the
1945    first element in a list is represented directly.
1946  </td>
1947</tr>
1948<tr>
1949  <td>access_flags</td>
1950  <td>uleb128</td>
1951  <td>access flags for the method (<code>public</code>, <code>final</code>,
1952    etc.). See "<code>access_flags</code> Definitions" for details.
1953  </td>
1954</tr>
1955<tr>
1956  <td>code_off</td>
1957  <td>uleb128</td>
1958  <td>offset from the start of the file to the code structure for this
1959    method, or <code>0</code> if this method is either <code>abstract</code>
1960    or <code>native</code>. The offset should be to a location in the
1961    <code>data</code> section. The format of the data is specified by
1962    "<code>code_item</code>" below.
1963  </td>
1964</tr>
1965</tbody>
1966</table>
1967
1968<h2><code>type_list</code></h2>
1969<h4>referenced from <code>class_def_item</code> and
1970<code>proto_id_item</code></h4>
1971<h4>appears in the <code>data</code> section</h4>
1972<h4>alignment: 4 bytes</h4>
1973
1974<table class="format">
1975<thead>
1976<tr>
1977  <th>Name</th>
1978  <th>Format</th>
1979  <th>Description</th>
1980</tr>
1981</thead>
1982<tbody>
1983<tr>
1984  <td>size</td>
1985  <td>uint</td>
1986  <td>size of the list, in entries</td>
1987</tr>
1988<tr>
1989  <td>list</td>
1990  <td>type_item[size]</td>
1991  <td>elements of the list</td>
1992</tr>
1993</tbody>
1994</table>
1995
1996<h3><code>type_item</code> Format</h3>
1997
1998<table class="format">
1999<thead>
2000<tr>
2001  <th>Name</th>
2002  <th>Format</th>
2003  <th>Description</th>
2004</tr>
2005</thead>
2006<tbody>
2007<tr>
2008  <td>type_idx</td>
2009  <td>ushort</td>
2010  <td>index into the <code>type_ids</code> list</td>
2011</tr>
2012</tbody>
2013</table>
2014
2015<h2><code>code_item</code></h2>
2016<h4>referenced from <code>method_item</code></h4>
2017<h4>appears in the <code>data</code> section</h4>
2018<h4>alignment: 4 bytes</h4>
2019
2020<table class="format">
2021<thead>
2022<tr>
2023  <th>Name</th>
2024  <th>Format</th>
2025  <th>Description</th>
2026</tr>
2027</thead>
2028<tbody>
2029<tr>
2030  <td>registers_size</td>
2031  <td>ushort</td>
2032  <td>the number of registers used by this code</td>
2033</tr>
2034<tr>
2035  <td>ins_size</td>
2036  <td>ushort</td>
2037  <td>the number of words of incoming arguments to the method that this
2038    code is for</td>
2039</tr>
2040<tr>
2041  <td>outs_size</td>
2042  <td>ushort</td>
2043  <td>the number of words of outgoing argument space required by this
2044    code for method invocation
2045  </td>
2046</tr>
2047<tr>
2048  <td>tries_size</td>
2049  <td>ushort</td>
2050  <td>the number of <code>try_item</code>s for this instance. If non-zero,
2051    then these appear as the <code>tries</code> array just after the
2052    <code>insns</code> in this instance.
2053  </td>
2054</tr>
2055<tr>
2056  <td>debug_info_off</td>
2057  <td>uint</td>
2058  <td>offset from the start of the file to the debug info (line numbers +
2059    local variable info) sequence for this code, or <code>0</code> if
2060    there simply is no information. The offset, if non-zero, should be
2061    to a location in the <code>data</code> section. The format of
2062    the data is specified by "<code>debug_info_item</code>" below.
2063  </td>
2064</tr>
2065<tr>
2066  <td>insns_size</td>
2067  <td>uint</td>
2068  <td>size of the instructions list, in 16-bit code units</td>
2069</tr>
2070<tr>
2071  <td>insns</td>
2072  <td>ushort[insns_size]</td>
2073  <td>actual array of bytecode. The format of code in an <code>insns</code>
2074    array is specified by the companion document
2075    <a href="dalvik-bytecode.html">"Bytecode for the Dalvik VM"</a>. Note
2076    that though this is defined as an array of <code>ushort</code>, there
2077    are some internal structures that prefer four-byte alignment. Also,
2078    if this happens to be in an endian-swapped file, then the swapping is
2079    <i>only</i> done on individual <code>ushort</code>s and not on the
2080    larger internal structures.
2081  </td>
2082</tr>
2083<tr>
2084  <td>padding</td>
2085  <td>ushort <i>(optional)</i> = 0</td>
2086  <td>two bytes of padding to make <code>tries</code> four-byte aligned.
