cmake/Utilities/cmcurl/lib/content_encoding.c

1053 lines
29 KiB
C
Raw Normal View History

/***************************************************************************
* _ _ ____ _
* Project ___| | | | _ \| |
* / __| | | | |_) | |
* | (__| |_| | _ <| |___
* \___|\___/|_| \_\_____|
*
2023-07-02 19:51:09 +02:00
* Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
*
* This software is licensed as described in the file COPYING, which
* you should have received as part of this distribution. The terms
2021-09-14 00:13:48 +02:00
* are also available at https://curl.se/docs/copyright.html.
*
* You may opt to use, copy, modify, merge, publish, distribute and/or sell
* copies of the Software, and permit persons to whom the Software is
* furnished to do so, under the terms of the COPYING file.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
2022-11-16 20:14:03 +01:00
* SPDX-License-Identifier: curl
*
***************************************************************************/
2015-04-27 22:25:09 +02:00
#include "curl_setup.h"
#include "urldata.h"
#include <curl/curl.h>
2018-04-23 21:13:27 +02:00
#include <stddef.h>
2022-11-16 20:14:03 +01:00
#ifdef HAVE_LIBZ
#include <cm3p/zlib.h>
2018-04-23 21:13:27 +02:00
#endif
#ifdef HAVE_BROTLI
2023-07-02 19:51:09 +02:00
#if defined(__GNUC__)
/* Ignore -Wvla warnings in brotli headers */
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wvla"
#endif
2018-04-23 21:13:27 +02:00
#include <brotli/decode.h>
2023-07-02 19:51:09 +02:00
#if defined(__GNUC__)
#pragma GCC diagnostic pop
#endif
2018-04-23 21:13:27 +02:00
#endif
2021-09-14 00:13:48 +02:00
#ifdef HAVE_ZSTD
#include <zstd.h>
#endif
#include "sendf.h"
2018-04-23 21:13:27 +02:00
#include "http.h"
#include "content_encoding.h"
2017-07-20 19:35:53 +02:00
#include "strdup.h"
2018-04-23 21:13:27 +02:00
#include "strcase.h"
2023-07-02 19:51:09 +02:00
/* The last 3 #include files should be in this order */
#include "curl_printf.h"
2015-04-27 22:25:09 +02:00
#include "curl_memory.h"
#include "memdebug.h"
2018-04-23 21:13:27 +02:00
#define CONTENT_ENCODING_DEFAULT "identity"
#ifndef CURL_DISABLE_HTTP
2024-04-14 22:45:38 +02:00
/* allow no more than 5 "chained" compression steps */
#define MAX_ENCODE_STACK 5
2018-04-23 21:13:27 +02:00
#define DSIZ CURL_MAX_WRITE_SIZE /* buffer size for decompressed data */
#ifdef HAVE_LIBZ
/* Comment this out if zlib is always going to be at least ver. 1.2.0.4
(doing so will reduce code size slightly). */
#define OLD_ZLIB_SUPPORT 1
#define GZIP_MAGIC_0 0x1f
#define GZIP_MAGIC_1 0x8b
/* gzip flag byte */
#define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */
#define HEAD_CRC 0x02 /* bit 1 set: header CRC present */
#define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */
#define ORIG_NAME 0x08 /* bit 3 set: original file name present */
#define COMMENT 0x10 /* bit 4 set: file comment present */
#define RESERVED 0xE0 /* bits 5..7: reserved */
2018-04-23 21:13:27 +02:00
typedef enum {
2018-10-28 12:09:07 +01:00
ZLIB_UNINIT, /* uninitialized */
ZLIB_INIT, /* initialized */
ZLIB_INFLATING, /* inflating started. */
ZLIB_EXTERNAL_TRAILER, /* reading external trailer */
ZLIB_GZIP_HEADER, /* reading gzip header */
ZLIB_GZIP_INFLATING, /* inflating gzip stream */
ZLIB_INIT_GZIP /* initialized in transparent gzip mode */
2018-04-23 21:13:27 +02:00
} zlibInitState;
2022-11-16 20:14:03 +01:00
/* Deflate and gzip writer. */
struct zlib_writer {
2024-04-14 22:45:38 +02:00
struct Curl_cwriter super;
2018-04-23 21:13:27 +02:00
zlibInitState zlib_init; /* zlib init state */
uInt trailerlen; /* Remaining trailer byte count. */
z_stream z; /* State structure for zlib. */
2020-08-30 11:54:41 +02:00
};
2018-04-23 21:13:27 +02:00
2015-04-27 22:25:09 +02:00
static voidpf
zalloc_cb(voidpf opaque, unsigned int items, unsigned int size)
{
(void) opaque;
/* not a typo, keep it calloc() */
return (voidpf) calloc(items, size);
}
static void
zfree_cb(voidpf opaque, voidpf ptr)
{
(void) opaque;
free(ptr);
}
static CURLcode
2021-09-14 00:13:48 +02:00
process_zlib_error(struct Curl_easy *data, z_stream *z)
{
2015-04-27 22:25:09 +02:00
if(z->msg)
2017-07-20 19:35:53 +02:00
failf(data, "Error while processing content unencoding: %s",
z->msg);
else
2017-07-20 19:35:53 +02:00
failf(data, "Error while processing content unencoding: "
"Unknown failure within decompression software.");
return CURLE_BAD_CONTENT_ENCODING;
}
static CURLcode
2021-09-14 00:13:48 +02:00
exit_zlib(struct Curl_easy *data,
2018-04-23 21:13:27 +02:00
z_stream *z, zlibInitState *zlib_init, CURLcode result)
{
2018-04-23 21:13:27 +02:00
if(*zlib_init == ZLIB_GZIP_HEADER)
Curl_safefree(z->next_in);
if(*zlib_init != ZLIB_UNINIT) {
if(inflateEnd(z) != Z_OK && result == CURLE_OK)
2021-09-14 00:13:48 +02:00
result = process_zlib_error(data, z);
2018-04-23 21:13:27 +02:00
*zlib_init = ZLIB_UNINIT;
}
return result;
}
2021-09-14 00:13:48 +02:00
static CURLcode process_trailer(struct Curl_easy *data,
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp)
{
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z;
CURLcode result = CURLE_OK;
uInt len = z->avail_in < zp->trailerlen? z->avail_in: zp->trailerlen;
/* Consume expected trailer bytes. Terminate stream if exhausted.
