Line | Branch | Exec | Source |
---|---|---|---|
1 | // | ||
2 | // Copyright (c) 2019 Vinnie Falco (vinnie.falco@gmail.com) | ||
3 | // | ||
4 | // Distributed under the Boost Software License, Version 1.0. (See accompanying | ||
5 | // file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) | ||
6 | // | ||
7 | // Official repository: https://github.com/cppalliance/http_proto | ||
8 | // | ||
9 | |||
10 | #ifndef BOOST_HTTP_PROTO_PARSER_HPP | ||
11 | #define BOOST_HTTP_PROTO_PARSER_HPP | ||
12 | |||
13 | #include <boost/http_proto/detail/config.hpp> | ||
14 | #include <boost/http_proto/error.hpp> | ||
15 | #include <boost/http_proto/header_limits.hpp> | ||
16 | #include <boost/http_proto/sink.hpp> | ||
17 | #include <boost/http_proto/detail/header.hpp> | ||
18 | #include <boost/http_proto/detail/type_traits.hpp> | ||
19 | #include <boost/http_proto/detail/workspace.hpp> | ||
20 | #include <boost/buffers/circular_buffer.hpp> | ||
21 | #include <boost/buffers/flat_buffer.hpp> | ||
22 | #include <boost/buffers/mutable_buffer_pair.hpp> | ||
23 | #include <boost/buffers/mutable_buffer_span.hpp> | ||
24 | #include <boost/buffers/type_traits.hpp> | ||
25 | #include <boost/buffers/any_dynamic_buffer.hpp> | ||
26 | #include <boost/url/grammar/error.hpp> | ||
27 | #include <cstddef> | ||
28 | #include <cstdint> | ||
29 | #include <functional> | ||
30 | #include <memory> | ||
31 | #include <utility> | ||
32 | |||
33 | namespace boost { | ||
34 | namespace http_proto { | ||
35 | |||
36 | #ifndef BOOST_HTTP_PROTO_DOCS | ||
37 | class parser_service; | ||
38 | class request_parser; | ||
39 | class response_parser; | ||
40 | class context; | ||
41 | namespace detail { | ||
42 | class filter; | ||
43 | } // detail | ||
44 | #endif | ||
45 | |||
46 | /** A parser for HTTP/1 messages. | ||
47 | |||
48 | The parser is strict. Any malformed | ||
49 | inputs according to the documented | ||
50 | HTTP ABNFs is treated as an | ||
51 | unrecoverable error. | ||
52 | */ | ||
53 | class BOOST_SYMBOL_VISIBLE | ||
54 | parser | ||
55 | { | ||
56 | BOOST_HTTP_PROTO_DECL | ||
57 | parser(context& ctx, detail::kind); | ||
58 | |||
59 | public: | ||
60 | /** Parser configuration settings | ||
61 | |||
62 | @see | ||
63 | @li <a href="https://stackoverflow.com/questions/686217/maximum-on-http-header-values" | ||
64 | >Maximum on HTTP header values (Stackoverflow)</a> | ||
65 | */ | ||
66 | struct config_base | ||
67 | { | ||
68 | header_limits headers; | ||
69 | |||
70 | /** Largest allowed size for a content body. | ||
71 | |||
72 | The size of the body is measured | ||
73 | after removing any transfer encodings, | ||
74 | including a chunked encoding. | ||
75 | */ | ||
76 | std::uint64_t body_limit = 64 * 1024; | ||
77 | |||
78 | /** True if parser can decode deflate transfer and content encodings. | ||
79 | |||
80 | The deflate decoder must already be | ||
81 | installed thusly, or else an exception | ||
82 | is thrown. | ||
83 | |||
84 | @par Install Deflate Decoder | ||
85 | @code | ||
86 | deflate_decoder_service::config cfg; | ||
87 | cfg.install( ctx ); | ||
88 | @endcode | ||
89 | */ | ||
90 | bool apply_deflate_decoder = false; | ||
91 | |||
92 | /** Minimum space for payload buffering. | ||
93 | |||
94 | This value controls the following | ||
95 | settings: | ||
96 | |||
97 | @li The smallest allocated size of | ||
