http.c 17.9 KB
Newer Older
1
/*
2
 * HTTP protocol for avconv client
3
 * Copyright (c) 2000, 2001 Fabrice Bellard
4
 *
5
 * This file is part of Libav.
6
 *
7
 * Libav is free software; you can redistribute it and/or
8 9 10 11
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
12
 * Libav is distributed in the hope that it will be useful,
13 14 15 16 17
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with Libav; if not, write to the Free Software
19 20
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */
21 22

#include "libavutil/avstring.h"
23 24
#include "avformat.h"
#include <unistd.h>
25
#include "internal.h"
26
#include "network.h"
27
#include "http.h"
28
#include "os_support.h"
29
#include "httpauth.h"
30
#include "url.h"
31
#include "libavutil/opt.h"
32

33
/* XXX: POST protocol is not completely implemented because avconv uses
Diego Biurrun's avatar
Diego Biurrun committed
34
   only a subset of it. */
35 36 37 38 39 40

/* used for protocol handling */
#define BUFFER_SIZE 1024
#define MAX_REDIRECTS 8

typedef struct {
41
    const AVClass *class;
42 43 44 45
    URLContext *hd;
    unsigned char buffer[BUFFER_SIZE], *buf_ptr, *buf_end;
    int line_count;
    int http_code;
46
    int64_t chunksize;      /**< Used if "Transfer-Encoding: chunked" otherwise -1. */
47
    int64_t off, filesize;
48
    char location[MAX_URL_SIZE];
49
    HTTPAuthState auth_state;
50
    HTTPAuthState proxy_auth_state;
51
    char *headers;
52
    int willclose;          /**< Set if the server correctly handles Connection: close and will close the connection after feeding us the content. */
53
    int chunked_post;
54 55
} HTTPContext;

56
#define OFFSET(x) offsetof(HTTPContext, x)
57 58
#define D AV_OPT_FLAG_DECODING_PARAM
#define E AV_OPT_FLAG_ENCODING_PARAM
59
static const AVOption options[] = {
60
{"chunked_post", "use chunked transfer-encoding for posts", OFFSET(chunked_post), AV_OPT_TYPE_INT, {.dbl = 1}, 0, 1, E },
61
{"headers", "custom HTTP headers, can override built in default headers", OFFSET(headers), AV_OPT_TYPE_STRING, { 0 }, 0, 0, D|E },
62 63
{NULL}
};
64 65 66 67 68 69
#define HTTP_CLASS(flavor)\
static const AVClass flavor ## _context_class = {\
    .class_name     = #flavor,\
    .item_name      = av_default_item_name,\
    .option         = options,\
    .version        = LIBAVUTIL_VERSION_INT,\
70 71
};

72 73 74
HTTP_CLASS(http);
HTTP_CLASS(https);

75 76 77
static int http_connect(URLContext *h, const char *path, const char *local_path,
                        const char *hoststr, const char *auth,
                        const char *proxyauth, int *new_location);
78

79 80 81 82
void ff_http_init_auth_state(URLContext *dest, const URLContext *src)
{
    memcpy(&((HTTPContext*)dest->priv_data)->auth_state,
           &((HTTPContext*)src->priv_data)->auth_state, sizeof(HTTPAuthState));
83 84 85
    memcpy(&((HTTPContext*)dest->priv_data)->proxy_auth_state,
           &((HTTPContext*)src->priv_data)->proxy_auth_state,
           sizeof(HTTPAuthState));
86 87
}

88 89 90
/* return non zero if error */
static int http_open_cnx(URLContext *h)
{
91
    const char *path, *proxy_path, *lower_proto = "tcp", *local_path;
92
    char hostname[1024], hoststr[1024], proto[10];
93
    char auth[1024], proxyauth[1024];
94
    char path1[1024];
95
    char buf[1024], urlbuf[1024];
96
    int port, use_proxy, err, location_changed = 0, redirects = 0;
97
    HTTPAuthType cur_auth_type, cur_proxy_auth_type;
98 99 100 101 102
    HTTPContext *s = h->priv_data;
    URLContext *hd = NULL;

    proxy_path = getenv("http_proxy");
    use_proxy = (proxy_path != NULL) && !getenv("no_proxy") &&
103
        av_strstart(proxy_path, "http://", NULL);
104 105 106 107

    /* fill the dest addr */
 redo:
    /* needed in any case to build the host string */
108 109
    av_url_split(proto, sizeof(proto), auth, sizeof(auth),
                 hostname, sizeof(hostname), &port,
Martin Storsjö's avatar
Martin Storsjö committed
110
                 path1, sizeof(path1), s->location);
111
    ff_url_join(hoststr, sizeof(hoststr), NULL, NULL, hostname, port, NULL);
112