2087    This element is only present if <code>tries_size</code> is non-zero
2088    and <code>insns_size</code> is odd.
2089  </td>
2090</tr>
2091<tr>
2092  <td>tries</td>
2093  <td>try_item[tries_size] <i>(optional)</i></td>
2094  <td>array indicating where in the code exceptions may be caught and
2095    how to handle them. Elements of the array must be non-overlapping in
2096    range and in order from low to high address. This element is only
2097    present if <code>tries_size</code> is non-zero.
2098  </td>
2099</tr>
2100<tr>
2101  <td>handlers</td>
2102  <td>encoded_catch_handler_list <i>(optional)</i></td>
2103  <td>bytes representing a list of lists of catch types and associated
2104    handler addresses. Each <code>try_item</code> has a byte-wise offset
2105    into this structure. This element is only present if
2106    <code>tries_size</code> is non-zero.
2107  </td>
2108</tr>
2109</tbody>
2110</table>
2111
2112<h3><code>try_item</code> Format </h3>
2113
2114<table class="format">
2115<thead>
2116<tr>
2117  <th>Name</th>
2118  <th>Format</th>
2119  <th>Description</th>
2120</tr>
2121</thead>
2122<tbody>
2123<tr>
2124  <td>start_addr</td>
2125  <td>uint</td>
2126  <td>start address of the block of code covered by this entry. The address
2127    is a count of 16-bit code units to the start of the first covered
2128    instruction.
2129  </td>
2130</tr>
2131<tr>
2132  <td>insn_count</td>
2133  <td>ushort</td>
2134  <td>number of 16-bit code units covered by this entry. The last code
2135    unit covered (inclusive) is <code>start_addr + insn_count - 1</code>.
2136  </td>
2137</tr>
2138<tr>
2139  <td>handler_off</td>
2140  <td>ushort</td>
2141  <td>offset in bytes from the start of the associated encoded handler data
2142    to the <code>catch_handler_item</code> for this entry
2143  </td>
2144</tr>
2145</tbody>
2146</table>
2147
2148<h3><code>encoded_catch_handler_list</code> Format</h3>
2149
2150<table class="format">
2151<thead>
2152<tr>
2153  <th>Name</th>
2154  <th>Format</th>
2155  <th>Description</th>
2156</tr>
2157</thead>
2158<tbody>
2159<tr>
2160  <td>size</td>
2161  <td>uleb128</td>
2162  <td>size of this list, in entries</td>
2163</tr>
2164<tr>
2165  <td>list</td>
2166  <td>encoded_catch_handler[handlers_size]</td>
2167  <td>actual list of handler lists, represented directly (not as offsets),
2168    and concatenated sequentially</td>
2169</tr>
2170</tbody>
2171</table>
2172
2173<h3><code>encoded_catch_handler</code> Format</h3>
2174
2175<table class="format">
2176<thead>
2177<tr>
2178  <th>Name</th>
2179  <th>Format</th>
2180  <th>Description</th>
2181</tr>
2182</thead>
2183<tbody>
2184<tr>
2185  <td>size</td>
2186  <td>sleb128</td>
2187  <td>number of catch types in this list. If non-positive, then this is
2188    the negative of the number of catch types, and the catches are followed
2189    by a catch-all handler. For example: A <code>size</code> of <code>0</code>
2190    means that there is a catch-all but no explicitly typed catches.
2191    A <code>size</code> of <code>2</code> means that there are two explicitly
2192    typed catches and no catch-all. And a <code>size</code> of <code>-1</code>
2193    means that there is one typed catch along with a catch-all.
2194  </td>
2195</tr>
2196<tr>
2197  <td>handlers</td>
2198  <td>encoded_type_addr_pair[abs(size)]</td>
2199  <td>stream of <code>abs(size)</code> encoded items, one for each caught
2200    type, in the order that the types should be tested.
2201  </td>
2202</tr>
2203<tr>
2204  <td>catch_all_addr</td>
2205  <td>uleb128 <i>(optional)</i></td>
2206  <td>bytecode address of the catch-all handler. This element is only
2207    present if <code>size</code> is non-positive.
