descriptor_database.h 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369
  1. // Protocol Buffers - Google's data interchange format
  2. // Copyright 2008 Google Inc. All rights reserved.
  3. // https://developers.google.com/protocol-buffers/
  4. //
  5. // Redistribution and use in source and binary forms, with or without
  6. // modification, are permitted provided that the following conditions are
  7. // met:
  8. //
  9. // * Redistributions of source code must retain the above copyright
  10. // notice, this list of conditions and the following disclaimer.
  11. // * Redistributions in binary form must reproduce the above
  12. // copyright notice, this list of conditions and the following disclaimer
  13. // in the documentation and/or other materials provided with the
  14. // distribution.
  15. // * Neither the name of Google Inc. nor the names of its
  16. // contributors may be used to endorse or promote products derived from
  17. // this software without specific prior written permission.
  18. //
  19. // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  20. // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  21. // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  22. // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  23. // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  24. // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  25. // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  26. // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  27. // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  28. // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  29. // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  30. // Author: kenton@google.com (Kenton Varda)
  31. // Based on original Protocol Buffers design by
  32. // Sanjay Ghemawat, Jeff Dean, and others.
  33. //
  34. // Interface for manipulating databases of descriptors.
  35. #ifndef GOOGLE_PROTOBUF_DESCRIPTOR_DATABASE_H__
  36. #define GOOGLE_PROTOBUF_DESCRIPTOR_DATABASE_H__
  37. #include <map>
  38. #include <string>
  39. #include <utility>
  40. #include <vector>
  41. #include <google/protobuf/stubs/common.h>
  42. #include <google/protobuf/descriptor.h>
  43. namespace google {
  44. namespace protobuf {
  45. // Defined in this file.
  46. class DescriptorDatabase;
  47. class SimpleDescriptorDatabase;
  48. class EncodedDescriptorDatabase;
  49. class DescriptorPoolDatabase;
  50. class MergedDescriptorDatabase;
  51. // Abstract interface for a database of descriptors.
  52. //
  53. // This is useful if you want to create a DescriptorPool which loads
  54. // descriptors on-demand from some sort of large database. If the database
  55. // is large, it may be inefficient to enumerate every .proto file inside it
  56. // calling DescriptorPool::BuildFile() for each one. Instead, a DescriptorPool
  57. // can be created which wraps a DescriptorDatabase and only builds particular
  58. // descriptors when they are needed.
  59. class LIBPROTOBUF_EXPORT DescriptorDatabase {
  60. public:
  61. inline DescriptorDatabase() {}
  62. virtual ~DescriptorDatabase();
  63. // Find a file by file name. Fills in in *output and returns true if found.
  64. // Otherwise, returns false, leaving the contents of *output undefined.
  65. virtual bool FindFileByName(const string& filename,
  66. FileDescriptorProto* output) = 0;
  67. // Find the file that declares the given fully-qualified symbol name.
  68. // If found, fills in *output and returns true, otherwise returns false
  69. // and leaves *output undefined.
  70. virtual bool FindFileContainingSymbol(const string& symbol_name,
  71. FileDescriptorProto* output) = 0;
  72. // Find the file which defines an extension extending the given message type
  73. // with the given field number. If found, fills in *output and returns true,
  74. // otherwise returns false and leaves *output undefined. containing_type
  75. // must be a fully-qualified type name.
  76. virtual bool FindFileContainingExtension(const string& containing_type,
  77. int field_number,
  78. FileDescriptorProto* output) = 0;
  79. // Finds the tag numbers used by all known extensions of
  80. // extendee_type, and appends them to output in an undefined
  81. // order. This method is best-effort: it's not guaranteed that the
  82. // database will find all extensions, and it's not guaranteed that
  83. // FindFileContainingExtension will return true on all of the found
  84. // numbers. Returns true if the search was successful, otherwise
  85. // returns false and leaves output unchanged.
  86. //
  87. // This method has a default implementation that always returns
  88. // false.
