<?php

[0] Fix | Delete

[1] Fix | Delete

// SPDX-FileCopyrightText: 2004-2023 Ryan Parman, Sam Sneddon, Ryan McCue

[2] Fix | Delete

// SPDX-License-Identifier: BSD-3-Clause

[3] Fix | Delete

[4] Fix | Delete

declare(strict_types=1);

[5] Fix | Delete

[6] Fix | Delete

namespace SimplePie\HTTP;

[7] Fix | Delete

[8] Fix | Delete

/**

[9] Fix | Delete

* HTTP Response Parser

[10] Fix | Delete

* @template Psr7Compatible of bool

[11] Fix | Delete

*/

[12] Fix | Delete

class Parser

[13] Fix | Delete

{

[14] Fix | Delete

/**

[15] Fix | Delete

* HTTP Version

[16] Fix | Delete

*

[17] Fix | Delete

* @var float

[18] Fix | Delete

*/

[19] Fix | Delete

public $http_version = 0.0;

[20] Fix | Delete

[21] Fix | Delete

/**

[22] Fix | Delete

* Status code

[23] Fix | Delete

*

[24] Fix | Delete

* @var int

[25] Fix | Delete

*/

[26] Fix | Delete

public $status_code = 0;

[27] Fix | Delete

[28] Fix | Delete

/**

[29] Fix | Delete

* Reason phrase

[30] Fix | Delete

*

[31] Fix | Delete

* @var string

[32] Fix | Delete

*/

[33] Fix | Delete

public $reason = '';

[34] Fix | Delete

[35] Fix | Delete

/**

[36] Fix | Delete

* @var Psr7Compatible whether headers are compatible with PSR-7 format.

[37] Fix | Delete

*/

[38] Fix | Delete

private $psr7Compatible;

[39] Fix | Delete

[40] Fix | Delete

/**

[41] Fix | Delete

* Key/value pairs of the headers

[42] Fix | Delete

*

[43] Fix | Delete

* @var (Psr7Compatible is true ? array<string, non-empty-array<string>> : array<string, string>)

[44] Fix | Delete

*/

[45] Fix | Delete

public $headers = [];

[46] Fix | Delete

[47] Fix | Delete

/**

[48] Fix | Delete

* Body of the response

[49] Fix | Delete

*

[50] Fix | Delete

* @var string

[51] Fix | Delete

*/

[52] Fix | Delete

public $body = '';

[53] Fix | Delete

[54] Fix | Delete

private const STATE_HTTP_VERSION = 'http_version';

[55] Fix | Delete

[56] Fix | Delete

private const STATE_STATUS = 'status';

[57] Fix | Delete

[58] Fix | Delete

private const STATE_REASON = 'reason';

[59] Fix | Delete

[60] Fix | Delete

private const STATE_NEW_LINE = 'new_line';

[61] Fix | Delete

[62] Fix | Delete

private const STATE_BODY = 'body';

[63] Fix | Delete

[64] Fix | Delete

private const STATE_NAME = 'name';

[65] Fix | Delete

[66] Fix | Delete

private const STATE_VALUE = 'value';

[67] Fix | Delete

[68] Fix | Delete

private const STATE_VALUE_CHAR = 'value_char';

[69] Fix | Delete

[70] Fix | Delete

private const STATE_QUOTE = 'quote';

[71] Fix | Delete

[72] Fix | Delete

private const STATE_QUOTE_ESCAPED = 'quote_escaped';

[73] Fix | Delete

[74] Fix | Delete

private const STATE_QUOTE_CHAR = 'quote_char';

[75] Fix | Delete

[76] Fix | Delete

private const STATE_CHUNKED = 'chunked';

[77] Fix | Delete

[78] Fix | Delete

private const STATE_EMIT = 'emit';

[79] Fix | Delete

[80] Fix | Delete

private const STATE_ERROR = false;

[81] Fix | Delete

[82] Fix | Delete

/**

[83] Fix | Delete

* Current state of the state machine

[84] Fix | Delete

*

[85] Fix | Delete

* @var self::STATE_*

[86] Fix | Delete

*/

[87] Fix | Delete

protected $state = self::STATE_HTTP_VERSION;