2208  </td>
2209</tr>
2210</tbody>
2211</table>
2212
2213<h3><code>encoded_type_addr_pair</code> Format</h3>
2214
2215<table class="format">
2216<thead>
2217<tr>
2218  <th>Name</th>
2219  <th>Format</th>
2220  <th>Description</th>
2221</tr>
2222</thead>
2223<tbody>
2224<tr>
2225  <td>type_idx</td>
2226  <td>uleb128</td>
2227  <td>index into the <code>type_ids</code> list for the type of the
2228    exception to catch
2229  </td>
2230</tr>
2231<tr>
2232  <td>addr</td>
2233  <td>uleb128</td>
2234  <td>bytecode address of the associated exception handler</td>
2235</tr>
2236</tbody>
2237</table>
2238
2239<h2><code>debug_info_item</code></h2>
2240<h4>referenced from <code>code_item</code></h4>
2241<h4>appears in the <code>data</code> section</h4>
2242<h4>alignment: none (byte-aligned)</h4>
2243
2244<p>Each <code>debug_info_item</code> defines a DWARF3-inspired byte-coded
2245state machine that, when interpreted, emits the positions
2246table and (potentially) the local variable information for a
2247<code>code_item</code>. The sequence begins with a variable-length
2248header (the length of which depends on the number of method
2249parameters), is followed by the state machine bytecodes, and ends
2250with an <code>DBG_END_SEQUENCE</code> byte.</p>
2251
2252<p>The state machine consists of five registers. The
2253<code>address</code> register represents the instruction offset in the
2254associated <code>insns_item</code> in 16-bit code units. The
2255<code>address</code> register starts at <code>0</code> at the beginning of each
2256<code>debug_info</code> sequence and may only monotonically increase.
2257The <code>line</code> register represents what source line number
2258should be associated with the next positions table entry emitted by
2259the state machine. It is initialized in the sequence header, and may
2260change in positive or negative directions but must never be less than
2261<code>1</code>. The <code>source_file</code> register represents the
2262source file that the line number entries refer to. It is initialized to
2263the value of <code>source_file_idx</code> in <code>class_def_item</code>.
2264The other two variables, <code>prologue_end</code> and
2265<code>epilogue_begin</code>, are boolean flags (initialized to
2266<code>false</code>) that indicate whether the next position emitted
2267should be considered a method prologue or epilogue. The state machine
2268must also track the name and type of the last local variable live in
2269each register for the <code>DBG_RESTART_LOCAL</code> code.</p>
2270
2271<p>The header is as follows:</p>
2272
2273<table class="format">
2274<thead>
2275<tr>
2276  <th>Name</th>
2277  <th>Format</th>
2278  <th>Description</th>
2279</tr>
2280</thead>
2281<tbody>
2282<tr>
2283 <td>line_start</td>
2284 <td>uleb128</td>
2285 <td>the initial value for the state machine's <code>line</code> register.
2286    Does not represent an actual positions entry.
2287 </td>
2288</tr>
2289<tr>
2290 <td>parameters_size</td>
2291 <td>uleb128</td>
2292 <td>the number of parameter names that are encoded. There should be
2293   one per method parameter, excluding an instance method's <code>this</code>,
2294   if any.
2295 </td>
2296</tr>
2297<tr>
2298 <td>parameter_names</td>
2299 <td>uleb128p1[parameters_size]</td>
2300 <td>string index of the method parameter name. An encoded value of
2301   <code>NO_INDEX</code> indicates that no name
2302   is available for the associated parameter. The type descriptor
2303   and signature are implied from the method descriptor and signature.
2304 </td>
2305</tr>
2306</tbody>
2307</table>
2308
2309<p>The byte code values are as follows:</p>
2310
2311<table class="debugByteCode">
2312<thead>
2313<tr>
2314  <th>Name</th>
2315  <th>Value</th>
2316  <th>Format</th>
2317  <th>Arguments</th>
2318  <th>Description</th>
2319</tr>
2320</thead>
2321<tbody>
2322<tr>
2323  <td>DBG_END_SEQUENCE</td>
2324  <td>0x00</td>
2325  <td></td>
2326  <td><i>(none)</i></td>
2327  <td>terminates a debug info sequence for a <code>code_item</code></td>
2328</tr>
2329<tr>
2330  <td>DBG_ADVANCE_PC</td>
2331  <td>0x01</td>
2332  <td>uleb128&nbsp;addr_diff</td>
2333  <td><code>addr_diff</code>: amount to add to address register</td>
2334  <td>advances the address register without emitting a positions entry</td>
2335</tr>
2336<tr>
2337  <td>DBG_ADVANCE_LINE</td>
2338  <td>0x02</td>
2339  <td>sleb128&nbsp;line_diff</td>
2340  <td><code>line_diff</code>: amount to change line register by</td>
2341  <td>advances the line register without emitting a positions entry</td>
2342</tr>
2343<tr>
2344  <td>DBG_START_LOCAL</td>
2345  <td>0x03</td>
2346  <td>uleb128&nbsp;register_num<br/>
2347    uleb128p1&nbsp;name_idx<br/>
2348    uleb128p1&nbsp;type_idx
2349  </td>
2350  <td><code>register_num</code>: register that will contain local<br/>
2351    <code>name_idx</code>: string index of the name<br/>
2352    <code>type_idx</code>: type index of the type
2353  </td>
2354  <td>introduces a local variable at the current address. Either
2355    <code>name_idx</code> or <code>type_idx</code> may be
2356    <code>NO_INDEX</code> to indicate that that value is unknown.