  89. virtual bool FindAllExtensionNumbers(const string& /* extendee_type */,
  90. vector<int>* /* output */) {
  91. return false;
  92. }
  93. private:
  94. GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(DescriptorDatabase);
  95. };
  96. // A DescriptorDatabase into which you can insert files manually.
  97. //
  98. // FindFileContainingSymbol() is fully-implemented. When you add a file, its
  99. // symbols will be indexed for this purpose. Note that the implementation
  100. // may return false positives, but only if it isn't possible for the symbol
  101. // to be defined in any other file. In particular, if a file defines a symbol
  102. // "Foo", then searching for "Foo.[anything]" will match that file. This way,
  103. // the database does not need to aggressively index all children of a symbol.
  104. //
  105. // FindFileContainingExtension() is mostly-implemented. It works if and only
  106. // if the original FieldDescriptorProto defining the extension has a
  107. // fully-qualified type name in its "extendee" field (i.e. starts with a '.').
  108. // If the extendee is a relative name, SimpleDescriptorDatabase will not
  109. // attempt to resolve the type, so it will not know what type the extension is
  110. // extending. Therefore, calling FindFileContainingExtension() with the
  111. // extension's containing type will never actually find that extension. Note
  112. // that this is an unlikely problem, as all FileDescriptorProtos created by the
  113. // protocol compiler (as well as ones created by calling
  114. // FileDescriptor::CopyTo()) will always use fully-qualified names for all
  115. // types. You only need to worry if you are constructing FileDescriptorProtos
  116. // yourself, or are calling compiler::Parser directly.
  117. class LIBPROTOBUF_EXPORT SimpleDescriptorDatabase : public DescriptorDatabase {
  118. public:
  119. SimpleDescriptorDatabase();
  120. ~SimpleDescriptorDatabase();
  121. // Adds the FileDescriptorProto to the database, making a copy. The object
  122. // can be deleted after Add() returns. Returns false if the file conflicted
  123. // with a file already in the database, in which case an error will have
  124. // been written to GOOGLE_LOG(ERROR).
  125. bool Add(const FileDescriptorProto& file);
  126. // Adds the FileDescriptorProto to the database and takes ownership of it.
  127. bool AddAndOwn(const FileDescriptorProto* file);
  128. // implements DescriptorDatabase -----------------------------------
  129. bool FindFileByName(const string& filename,
  130. FileDescriptorProto* output);
  131. bool FindFileContainingSymbol(const string& symbol_name,
  132. FileDescriptorProto* output);
  133. bool FindFileContainingExtension(const string& containing_type,
  134. int field_number,
  135. FileDescriptorProto* output);
  136. bool FindAllExtensionNumbers(const string& extendee_type,
  137. vector<int>* output);
  138. private:
  139. // So that it can use DescriptorIndex.
  140. friend class EncodedDescriptorDatabase;
  141. // An index mapping file names, symbol names, and extension numbers to
  142. // some sort of values.
  143. template <typename Value>
  144. class DescriptorIndex {
  145. public:
  146. // Helpers to recursively add particular descriptors and all their contents
  147. // to the index.
  148. bool AddFile(const FileDescriptorProto& file,
  149. Value value);
  150. bool AddSymbol(const string& name, Value value);
  151. bool AddNestedExtensions(const DescriptorProto& message_type,
  152. Value value);
  153. bool AddExtension(const FieldDescriptorProto& field,
  154. Value value);
  155. Value FindFile(const string& filename);
  156. Value FindSymbol(const string& name);
  157. Value FindExtension(const string& containing_type, int field_number);
  158. bool FindAllExtensionNumbers(const string& containing_type,
  159. vector<int>* output);
  160. private:
  161. map<string, Value> by_name_;
  162. map<string, Value> by_symbol_;
  163. map<pair<string, int>, Value> by_extension_;
  164. // Invariant: The by_symbol_ map does not contain any symbols which are
  165. // prefixes of other symbols in the map. For example, "foo.bar" is a
  166. // prefix of "foo.bar.baz" (but is not a prefix of "foo.barbaz").
  167. //
  168. // This invariant is important because it means that given a symbol name,
  169. // we can find a key in the map which is a prefix of the symbol in O(lg n)
  170. // time, and we know that there is at most one such key.