Issue an error if unexpected bytes follow. */
zp->trailerlen -= len;
z->avail_in -= len;
z->next_in += len;
if(z->avail_in)
result = CURLE_WRITE_ERROR;
if(result || !zp->trailerlen)
2021-09-14 00:13:48 +02:00
result = exit_zlib(data, z, &zp->zlib_init, result);
2018-04-23 21:13:27 +02:00
else {
2018-10-28 12:09:07 +01:00
/* Only occurs for gzip with zlib < 1.2.0.4 or raw deflate. */
zp->zlib_init = ZLIB_EXTERNAL_TRAILER;
2018-04-23 21:13:27 +02:00
}
return result;
}
2021-09-14 00:13:48 +02:00
static CURLcode inflate_stream(struct Curl_easy *data,
2024-04-14 22:45:38 +02:00
struct Curl_cwriter *writer, int type,
2020-08-30 11:54:41 +02:00
zlibInitState started)
2018-04-23 21:13:27 +02:00
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
uInt nread = z->avail_in;
Bytef *orig_in = z->next_in;
2018-04-23 21:13:27 +02:00
bool done = FALSE;
CURLcode result = CURLE_OK; /* Curl_client_write status */
char *decomp; /* Put the decompressed data here. */
2018-04-23 21:13:27 +02:00
/* Check state. */
if(zp->zlib_init != ZLIB_INIT &&
zp->zlib_init != ZLIB_INFLATING &&
zp->zlib_init != ZLIB_INIT_GZIP &&
zp->zlib_init != ZLIB_GZIP_INFLATING)
2021-09-14 00:13:48 +02:00
return exit_zlib(data, z, &zp->zlib_init, CURLE_WRITE_ERROR);
2018-04-23 21:13:27 +02:00
/* Dynamically allocate a buffer for decompression because it's uncommonly
large to hold on the stack */
2015-04-27 22:25:09 +02:00
decomp = malloc(DSIZ);
2021-09-14 00:13:48 +02:00
if(!decomp)
return exit_zlib(data, z, &zp->zlib_init, CURLE_OUT_OF_MEMORY);
/* because the buffer size is fixed, iteratively decompress and transfer to
2024-04-14 22:45:38 +02:00
the client via next_write function. */
2018-04-23 21:13:27 +02:00
while(!done) {
2018-10-28 12:09:07 +01:00
int status; /* zlib status */
2018-04-23 21:13:27 +02:00
done = TRUE;
/* (re)set buffer for decompressed output for every iteration */
2018-04-23 21:13:27 +02:00
z->next_out = (Bytef *) decomp;
z->avail_out = DSIZ;
2018-10-28 12:09:07 +01:00
#ifdef Z_BLOCK
/* Z_BLOCK is only available in zlib ver. >= 1.2.0.5 */
2018-04-23 21:13:27 +02:00
status = inflate(z, Z_BLOCK);
2018-10-28 12:09:07 +01:00
#else
/* fallback for zlib ver. < 1.2.0.5 */
status = inflate(z, Z_SYNC_FLUSH);
#endif
2018-04-23 21:13:27 +02:00
/* Flush output data if some. */
if(z->avail_out != DSIZ) {
if(status == Z_OK || status == Z_STREAM_END) {
zp->zlib_init = started; /* Data started. */
2024-04-14 22:45:38 +02:00
result = Curl_cwriter_write(data, writer->next, type, decomp,
2018-04-23 21:13:27 +02:00
DSIZ - z->avail_out);
2015-04-27 22:25:09 +02:00
if(result) {
2021-09-14 00:13:48 +02:00
exit_zlib(data, z, &zp->zlib_init, result);
2018-04-23 21:13:27 +02:00
break;
}
}
}
2018-04-23 21:13:27 +02:00
/* Dispatch by inflate() status. */
switch(status) {
case Z_OK:
/* Always loop: there may be unflushed latched data in zlib state. */
done = FALSE;
break;
case Z_BUF_ERROR:
/* No more data to flush: just exit loop. */
break;
case Z_STREAM_END:
2021-09-14 00:13:48 +02:00
result = process_trailer(data, zp);
2018-04-23 21:13:27 +02:00
break;
case Z_DATA_ERROR:
/* some servers seem to not generate zlib headers, so this is an attempt
to fix and continue anyway */
2018-04-23 21:13:27 +02:00
if(zp->zlib_init == ZLIB_INIT) {
/* Do not use inflateReset2(): only available since zlib 1.2.3.4. */
(void) inflateEnd(z); /* don't care about the return code */
if(inflateInit2(z, -MAX_WBITS) == Z_OK) {
z->next_in = orig_in;
z->avail_in = nread;
zp->zlib_init = ZLIB_INFLATING;
2018-10-28 12:09:07 +01:00
zp->trailerlen = 4; /* Tolerate up to 4 unknown trailer bytes. */
2018-04-23 21:13:27 +02:00
done = FALSE;
break;
}
zp->zlib_init = ZLIB_UNINIT; /* inflateEnd() already called. */
