1 | /*************************************************************************** |
2 | * _ _ ____ _ |
3 | * Project ___| | | | _ \| | |
4 | * / __| | | | |_) | | |
5 | * | (__| |_| | _ <| |___ |
6 | * \___|\___/|_| \_\_____| |
7 | * |
8 | * Copyright (C) 1998 - 2021, Daniel Stenberg, <daniel@haxx.se>, et al. |
9 | * |
10 | * This software is licensed as described in the file COPYING, which |
11 | * you should have received as part of this distribution. The terms |
12 | * are also available at https://curl.se/docs/copyright.html. |
13 | * |
14 | * You may opt to use, copy, modify, merge, publish, distribute and/or sell |
15 | * copies of the Software, and permit persons to whom the Software is |
16 | * furnished to do so, under the terms of the COPYING file. |
17 | * |
18 | * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY |
19 | * KIND, either express or implied. |
20 | * |
21 | ***************************************************************************/ |
22 | |
23 | #include "curl_setup.h" |
24 | |
25 | #ifndef CURL_DISABLE_HTTP |
26 | |
27 | #include "urldata.h" /* it includes http_chunks.h */ |
28 | #include "sendf.h" /* for the client write stuff */ |
29 | #include "dynbuf.h" |
30 | #include "content_encoding.h" |
31 | #include "http.h" |
32 | #include "non-ascii.h" /* for Curl_convert_to_network prototype */ |
33 | #include "strtoofft.h" |
34 | #include "warnless.h" |
35 | |
36 | /* The last #include files should be: */ |
37 | #include "curl_memory.h" |
38 | #include "memdebug.h" |
39 | |
40 | /* |
41 | * Chunk format (simplified): |
42 | * |
43 | * <HEX SIZE>[ chunk extension ] CRLF |
44 | * <DATA> CRLF |
45 | * |
46 | * Highlights from RFC2616 section 3.6 say: |
47 | |
48 | The chunked encoding modifies the body of a message in order to |
49 | transfer it as a series of chunks, each with its own size indicator, |
50 | followed by an OPTIONAL trailer containing entity-header fields. This |
51 | allows dynamically produced content to be transferred along with the |
52 | information necessary for the recipient to verify that it has |
53 | received the full message. |
54 | |
55 | Chunked-Body = *chunk |
56 | last-chunk |
57 | trailer |
58 | CRLF |
59 | |
60 | chunk = chunk-size [ chunk-extension ] CRLF |
61 | chunk-data CRLF |
62 | chunk-size = 1*HEX |
63 | last-chunk = 1*("0") [ chunk-extension ] CRLF |
64 | |
65 | chunk-extension= *( ";" chunk-ext-name [ "=" chunk-ext-val ] ) |
66 | chunk-ext-name = token |
67 | chunk-ext-val = token | quoted-string |
68 | chunk-data = chunk-size(OCTET) |
69 | trailer = *(entity-header CRLF) |
70 | |
71 | The chunk-size field is a string of hex digits indicating the size of |
72 | the chunk. The chunked encoding is ended by any chunk whose size is |
73 | zero, followed by the trailer, which is terminated by an empty line. |
74 | |
75 | */ |
76 | |
77 | #ifdef CURL_DOES_CONVERSIONS |
78 | /* Check for an ASCII hex digit. |
79 | We avoid the use of ISXDIGIT to accommodate non-ASCII hosts. */ |
80 | static bool isxdigit_ascii(char digit) |
81 | { |
82 | return (digit >= 0x30 && digit <= 0x39) /* 0-9 */ |
83 | || (digit >= 0x41 && digit <= 0x46) /* A-F */ |
84 | || (digit >= 0x61 && digit <= 0x66); /* a-f */ |
85 | } |
86 | #else |
87 | #define isxdigit_ascii(x) Curl_isxdigit(x) |
88 | #endif |
89 | |
90 | void Curl_httpchunk_init(struct Curl_easy *data) |
91 | { |
92 | struct connectdata *conn = data->conn; |
93 | struct Curl_chunker *chunk = &conn->chunk; |
94 | chunk->hexindex = 0; /* start at 0 */ |
95 | chunk->state = CHUNK_HEX; /* we get hex first! */ |
96 | Curl_dyn_init(&conn->trailer, DYN_H1_TRAILER); |
97 | } |
98 | |
99 | /* |
100 | * chunk_read() returns a OK for normal operations, or a positive return code |
101 | * for errors. STOP means this sequence of chunks is complete. The 'wrote' |
102 | * argument is set to tell the caller how many bytes we actually passed to the |