98 | the buffers used for reading | ||
99 | and decoding the payload. | ||
100 | |||
101 | @li The lowest guaranteed size of | ||
102 | an in-place body. | ||
103 | |||
104 | @li The largest size used to reserve | ||
105 | space in dynamic buffer bodies | ||
106 | when the payload size is not | ||
107 | known ahead of time. | ||
108 | |||
109 | This cannot be zero, and this cannot | ||
110 | be greater than @ref body_limit. | ||
111 | */ | ||
112 | std::size_t min_buffer = 4096; | ||
113 | |||
114 | /** Largest permissible output size in prepare. | ||
115 | |||
116 | This cannot be zero. | ||
117 | */ | ||
118 | std::size_t max_prepare = std::size_t(-1); | ||
119 | |||
120 | /** Space to reserve for type-erasure. | ||
121 | */ | ||
122 | std::size_t max_type_erase = 1024; | ||
123 | }; | ||
124 | |||
125 | using mutable_buffers_type = | ||
126 | buffers::mutable_buffer_span; | ||
127 | |||
128 | using const_buffers_type = | ||
129 | buffers::const_buffer_span; | ||
130 | |||
131 | struct stream; | ||
132 | |||
133 | //-------------------------------------------- | ||
134 | // | ||
135 | // Special Members | ||
136 | // | ||
137 | //-------------------------------------------- | ||
138 | |||
139 | /** Destructor. | ||
140 | */ | ||
141 | BOOST_HTTP_PROTO_DECL | ||
142 | ~parser(); | ||
143 | |||
144 | /** Constructor (deleted) | ||
145 | */ | ||
146 | parser(parser&&) = delete; | ||
147 | |||
148 | /** Assignment (deleted) | ||
149 | */ | ||
150 | parser& operator=(parser&&) = delete; | ||
151 | |||
152 | //-------------------------------------------- | ||
153 | // | ||
154 | // Observers | ||
155 | // | ||
156 | //-------------------------------------------- | ||
157 | |||
158 | #if 0 | ||
159 | /** Return true if any input was committed. | ||
160 | */ | ||
161 | bool | ||
162 | got_some() const noexcept | ||
163 | { | ||
164 | return st_ != state::need_start; | ||
165 | } | ||
166 | #endif | ||
167 | |||
168 | /** Return true if the complete header was parsed. | ||
169 | */ | ||
170 | bool | ||
171 | 53704 | got_header() const noexcept | |
172 | { | ||
173 | 53704 | return st_ > state::header; | |
174 | } | ||
175 | |||
176 | /** Returns `true` if a complete message has been parsed. | ||
177 | |||
178 | Calling @ref reset prepares the parser | ||
179 | to process the next message in the stream. | ||
180 | |||
181 | */ | ||
182 | bool | ||
183 | 52344 | is_complete() const noexcept | |
184 | { | ||
185 | 52344 | return st_ == state::complete; | |
186 | } | ||
187 | |||
188 | /** Returns `true` if the end of the stream was reached. | ||
189 | |||
190 | The end of the stream is encountered | ||
191 | when one of the following conditions | ||
192 | occurs: | ||
193 | |||
194 | @li @ref commit_eof was called and there | ||
195 | is no more data left to parse, or | ||
196 | |||
197 | @li An unrecoverable error occurred | ||
198 | during parsing. | ||
199 | |||
200 | When the end of stream is reached, the | ||
201 | function @ref reset must be called | ||
202 | to start parsing a new stream. | ||
203 | */ | ||
204 | bool | ||
205 | 714 | is_end_of_stream() const noexcept | |
206 | { | ||
207 | return | ||
208 |
2/2✓ Branch 0 taken 582 times.
✓ Branch 1 taken 132 times.
|
1296 | st_ == state::reset || |
209 |
1/2✓ Branch 0 taken 582 times.
✗ Branch 1 not taken.
|
582 | ( st_ == state::complete && |
210 |
2/2✓ Branch 0 taken 342 times.