}
2022-03-29 21:10:50 +02:00
result = exit_zlib(data, z, &zp->zlib_init, process_zlib_error(data, z));
break;
2018-04-23 21:13:27 +02:00
default:
2021-09-14 00:13:48 +02:00
result = exit_zlib(data, z, &zp->zlib_init, process_zlib_error(data, z));
2018-04-23 21:13:27 +02:00
break;
}
}
2018-04-23 21:13:27 +02:00
free(decomp);
/* We're about to leave this call so the `nread' data bytes won't be seen
again. If we are in a state that would wrongly allow restart in raw mode
at the next call, assume output has already started. */
if(nread && zp->zlib_init == ZLIB_INIT)
zp->zlib_init = started; /* Cannot restart anymore. */
return result;
}
2018-04-23 21:13:27 +02:00
/* Deflate handler. */
2024-04-14 22:45:38 +02:00
static CURLcode deflate_do_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
2018-04-23 21:13:27 +02:00
/* Initialize zlib */
z->zalloc = (alloc_func) zalloc_cb;
z->zfree = (free_func) zfree_cb;
if(inflateInit(z) != Z_OK)
2021-09-14 00:13:48 +02:00
return process_zlib_error(data, z);
2018-04-23 21:13:27 +02:00
zp->zlib_init = ZLIB_INIT;
return CURLE_OK;
}
2024-04-14 22:45:38 +02:00
static CURLcode deflate_do_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
2018-04-23 21:13:27 +02:00
const char *buf, size_t nbytes)
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
2024-04-14 22:45:38 +02:00
if(!(type & CLIENTWRITE_BODY))
return Curl_cwriter_write(data, writer->next, type, buf, nbytes);
/* Set the compressed input when this function is called */
2018-04-23 21:13:27 +02:00
z->next_in = (Bytef *) buf;
z->avail_in = (uInt) nbytes;
2018-10-28 12:09:07 +01:00
if(zp->zlib_init == ZLIB_EXTERNAL_TRAILER)
2021-09-14 00:13:48 +02:00
return process_trailer(data, zp);
2018-10-28 12:09:07 +01:00
/* Now uncompress the data */
2024-04-14 22:45:38 +02:00
return inflate_stream(data, writer, type, ZLIB_INFLATING);
2018-04-23 21:13:27 +02:00
}
2024-04-14 22:45:38 +02:00
static void deflate_do_close(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
2021-09-14 00:13:48 +02:00
exit_zlib(data, z, &zp->zlib_init, CURLE_OK);
2018-04-23 21:13:27 +02:00
}
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype deflate_encoding = {
2018-04-23 21:13:27 +02:00
"deflate",
NULL,
2024-04-14 22:45:38 +02:00
deflate_do_init,
deflate_do_write,
deflate_do_close,
2022-11-16 20:14:03 +01:00
sizeof(struct zlib_writer)
2018-04-23 21:13:27 +02:00
};
/* Gzip handler. */
2024-04-14 22:45:38 +02:00
static CURLcode gzip_do_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
/* Initialize zlib */
z->zalloc = (alloc_func) zalloc_cb;
z->zfree = (free_func) zfree_cb;
if(strcmp(zlibVersion(), "1.2.0.4") >= 0) {
/* zlib ver. >= 1.2.0.4 supports transparent gzip decompressing */
if(inflateInit2(z, MAX_WBITS + 32) != Z_OK) {
2021-09-14 00:13:48 +02:00
return process_zlib_error(data, z);
2018-04-23 21:13:27 +02:00
}
zp->zlib_init = ZLIB_INIT_GZIP; /* Transparent gzip decompress state */
}
else {
/* we must parse the gzip header and trailer ourselves */
if(inflateInit2(z, -MAX_WBITS) != Z_OK) {
2021-09-14 00:13:48 +02:00
return process_zlib_error(data, z);
2018-04-23 21:13:27 +02:00
}
zp->trailerlen = 8; /* A CRC-32 and a 32-bit input size (RFC 1952, 2.2) */
zp->zlib_init = ZLIB_INIT; /* Initial call state */
}
return CURLE_OK;
}
#ifdef OLD_ZLIB_SUPPORT
/* Skip over the gzip header */
2024-04-14 22:45:38 +02:00
typedef enum {
GZIP_OK,
GZIP_BAD,
GZIP_UNDERFLOW
2024-04-14 22:45:38 +02:00
} gzip_status;
static gzip_status check_gzip_header(unsigned char const *data, ssize_t len,
ssize_t *headerlen)
{
int method, flags;
const ssize_t totallen = len;
/* The shortest header is 10 bytes */
2015-04-27 22:25:09 +02:00