2357  </td>
2358</tr>
2359<tr>
2360  <td>DBG_START_LOCAL_EXTENDED</td>
2361  <td>0x04</td>
2362  <td>uleb128&nbsp;register_num<br/>
2363    uleb128p1&nbsp;name_idx<br/>
2364    uleb128p1&nbsp;type_idx<br/>
2365    uleb128p1&nbsp;sig_idx
2366  </td>
2367  <td><code>register_num</code>: register that will contain local<br/>
2368    <code>name_idx</code>: string index of the name<br/>
2369    <code>type_idx</code>: type index of the type<br/>
2370    <code>sig_idx</code>: string index of the type signature
2371  </td>
2372  <td>introduces a local with a type signature at the current address.
2373    Any of <code>name_idx</code>, <code>type_idx</code>, or
2374    <code>sig_idx</code> may be <code>NO_INDEX</code>
2375    to indicate that that value is unknown. (If <code>sig_idx</code> is
2376    <code>-1</code>, though, the same data could be represented more
2377    efficiently using the opcode <code>DBG_START_LOCAL</code>.)
2378    <p><b>Note:</b> See the discussion under
2379    "<code>dalvik.annotation.Signature</code>" below for caveats about
2380    handling signatures.</p>
2381  </td>
2382</tr>
2383<tr>
2384  <td>DBG_END_LOCAL</td>
2385  <td>0x05</td>
2386  <td>uleb128&nbsp;register_num</td>
2387  <td><code>register_num</code>: register that contained local</td>
2388  <td>marks a currently-live local variable as out of scope at the current
2389    address
2390  </td>
2391</tr>
2392<tr>
2393  <td>DBG_RESTART_LOCAL</td>
2394  <td>0x06</td>
2395  <td>uleb128&nbsp;register_num</td>
2396  <td><code>register_num</code>: register to restart</td>
2397  <td>re-introduces a local variable at the current address. The name
2398    and type are the same as the last local that was live in the specified
2399    register.
2400  </td>
2401</tr>
2402<tr>
2403  <td>DBG_SET_PROLOGUE_END</td>
2404  <td>0x07</td>
2405  <td></td>
2406  <td><i>(none)</i></td>
2407  <td>sets the <code>prologue_end</code> state machine register,
2408    indicating that the next position entry that is added should be
2409    considered the end of a method prologue (an appropriate place for
2410    a method breakpoint). The <code>prologue_end</code> register is
2411    cleared by any special (<code>&gt;= 0x0a</code>) opcode.
2412  </td>
2413</tr>
2414<tr>
2415  <td>DBG_SET_EPILOGUE_BEGIN</td>
2416  <td>0x08</td>
2417  <td></td>
2418  <td><i>(none)</i></td>
2419  <td>sets the <code>epilogue_begin</code> state machine register,
2420    indicating that the next position entry that is added should be
2421    considered the beginning of a method epilogue (an appropriate place
2422    to suspend execution before method exit).
2423    The <code>epilogue_begin</code> register is cleared by any special
2424    (<code>&gt;= 0x0a</code>) opcode.
2425  </td>
2426</tr>
2427<tr>
2428  <td>DBG_SET_FILE</td>
2429  <td>0x09</td>
2430  <td>uleb128p1&nbsp;name_idx</td>
2431  <td><code>name_idx</code>: string index of source file name;
2432    <code>NO_INDEX</code> if unknown
2433  </td>
2434  <td>indicates that all subsequent line number entries make reference to this
2435    source file name, instead of the default name specified in
2436    <code>code_item</code>
2437  </td>
2438</tr>
2439<tr>
2440  <td><i>Special Opcodes</i></td>
2441  <!-- When updating the range below, make sure to search for other
2442  instances of 0x0a in this section. -->
2443  <td>0x0a&hellip;0xff</td>
2444  <td></td>
2445  <td><i>(none)</i></td>
2446  <td>advances the <code>line</code> and <code>address</code> registers,
2447    emits a position entry, and clears <code>prologue_end</code> and
2448    <code>epilogue_begin</code>. See below for description.
2449  </td>
2450</tr>
2451</tbody>
2452</table>
2453
2454<h3>Special Opcodes</h3>
2455
2456<p>Opcodes with values between <code>0x0a</code> and <code>0xff</code>
2457(inclusive) move both the <code>line</code> and <code>address</code>
2458registers by a small amount and then emit a new position table entry.
