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/workspace.hpp> | ||
19 | #include <boost/buffers/circular_buffer.hpp> | ||
20 | #include <boost/buffers/flat_buffer.hpp> | ||
21 | #include <boost/buffers/mutable_buffer_pair.hpp> | ||
22 | #include <boost/buffers/mutable_buffer_span.hpp> | ||
23 | #include <boost/buffers/type_traits.hpp> | ||
24 | #include <boost/buffers/any_dynamic_buffer.hpp> | ||
25 | #include <boost/url/grammar/error.hpp> | ||
26 | #include <cstddef> | ||
27 | #include <cstdint> | ||
28 | #include <memory> | ||
29 | #include <utility> | ||
30 | |||
31 | namespace boost { | ||
32 | namespace http_proto { | ||
33 | |||
34 | #ifndef BOOST_HTTP_PROTO_DOCS | ||
35 | class parser_service; | ||
36 | class filter; | ||
37 | class request_parser; | ||
38 | class response_parser; | ||
39 | class context; | ||
40 | |||
41 | #endif | ||
42 | |||
43 | /** A parser for HTTP/1 messages. | ||
44 | |||
45 | The parser is strict. Any malformed | ||
46 | inputs according to the documented | ||
47 | HTTP ABNFs is treated as an | ||
48 | unrecoverable error. | ||
49 | */ | ||
50 | class BOOST_SYMBOL_VISIBLE | ||
51 | parser | ||
52 | { | ||
53 | BOOST_HTTP_PROTO_DECL | ||
54 | parser(context& ctx, detail::kind); | ||
55 | |||
56 | public: | ||
57 | /** Parser configuration settings | ||
58 | |||
59 | @see | ||
60 | @li <a href="https://stackoverflow.com/questions/686217/maximum-on-http-header-values" | ||
61 | >Maximum on HTTP header values (Stackoverflow)</a> | ||
62 | */ | ||
63 | struct config_base | ||
64 | { | ||
65 | header_limits headers; | ||
66 | |||
67 | /** Largest allowed size for a content body. | ||
68 | |||
69 | The size of the body is measured | ||
70 | after removing any transfer encodings, | ||
71 | including a chunked encoding. | ||
72 | */ | ||
73 | std::uint64_t body_limit = 64 * 1024; | ||
74 | |||
75 | /** True if parser can decode deflate transfer and content encodings. | ||
76 | |||
77 | The deflate decoder must already be | ||
78 | installed thusly, or else an exception | ||
79 | is thrown. | ||
80 | |||
81 | @par Install Deflate Decoder | ||
82 | @code | ||
83 | deflate_decoder_service::config cfg; | ||
84 | cfg.install( ctx ); | ||
85 | @endcode | ||
86 | */ | ||
87 | bool apply_deflate_decoder = false; | ||
88 | |||
89 | /** Minimum space for payload buffering. | ||
90 | |||
91 | This value controls the following | ||
92 | settings: | ||
93 | |||
94 | @li The smallest allocated size of | ||
95 | the buffers used for reading | ||
96 | and decoding the payload. | ||
97 | |||
98 | @li The lowest guaranteed size of | ||
99 | an in-place body. | ||
100 | |||
101 | @li The largest size used to reserve | ||
102 | space in dynamic buffer bodies | ||
103 | when the payload size is not | ||
104 | known ahead of time. | ||
105 | |||
106 | This cannot be zero, and this cannot | ||
107 | be greater than @ref body_limit. | ||
108 | */ | ||
109 | std::size_t min_buffer = 4096; | ||
110 | |||
111 | /** Largest permissible output size in prepare. | ||
112 | |||
113 | This cannot be zero. | ||
114 | */ | ||
115 | std::size_t max_prepare = std::size_t(-1); | ||
116 | |||
117 | /** Space to reserve for type-erasure. | ||
118 | */ | ||
119 | std::size_t max_type_erase = 1024; | ||
120 | }; | ||
121 | |||
122 | using mutable_buffers_type = | ||
123 | buffers::mutable_buffer_span; | ||
124 | |||
125 | using const_buffers_type = | ||
126 | buffers::const_buffer_span; | ||
127 | |||
128 | struct stream; | ||
129 | |||
130 | //-------------------------------------------- | ||
131 | // | ||
132 | // Special Members | ||
133 | // | ||
134 | //-------------------------------------------- | ||
135 | |||
136 | /** Destructor. | ||
137 | */ | ||
138 | BOOST_HTTP_PROTO_DECL | ||
139 | ~parser(); | ||
140 | |||