if(len < 10)
return GZIP_UNDERFLOW;
2015-04-27 22:25:09 +02:00
if((data[0] != GZIP_MAGIC_0) || (data[1] != GZIP_MAGIC_1))
return GZIP_BAD;
method = data[2];
flags = data[3];
2015-04-27 22:25:09 +02:00
if(method != Z_DEFLATED || (flags & RESERVED) != 0) {
/* Can't handle this compression method or unknown flag */
return GZIP_BAD;
}
/* Skip over time, xflags, OS code and all previous bytes */
len -= 10;
data += 10;
2015-04-27 22:25:09 +02:00
if(flags & EXTRA_FIELD) {
ssize_t extra_len;
2015-04-27 22:25:09 +02:00
if(len < 2)
return GZIP_UNDERFLOW;
extra_len = (data[1] << 8) | data[0];
2018-01-26 17:06:56 +01:00
if(len < (extra_len + 2))
return GZIP_UNDERFLOW;
len -= (extra_len + 2);
data += (extra_len + 2);
}
2015-04-27 22:25:09 +02:00
if(flags & ORIG_NAME) {
/* Skip over NUL-terminated file name */
2015-04-27 22:25:09 +02:00
while(len && *data) {
--len;
++data;
}
2015-04-27 22:25:09 +02:00
if(!len || *data)
return GZIP_UNDERFLOW;
/* Skip over the NUL */
--len;
++data;
}
2015-04-27 22:25:09 +02:00
if(flags & COMMENT) {
/* Skip over NUL-terminated comment */
2015-04-27 22:25:09 +02:00
while(len && *data) {
--len;
++data;
}
2015-04-27 22:25:09 +02:00
if(!len || *data)
return GZIP_UNDERFLOW;
/* Skip over the NUL */
--len;
}
2015-04-27 22:25:09 +02:00
if(flags & HEAD_CRC) {
if(len < 2)
return GZIP_UNDERFLOW;
len -= 2;
}
*headerlen = totallen - len;
return GZIP_OK;
}
#endif
2024-04-14 22:45:38 +02:00
static CURLcode gzip_do_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
2018-04-23 21:13:27 +02:00
const char *buf, size_t nbytes)
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
2024-04-14 22:45:38 +02:00
if(!(type & CLIENTWRITE_BODY))
return Curl_cwriter_write(data, writer->next, type, buf, nbytes);
2018-04-23 21:13:27 +02:00
if(zp->zlib_init == ZLIB_INIT_GZIP) {
2015-04-27 22:25:09 +02:00
/* Let zlib handle the gzip decompression entirely */
2018-04-23 21:13:27 +02:00
z->next_in = (Bytef *) buf;
z->avail_in = (uInt) nbytes;
2015-04-27 22:25:09 +02:00
/* Now uncompress the data */
2024-04-14 22:45:38 +02:00
return inflate_stream(data, writer, type, ZLIB_INIT_GZIP);
}
#ifndef OLD_ZLIB_SUPPORT
/* Support for old zlib versions is compiled away and we are running with
an old version, so return an error. */
2021-09-14 00:13:48 +02:00
return exit_zlib(data, z, &zp->zlib_init, CURLE_WRITE_ERROR);
#else
/* This next mess is to get around the potential case where there isn't
* enough data passed in to skip over the gzip header. If that happens, we
* malloc a block and copy what we have then wait for the next call. If
* there still isn't enough (this is definitely a worst-case scenario), we
* make the block bigger, copy the next part in and keep waiting.
*
* This is only required with zlib versions < 1.2.0.4 as newer versions
* can handle the gzip header themselves.
*/
2018-04-23 21:13:27 +02:00
switch(zp->zlib_init) {
/* Skip over gzip header? */
case ZLIB_INIT:
{
/* Initial call state */
ssize_t hlen;
2018-04-23 21:13:27 +02:00
switch(check_gzip_header((unsigned char *) buf, nbytes, &hlen)) {
case GZIP_OK:
2018-04-23 21:13:27 +02:00
z->next_in = (Bytef *) buf + hlen;
z->avail_in = (uInt) (nbytes - hlen);
zp->zlib_init = ZLIB_GZIP_INFLATING; /* Inflating stream state */
break;
case GZIP_UNDERFLOW:
/* We need more data so we can find the end of the gzip header. It's
* possible that the memory block we malloc here will never be freed if
* the transfer abruptly aborts after this point. Since it's unlikely
* that circumstances will be right for this code path to be followed in
* the first place, and it's even more unlikely for a transfer to fail
* immediately afterwards, it should seldom be a problem.