113 114 115 116 117
    if (path1[0] == '\0')
        path = "/";
    else
        path = path1;
    local_path = path;
118
    if (use_proxy) {
119 120 121 122 123 124 125
        /* Reassemble the request URL without auth string - we don't
         * want to leak the auth to the proxy. */
        ff_url_join(urlbuf, sizeof(urlbuf), proto, NULL, hostname, port, "%s",
                    path1);
        path = urlbuf;
        av_url_split(NULL, 0, proxyauth, sizeof(proxyauth),
                     hostname, sizeof(hostname), &port, NULL, 0, proxy_path);
126
    }
127 128 129 130 131
    if (!strcmp(proto, "https")) {
        lower_proto = "tls";
        if (port < 0)
            port = 443;
    }
132 133 134
    if (port < 0)
        port = 80;

135
    ff_url_join(buf, sizeof(buf), lower_proto, NULL, hostname, port, NULL);
136 137
    err = ffurl_open(&hd, buf, AVIO_FLAG_READ_WRITE,
                     &h->interrupt_callback, NULL);
138 139 140 141
    if (err < 0)
        goto fail;

    s->hd = hd;
142
    cur_auth_type = s->auth_state.auth_type;
143 144
    cur_proxy_auth_type = s->auth_state.auth_type;
    if (http_connect(h, path, local_path, hoststr, auth, proxyauth, &location_changed) < 0)
145
        goto fail;
146 147
    if (s->http_code == 401) {
        if (cur_auth_type == HTTP_AUTH_NONE && s->auth_state.auth_type != HTTP_AUTH_NONE) {
148
            ffurl_close(hd);
149 150 151 152
            goto redo;
        } else
            goto fail;
    }
153 154 155 156 157 158 159 160
    if (s->http_code == 407) {
        if (cur_proxy_auth_type == HTTP_AUTH_NONE &&
            s->proxy_auth_state.auth_type != HTTP_AUTH_NONE) {
            ffurl_close(hd);
            goto redo;
        } else
            goto fail;
    }
161 162
    if ((s->http_code == 301 || s->http_code == 302 || s->http_code == 303 || s->http_code == 307)
        && location_changed == 1) {
163
        /* url moved, get next */
164
        ffurl_close(hd);
165
        if (redirects++ >= MAX_REDIRECTS)
166
            return AVERROR(EIO);
167 168 169 170 171 172
        location_changed = 0;
        goto redo;
    }
    return 0;
 fail:
    if (hd)
173
        ffurl_close(hd);
174
    s->hd = NULL;
175
    return AVERROR(EIO);
176 177 178 179
}

static int http_open(URLContext *h, const char *uri, int flags)
{
180
    HTTPContext *s = h->priv_data;
181 182 183 184

    h->is_streamed = 1;

    s->filesize = -1;
185
    av_strlcpy(s->location, uri, sizeof(s->location));
186

187 188 189
    if (s->headers) {
        int len = strlen(s->headers);
        if (len < 2 || strcmp("\r\n", s->headers + len - 2))
190
            av_log(h, AV_LOG_WARNING, "No trailing CRLF found in HTTP header.\n");
191 192
    }

193
    return http_open_cnx(h);
194 195 196 197 198
}
static int http_getc(HTTPContext *s)
{
    int len;
    if (s->buf_ptr >= s->buf_end) {
199
        len = ffurl_read(s->hd, s->buffer, BUFFER_SIZE);
200
        if (len < 0) {
201
            return AVERROR(EIO);
202 203 204 205 206 207 208 209 210 211
        } else if (len == 0) {
            return -1;
        } else {
            s->buf_ptr = s->buffer;
            s->buf_end = s->buffer + len;
        }
    }
    return *s->buf_ptr++;
}

212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
static int http_get_line(HTTPContext *s, char *line, int line_size)
{
    int ch;
    char *q;

    q = line;
    for(;;) {
        ch = http_getc(s);
        if (ch < 0)
            return AVERROR(EIO);
        if (ch == '\n') {
            /* process line */
            if (q > line && q[-1] == '\r')
                q--;
            *q = '\0';

            return 0;
        } else {
            if ((q - line) < line_size - 1)
                *q++ = ch;
        }
    }
}