  171. //
  172. // The prefix lookup algorithm works like so:
  173. // 1) Find the last key in the map which is less than or equal to the
  174. // search key.
  175. // 2) If the found key is a prefix of the search key, then return it.
  176. // Otherwise, there is no match.
  177. //
  178. // I am sure this algorithm has been described elsewhere, but since I
  179. // wasn't able to find it quickly I will instead prove that it works
  180. // myself. The key to the algorithm is that if a match exists, step (1)
  181. // will find it. Proof:
  182. // 1) Define the "search key" to be the key we are looking for, the "found
  183. // key" to be the key found in step (1), and the "match key" to be the
  184. // key which actually matches the serach key (i.e. the key we're trying
  185. // to find).
  186. // 2) The found key must be less than or equal to the search key by
  187. // definition.
  188. // 3) The match key must also be less than or equal to the search key
  189. // (because it is a prefix).
  190. // 4) The match key cannot be greater than the found key, because if it
  191. // were, then step (1) of the algorithm would have returned the match
  192. // key instead (since it finds the *greatest* key which is less than or
  193. // equal to the search key).
  194. // 5) Therefore, the found key must be between the match key and the search
  195. // key, inclusive.
  196. // 6) Since the search key must be a sub-symbol of the match key, if it is
  197. // not equal to the match key, then search_key[match_key.size()] must
  198. // be '.'.
  199. // 7) Since '.' sorts before any other character that is valid in a symbol
  200. // name, then if the found key is not equal to the match key, then
  201. // found_key[match_key.size()] must also be '.', because any other value
  202. // would make it sort after the search key.
  203. // 8) Therefore, if the found key is not equal to the match key, then the
  204. // found key must be a sub-symbol of the match key. However, this would
  205. // contradict our map invariant which says that no symbol in the map is
  206. // a sub-symbol of any other.
  207. // 9) Therefore, the found key must match the match key.
  208. //
  209. // The above proof assumes the match key exists. In the case that the
  210. // match key does not exist, then step (1) will return some other symbol.
  211. // That symbol cannot be a super-symbol of the search key since if it were,
  212. // then it would be a match, and we're assuming the match key doesn't exist.
  213. // Therefore, step 2 will correctly return no match.
  214. // Find the last entry in the by_symbol_ map whose key is less than or
  215. // equal to the given name.
  216. typename map<string, Value>::iterator FindLastLessOrEqual(
  217. const string& name);
  218. // True if either the arguments are equal or super_symbol identifies a
  219. // parent symbol of sub_symbol (e.g. "foo.bar" is a parent of
  220. // "foo.bar.baz", but not a parent of "foo.barbaz").
  221. bool IsSubSymbol(const string& sub_symbol, const string& super_symbol);
  222. // Returns true if and only if all characters in the name are alphanumerics,
  223. // underscores, or periods.
  224. bool ValidateSymbolName(const string& name);
  225. };
  226. DescriptorIndex<const FileDescriptorProto*> index_;
  227. vector<const FileDescriptorProto*> files_to_delete_;
  228. // If file is non-NULL, copy it into *output and return true, otherwise
  229. // return false.
  230. bool MaybeCopy(const FileDescriptorProto* file,
  231. FileDescriptorProto* output);
  232. GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(SimpleDescriptorDatabase);
  233. };
  234. // Very similar to SimpleDescriptorDatabase, but stores all the descriptors
  235. // as raw bytes and generally tries to use as little memory as possible.
  236. //
  237. // The same caveats regarding FindFileContainingExtension() apply as with
  238. // SimpleDescriptorDatabase.
  239. class LIBPROTOBUF_EXPORT EncodedDescriptorDatabase : public DescriptorDatabase {
  240. public:
  241. EncodedDescriptorDatabase();
  242. ~EncodedDescriptorDatabase();
  243. // Adds the FileDescriptorProto to the database. The descriptor is provided
  244. // in encoded form. The database does not make a copy of the bytes, nor
  245. // does it take ownership; it's up to the caller to make sure the bytes
  246. // remain valid for the life of the database. Returns false and logs an error
  247. // if the bytes are not a valid FileDescriptorProto or if the file conflicted
  248. // with a file already in the database.