[88] Fix | Delete

[89] Fix | Delete

/**

[90] Fix | Delete

* Input data

[91] Fix | Delete

*

[92] Fix | Delete

* @var string

[93] Fix | Delete

*/

[94] Fix | Delete

protected $data = '';

[95] Fix | Delete

[96] Fix | Delete

/**

[97] Fix | Delete

* Input data length (to avoid calling strlen() everytime this is needed)

[98] Fix | Delete

*

[99] Fix | Delete

* @var int

[100] Fix | Delete

*/

[101] Fix | Delete

protected $data_length = 0;

[102] Fix | Delete

[103] Fix | Delete

/**

[104] Fix | Delete

* Current position of the pointer

[105] Fix | Delete

*

[106] Fix | Delete

* @var int

[107] Fix | Delete

*/

[108] Fix | Delete

protected $position = 0;

[109] Fix | Delete

[110] Fix | Delete

/**

[111] Fix | Delete

* Name of the header currently being parsed

[112] Fix | Delete

*

[113] Fix | Delete

* @var string

[114] Fix | Delete

*/

[115] Fix | Delete

protected $name = '';

[116] Fix | Delete

[117] Fix | Delete

/**

[118] Fix | Delete

* Value of the header currently being parsed

[119] Fix | Delete

*

[120] Fix | Delete

* @var string

[121] Fix | Delete

*/

[122] Fix | Delete

protected $value = '';

[123] Fix | Delete

[124] Fix | Delete

/**

[125] Fix | Delete

* Create an instance of the class with the input data

[126] Fix | Delete

*

[127] Fix | Delete

* @param string $data Input data

[128] Fix | Delete

* @param Psr7Compatible $psr7Compatible Whether the data types are in format compatible with PSR-7.

[129] Fix | Delete

*/

[130] Fix | Delete

public function __construct(string $data, bool $psr7Compatible = false)

[131] Fix | Delete

{

[132] Fix | Delete

$this->data = $data;

[133] Fix | Delete

$this->data_length = strlen($this->data);

[134] Fix | Delete

$this->psr7Compatible = $psr7Compatible;

[135] Fix | Delete

}

[136] Fix | Delete

[137] Fix | Delete

/**

[138] Fix | Delete

* Parse the input data

[139] Fix | Delete

*

[140] Fix | Delete

* @return bool true on success, false on failure

[141] Fix | Delete

*/

[142] Fix | Delete

public function parse()

[143] Fix | Delete

{

[144] Fix | Delete

while ($this->state && $this->state !== self::STATE_EMIT && $this->has_data()) {

[145] Fix | Delete

$state = $this->state;

[146] Fix | Delete

$this->$state();

[147] Fix | Delete

}

[148] Fix | Delete

$this->data = '';

[149] Fix | Delete

if ($this->state === self::STATE_EMIT || $this->state === self::STATE_BODY) {

[150] Fix | Delete

return true;

[151] Fix | Delete

}

[152] Fix | Delete

[153] Fix | Delete

// Reset the parser state.

[154] Fix | Delete

$this->http_version = 0.0;

[155] Fix | Delete

$this->status_code = 0;

[156] Fix | Delete

$this->reason = '';

[157] Fix | Delete

$this->headers = [];

[158] Fix | Delete

$this->body = '';

[159] Fix | Delete

return false;

[160] Fix | Delete

}

[161] Fix | Delete

[162] Fix | Delete

/**

[163] Fix | Delete

* Check whether there is data beyond the pointer

[164] Fix | Delete

*

[165] Fix | Delete

* @return bool true if there is further data, false if not

[166] Fix | Delete

*/

[167] Fix | Delete

protected function has_data()

[168] Fix | Delete

{

[169] Fix | Delete

return (bool) ($this->position < $this->data_length);

[170] Fix | Delete

}

[171] Fix | Delete

[172] Fix | Delete

/**

[173] Fix | Delete

* See if the next character is LWS

[174] Fix | Delete

*

[175] Fix | Delete

* @return bool true if the next character is LWS, false if not

[176] Fix | Delete

*/

[177] Fix | Delete

protected function is_linear_whitespace()

[178] Fix | Delete

{

[179] Fix | Delete

return (bool) ($this->data[$this->position] === "\x09"

[180] Fix | Delete

|| $this->data[$this->position] === "\x20"