*/
2018-04-23 21:13:27 +02:00
z->avail_in = (uInt) nbytes;
z->next_in = malloc(z->avail_in);
2021-09-14 00:13:48 +02:00
if(!z->next_in) {
return exit_zlib(data, z, &zp->zlib_init, CURLE_OUT_OF_MEMORY);
}
2018-04-23 21:13:27 +02:00
memcpy(z->next_in, buf, z->avail_in);
zp->zlib_init = ZLIB_GZIP_HEADER; /* Need more gzip header data state */
/* We don't have any data to inflate yet */
return CURLE_OK;
case GZIP_BAD:
default:
2021-09-14 00:13:48 +02:00
return exit_zlib(data, z, &zp->zlib_init, process_zlib_error(data, z));
}
}
break;
case ZLIB_GZIP_HEADER:
{
/* Need more gzip header data state */
ssize_t hlen;
2018-04-23 21:13:27 +02:00
z->avail_in += (uInt) nbytes;
2017-07-20 19:35:53 +02:00
z->next_in = Curl_saferealloc(z->next_in, z->avail_in);
2021-09-14 00:13:48 +02:00
if(!z->next_in) {
return exit_zlib(data, z, &zp->zlib_init, CURLE_OUT_OF_MEMORY);
}
/* Append the new block of data to the previous one */
2018-04-23 21:13:27 +02:00
memcpy(z->next_in + z->avail_in - nbytes, buf, nbytes);
2017-07-20 19:35:53 +02:00
switch(check_gzip_header(z->next_in, z->avail_in, &hlen)) {
case GZIP_OK:
/* This is the zlib stream data */
free(z->next_in);
/* Don't point into the malloced block since we just freed it */
2018-04-23 21:13:27 +02:00
z->next_in = (Bytef *) buf + hlen + nbytes - z->avail_in;
z->avail_in = (uInt) (z->avail_in - hlen);
zp->zlib_init = ZLIB_GZIP_INFLATING; /* Inflating stream state */
break;
case GZIP_UNDERFLOW:
/* We still don't have any data to inflate! */
return CURLE_OK;
case GZIP_BAD:
default:
2021-09-14 00:13:48 +02:00
return exit_zlib(data, z, &zp->zlib_init, process_zlib_error(data, z));
}
}
break;
2018-10-28 12:09:07 +01:00
case ZLIB_EXTERNAL_TRAILER:
2018-04-23 21:13:27 +02:00
z->next_in = (Bytef *) buf;
z->avail_in = (uInt) nbytes;
2021-09-14 00:13:48 +02:00
return process_trailer(data, zp);
2018-04-23 21:13:27 +02:00
case ZLIB_GZIP_INFLATING:
default:
/* Inflating stream state */
2018-04-23 21:13:27 +02:00
z->next_in = (Bytef *) buf;
z->avail_in = (uInt) nbytes;
break;
}
2015-04-27 22:25:09 +02:00
if(z->avail_in == 0) {
/* We don't have any data to inflate; wait until next time */
return CURLE_OK;
}
/* We've parsed the header, now uncompress the data */
2024-04-14 22:45:38 +02:00
return inflate_stream(data, writer, type, ZLIB_GZIP_INFLATING);
#endif
}
2015-04-27 22:25:09 +02:00
2024-04-14 22:45:38 +02:00
static void gzip_do_close(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2022-11-16 20:14:03 +01:00
struct zlib_writer *zp = (struct zlib_writer *) writer;
2018-04-23 21:13:27 +02:00
z_stream *z = &zp->z; /* zlib state structure */
2021-09-14 00:13:48 +02:00
exit_zlib(data, z, &zp->zlib_init, CURLE_OK);
2018-04-23 21:13:27 +02:00
}
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype gzip_encoding = {
2018-04-23 21:13:27 +02:00
"gzip",
"x-gzip",
2024-04-14 22:45:38 +02:00
gzip_do_init,
gzip_do_write,
gzip_do_close,
2022-11-16 20:14:03 +01:00
sizeof(struct zlib_writer)
2018-04-23 21:13:27 +02:00
};
#endif /* HAVE_LIBZ */
#ifdef HAVE_BROTLI
2022-11-16 20:14:03 +01:00
/* Brotli writer. */
struct brotli_writer {
2024-04-14 22:45:38 +02:00
struct Curl_cwriter super;
2018-04-23 21:13:27 +02:00
BrotliDecoderState *br; /* State structure for brotli. */
2020-08-30 11:54:41 +02:00
};
2018-04-23 21:13:27 +02:00
static CURLcode brotli_map_error(BrotliDecoderErrorCode be)
{
switch(be) {
case BROTLI_DECODER_ERROR_FORMAT_EXUBERANT_NIBBLE:
case BROTLI_DECODER_ERROR_FORMAT_EXUBERANT_META_NIBBLE:
case BROTLI_DECODER_ERROR_FORMAT_SIMPLE_HUFFMAN_ALPHABET:
case BROTLI_DECODER_ERROR_FORMAT_SIMPLE_HUFFMAN_SAME:
case BROTLI_DECODER_ERROR_FORMAT_CL_SPACE:
case BROTLI_DECODER_ERROR_FORMAT_HUFFMAN_SPACE:
case BROTLI_DECODER_ERROR_FORMAT_CONTEXT_MAP_REPEAT:
case BROTLI_DECODER_ERROR_FORMAT_BLOCK_LENGTH_1:
case BROTLI_DECODER_ERROR_FORMAT_BLOCK_LENGTH_2:
case BROTLI_DECODER_ERROR_FORMAT_TRANSFORM:
case BROTLI_DECODER_ERROR_FORMAT_DICTIONARY:
case BROTLI_DECODER_ERROR_FORMAT_WINDOW_BITS:
case BROTLI_DECODER_ERROR_FORMAT_PADDING_1:
case BROTLI_DECODER_ERROR_FORMAT_PADDING_2:
#ifdef BROTLI_DECODER_ERROR_COMPOUND_DICTIONARY
case BROTLI_DECODER_ERROR_COMPOUND_DICTIONARY:
#endif
#ifdef BROTLI_DECODER_ERROR_DICTIONARY_NOT_SET
case BROTLI_DECODER_ERROR_DICTIONARY_NOT_SET:
#endif
case BROTLI_DECODER_ERROR_INVALID_ARGUMENTS:
return CURLE_BAD_CONTENT_ENCODING;
case BROTLI_DECODER_ERROR_ALLOC_CONTEXT_MODES:
case BROTLI_DECODER_ERROR_ALLOC_TREE_GROUPS:
case BROTLI_DECODER_ERROR_ALLOC_CONTEXT_MAP:
case BROTLI_DECODER_ERROR_ALLOC_RING_BUFFER_1:
case BROTLI_DECODER_ERROR_ALLOC_RING_BUFFER_2:
case BROTLI_DECODER_ERROR_ALLOC_BLOCK_TYPE_TREES:
return CURLE_OUT_OF_MEMORY;
default:
break;
}
return CURLE_WRITE_ERROR;
}
2024-04-14 22:45:38 +02:00
static CURLcode brotli_do_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2022-11-16 20:14:03 +01:00
struct brotli_writer *bp = (struct brotli_writer *) writer;
2021-09-14 00:13:48 +02:00
(void) data;
2018-04-23 21:13:27 +02:00
bp->br = BrotliDecoderCreateInstance(NULL, NULL, NULL);
return bp->br? CURLE_OK: CURLE_OUT_OF_MEMORY;
}
2024-04-14 22:45:38 +02:00
static CURLcode brotli_do_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
2018-04-23 21:13:27 +02:00
const char *buf, size_t nbytes)
{
2022-11-16 20:14:03 +01:00
struct brotli_writer *bp = (struct brotli_writer *) writer;
2018-04-23 21:13:27 +02:00
const uint8_t *src = (const uint8_t *) buf;
char *decomp;
uint8_t *dst;
size_t dstleft;
CURLcode result = CURLE_OK;
BrotliDecoderResult r = BROTLI_DECODER_RESULT_NEEDS_MORE_OUTPUT;
2024-04-14 22:45:38 +02:00
if(!(type & CLIENTWRITE_BODY))
return Curl_cwriter_write(data, writer->next, type, buf, nbytes);
2018-04-23 21:13:27 +02:00
if(!bp->br)
return CURLE_WRITE_ERROR; /* Stream already ended. */
decomp = malloc(DSIZ);
if(!decomp)
return CURLE_OUT_OF_MEMORY;
while((nbytes || r == BROTLI_DECODER_RESULT_NEEDS_MORE_OUTPUT) &&
result == CURLE_OK) {
dst = (uint8_t *) decomp;
dstleft = DSIZ;
r = BrotliDecoderDecompressStream(bp->br,
&nbytes, &src, &dstleft, &dst, NULL);
2024-04-14 22:45:38 +02:00
result = Curl_cwriter_write(data, writer->next, type,
2018-04-23 21:13:27 +02:00
decomp, DSIZ - dstleft);
if(result)
break;
switch(r) {
case BROTLI_DECODER_RESULT_NEEDS_MORE_OUTPUT:
case BROTLI_DECODER_RESULT_NEEDS_MORE_INPUT:
break;
case BROTLI_DECODER_RESULT_SUCCESS:
BrotliDecoderDestroyInstance(bp->br);
bp->br = NULL;
if(nbytes)
result = CURLE_WRITE_ERROR;
break;
default:
result = brotli_map_error(BrotliDecoderGetErrorCode(bp->br));
break;
}
}
free(decomp);
return result;
}
2024-04-14 22:45:38 +02:00
static void brotli_do_close(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2022-11-16 20:14:03 +01:00
struct brotli_writer *bp = (struct brotli_writer *) writer;
2021-09-14 00:13:48 +02:00
(void) data;
2018-04-23 21:13:27 +02:00
if(bp->br) {
BrotliDecoderDestroyInstance(bp->br);
bp->br = NULL;
}
}
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype brotli_encoding = {
2018-04-23 21:13:27 +02:00
"br",
NULL,
2024-04-14 22:45:38 +02:00
brotli_do_init,
brotli_do_write,
brotli_do_close,
2022-11-16 20:14:03 +01:00
sizeof(struct brotli_writer)
2018-04-23 21:13:27 +02:00
};
#endif
2021-09-14 00:13:48 +02:00
#ifdef HAVE_ZSTD
2022-11-16 20:14:03 +01:00
/* Zstd writer. */
struct zstd_writer {
2024-04-14 22:45:38 +02:00
struct Curl_cwriter super;
2021-09-14 00:13:48 +02:00
ZSTD_DStream *zds; /* State structure for zstd. */
void *decomp;
};
2024-04-14 22:45:38 +02:00
static CURLcode zstd_do_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
2021-09-14 00:13:48 +02:00
{
2022-11-16 20:14:03 +01:00
struct zstd_writer *zp = (struct zstd_writer *) writer;
2021-09-14 00:13:48 +02:00
(void)data;
zp->zds = ZSTD_createDStream();
zp->decomp = NULL;
return zp->zds ? CURLE_OK : CURLE_OUT_OF_MEMORY;
}
2024-04-14 22:45:38 +02:00
static CURLcode zstd_do_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
2021-09-14 00:13:48 +02:00
const char *buf, size_t nbytes)
{
CURLcode result = CURLE_OK;
2022-11-16 20:14:03 +01:00
struct zstd_writer *zp = (struct zstd_writer *) writer;
2021-09-14 00:13:48 +02:00
ZSTD_inBuffer in;
ZSTD_outBuffer out;
size_t errorCode;
2024-04-14 22:45:38 +02:00
if(!(type & CLIENTWRITE_BODY))
return Curl_cwriter_write(data, writer->next, type, buf, nbytes);
2021-09-14 00:13:48 +02:00
if(!zp->decomp) {
zp->decomp = malloc(DSIZ);
if(!zp->decomp)
return CURLE_OUT_OF_MEMORY;
}
in.pos = 0;
in.src = buf;
in.size = nbytes;
for(;;) {
out.pos = 0;
out.dst = zp->decomp;
out.size = DSIZ;
errorCode = ZSTD_decompressStream(zp->zds, &out, &in);
if(ZSTD_isError(errorCode)) {
return CURLE_BAD_CONTENT_ENCODING;
}
if(out.pos > 0) {
2024-04-14 22:45:38 +02:00
result = Curl_cwriter_write(data, writer->next, type,
2021-09-14 00:13:48 +02:00
zp->decomp, out.pos);
if(result)
break;
}
if((in.pos == nbytes) && (out.pos < out.size))
break;
}
return result;
}
2024-04-14 22:45:38 +02:00
static void zstd_do_close(struct Curl_easy *data,
struct Curl_cwriter *writer)
2021-09-14 00:13:48 +02:00
{
2022-11-16 20:14:03 +01:00
struct zstd_writer *zp = (struct zstd_writer *) writer;
2021-09-14 00:13:48 +02:00
(void)data;
if(zp->decomp) {
free(zp->decomp);
zp->decomp = NULL;
}