2459The formula for the increments are as follows:</p>
2460
2461<pre>
2462DBG_FIRST_SPECIAL = 0x0a  // the smallest special opcode
2463DBG_LINE_BASE   = -4      // the smallest line number increment
2464DBG_LINE_RANGE  = 15      // the number of line increments represented
2465
2466adjusted_opcode = opcode - DBG_FIRST_SPECIAL
2467
2468line += DBG_LINE_BASE + (adjusted_opcode % DBG_LINE_RANGE)
2469address += (adjusted_opcode / DBG_LINE_RANGE)
2470</pre>
2471
2472<h2><code>annotations_directory_item</code></h2>
2473<h4>referenced from <code>class_def_item</code></h4>
2474<h4>appears in the <code>data</code> section</h4>
2475<h4>alignment: 4 bytes</h4>
2476
2477<table class="format">
2478<thead>
2479<tr>
2480  <th>Name</th>
2481  <th>Format</th>
2482  <th>Description</th>
2483</tr>
2484</thead>
2485<tbody>
2486<tr>
2487  <td>class_annotations_off</td>
2488  <td>uint</td>
2489  <td>offset from the start of the file to the annotations made directly
2490    on the class, or <code>0</code> if the class has no direct annotations.
2491    The offset, if non-zero, should be to a location in the
2492    <code>data</code> section. The format of the data is specified
2493    by "<code>annotation_set_item</code>" below.
2494  </td>
2495</tr>
2496<tr>
2497  <td>fields_size</td>
2498  <td>uint</td>
2499  <td>count of fields annotated by this item</td>
2500</tr>
2501<tr>
2502  <td>annotated_methods_size</td>
2503  <td>uint</td>
2504  <td>count of methods annotated by this item</td>
2505</tr>
2506<tr>
2507  <td>annotated_parameters_size</td>
2508  <td>uint</td>
2509  <td>count of method parameter lists annotated by this item</td>
2510</tr>
2511<tr>
2512  <td>field_annotations</td>
2513  <td>field_annotation[fields_size] <i>(optional)</i></td>
2514  <td>list of associated field annotations. The elements of the list must
2515    be sorted in increasing order, by <code>field_idx</code>.
2516  </td>
2517</tr>
2518<tr>
2519  <td>method_annotations</td>
2520  <td>method_annotation[methods_size] <i>(optional)</i></td>
2521  <td>list of associated method annotations. The elements of the list must
2522    be sorted in increasing order, by <code>method_idx</code>.
2523  </td>
2524</tr>
2525<tr>
2526  <td>parameter_annotations</td>
2527  <td>parameter_annotation[parameters_size] <i>(optional)</i></td>
2528  <td>list of associated method parameter annotations. The elements of the
2529    list must be sorted in increasing order, by <code>method_idx</code>.
2530  </td>
2531</tr>
2532</tbody>
2533</table>
2534
2535<p><b>Note:</b> All elements' <code>field_id</code>s and
2536<code>method_id</code>s must refer to the same defining class.</p>
2537
2538<h3><code>field_annotation</code> Format</h3>
2539
2540<table class="format">
2541<thead>
2542<tr>
2543  <th>Name</th>
2544  <th>Format</th>
2545  <th>Description</th>
2546</tr>
2547</thead>
2548<tbody>
2549<tr>
2550  <td>field_idx</td>
2551  <td>uint</td>
2552  <td>index into the <code>field_ids</code> list for the identity of the
2553    field being annotated
2554  </td>
2555</tr>
2556<tr>
2557  <td>annotations_off</td>
2558  <td>uint</td>
2559  <td>offset from the start of the file to the list of annotations for
2560    the field. The offset should be to a location in the <code>data</code>
2561    section. The format of the data is specified by
2562    "<code>annotation_set_item</code>" below.
2563  </td>
2564</tr>
2565</tbody>
2566</table>
2567
2568<h3><code>method_annotation</code> Format</h3>
2569
2570<table class="format">
2571<thead>
2572<tr>
2573  <th>Name</th>
2574  <th>Format</th>
2575  <th>Description</th>
2576</tr>
2577</thead>
2578<tbody>
2579<tr>
2580  <td>method_idx</td>
2581  <td>uint</td>
2582  <td>index into the <code>method_ids</code> list for the identity of the
2583    method being annotated
2584  </td>
2585</tr>
2586<tr>
2587  <td>annotations_off</td>
2588  <td>uint</td>
2589  <td>offset from the start of the file to the list of annotations for
2590    the method. The offset should be to a location in the
2591    <code>data</code> section. The format of the data is specified by
2592    "<code>annotation_set_item</code>" below.