[181] Fix | Delete

|| ($this->data[$this->position] === "\x0A"

[182] Fix | Delete

&& isset($this->data[$this->position + 1])

[183] Fix | Delete

&& ($this->data[$this->position + 1] === "\x09" || $this->data[$this->position + 1] === "\x20")));

[184] Fix | Delete

}

[185] Fix | Delete

[186] Fix | Delete

/**

[187] Fix | Delete

* Parse the HTTP version

[188] Fix | Delete

* @return void

[189] Fix | Delete

*/

[190] Fix | Delete

protected function http_version()

[191] Fix | Delete

{

[192] Fix | Delete

if (strpos($this->data, "\x0A") !== false && strtoupper(substr($this->data, 0, 5)) === 'HTTP/') {

[193] Fix | Delete

$len = strspn($this->data, '0123456789.', 5);

[194] Fix | Delete

$http_version = substr($this->data, 5, $len);

[195] Fix | Delete

$this->position += 5 + $len;

[196] Fix | Delete

if (substr_count($http_version, '.') <= 1) {

[197] Fix | Delete

$this->http_version = (float) $http_version;

[198] Fix | Delete

$this->position += strspn($this->data, "\x09\x20", $this->position);

[199] Fix | Delete

$this->state = self::STATE_STATUS;

[200] Fix | Delete

} else {

[201] Fix | Delete

$this->state = self::STATE_ERROR;

[202] Fix | Delete

}

[203] Fix | Delete

} else {

[204] Fix | Delete

$this->state = self::STATE_ERROR;

[205] Fix | Delete

}

[206] Fix | Delete

}

[207] Fix | Delete

[208] Fix | Delete

/**

[209] Fix | Delete

* Parse the status code

[210] Fix | Delete

* @return void

[211] Fix | Delete

*/

[212] Fix | Delete

protected function status()

[213] Fix | Delete

{

[214] Fix | Delete

if ($len = strspn($this->data, '0123456789', $this->position)) {

[215] Fix | Delete

$this->status_code = (int) substr($this->data, $this->position, $len);

[216] Fix | Delete

$this->position += $len;

[217] Fix | Delete

$this->state = self::STATE_REASON;

[218] Fix | Delete

} else {

[219] Fix | Delete

$this->state = self::STATE_ERROR;

[220] Fix | Delete

}

[221] Fix | Delete

}

[222] Fix | Delete

[223] Fix | Delete

/**

[224] Fix | Delete

* Parse the reason phrase

[225] Fix | Delete

* @return void

[226] Fix | Delete

*/

[227] Fix | Delete

protected function reason()

[228] Fix | Delete

{

[229] Fix | Delete

$len = strcspn($this->data, "\x0A", $this->position);

[230] Fix | Delete

$this->reason = trim(substr($this->data, $this->position, $len), "\x09\x0D\x20");

[231] Fix | Delete

$this->position += $len + 1;

[232] Fix | Delete

$this->state = self::STATE_NEW_LINE;

[233] Fix | Delete

}

[234] Fix | Delete

[235] Fix | Delete

private function add_header(string $name, string $value): void

[236] Fix | Delete

{

[237] Fix | Delete

if ($this->psr7Compatible) {

[238] Fix | Delete

// For PHPStan: should be enforced by template parameter but PHPStan is not smart enough.

[239] Fix | Delete

/** @var array<string, non-empty-array<string>> */

[240] Fix | Delete

$headers = &$this->headers;

[241] Fix | Delete

$headers[$name][] = $value;

[242] Fix | Delete

} else {

[243] Fix | Delete

// For PHPStan: should be enforced by template parameter but PHPStan is not smart enough.

[244] Fix | Delete

/** @var array<string, string>) */

[245] Fix | Delete

$headers = &$this->headers;

[246] Fix | Delete

$headers[$name] .= ', ' . $value;

[247] Fix | Delete

}

[248] Fix | Delete

}

[249] Fix | Delete

[250] Fix | Delete

private function replace_header(string $name, string $value): void

[251] Fix | Delete

{

[252] Fix | Delete

if ($this->psr7Compatible) {

[253] Fix | Delete

// For PHPStan: should be enforced by template parameter but PHPStan is not smart enough.