✓ Branch 1 taken 240 times.
|
1296 | got_eof_); |
211 | } | ||
212 | |||
213 | //-------------------------------------------- | ||
214 | // | ||
215 | // Modifiers | ||
216 | // | ||
217 | //-------------------------------------------- | ||
218 | |||
219 | /** Prepare for a new stream. | ||
220 | */ | ||
221 | BOOST_HTTP_PROTO_DECL | ||
222 | void | ||
223 | reset() noexcept; | ||
224 | |||
225 | /** Prepare for the next message on the stream. | ||
226 | */ | ||
227 | void | ||
228 | 9872 | start() | |
229 | { | ||
230 | 9872 | start_impl(false); | |
231 | 9867 | } | |
232 | |||
233 | private: | ||
234 | // New message on the current stream | ||
235 | BOOST_HTTP_PROTO_DECL void | ||
236 | start_impl(bool head_response); | ||
237 | public: | ||
238 | |||
239 | /** Return the input buffer | ||
240 | */ | ||
241 | BOOST_HTTP_PROTO_DECL | ||
242 | mutable_buffers_type | ||
243 | prepare(); | ||
244 | |||
245 | /** Commit bytes to the input buffer | ||
246 | */ | ||
247 | BOOST_HTTP_PROTO_DECL | ||
248 | void | ||
249 | commit( | ||
250 | std::size_t n); | ||
251 | |||
252 | /** Indicate there will be no more input | ||
253 | |||
254 | @par Postconditions | ||
255 | All buffer sequences previously obtained | ||
256 | by calling @ref prepare are invalidated. | ||
257 | */ | ||
258 | BOOST_HTTP_PROTO_DECL | ||
259 | void | ||
260 | commit_eof(); | ||
261 | |||
262 | /** Parse pending input data | ||
263 | */ | ||
264 | // VFALCO return result<void>? | ||
265 | BOOST_HTTP_PROTO_DECL | ||
266 | void | ||
267 | parse( | ||
268 | system::error_code& ec); | ||
269 | |||
270 | /** Attach a body. | ||
271 | |||
272 | This function attaches the specified elastic | ||
273 | buffer as the storage for the message body. | ||
274 | The parser acquires ownership of the object | ||
275 | `eb` and destroys it when: | ||
276 | |||
277 | @li @ref is_complete returns `true`, or | ||
278 | @li @ref reset is called, or | ||
279 | @li an unrecoverable parsing error occurs, or | ||
280 | @li the parser is destroyed. | ||
281 | */ | ||
282 | // VFALCO Should this function have | ||
283 | // error_code& ec and call parse? | ||
284 | template<class ElasticBuffer> | ||
285 | #ifndef BOOST_HTTP_PROTO_DOCS | ||
286 | typename std::enable_if< | ||
287 | ! detail::is_reference_wrapper< | ||
288 | ElasticBuffer>::value && | ||
289 | ! is_sink<ElasticBuffer>::value>::type | ||
290 | #else | ||
291 | void | ||
292 | #endif | ||
293 | set_body(ElasticBuffer&& eb); | ||
294 | |||
295 | /** Attach a body. | ||
296 | |||
297 | This function attaches the specified elastic | ||
298 | buffer reference as the storage for the message body. | ||
299 | Ownership is not transferred; the caller must | ||
300 | ensure that the lifetime of the object | ||
301 | reference by `eb` extends until: | ||
302 | |||
303 | @li @ref is_complete returns `true`, or | ||
304 | @li @ref reset is called, or | ||
305 | @li an unrecoverable parsing error occurs, or | ||
306 | @li the parser is destroyed. | ||
307 | */ | ||
308 | template<class ElasticBuffer> | ||
309 | void set_body( | ||
310 | std::reference_wrapper<ElasticBuffer> eb); | ||
311 | |||
312 | /** Attach a body | ||
313 | */ | ||
314 | template<class Sink> | ||
315 | #ifndef BOOST_HTTP_PROTO_DOCS | ||
316 | typename std::enable_if< | ||
317 | is_sink<Sink>::value, | ||
318 | typename std::decay<Sink>::type | ||
319 | >::type& | ||
320 | #else | ||
321 | typename std::decay<Sink>::type& | ||
322 | #endif | ||
323 | set_body(Sink&& sink); | ||
324 | |||
325 | /** Return the available body data. | ||
326 | |||
327 | The returned buffer span will be invalidated if any member | ||
328 | function of the parser is subsequently called. | ||
329 | */ | ||
330 | BOOST_HTTP_PROTO_DECL | ||