2593  </td>
2594</tr>
2595</tbody>
2596</table>
2597
2598<h3><code>parameter_annotation</code> Format</h2>
2599
2600<table class="format">
2601<thead>
2602<tr>
2603  <th>Name</th>
2604  <th>Format</th>
2605  <th>Description</th>
2606</tr>
2607</thead>
2608<tbody>
2609<tr>
2610  <td>method_idx</td>
2611  <td>uint</td>
2612  <td>index into the <code>method_ids</code> list for the identity of the
2613    method whose parameters are being annotated
2614  </td>
2615</tr>
2616<tr>
2617  <td>annotations_off</td>
2618  <td>uint</td>
2619  <td>offset from the start of the file to the list of annotations for
2620    the method parameters. The offset should be to a location in the
2621    <code>data</code> section. The format of the data is specified by
2622    "<code>annotation_set_ref_list</code>" below.
2623  </td>
2624</tr>
2625</tbody>
2626</table>
2627
2628<h2><code>annotation_set_ref_list</code></h2>
2629<h4>referenced from <code>parameter_annotations_item</code></h4>
2630<h4>appears in the <code>data</code> section</h4>
2631<h4>alignment: 4 bytes</h4>
2632
2633<table class="format">
2634<thead>
2635<tr>
2636  <th>Name</th>
2637  <th>Format</th>
2638  <th>Description</th>
2639</tr>
2640</thead>
2641<tbody>
2642<tr>
2643  <td>size</td>
2644  <td>uint</td>
2645  <td>size of the list, in entries</td>
2646</tr>
2647<tr>
2648  <td>list</td>
2649  <td>annotation_set_ref_item[size]</td>
2650  <td>elements of the list</td>
2651</tr>
2652</tbody>
2653</table>
2654
2655<h3><code>annotation_set_ref_item</code> Format</h3>
2656
2657<table class="format">
2658<thead>
2659<tr>
2660  <th>Name</th>
2661  <th>Format</th>
2662  <th>Description</th>
2663</tr>
2664</thead>
2665<tbody>
2666<tr>
2667  <td>annotations_off</td>
2668  <td>uint</td>
2669  <td>offset from the start of the file to the referenced annotation set
2670    or <code>0</code> if there are no annotations for this element.
2671    The offset, if non-zero, should be to a location in the <code>data</code>
2672    section. The format of the data is specified by
2673    "<code>annotation_set_item</code>" below.
2674  </td>
2675</tr>
2676</tbody>
2677</table>
2678
2679<h2><code>annotation_set_item</code></h2>
2680<h4>referenced from <code>annotations_directory_item</code>,
2681<code>field_annotations_item</code>,
2682<code>method_annotations_item</code>, and
2683<code>annotation_set_ref_item</code></h4>
2684<h4>appears in the <code>data</code> section</h4>
2685<h4>alignment: 4 bytes</h4>
2686
2687<table class="format">
2688<thead>
2689<tr>
2690  <th>Name</th>
2691  <th>Format</th>
2692  <th>Description</th>
2693</tr>
2694</thead>
2695<tbody>
2696<tr>
2697  <td>size</td>
2698  <td>uint</td>
2699  <td>size of the set, in entries</td>
2700</tr>
2701<tr>
2702  <td>entries</td>
2703  <td>annotation_off_item[size]</td>
2704  <td>elements of the set. The elements must be sorted in increasing order,
2705    by <code>type_idx</code>.
2706  </td>
2707</tr>
2708</tbody>
2709</table>
2710
2711<h3><code>annotation_off_item</code> Format</h3>
2712
2713<table class="format">
2714<thead>
2715<tr>
2716  <th>Name</th>
2717  <th>Format</th>
2718  <th>Description</th>
2719</tr>
2720</thead>
2721<tbody>
2722<tr>
2723  <td>annotation_off</td>
2724  <td>uint</td>
2725  <td>offset from the start of the file to an annotation.
2726    The offset should be to a location in the <code>data</code> section,
2727    and the format of the data at that location is specified by
2728    "<code>annotation_item</code>" below.
2729  </td>
2730</tr>
2731</tbody>
2732</table>
2733
2734
2735<h2><code>annotation_item</code></h2>
2736<h4>referenced from <code>annotation_set_item</code></h4>
2737<h4>appears in the <code>data</code> section</h4>
2738<h4>alignment: none (byte-aligned)</h4>
2739
2740<table class="format">
2741<thead>
2742<tr>
2743  <th>Name</th>
2744  <th>Format</th>
2745  <th>Description</th>
2746</tr>
2747</thead>
2748<tbody>
2749<tr>
2750  <td>visibility</td>
2751  <td>ubyte</td>
2752  <td>intended visibility of this annotation (see below)</td>
2753</tr>
2754<tr>
2755  <td>annotation</td>
2756  <td>encoded_annotation</td>
2757  <td>encoded annotation contents, in the format described by
2758    "<code>encoded_annotation</code> Format" under
2759    "<code>encoded_value</code> Encoding" above.