if(zp->zds) {
ZSTD_freeDStream(zp->zds);
zp->zds = NULL;
}
}
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype zstd_encoding = {
2021-09-14 00:13:48 +02:00
"zstd",
NULL,
2024-04-14 22:45:38 +02:00
zstd_do_init,
zstd_do_write,
zstd_do_close,
2022-11-16 20:14:03 +01:00
sizeof(struct zstd_writer)
2021-09-14 00:13:48 +02:00
};
#endif
2018-04-23 21:13:27 +02:00
/* Identity handler. */
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype identity_encoding = {
2018-04-23 21:13:27 +02:00
"identity",
2018-08-09 18:06:22 +02:00
"none",
2024-04-14 22:45:38 +02:00
Curl_cwriter_def_init,
Curl_cwriter_def_write,
Curl_cwriter_def_close,
sizeof(struct Curl_cwriter)
2018-04-23 21:13:27 +02:00
};
2024-04-14 22:45:38 +02:00
/* supported general content decoders. */
static const struct Curl_cwtype * const general_unencoders[] = {
2018-04-23 21:13:27 +02:00
&identity_encoding,
#ifdef HAVE_LIBZ
&deflate_encoding,
&gzip_encoding,
#endif
#ifdef HAVE_BROTLI
&brotli_encoding,
2021-09-14 00:13:48 +02:00
#endif
#ifdef HAVE_ZSTD
&zstd_encoding,
2018-04-23 21:13:27 +02:00
#endif
NULL
};
2024-04-14 22:45:38 +02:00
/* supported content decoders only for transfer encodings */
static const struct Curl_cwtype * const transfer_unencoders[] = {
#ifndef CURL_DISABLE_HTTP
&Curl_httpchunk_unencoder,
#endif
NULL
};
2018-04-23 21:13:27 +02:00
2024-04-14 22:45:38 +02:00
/* Provide a list of comma-separated names of supported encodings.
*/
void Curl_all_content_encodings(char *buf, size_t blen)
2018-04-23 21:13:27 +02:00
{
size_t len = 0;
2024-04-14 22:45:38 +02:00
const struct Curl_cwtype * const *cep;
const struct Curl_cwtype *ce;
DEBUGASSERT(buf);
DEBUGASSERT(blen);
buf[0] = 0;
2018-04-23 21:13:27 +02:00
2024-04-14 22:45:38 +02:00
for(cep = general_unencoders; *cep; cep++) {
2018-04-23 21:13:27 +02:00
ce = *cep;
if(!strcasecompare(ce->name, CONTENT_ENCODING_DEFAULT))
len += strlen(ce->name) + 2;
}
2024-04-14 22:45:38 +02:00
if(!len) {
if(blen >= sizeof(CONTENT_ENCODING_DEFAULT))
strcpy(buf, CONTENT_ENCODING_DEFAULT);
}
else if(blen > len) {
char *p = buf;
for(cep = general_unencoders; *cep; cep++) {
2018-04-23 21:13:27 +02:00
ce = *cep;
if(!strcasecompare(ce->name, CONTENT_ENCODING_DEFAULT)) {
strcpy(p, ce->name);
p += strlen(p);
*p++ = ',';
*p++ = ' ';
}
}
p[-2] = '\0';
}
}
/* Deferred error dummy writer. */
2024-04-14 22:45:38 +02:00
static CURLcode error_do_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2023-12-07 09:12:54 +01:00
(void)data;
(void)writer;
return CURLE_OK;
2018-04-23 21:13:27 +02:00
}
2024-04-14 22:45:38 +02:00
static CURLcode error_do_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
2018-04-23 21:13:27 +02:00
const char *buf, size_t nbytes)
{
2024-04-14 22:45:38 +02:00
char all[256];
(void)Curl_all_content_encodings(all, sizeof(all));
2018-04-23 21:13:27 +02:00
(void) writer;
(void) buf;
(void) nbytes;
2024-04-14 22:45:38 +02:00
if(!(type & CLIENTWRITE_BODY))
return Curl_cwriter_write(data, writer->next, type, buf, nbytes);
2021-09-14 00:13:48 +02:00
failf(data, "Unrecognized content encoding type. "
"libcurl understands %s content encodings.", all);
2018-04-23 21:13:27 +02:00
return CURLE_BAD_CONTENT_ENCODING;
}
2024-04-14 22:45:38 +02:00
static void error_do_close(struct Curl_easy *data,
struct Curl_cwriter *writer)
2018-04-23 21:13:27 +02:00
{
2021-09-14 00:13:48 +02:00
(void) data;
2018-04-23 21:13:27 +02:00
(void) writer;
}
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype error_writer = {
"ce-error",
2018-04-23 21:13:27 +02:00
NULL,
2024-04-14 22:45:38 +02:00
error_do_init,
error_do_write,
error_do_close,
sizeof(struct Curl_cwriter)
2018-04-23 21:13:27 +02:00
};
/* Find the content encoding by name. */
2024-04-14 22:45:38 +02:00
static const struct Curl_cwtype *find_unencode_writer(const char *name,
size_t len,
Curl_cwriter_phase phase)
2018-04-23 21:13:27 +02:00
{
2024-04-14 22:45:38 +02:00
const struct Curl_cwtype * const *cep;
if(phase == CURL_CW_TRANSFER_DECODE) {
for(cep = transfer_unencoders; *cep; cep++) {
const struct Curl_cwtype *ce = *cep;
if((strncasecompare(name, ce->name, len) && !ce->name[len]) ||
(ce->alias && strncasecompare(name, ce->alias, len)
&& !ce->alias[len]))
return ce;
}
}
/* look among the general decoders */
for(cep = general_unencoders; *cep; cep++) {
const struct Curl_cwtype *ce = *cep;
2018-04-23 21:13:27 +02:00
if((strncasecompare(name, ce->name, len) && !ce->name[len]) ||
(ce->alias && strncasecompare(name, ce->alias, len) && !ce->alias[len]))
return ce;
}
return NULL;
}
/* Set-up the unencoding stack from the Content-Encoding header value.