103 | * client (for byte-counting and whatever). |
104 | * |
105 | * The states and the state-machine is further explained in the header file. |
106 | * |
107 | * This function always uses ASCII hex values to accommodate non-ASCII hosts. |
108 | * For example, 0x0d and 0x0a are used instead of '\r' and '\n'. |
109 | */ |
110 | CHUNKcode Curl_httpchunk_read(struct Curl_easy *data, |
111 | char *datap, |
112 | ssize_t datalen, |
113 | ssize_t *wrotep, |
114 | CURLcode *) |
115 | { |
116 | CURLcode result = CURLE_OK; |
117 | struct connectdata *conn = data->conn; |
118 | struct Curl_chunker *ch = &conn->chunk; |
119 | struct SingleRequest *k = &data->req; |
120 | size_t piece; |
121 | curl_off_t length = (curl_off_t)datalen; |
122 | size_t *wrote = (size_t *)wrotep; |
123 | |
124 | *wrote = 0; /* nothing's written yet */ |
125 | |
126 | /* the original data is written to the client, but we go on with the |
127 | chunk read process, to properly calculate the content length*/ |
128 | if(data->set.http_te_skip && !k->ignorebody) { |
129 | result = Curl_client_write(data, CLIENTWRITE_BODY, datap, datalen); |
130 | if(result) { |
131 | *extrap = result; |
132 | return CHUNKE_PASSTHRU_ERROR; |
133 | } |
134 | } |
135 | |
136 | while(length) { |
137 | switch(ch->state) { |
138 | case CHUNK_HEX: |
139 | if(isxdigit_ascii(*datap)) { |
140 | if(ch->hexindex < CHUNK_MAXNUM_LEN) { |
141 | ch->hexbuffer[ch->hexindex] = *datap; |
142 | datap++; |
143 | length--; |
144 | ch->hexindex++; |
145 | } |
146 | else { |
147 | return CHUNKE_TOO_LONG_HEX; /* longer hex than we support */ |
148 | } |
149 | } |
150 | else { |
151 | char *endptr; |
152 | if(0 == ch->hexindex) |
153 | /* This is illegal data, we received junk where we expected |
154 | a hexadecimal digit. */ |
155 | return CHUNKE_ILLEGAL_HEX; |
156 | |
157 | /* length and datap are unmodified */ |
158 | ch->hexbuffer[ch->hexindex] = 0; |
159 | |
160 | /* convert to host encoding before calling strtoul */ |
161 | result = Curl_convert_from_network(data, ch->hexbuffer, ch->hexindex); |
162 | if(result) { |
163 | /* Curl_convert_from_network calls failf if unsuccessful */ |
164 | /* Treat it as a bad hex character */ |
165 | return CHUNKE_ILLEGAL_HEX; |
166 | } |
167 | |
168 | if(curlx_strtoofft(ch->hexbuffer, &endptr, 16, &ch->datasize)) |
169 | return CHUNKE_ILLEGAL_HEX; |
170 | ch->state = CHUNK_LF; /* now wait for the CRLF */ |
171 | } |
172 | break; |
173 | |
174 | case CHUNK_LF: |
175 | /* waiting for the LF after a chunk size */ |
176 | if(*datap == 0x0a) { |
177 | /* we're now expecting data to come, unless size was zero! */ |
178 | if(0 == ch->datasize) { |
179 | ch->state = CHUNK_TRAILER; /* now check for trailers */ |
180 | } |
181 | else |
182 | ch->state = CHUNK_DATA; |
183 | } |
184 | |
185 | datap++; |
186 | length--; |
187 | break; |
188 | |
189 | case CHUNK_DATA: |
190 | /* We expect 'datasize' of data. We have 'length' right now, it can be |
191 | more or less than 'datasize'. Get the smallest piece. |
192 | */ |
193 | piece = curlx_sotouz((ch->datasize >= length)?length:ch->datasize); |
194 | |
195 | /* Write the data portion available */ |
196 | if(!data->set.http_te_skip && !k->ignorebody) { |
197 | if(!data->set.http_ce_skip && k->writer_stack) |
198 | result = Curl_unencode_write(data, k->writer_stack, datap, piece); |
199 | else |
200 | result = Curl_client_write(data, CLIENTWRITE_BODY, datap, piece); |
201 | |
202 | if(result) { |
203 | *extrap = result; |
204 | return CHUNKE_PASSTHRU_ERROR; |
205 | } |
206 | } |
207 | |
208 | *wrote += piece; |
209 | ch->datasize -= piece; /* decrease amount left to expect */ |
210 | datap += piece; /* move read pointer forward */ |
211 | length -= piece; /* decrease space left in this round */ |
212 | |
213 | if(0 == ch->datasize) |
214 | /* end of data this round, we now expect a trailing CRLF */ |