2760  </td>
2761</tr>
2762</tbody>
2763</table>
2764
2765<h3>Visibility values</h3>
2766
2767<p>These are the options for the <code>visibility</code> field in an
2768<code>annotation_item</code>:</p>
2769
2770<table class="format">
2771<thead>
2772<tr>
2773  <th>Name</th>
2774  <th>Value</th>
2775  <th>Description</th>
2776</tr>
2777</thead>
2778<tbody>
2779<tr>
2780  <td>VISIBILITY_BUILD</td>
2781  <td>0x00</td>
2782  <td>intended only to be visible at build time (e.g., during compilation
2783    of other code)
2784  </td>
2785</tr>
2786<tr>
2787  <td>VISIBILITY_RUNTIME</td>
2788  <td>0x01</td>
2789  <td>intended to visible at runtime</td>
2790</tr>
2791<tr>
2792  <td>VISIBILITY_SYSTEM</td>
2793  <td>0x02</td>
2794  <td>intended to visible at runtime, but only to the underlying system
2795    (and not to regular user code)
2796  </td>
2797</tr>
2798</tbody>
2799</table>
2800
2801<h2><code>encoded_array_item</code></h2>
2802<h4>referenced from <code>class_def_item</code></h4>
2803<h4>appears in the <code>data</code> section</h4>
2804<h4>alignment: none (byte-aligned)</h4>
2805
2806<table class="format">
2807<thead>
2808<tr>
2809  <th>Name</th>
2810  <th>Format</th>
2811  <th>Description</th>
2812</tr>
2813</thead>
2814<tbody>
2815<tr>
2816  <td>value</td>
2817  <td>encoded_array</td>
2818  <td>bytes representing the encoded array value, in the format specified
2819    by "<code>encoded_array</code> Format" under "<code>encoded_value</code>
2820    Encoding" above.
2821  </td>
2822</tr>
2823</tbody>
2824</table>
2825
2826<h1>System Annotations</h1>
2827
2828<p>System annotations are used to represent various pieces of reflective
2829information about classes (and methods and fields). This information is
2830generally only accessed indirectly by client (non-system) code.</p>
2831
2832<p>System annotations are represented in <code>.dex</code> files as
2833annotations with visibility set to <code>VISIBILITY_SYSTEM</code>.
2834
2835<h2><code>dalvik.annotation.AnnotationDefault</code></h2>
2836<h4>appears on methods in annotation interfaces</h4>
2837
2838<p>An <code>AnnotationDefault</code> annotation is attached to each
2839annotation interface which wishes to indicate default bindings.</p>
2840
2841<table class="format">
2842<thead>
2843<tr>
2844  <th>Name</th>
2845  <th>Format</th>
2846  <th>Description</th>
2847</tr>
2848</thead>
2849<tbody>
2850<tr>
2851  <td>value</td>
2852  <td>Annotation</td>
2853  <td>the default bindings for this annotation, represented as an annotation
2854    of this type. The annotation need not include all names defined by the
2855    annotation; missing names simply do not have defaults.
2856  </td>
2857</tr>
2858</tbody>
2859</table>
2860
2861<h2><code>dalvik.annotation.EnclosingClass</code></h2>
2862<h4>appears on classes</h4>
2863
2864<p>An <code>EnclosingClass</code> annotation is attached to each class
2865which is either defined as a member of another class, per se, or is
2866anonymous but not defined within a method body (e.g., a synthetic
2867inner class). Every class that has this annotation must also have an
2868<code>InnerClass</code> annotation. Additionally, a class may not have
2869both an <code>EnclosingClass</code> and an
2870<code>EnclosingMethod</code> annotation.</p>
2871
2872<table class="format">
2873<thead>
2874<tr>
2875  <th>Name</th>
2876  <th>Format</th>
2877  <th>Description</th>
2878</tr>
2879</thead>
2880<tbody>
2881<tr>
2882  <td>value</td>
2883  <td>Class</td>
2884  <td>the class which most closely lexically scopes this class</td>
2885</tr>
2886</tbody>
2887</table>
2888
2889<h2><code>dalvik.annotation.EnclosingMethod</code></h2>
2890<h4>appears on classes</h4>
2891
2892<p>An <code>EnclosingMethod</code> annotation is attached to each class
2893which is defined inside a method body. Every class that has this
2894annotation must also have an <code>InnerClass</code> annotation.