* See RFC 7231 section 3.1.2.2. */
2021-09-14 00:13:48 +02:00
CURLcode Curl_build_unencoding_stack(struct Curl_easy *data,
2023-07-02 19:51:09 +02:00
const char *enclist, int is_transfer)
2018-04-23 21:13:27 +02:00
{
2024-04-14 22:45:38 +02:00
Curl_cwriter_phase phase = is_transfer?
CURL_CW_TRANSFER_DECODE:CURL_CW_CONTENT_DECODE;
2023-12-07 09:12:54 +01:00
CURLcode result;
2018-04-23 21:13:27 +02:00
do {
const char *name;
size_t namelen;
/* Parse a single encoding name. */
2022-11-16 20:14:03 +01:00
while(ISBLANK(*enclist) || *enclist == ',')
2018-04-23 21:13:27 +02:00
enclist++;
name = enclist;
for(namelen = 0; *enclist && *enclist != ','; enclist++)
if(!ISSPACE(*enclist))
namelen = enclist - name + 1;
2024-04-14 22:45:38 +02:00
if(namelen) {
const struct Curl_cwtype *cwt;
struct Curl_cwriter *writer;
/* if we skip the decoding in this phase, do not look further.
* Exception is "chunked" transfer-encoding which always must happen */
if((is_transfer && !data->set.http_transfer_encoding &&
(namelen != 7 || !strncasecompare(name, "chunked", 7))) ||
(!is_transfer && data->set.http_ce_skip)) {
2023-07-02 19:51:09 +02:00
/* not requested, ignore */
return CURLE_OK;
2024-04-14 22:45:38 +02:00
}
2018-04-23 21:13:27 +02:00
2024-04-14 22:45:38 +02:00
if(Curl_cwriter_count(data, phase) + 1 >= MAX_ENCODE_STACK) {
failf(data, "Reject response due to more than %u content encodings",
MAX_ENCODE_STACK);
return CURLE_BAD_CONTENT_ENCODING;
}
2018-04-23 21:13:27 +02:00
2024-04-14 22:45:38 +02:00
cwt = find_unencode_writer(name, namelen, phase);
if(!cwt)
cwt = &error_writer; /* Defer error at use. */
result = Curl_cwriter_create(&writer, data, cwt, phase);
2023-12-07 09:12:54 +01:00
if(result)
return result;
2024-04-14 22:45:38 +02:00
result = Curl_cwriter_add(data, writer);
2023-12-07 09:12:54 +01:00
if(result) {
2024-04-14 22:45:38 +02:00
Curl_cwriter_free(data, writer);
2023-12-07 09:12:54 +01:00
return result;
2023-07-02 19:51:09 +02:00
}
2018-04-23 21:13:27 +02:00
}
} while(*enclist);
return CURLE_OK;
}
#else
/* Stubs for builds without HTTP. */
2021-09-14 00:13:48 +02:00
CURLcode Curl_build_unencoding_stack(struct Curl_easy *data,
2023-07-02 19:51:09 +02:00
const char *enclist, int is_transfer)
2018-04-23 21:13:27 +02:00
{
2021-09-14 00:13:48 +02:00
(void) data;
2018-04-23 21:13:27 +02:00
(void) enclist;
2023-07-02 19:51:09 +02:00
(void) is_transfer;
2018-04-23 21:13:27 +02:00
return CURLE_NOT_BUILT_IN;
}
2024-04-14 22:45:38 +02:00
void Curl_all_content_encodings(char *buf, size_t blen)
2018-04-23 21:13:27 +02:00
{
2024-04-14 22:45:38 +02:00
DEBUGASSERT(buf);
DEBUGASSERT(blen);
if(blen < sizeof(CONTENT_ENCODING_DEFAULT))
buf[0] = 0;
else
strcpy(buf, CONTENT_ENCODING_DEFAULT);
2018-04-23 21:13:27 +02:00
}
#endif /* CURL_DISABLE_HTTP */