215 | ch->state = CHUNK_POSTLF; |
216 | break; |
217 | |
218 | case CHUNK_POSTLF: |
219 | if(*datap == 0x0a) { |
220 | /* The last one before we go back to hex state and start all over. */ |
221 | Curl_httpchunk_init(data); /* sets state back to CHUNK_HEX */ |
222 | } |
223 | else if(*datap != 0x0d) |
224 | return CHUNKE_BAD_CHUNK; |
225 | datap++; |
226 | length--; |
227 | break; |
228 | |
229 | case CHUNK_TRAILER: |
230 | if((*datap == 0x0d) || (*datap == 0x0a)) { |
231 | char *tr = Curl_dyn_ptr(&conn->trailer); |
232 | /* this is the end of a trailer, but if the trailer was zero bytes |
233 | there was no trailer and we move on */ |
234 | |
235 | if(tr) { |
236 | size_t trlen; |
237 | result = Curl_dyn_add(&conn->trailer, (char *)"\x0d\x0a" ); |
238 | if(result) |
239 | return CHUNKE_OUT_OF_MEMORY; |
240 | |
241 | tr = Curl_dyn_ptr(&conn->trailer); |
242 | trlen = Curl_dyn_len(&conn->trailer); |
243 | /* Convert to host encoding before calling Curl_client_write */ |
244 | result = Curl_convert_from_network(data, tr, trlen); |
245 | if(result) |
246 | /* Curl_convert_from_network calls failf if unsuccessful */ |
247 | /* Treat it as a bad chunk */ |
248 | return CHUNKE_BAD_CHUNK; |
249 | |
250 | if(!data->set.http_te_skip) { |
251 | result = Curl_client_write(data, CLIENTWRITE_HEADER, tr, trlen); |
252 | if(result) { |
253 | *extrap = result; |
254 | return CHUNKE_PASSTHRU_ERROR; |
255 | } |
256 | } |
257 | Curl_dyn_reset(&conn->trailer); |
258 | ch->state = CHUNK_TRAILER_CR; |
259 | if(*datap == 0x0a) |
260 | /* already on the LF */ |
261 | break; |
262 | } |
263 | else { |
264 | /* no trailer, we're on the final CRLF pair */ |
265 | ch->state = CHUNK_TRAILER_POSTCR; |
266 | break; /* don't advance the pointer */ |
267 | } |
268 | } |
269 | else { |
270 | result = Curl_dyn_addn(&conn->trailer, datap, 1); |
271 | if(result) |
272 | return CHUNKE_OUT_OF_MEMORY; |
273 | } |
274 | datap++; |
275 | length--; |
276 | break; |
277 | |
278 | case CHUNK_TRAILER_CR: |
279 | if(*datap == 0x0a) { |
280 | ch->state = CHUNK_TRAILER_POSTCR; |
281 | datap++; |
282 | length--; |
283 | } |
284 | else |
285 | return CHUNKE_BAD_CHUNK; |
286 | break; |
287 | |
288 | case CHUNK_TRAILER_POSTCR: |
289 | /* We enter this state when a CR should arrive so we expect to |
290 | have to first pass a CR before we wait for LF */ |
291 | if((*datap != 0x0d) && (*datap != 0x0a)) { |
292 | /* not a CR then it must be another header in the trailer */ |
293 | ch->state = CHUNK_TRAILER; |
294 | break; |
295 | } |
296 | if(*datap == 0x0d) { |
297 | /* skip if CR */ |
298 | datap++; |
299 | length--; |
300 | } |
301 | /* now wait for the final LF */ |
302 | ch->state = CHUNK_STOP; |
303 | break; |
304 | |
305 | case CHUNK_STOP: |
306 | if(*datap == 0x0a) { |
307 | length--; |
308 | |
309 | /* Record the length of any data left in the end of the buffer |
310 | even if there's no more chunks to read */ |
311 | ch->datasize = curlx_sotouz(length); |
312 | |
313 | return CHUNKE_STOP; /* return stop */ |
314 | } |
315 | else |
316 | return CHUNKE_BAD_CHUNK; |
317 | } |
318 | } |
319 | return CHUNKE_OK; |
320 | } |
321 | |
322 | const char *Curl_chunked_strerror(CHUNKcode code) |
323 | { |
324 | switch(code) { |
325 | default: |
326 | return "OK" ; |
327 | case CHUNKE_TOO_LONG_HEX: |
328 | return "Too long hexadecimal number" ; |
329 | case CHUNKE_ILLEGAL_HEX: |
330 | return "Illegal or missing hexadecimal sequence" ; |
331 | case CHUNKE_BAD_CHUNK: |
332 | return "Malformed encoding found" ; |
333 | case CHUNKE_PASSTHRU_ERROR: |
334 | DEBUGASSERT(0); /* never used */ |
335 | return "" ; |
336 | case CHUNKE_BAD_ENCODING: |
337 | return "Bad content-encoding found" ; |
338 | case CHUNKE_OUT_OF_MEMORY: |
339 | return "Out of memory" ; |
340 | } |
341 | } |
342 | |
343 | #endif /* CURL_DISABLE_HTTP */ |
344 | |