[254] Fix | Delete

/** @var array<string, non-empty-array<string>> */

[255] Fix | Delete

$headers = &$this->headers;

[256] Fix | Delete

$headers[$name] = [$value];

[257] Fix | Delete

} else {

[258] Fix | Delete

// For PHPStan: should be enforced by template parameter but PHPStan is not smart enough.

[259] Fix | Delete

/** @var array<string, string>) */

[260] Fix | Delete

$headers = &$this->headers;

[261] Fix | Delete

$headers[$name] = $value;

[262] Fix | Delete

}

[263] Fix | Delete

}

[264] Fix | Delete

[265] Fix | Delete

/**

[266] Fix | Delete

* Deal with a new line, shifting data around as needed

[267] Fix | Delete

* @return void

[268] Fix | Delete

*/

[269] Fix | Delete

protected function new_line()

[270] Fix | Delete

{

[271] Fix | Delete

$this->value = trim($this->value, "\x0D\x20");

[272] Fix | Delete

if ($this->name !== '' && $this->value !== '') {

[273] Fix | Delete

$this->name = strtolower($this->name);

[274] Fix | Delete

// We should only use the last Content-Type header. c.f. issue #1

[275] Fix | Delete

if (isset($this->headers[$this->name]) && $this->name !== 'content-type') {

[276] Fix | Delete

$this->add_header($this->name, $this->value);

[277] Fix | Delete

} else {

[278] Fix | Delete

$this->replace_header($this->name, $this->value);

[279] Fix | Delete

}

[280] Fix | Delete

}

[281] Fix | Delete

$this->name = '';

[282] Fix | Delete

$this->value = '';

[283] Fix | Delete

if (substr($this->data[$this->position], 0, 2) === "\x0D\x0A") {

[284] Fix | Delete

$this->position += 2;

[285] Fix | Delete

$this->state = self::STATE_BODY;

[286] Fix | Delete

} elseif ($this->data[$this->position] === "\x0A") {

[287] Fix | Delete

$this->position++;

[288] Fix | Delete

$this->state = self::STATE_BODY;

[289] Fix | Delete

} else {

[290] Fix | Delete

$this->state = self::STATE_NAME;

[291] Fix | Delete

}

[292] Fix | Delete

}

[293] Fix | Delete

[294] Fix | Delete

/**

[295] Fix | Delete

* Parse a header name

[296] Fix | Delete

* @return void

[297] Fix | Delete

*/

[298] Fix | Delete

protected function name()

[299] Fix | Delete

{

[300] Fix | Delete

$len = strcspn($this->data, "\x0A:", $this->position);

[301] Fix | Delete

if (isset($this->data[$this->position + $len])) {

[302] Fix | Delete

if ($this->data[$this->position + $len] === "\x0A") {

[303] Fix | Delete

$this->position += $len;

[304] Fix | Delete

$this->state = self::STATE_NEW_LINE;

[305] Fix | Delete

} else {

[306] Fix | Delete

$this->name = substr($this->data, $this->position, $len);

[307] Fix | Delete

$this->position += $len + 1;

[308] Fix | Delete

$this->state = self::STATE_VALUE;

[309] Fix | Delete

}

[310] Fix | Delete

} else {

[311] Fix | Delete

$this->state = self::STATE_ERROR;

[312] Fix | Delete

}

[313] Fix | Delete

}

[314] Fix | Delete

[315] Fix | Delete

/**

[316] Fix | Delete

* Parse LWS, replacing consecutive LWS characters with a single space

[317] Fix | Delete

* @return void

[318] Fix | Delete

*/

[319] Fix | Delete

protected function linear_whitespace()

[320] Fix | Delete

{

[321] Fix | Delete

do {

[322] Fix | Delete

if (substr($this->data, $this->position, 2) === "\x0D\x0A") {

[323] Fix | Delete

$this->position += 2;

[324] Fix | Delete

} elseif ($this->data[$this->position] === "\x0A") {

[325] Fix | Delete

$this->position++;

[326] Fix | Delete

}

[327] Fix | Delete

$this->position += strspn($this->data, "\x09\x20", $this->position);

[328] Fix | Delete

} while ($this->has_data() && $this->is_linear_whitespace());

[329] Fix | Delete

$this->value .= "\x20";

[330] Fix | Delete

}

[331] Fix | Delete

[332] Fix | Delete

/**

[333] Fix | Delete

* See what state to move to while within non-quoted header values

[334] Fix | Delete

* @return void

[335] Fix | Delete

*/

[336] Fix | Delete

protected function value()

[337] Fix | Delete

{

[338] Fix | Delete

if ($this->is_linear_whitespace()) {

[339] Fix | Delete

$this->linear_whitespace();

[340] Fix | Delete

} else {

[341] Fix | Delete

switch ($this->data[$this->position]) {

[342] Fix | Delete

case '"':

[343] Fix | Delete

// Workaround for ETags: we have to include the quotes as

[344] Fix | Delete

// part of the tag.