236 237 238 239
static int process_line(URLContext *h, char *line, int line_count,
                        int *new_location)
{
    HTTPContext *s = h->priv_data;
240
    char *tag, *p, *end;
241 242 243 244 245 246 247 248 249 250 251

    /* end of header */
    if (line[0] == '\0')
        return 0;

    p = line;
    if (line_count == 0) {
        while (!isspace(*p) && *p != '\0')
            p++;
        while (isspace(*p))
            p++;
252
        s->http_code = strtol(p, &end, 10);
253

Luca Barbato's avatar
Luca Barbato committed
254
        av_dlog(NULL, "http_code=%d\n", s->http_code);
255

256 257
        /* error codes are 4xx and 5xx, but regard 401 as a success, so we
         * don't abort until all headers have been parsed. */
258
        if (s->http_code >= 400 && s->http_code < 600 && (s->http_code != 401
259 260
            || s->auth_state.auth_type != HTTP_AUTH_NONE) &&
            (s->http_code != 407 || s->proxy_auth_state.auth_type != HTTP_AUTH_NONE)) {
261
            end += strspn(end, SPACE_CHARS);
262
            av_log(h, AV_LOG_WARNING, "HTTP error %d %s\n",
263
                   s->http_code, end);
Alex Beregszaszi's avatar
Alex Beregszaszi committed
264
            return -1;
265
        }
266 267 268 269 270 271 272 273 274 275 276
    } else {
        while (*p != '\0' && *p != ':')
            p++;
        if (*p != ':')
            return 1;

        *p = '\0';
        tag = line;
        p++;
        while (isspace(*p))
            p++;
277
        if (!av_strcasecmp(tag, "Location")) {
278 279
            strcpy(s->location, p);
            *new_location = 1;
280
        } else if (!av_strcasecmp (tag, "Content-Length") && s->filesize == -1) {
281
            s->filesize = atoll(p);
282
        } else if (!av_strcasecmp (tag, "Content-Range")) {
283 284 285 286 287 288 289 290 291
            /* "bytes $from-$to/$document_size" */
            const char *slash;
            if (!strncmp (p, "bytes ", 6)) {
                p += 6;
                s->off = atoll(p);
                if ((slash = strchr(p, '/')) && strlen(slash) > 0)
                    s->filesize = atoll(slash+1);
            }
            h->is_streamed = 0; /* we _can_ in fact seek */
292
        } else if (!av_strcasecmp(tag, "Accept-Ranges") && !strncmp(p, "bytes", 5)) {
293
            h->is_streamed = 0;
294
        } else if (!av_strcasecmp (tag, "Transfer-Encoding") && !av_strncasecmp(p, "chunked", 7)) {
295 296
            s->filesize = -1;
            s->chunksize = 0;
297
        } else if (!av_strcasecmp (tag, "WWW-Authenticate")) {
298
            ff_http_auth_handle_header(&s->auth_state, tag, p);
299
        } else if (!av_strcasecmp (tag, "Authentication-Info")) {
300
            ff_http_auth_handle_header(&s->auth_state, tag, p);
301 302
        } else if (!av_strcasecmp (tag, "Proxy-Authenticate")) {
            ff_http_auth_handle_header(&s->proxy_auth_state, tag, p);
303
        } else if (!av_strcasecmp (tag, "Connection")) {
304 305
            if (!strcmp(p, "close"))
                s->willclose = 1;
306 307 308 309 310
        }
    }
    return 1;
}

311 312 313
static inline int has_header(const char *str, const char *header)
{
    /* header + 2 to skip over CRLF prefix. (make sure you have one!) */
314 315
    if (!str)
        return 0;
316 317 318
    return av_stristart(str, header + 2, NULL) || av_stristr(str, header);
}

319 320 321
static int http_connect(URLContext *h, const char *path, const char *local_path,
                        const char *hoststr, const char *auth,
                        const char *proxyauth, int *new_location)
322 323
{
    HTTPContext *s = h->priv_data;
324 325
    int post, err;
    char line[1024];
326
    char headers[1024] = "";
327
    char *authstr = NULL, *proxyauthstr = NULL;
328
    int64_t off = s->off;
329
    int len = 0;
330
    const char *method;
331 332 333


    /* send http header */
334
    post = h->flags & AVIO_FLAG_WRITE;
335 336 337 338 339
    method = post ? "POST" : "GET";
    authstr = ff_http_auth_create_response(&s->auth_state, auth, local_path,
                                           method);
    proxyauthstr = ff_http_auth_create_response(&s->proxy_auth_state, proxyauth,
                                                local_path, method);
340 341 342 343 344 345 346 347