2895Additionally, a class may not have both an <code>EnclosingClass</code>
2896and an <code>EnclosingMethod</code> annotation.</p>
2897
2898<table class="format">
2899<thead>
2900<tr>
2901  <th>Name</th>
2902  <th>Format</th>
2903  <th>Description</th>
2904</tr>
2905</thead>
2906<tbody>
2907<tr>
2908  <td>value</td>
2909  <td>Method</td>
2910  <td>the method which most closely lexically scopes this class</td>
2911</tr>
2912</tbody>
2913</table>
2914
2915<h2><code>dalvik.annotation.InnerClass</code></h2>
2916<h4>appears on classes</h4>
2917
2918<p>An <code>InnerClass</code> annotation is attached to each class
2919which is defined in the lexical scope of another class's definition.
2920Any class which has this annotation must also have <i>either</i> an
2921<code>EnclosingClass</code> annotation <i>or</i> an
2922<code>EnclosingMethod</code> annotation.</p>
2923
2924<table class="format">
2925<thead>
2926<tr>
2927  <th>Name</th>
2928  <th>Format</th>
2929  <th>Description</th>
2930</tr>
2931</thead>
2932<tbody>
2933<tr>
2934  <td>name</td>
2935  <td>String</td>
2936  <td>the originally declared simple name of this class (not including any
2937    package prefix). If this class is anonymous, then the name is
2938    <code>null</code>.
2939  </td>
2940</tr>
2941<tr>
2942  <td>accessFlags</td>
2943  <td>int</td>
2944  <td>the originally declared access flags of the class (which may differ
2945    from the effective flags because of a mismatch between the execution
2946    models of the source language and target virtual machine)
2947  </td>
2948</tr>
2949</tbody>
2950</table>
2951
2952<h2><code>dalvik.annotation.MemberClasses</code></h2>
2953<h4>appears on classes</h4>
2954
2955<p>A <code>MemberClasses</code> annotation is attached to each class
2956which declares member classes. (A member class is a direct inner class
2957that has a name.)</p>
2958
2959<table class="format">
2960<thead>
2961<tr>
2962  <th>Name</th>
2963  <th>Format</th>
2964  <th>Description</th>
2965</tr>
2966</thead>
2967<tbody>
2968<tr>
2969  <td>value</td>
2970  <td>Class[]</td>
2971  <td>array of the member classes</td>
2972</tr>
2973</tbody>
2974</table>
2975
2976<h2><code>dalvik.annotation.Signature</code></h2>
2977<h4>appears on classes, fields, and methods</h4>
2978
2979<p>A <code>Signature</code> annotation is attached to each class,
2980field, or method which is defined in terms of a more complicated type
2981than is representable by a <code>type_id_item</code>. The
2982<code>.dex</code> format does not define the format for signatures; it
2983is merely meant to be able to represent whatever signatures a source
2984language requires for successful implementation of that language's
2985semantics. As such, signatures are not generally parsed (or verified)
2986by virtual machine implementations. The signatures simply get handed
2987off to higher-level APIs and tools (such as debuggers). Any use of a
2988signature, therefore, should be written so as not to make any
2989assumptions about only receiving valid signatures, explicitly guarding
2990itself against the possibility of coming across a syntactically
2991invalid signature.</p>
2992
2993<p>Because signature strings tend to have a lot of duplicated content,
2994a <code>Signature</code> annotation is defined as an <i>array</i> of
2995strings, where duplicated elements naturally refer to the same
2996underlying data, and the signature is taken to be the concatenation of
2997all the strings in the array. There are no rules about how to pull
2998apart a signature into separate strings; that is entirely up to the
2999tools that generate <code>.dex</code> files.</p>
3000
3001<table class="format">
3002<thead>
3003<tr>
3004  <th>Name</th>
3005  <th>Format</th>
3006  <th>Description</th>
3007</tr>
3008</thead>
3009<tbody>
3010<tr>
3011  <td>value</td>
3012  <td>String[]</td>
3013  <td>the signature of this class or member, as an array of strings that
3014    is to be concatenated together</td>
3015</tr>
3016</tbody>
3017</table>
3018
3019<h2><code>dalvik.annotation.Throws</code></h2>
3020<h4>appears on methods</h4>
3021
3022<p>A <code>Throws</code> annotation is attached to each method which is
3023declared to throw one or more exception types.</p>
3024
3025<table class="format">
3026<thead>
3027<tr>
3028  <th>Name</th>
3029  <th>Format</th>
3030  <th>Description</th>
3031</tr>
3032</thead>
3033<tbody>
3034<tr>
3035  <td>value</td>
3036  <td>Class[]</td>
3037  <td>the array of exception types thrown</td>
3038</tr>
3039</tbody>
3040</table>
3041
3042</body>
3043</html>
3044