1 /***************************************************************************
2 * _ _ ____ _
3 * Project ___| | | | _ \| |
4 * / __| | | | |_) | |
5 * | (__| |_| | _ <| |___
6 * \___|\___/|_| \_\_____|
7 *
8 * Copyright (C) 1998 - 2020, 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.haxx.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. */
Curl_isxdigit_ascii(char digit)80 static bool Curl_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 Curl_isxdigit_ascii(x) Curl_isxdigit(x)
88 #endif
89
Curl_httpchunk_init(struct connectdata * conn)90 void Curl_httpchunk_init(struct connectdata *conn)
91 {
92 struct Curl_chunker *chunk = &conn->chunk;
93 chunk->hexindex = 0; /* start at 0 */
94 chunk->dataleft = 0; /* no data left yet! */
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 */
Curl_httpchunk_read(struct connectdata * conn,char * datap,ssize_t datalen,ssize_t * wrotep,CURLcode * extrap)110 CHUNKcode Curl_httpchunk_read(struct connectdata *conn,
111 char *datap,
112 ssize_t datalen,
113 ssize_t *wrotep,
114 CURLcode *extrap)
115 {
116 CURLcode result = CURLE_OK;
117 struct Curl_easy *data = conn->data;
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(conn, 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(Curl_isxdigit_ascii(*datap)) {
140 if(ch->hexindex < MAXNUM_SIZE) {
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(conn->data, ch->hexbuffer,
162 ch->hexindex);
163 if(result) {
164 /* Curl_convert_from_network calls failf if unsuccessful */
165 /* Treat it as a bad hex character */
166 return CHUNKE_ILLEGAL_HEX;
167 }
168
169 if(curlx_strtoofft(ch->hexbuffer, &endptr, 16, &ch->datasize))
170 return CHUNKE_ILLEGAL_HEX;
171 ch->state = CHUNK_LF; /* now wait for the CRLF */
172 }
173 break;
174
175 case CHUNK_LF:
176 /* waiting for the LF after a chunk size */
177 if(*datap == 0x0a) {
178 /* we're now expecting data to come, unless size was zero! */
179 if(0 == ch->datasize) {
180 ch->state = CHUNK_TRAILER; /* now check for trailers */
181 }
182 else
183 ch->state = CHUNK_DATA;
184 }
185
186 datap++;
187 length--;
188 break;
189
190 case CHUNK_DATA:
191 /* We expect 'datasize' of data. We have 'length' right now, it can be
192 more or less than 'datasize'. Get the smallest piece.
193 */
194 piece = curlx_sotouz((ch->datasize >= length)?length:ch->datasize);
195
196 /* Write the data portion available */
197 if(!conn->data->set.http_te_skip && !k->ignorebody) {
198 if(!conn->data->set.http_ce_skip && k->writer_stack)
199 result = Curl_unencode_write(conn, k->writer_stack, datap, piece);
200 else
201 result = Curl_client_write(conn, CLIENTWRITE_BODY, datap, piece);
202
203 if(result) {
204 *extrap = result;
205 return CHUNKE_PASSTHRU_ERROR;
206 }
207 }
208
209 *wrote += piece;
210 ch->datasize -= piece; /* decrease amount left to expect */
211 datap += piece; /* move read pointer forward */
212 length -= piece; /* decrease space left in this round */
213
214 if(0 == ch->datasize)
215 /* end of data this round, we now expect a trailing CRLF */
216 ch->state = CHUNK_POSTLF;
217 break;
218
219 case CHUNK_POSTLF:
220 if(*datap == 0x0a) {
221 /* The last one before we go back to hex state and start all over. */