    /* set default headers if needed */
    if (!has_header(s->headers, "\r\nUser-Agent: "))
       len += av_strlcatf(headers + len, sizeof(headers) - len,
                          "User-Agent: %s\r\n", LIBAVFORMAT_IDENT);
    if (!has_header(s->headers, "\r\nAccept: "))
        len += av_strlcpy(headers + len, "Accept: */*\r\n",
                          sizeof(headers) - len);
348
    if (!has_header(s->headers, "\r\nRange: ") && !post)
349
        len += av_strlcatf(headers + len, sizeof(headers) - len,
350
                           "Range: bytes=%"PRId64"-\r\n", s->off);
351 352 353 354 355 356 357 358
    if (!has_header(s->headers, "\r\nConnection: "))
        len += av_strlcpy(headers + len, "Connection: close\r\n",
                          sizeof(headers)-len);
    if (!has_header(s->headers, "\r\nHost: "))
        len += av_strlcatf(headers + len, sizeof(headers) - len,
                           "Host: %s\r\n", hoststr);

    /* now add in custom headers */
359 360
    if (s->headers)
        av_strlcpy(headers + len, s->headers, sizeof(headers) - len);
361

362 363
    snprintf(s->buffer, sizeof(s->buffer),
             "%s %s HTTP/1.1\r\n"
364
             "%s"
365
             "%s"
366
             "%s"
367
             "%s%s"
368
             "\r\n",
369
             method,
370
             path,
371
             post && s->chunked_post ? "Transfer-Encoding: chunked\r\n" : "",
372
             headers,
373 374
             authstr ? authstr : "",
             proxyauthstr ? "Proxy-" : "", proxyauthstr ? proxyauthstr : "");
375

376
    av_freep(&authstr);
377
    av_freep(&proxyauthstr);
378
    if (ffurl_write(s->hd, s->buffer, strlen(s->buffer)) < 0)
379
        return AVERROR(EIO);
380 381 382 383 384 385

    /* init input buffer */
    s->buf_ptr = s->buffer;
    s->buf_end = s->buffer;
    s->line_count = 0;
    s->off = 0;
386
    s->filesize = -1;
387
    s->willclose = 0;
388
    if (post) {
389 390 391 392
        /* Pretend that it did work. We didn't read any header yet, since
         * we've still to send the POST data, but the code calling this
         * function will check http_code after we return. */
        s->http_code = 200;
393 394
        return 0;
    }
395
    s->chunksize = -1;
396 397 398

    /* wait for header */
    for(;;) {
399
        if (http_get_line(s, line, sizeof(line)) < 0)
400
            return AVERROR(EIO);
401

Luca Barbato's avatar
Luca Barbato committed
402
        av_dlog(NULL, "header='%s'\n", line);
403

404 405 406 407 408 409
        err = process_line(h, line, s->line_count, new_location);
        if (err < 0)
            return err;
        if (err == 0)
            break;
        s->line_count++;
410 411 412 413 414 415
    }

    return (off == s->off) ? 0 : -1;
}


416
static int http_buf_read(URLContext *h, uint8_t *buf, int size)
417 418 419
{
    HTTPContext *s = h->priv_data;
    int len;
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442
    /* read bytes from input buffer first */
    len = s->buf_end - s->buf_ptr;
    if (len > 0) {
        if (len > size)
            len = size;
        memcpy(buf, s->buf_ptr, len);
        s->buf_ptr += len;
    } else {
        if (!s->willclose && s->filesize >= 0 && s->off >= s->filesize)
            return AVERROR_EOF;
        len = ffurl_read(s->hd, buf, size);
    }
    if (len > 0) {
        s->off += len;
        if (s->chunksize > 0)
            s->chunksize -= len;
    }
    return len;
}

static int http_read(URLContext *h, uint8_t *buf, int size)
{
    HTTPContext *s = h->priv_data;
443

444 445 446 447 448 449 450 451 452 453 454 455
    if (s->chunksize >= 0) {
        if (!s->chunksize) {
            char line[32];

            for(;;) {
                do {
                    if (http_get_line(s, line, sizeof(line)) < 0)
                        return AVERROR(EIO);
                } while (!*line);    /* skip CR LF from last chunk */

                s->chunksize = strtoll(line, NULL, 16);

Luca Barbato's avatar
Luca Barbato committed
456
                av_dlog(NULL, "Chunked encoding data size: %"PRId64"'\n", s->chunksize);
457 458 459 460 461 462 463 464

                if (!s->chunksize)
                    return 0;
                break;
            }
        }
        size = FFMIN(size, s->chunksize);
    }
465
    return http_buf_read(h, buf, size);
466 467 468
}