141 | /** Constructor (deleted) | ||
142 | */ | ||
143 | parser(parser&&) = delete; | ||
144 | |||
145 | /** Assignment (deleted) | ||
146 | */ | ||
147 | parser& operator=(parser&&) = delete; | ||
148 | |||
149 | //-------------------------------------------- | ||
150 | // | ||
151 | // Observers | ||
152 | // | ||
153 | //-------------------------------------------- | ||
154 | |||
155 | #if 0 | ||
156 | /** Return true if any input was committed. | ||
157 | */ | ||
158 | bool | ||
159 | got_some() const noexcept | ||
160 | { | ||
161 | return st_ != state::need_start; | ||
162 | } | ||
163 | #endif | ||
164 | |||
165 | /** Return true if the complete header was parsed. | ||
166 | */ | ||
167 | bool | ||
168 | 2694 | got_header() const noexcept | |
169 | { | ||
170 | 2694 | return st_ > state::header; | |
171 | } | ||
172 | |||
173 | /** Returns `true` if a complete message has been parsed. | ||
174 | |||
175 | Calling @ref reset prepares the parser | ||
176 | to process the next message in the stream. | ||
177 | |||
178 | */ | ||
179 | bool | ||
180 | 2249 | is_complete() const noexcept | |
181 | { | ||
182 | 2249 | return st_ == state::complete; | |
183 | } | ||
184 | |||
185 | /** Returns `true` if the end of the stream was reached. | ||
186 | |||
187 | The end of the stream is encountered | ||
188 | when one of the following conditions | ||
189 | occurs: | ||
190 | |||
191 | @li @ref commit_eof was called and there | ||
192 | is no more data left to parse, or | ||
193 | |||
194 | @li An unrecoverable error occurred | ||
195 | during parsing. | ||
196 | |||
197 | When the end of stream is reached, the | ||
198 | function @ref reset must be called | ||
199 | to start parsing a new stream. | ||
200 | */ | ||
201 | bool | ||
202 | 714 | is_end_of_stream() const noexcept | |
203 | { | ||
204 | return | ||
205 |
2/2✓ Branch 0 taken 582 times.
✓ Branch 1 taken 132 times.
|
1296 | st_ == state::reset || |
206 |
1/2✓ Branch 0 taken 582 times.
✗ Branch 1 not taken.
|
582 | ( st_ == state::complete && |
207 |
2/2✓ Branch 0 taken 342 times.
✓ Branch 1 taken 240 times.
|
1296 | got_eof_); |
208 | } | ||
209 | |||
210 | //-------------------------------------------- | ||
211 | // | ||
212 | // Modifiers | ||
213 | // | ||
214 | //-------------------------------------------- | ||
215 | |||
216 | /** Prepare for a new stream. | ||
217 | */ | ||
218 | BOOST_HTTP_PROTO_DECL | ||
219 | void | ||
220 | reset() noexcept; | ||
221 | |||
222 | /** Prepare for the next message on the stream. | ||
223 | */ | ||
224 | void | ||
225 | 1488 | start() | |
226 | { | ||
227 | 1488 | start_impl(false); | |
228 | 1483 | } | |
229 | |||
230 | private: | ||
231 | // New message on the current stream | ||
232 | BOOST_HTTP_PROTO_DECL void | ||
233 | start_impl(bool head_response); | ||
234 | public: | ||
235 | |||
236 | /** Return the input buffer | ||
237 | */ | ||
238 | BOOST_HTTP_PROTO_DECL | ||
239 | mutable_buffers_type | ||
240 | prepare(); | ||
241 | |||
242 | /** Commit bytes to the input buffer | ||
243 | */ | ||
244 | BOOST_HTTP_PROTO_DECL | ||
245 | void | ||
246 | commit( | ||
247 | std::size_t n); | ||
248 | |||
249 | /** Indicate there will be no more input | ||
250 | |||
251 | @par Postconditions | ||
252 | All buffer sequences previously obtained | ||
253 | by calling @ref prepare are invalidated. | ||
254 | */ | ||
255 | BOOST_HTTP_PROTO_DECL | ||
256 | void | ||
257 | commit_eof(); | ||
258 | |||
259 | /** Parse pending input data | ||
260 | */ | ||
261 | BOOST_HTTP_PROTO_DECL | ||
262 | void | ||
263 | parse( | ||
264 | system::error_code& ec); | ||
265 | |||
266 | /** Attach a body | ||
267 | */ | ||
268 | // VFALCO Should this function have | ||
269 | // error_code& ec and call parse? | ||
270 | #ifndef BOOST_HTTP_PROTO_DOCS | ||