  249. bool Add(const void* encoded_file_descriptor, int size);
  250. // Like Add(), but makes a copy of the data, so that the caller does not
  251. // need to keep it around.
  252. bool AddCopy(const void* encoded_file_descriptor, int size);
  253. // Like FindFileContainingSymbol but returns only the name of the file.
  254. bool FindNameOfFileContainingSymbol(const string& symbol_name,
  255. string* output);
  256. // implements DescriptorDatabase -----------------------------------
  257. bool FindFileByName(const string& filename,
  258. FileDescriptorProto* output);
  259. bool FindFileContainingSymbol(const string& symbol_name,
  260. FileDescriptorProto* output);
  261. bool FindFileContainingExtension(const string& containing_type,
  262. int field_number,
  263. FileDescriptorProto* output);
  264. bool FindAllExtensionNumbers(const string& extendee_type,
  265. vector<int>* output);
  266. private:
  267. SimpleDescriptorDatabase::DescriptorIndex<pair<const void*, int> > index_;
  268. vector<void*> files_to_delete_;
  269. // If encoded_file.first is non-NULL, parse the data into *output and return
  270. // true, otherwise return false.
  271. bool MaybeParse(pair<const void*, int> encoded_file,
  272. FileDescriptorProto* output);
  273. GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(EncodedDescriptorDatabase);
  274. };
  275. // A DescriptorDatabase that fetches files from a given pool.
  276. class LIBPROTOBUF_EXPORT DescriptorPoolDatabase : public DescriptorDatabase {
  277. public:
  278. DescriptorPoolDatabase(const DescriptorPool& pool);
  279. ~DescriptorPoolDatabase();
  280. // implements DescriptorDatabase -----------------------------------
  281. bool FindFileByName(const string& filename,
  282. FileDescriptorProto* output);
  283. bool FindFileContainingSymbol(const string& symbol_name,
  284. FileDescriptorProto* output);
  285. bool FindFileContainingExtension(const string& containing_type,
  286. int field_number,
  287. FileDescriptorProto* output);
  288. bool FindAllExtensionNumbers(const string& extendee_type,
  289. vector<int>* output);
  290. private:
  291. const DescriptorPool& pool_;
  292. GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(DescriptorPoolDatabase);
  293. };
  294. // A DescriptorDatabase that wraps two or more others. It first searches the
  295. // first database and, if that fails, tries the second, and so on.
  296. class LIBPROTOBUF_EXPORT MergedDescriptorDatabase : public DescriptorDatabase {
  297. public:
  298. // Merge just two databases. The sources remain property of the caller.
  299. MergedDescriptorDatabase(DescriptorDatabase* source1,
  300. DescriptorDatabase* source2);
  301. // Merge more than two databases. The sources remain property of the caller.
  302. // The vector may be deleted after the constructor returns but the
  303. // DescriptorDatabases need to stick around.
  304. MergedDescriptorDatabase(const vector<DescriptorDatabase*>& sources);
  305. ~MergedDescriptorDatabase();
  306. // implements DescriptorDatabase -----------------------------------
  307. bool FindFileByName(const string& filename,
  308. FileDescriptorProto* output);
  309. bool FindFileContainingSymbol(const string& symbol_name,
  310. FileDescriptorProto* output);
  311. bool FindFileContainingExtension(const string& containing_type,
  312. int field_number,
  313. FileDescriptorProto* output);
  314. // Merges the results of calling all databases. Returns true iff any
  315. // of the databases returned true.
  316. bool FindAllExtensionNumbers(const string& extendee_type,
  317. vector<int>* output);
  318. private:
  319. vector<DescriptorDatabase*> sources_;
  320. GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(MergedDescriptorDatabase);
  321. };
  322. } // namespace protobuf
  323. } // namespace google
  324. #endif // GOOGLE_PROTOBUF_DESCRIPTOR_DATABASE_H__