/* used only when posting data */
469
static int http_write(URLContext *h, const uint8_t *buf, int size)
470
{
471
    char temp[11] = "";  /* 32-bit hex + CRLF + nul */
472 473
    int ret;
    char crlf[] = "\r\n";
474
    HTTPContext *s = h->priv_data;
475

476
    if (!s->chunked_post) {
Martin Storsjö's avatar
Martin Storsjö committed
477
        /* non-chunked data is sent without any special encoding */
478
        return ffurl_write(s->hd, buf, size);
479 480 481 482 483 484
    }

    /* silently ignore zero-size data since chunk encoding that would
     * signal EOF */
    if (size > 0) {
        /* upload data using chunked encoding */
Martin Storsjö's avatar
Martin Storsjö committed
485
        snprintf(temp, sizeof(temp), "%x\r\n", size);
486

487 488 489
        if ((ret = ffurl_write(s->hd, temp, strlen(temp))) < 0 ||
            (ret = ffurl_write(s->hd, buf, size)) < 0 ||
            (ret = ffurl_write(s->hd, crlf, sizeof(crlf) - 1)) < 0)
490 491 492
            return ret;
    }
    return size;
493 494 495 496
}

static int http_close(URLContext *h)
{
497 498
    int ret = 0;
    char footer[] = "0\r\n\r\n";
499
    HTTPContext *s = h->priv_data;
500 501

    /* signal end of chunked encoding if used */
502
    if ((h->flags & AVIO_FLAG_WRITE) && s->chunked_post) {
503
        ret = ffurl_write(s->hd, footer, sizeof(footer) - 1);
504 505 506
        ret = ret > 0 ? 0 : ret;
    }

507
    if (s->hd)
508
        ffurl_close(s->hd);
509
    return ret;
510 511
}

512
static int64_t http_seek(URLContext *h, int64_t off, int whence)
513 514 515
{
    HTTPContext *s = h->priv_data;
    URLContext *old_hd = s->hd;
516
    int64_t old_off = s->off;
517 518
    uint8_t old_buf[BUFFER_SIZE];
    int old_buf_size;
519 520 521 522 523 524 525

    if (whence == AVSEEK_SIZE)
        return s->filesize;
    else if ((s->filesize == -1 && whence == SEEK_END) || h->is_streamed)
        return -1;

    /* we save the old context in case the seek fails */
526 527
    old_buf_size = s->buf_end - s->buf_ptr;
    memcpy(old_buf, s->buf_ptr, old_buf_size);
528 529 530 531 532 533 534 535 536
    s->hd = NULL;
    if (whence == SEEK_CUR)
        off += s->off;
    else if (whence == SEEK_END)
        off += s->filesize;
    s->off = off;

    /* if it fails, continue on old connection */
    if (http_open_cnx(h) < 0) {
537 538 539
        memcpy(s->buffer, old_buf, old_buf_size);
        s->buf_ptr = s->buffer;
        s->buf_end = s->buffer + old_buf_size;
540 541 542 543
        s->hd = old_hd;
        s->off = old_off;
        return -1;
    }
544
    ffurl_close(old_hd);
545 546 547
    return off;
}

548 549 550 551
static int
http_get_file_handle(URLContext *h)
{
    HTTPContext *s = h->priv_data;
552
    return ffurl_get_file_handle(s->hd);
553 554
}

555
#if CONFIG_HTTP_PROTOCOL
556
URLProtocol ff_http_protocol = {
557 558 559 560 561 562
    .name                = "http",
    .url_open            = http_open,
    .url_read            = http_read,
    .url_write           = http_write,
    .url_seek            = http_seek,
    .url_close           = http_close,
563
    .url_get_file_handle = http_get_file_handle,
564
    .priv_data_size      = sizeof(HTTPContext),
565
    .priv_data_class     = &http_context_class,
566
};
567 568 569 570 571 572 573 574 575 576 577
#endif
#if CONFIG_HTTPS_PROTOCOL
URLProtocol ff_https_protocol = {
    .name                = "https",
    .url_open            = http_open,
    .url_read            = http_read,
    .url_write           = http_write,
    .url_seek            = http_seek,
    .url_close           = http_close,
    .url_get_file_handle = http_get_file_handle,
    .priv_data_size      = sizeof(HTTPContext),
578
    .priv_data_class     = &https_context_class,
579 580
};
#endif