271 | template< | ||
272 | class DynamicBuffer | ||
273 | , class = typename std::enable_if< | ||
274 | buffers::is_dynamic_buffer< | ||
275 | DynamicBuffer>::value | ||
276 | >::type | ||
277 | > | ||
278 | #else | ||
279 | template<class DynamicBuffer> | ||
280 | #endif | ||
281 | typename std::decay< | ||
282 | DynamicBuffer>::type& | ||
283 | set_body(DynamicBuffer&& b); | ||
284 | |||
285 | /** Attach a body | ||
286 | */ | ||
287 | template<class Sink> | ||
288 | #ifndef BOOST_HTTP_PROTO_DOCS | ||
289 | typename std::enable_if< | ||
290 | is_sink<Sink>::value, | ||
291 | typename std::decay<Sink>::type | ||
292 | >::type& | ||
293 | #else | ||
294 | typename std::decay<Sink>::type& | ||
295 | #endif | ||
296 | set_body(Sink&& sink); | ||
297 | |||
298 | /** Return the available body data and consume it. | ||
299 | |||
300 | The buffer referenced by the string view | ||
301 | will be invalidated if any member function | ||
302 | of the parser is called. | ||
303 | */ | ||
304 | BOOST_HTTP_PROTO_DECL | ||
305 | const_buffers_type | ||
306 | pull_some(); | ||
307 | |||
308 | /** Return the complete body as a contiguous character buffer. | ||
309 | */ | ||
310 | BOOST_HTTP_PROTO_DECL | ||
311 | core::string_view | ||
312 | body() const noexcept; | ||
313 | |||
314 | //-------------------------------------------- | ||
315 | |||
316 | /** Return any leftover data | ||
317 | |||
318 | This is used to forward unconsumed data | ||
319 | that could lie past the last message. | ||
320 | For example on a CONNECT request there | ||
321 | could be additional protocol-dependent | ||
322 | data that we want to retrieve. | ||
323 | */ | ||
324 | BOOST_HTTP_PROTO_DECL | ||
325 | core::string_view | ||
326 | release_buffered_data() noexcept; | ||
327 | |||
328 | private: | ||
329 | friend class request_parser; | ||
330 | friend class response_parser; | ||
331 | |||
332 | detail::header const* | ||
333 | safe_get_header() const; | ||
334 | bool is_plain() const noexcept; | ||
335 | void on_headers(system::error_code&); | ||
336 | BOOST_HTTP_PROTO_DECL void on_set_body(); | ||
337 | void init_dynamic(system::error_code&); | ||
338 | |||
339 | static constexpr unsigned buffers_N = 8; | ||
340 | |||
341 | enum class state | ||
342 | { | ||
343 | // order matters | ||
344 | reset, | ||
345 | start, | ||
346 | header, | ||
347 | body, | ||
348 | set_body, | ||
349 | complete | ||
350 | }; | ||
351 | |||
352 | enum class how | ||
353 | { | ||
354 | in_place, | ||
355 | dynamic, | ||
356 | sink, | ||
357 | pull | ||
358 | }; | ||
359 | |||
360 | context& ctx_; | ||
361 | parser_service& svc_; | ||
362 | detail::workspace ws_; | ||
363 | detail::header h_; | ||
364 | std::uint64_t body_avail_; | ||
365 | std::uint64_t body_total_; | ||
366 | std::uint64_t payload_remain_; | ||
367 | std::size_t nprepare_; | ||
368 | |||
369 | buffers::flat_buffer fb_; | ||
370 | buffers::circular_buffer cb0_; | ||
371 | buffers::circular_buffer cb1_; | ||
372 | buffers::circular_buffer* body_buf_; | ||
373 | buffers::mutable_buffer_pair mbp_; | ||
374 | buffers::any_dynamic_buffer* dyn_; | ||
375 | filter* filt_; | ||
376 | sink* sink_; | ||
377 | |||
378 | state st_; | ||
379 | how how_; | ||
380 | bool got_eof_; | ||
381 | // bool need_more_; | ||
382 | bool head_response_; | ||
383 | }; | ||
384 | |||
385 | //------------------------------------------------ | ||
386 | |||
387 | /** Install the parser service. | ||
388 | */ | ||
389 | BOOST_HTTP_PROTO_DECL | ||
390 | void | ||
391 | install_parser_service( | ||
392 | context& ctx, | ||
393 | parser::config_base const& cfg); | ||
394 | |||
395 | } // http_proto | ||
396 | } // boost | ||
397 | |||
398 | #include <boost/http_proto/impl/parser.hpp> | ||
399 | |||
400 | #endif | ||
401 |