331 | const_buffers_type | ||
332 | pull_body(); | ||
333 | |||
334 | /** Consumes bytes from the available body data. | ||
335 | */ | ||
336 | BOOST_HTTP_PROTO_DECL | ||
337 | void | ||
338 | consume_body(std::size_t n); | ||
339 | |||
340 | /** Return the complete body as a contiguous character buffer. | ||
341 | */ | ||
342 | BOOST_HTTP_PROTO_DECL | ||
343 | core::string_view | ||
344 | body() const noexcept; | ||
345 | |||
346 | //-------------------------------------------- | ||
347 | |||
348 | /** Return any leftover data | ||
349 | |||
350 | This is used to forward unconsumed data | ||
351 | that could lie past the last message. | ||
352 | For example on a CONNECT request there | ||
353 | could be additional protocol-dependent | ||
354 | data that we want to retrieve. | ||
355 | */ | ||
356 | // VFALCO rename to get_leftovers()? | ||
357 | BOOST_HTTP_PROTO_DECL | ||
358 | core::string_view | ||
359 | release_buffered_data() noexcept; | ||
360 | |||
361 | private: | ||
362 | friend class request_parser; | ||
363 | friend class response_parser; | ||
364 | |||
365 | detail::header const* | ||
366 | safe_get_header() const; | ||
367 | bool is_plain() const noexcept; | ||
368 | void on_headers(system::error_code&); | ||
369 | BOOST_HTTP_PROTO_DECL void on_set_body(); | ||
370 | void init_dynamic(system::error_code&); | ||
371 | |||
372 | static constexpr unsigned buffers_N = 8; | ||
373 | |||
374 | enum class state | ||
375 | { | ||
376 | // order matters | ||
377 | reset, | ||
378 | start, | ||
379 | header, | ||
380 | body, | ||
381 | set_body, | ||
382 | complete | ||
383 | }; | ||
384 | |||
385 | enum class how | ||
386 | { | ||
387 | in_place, | ||
388 | elastic, | ||
389 | sink | ||
390 | }; | ||
391 | |||
392 | context& ctx_; | ||
393 | parser_service& svc_; | ||
394 | |||
395 | detail::workspace ws_; | ||
396 | detail::header h_; | ||
397 | std::uint64_t body_avail_ = 0; | ||
398 | std::uint64_t body_total_ = 0; | ||
399 | std::uint64_t payload_remain_ = 0; | ||
400 | std::uint64_t chunk_remain_ = 0; | ||
401 | std::size_t nprepare_ = 0; | ||
402 | |||
403 | // used to store initial headers + any potential overread | ||
404 | buffers::flat_buffer fb_; | ||
405 | |||
406 | // used for raw input once headers are read | ||
407 | buffers::circular_buffer cb0_; | ||
408 | |||
409 | // used for transformed output, if applicable | ||
410 | // can be empty/null | ||
411 | buffers::circular_buffer cb1_; | ||
412 | |||
413 | // used to provide stable storage when returning | ||
414 | // `mutable_buffers_type` from relevant functions | ||
415 | buffers::mutable_buffer_pair mbp_; | ||
416 | |||
417 | // used to provide stable storage when returning | ||
418 | // `const_buffers_type` from relevant functions | ||
419 | buffers::const_buffer_pair cbp_; | ||
420 | |||
421 | buffers::circular_buffer* body_buf_ = nullptr; | ||
422 | buffers::any_dynamic_buffer* eb_ = nullptr; | ||
423 | detail::filter* filt_ = nullptr; | ||
424 | sink* sink_ = nullptr; | ||
425 | |||
426 | state st_ = state::start; | ||
427 | how how_ = how::in_place; | ||
428 | bool got_eof_ = false; | ||
429 | // bool need_more_; | ||
430 | bool head_response_ = false; | ||
431 | bool needs_chunk_close_ = false; | ||
432 | bool trailer_headers_ = false; | ||
433 | }; | ||
434 | |||
435 | //------------------------------------------------ | ||
436 | |||
437 | /** Install the parser service. | ||
438 | */ | ||
439 | BOOST_HTTP_PROTO_DECL | ||
440 | void | ||
441 | install_parser_service( | ||
442 | context& ctx, | ||
443 | parser::config_base const& cfg); | ||
444 | |||
445 | } // http_proto | ||
446 | } // boost | ||
447 | |||
448 | #include <boost/http_proto/impl/parser.hpp> | ||
449 | |||
450 | #endif | ||
451 |