222 Curl_httpchunk_init(conn); /* sets state back to CHUNK_HEX */
223 }
224 else if(*datap != 0x0d)
225 return CHUNKE_BAD_CHUNK;
226 datap++;
227 length--;
228 break;
229
230 case CHUNK_TRAILER:
231 if((*datap == 0x0d) || (*datap == 0x0a)) {
232 char *tr = Curl_dyn_ptr(&conn->trailer);
233 /* this is the end of a trailer, but if the trailer was zero bytes
234 there was no trailer and we move on */
235
236 if(tr) {
237 size_t trlen;
238 result = Curl_dyn_add(&conn->trailer, (char *)"\x0d\x0a");
239 if(result)
240 return CHUNKE_OUT_OF_MEMORY;
241
242 tr = Curl_dyn_ptr(&conn->trailer);
243 trlen = Curl_dyn_len(&conn->trailer);
244 /* Convert to host encoding before calling Curl_client_write */
245 result = Curl_convert_from_network(conn->data, tr, trlen);
246 if(result)
247 /* Curl_convert_from_network calls failf if unsuccessful */
248 /* Treat it as a bad chunk */
249 return CHUNKE_BAD_CHUNK;
250
251 if(!data->set.http_te_skip) {
252 result = Curl_client_write(conn, CLIENTWRITE_HEADER, tr, trlen);
253 if(result) {
254 *extrap = result;
255 return CHUNKE_PASSTHRU_ERROR;
256 }
257 }
258 Curl_dyn_reset(&conn->trailer);
259 ch->state = CHUNK_TRAILER_CR;
260 if(*datap == 0x0a)
261 /* already on the LF */
262 break;
263 }
264 else {
265 /* no trailer, we're on the final CRLF pair */
266 ch->state = CHUNK_TRAILER_POSTCR;
267 break; /* don't advance the pointer */
268 }
269 }
270 else {
271 result = Curl_dyn_addn(&conn->trailer, datap, 1);
272 if(result)
273 return CHUNKE_OUT_OF_MEMORY;
274 }
275 datap++;
276 length--;
277 break;
278
279 case CHUNK_TRAILER_CR:
280 if(*datap == 0x0a) {
281 ch->state = CHUNK_TRAILER_POSTCR;
282 datap++;
283 length--;
284 }
285 else
286 return CHUNKE_BAD_CHUNK;
287 break;
288
289 case CHUNK_TRAILER_POSTCR:
290 /* We enter this state when a CR should arrive so we expect to
291 have to first pass a CR before we wait for LF */
292 if((*datap != 0x0d) && (*datap != 0x0a)) {
293 /* not a CR then it must be another header in the trailer */
294 ch->state = CHUNK_TRAILER;
295 break;
296 }
297 if(*datap == 0x0d) {
298 /* skip if CR */
299 datap++;
300 length--;
301 }
302 /* now wait for the final LF */
303 ch->state = CHUNK_STOP;
304 break;
305
306 case CHUNK_STOP:
307 if(*datap == 0x0a) {
308 length--;
309
310 /* Record the length of any data left in the end of the buffer
311 even if there's no more chunks to read */
312 ch->dataleft = curlx_sotouz(length);
313
314 return CHUNKE_STOP; /* return stop */
315 }
316 else
317 return CHUNKE_BAD_CHUNK;
318 }
319 }
320 return CHUNKE_OK;
321 }
322
Curl_chunked_strerror(CHUNKcode code)323 const char *Curl_chunked_strerror(CHUNKcode code)
324 {
325 switch(code) {
326 default:
327 return "OK";
328 case CHUNKE_TOO_LONG_HEX:
329 return "Too long hexadecimal number";
330 case CHUNKE_ILLEGAL_HEX:
331 return "Illegal or missing hexadecimal sequence";
332 case CHUNKE_BAD_CHUNK:
333 return "Malformed encoding found";
334 case CHUNKE_PASSTHRU_ERROR:
335 DEBUGASSERT(0); /* never used */
336 return "";
337 case CHUNKE_BAD_ENCODING:
338 return "Bad content-encoding found";
339 case CHUNKE_OUT_OF_MEMORY:
340 return "Out of memory";
341 }
342 }
343
344 #endif /* CURL_DISABLE_HTTP */
345