[345] Fix | Delete

if (strtolower($this->name) === 'etag') {

[346] Fix | Delete

$this->value .= '"';

[347] Fix | Delete

$this->position++;

[348] Fix | Delete

$this->state = self::STATE_VALUE_CHAR;

[349] Fix | Delete

break;

[350] Fix | Delete

}

[351] Fix | Delete

$this->position++;

[352] Fix | Delete

$this->state = self::STATE_QUOTE;

[353] Fix | Delete

break;

[354] Fix | Delete

[355] Fix | Delete

case "\x0A":

[356] Fix | Delete

$this->position++;

[357] Fix | Delete

$this->state = self::STATE_NEW_LINE;

[358] Fix | Delete

break;

[359] Fix | Delete

[360] Fix | Delete

default:

[361] Fix | Delete

$this->state = self::STATE_VALUE_CHAR;

[362] Fix | Delete

break;

[363] Fix | Delete

}

[364] Fix | Delete

}

[365] Fix | Delete

}

[366] Fix | Delete

[367] Fix | Delete

/**

[368] Fix | Delete

* Parse a header value while outside quotes

[369] Fix | Delete

* @return void

[370] Fix | Delete

*/

[371] Fix | Delete

protected function value_char()

[372] Fix | Delete

{

[373] Fix | Delete

$len = strcspn($this->data, "\x09\x20\x0A\"", $this->position);

[374] Fix | Delete

$this->value .= substr($this->data, $this->position, $len);

[375] Fix | Delete

$this->position += $len;

[376] Fix | Delete

$this->state = self::STATE_VALUE;

[377] Fix | Delete

}

[378] Fix | Delete

[379] Fix | Delete

/**

[380] Fix | Delete

* See what state to move to while within quoted header values

[381] Fix | Delete

* @return void

[382] Fix | Delete

*/

[383] Fix | Delete

protected function quote()

[384] Fix | Delete

{

[385] Fix | Delete

if ($this->is_linear_whitespace()) {

[386] Fix | Delete

$this->linear_whitespace();

[387] Fix | Delete

} else {

[388] Fix | Delete

switch ($this->data[$this->position]) {

[389] Fix | Delete

case '"':

[390] Fix | Delete

$this->position++;

[391] Fix | Delete

$this->state = self::STATE_VALUE;

[392] Fix | Delete

break;

[393] Fix | Delete

[394] Fix | Delete

case "\x0A":

[395] Fix | Delete

$this->position++;

[396] Fix | Delete

$this->state = self::STATE_NEW_LINE;

[397] Fix | Delete

break;

[398] Fix | Delete

[399] Fix | Delete

case '\\':

[400] Fix | Delete

$this->position++;

[401] Fix | Delete

$this->state = self::STATE_QUOTE_ESCAPED;

[402] Fix | Delete

break;

[403] Fix | Delete

[404] Fix | Delete

default:

[405] Fix | Delete

$this->state = self::STATE_QUOTE_CHAR;

[406] Fix | Delete

break;

[407] Fix | Delete

}

[408] Fix | Delete

}

[409] Fix | Delete

}

[410] Fix | Delete

[411] Fix | Delete

/**

[412] Fix | Delete

* Parse a header value while within quotes

[413] Fix | Delete

* @return void

[414] Fix | Delete

*/

[415] Fix | Delete

protected function quote_char()

[416] Fix | Delete

{

[417] Fix | Delete

$len = strcspn($this->data, "\x09\x20\x0A\"\\", $this->position);

[418] Fix | Delete

$this->value .= substr($this->data, $this->position, $len);

[419] Fix | Delete

$this->position += $len;

[420] Fix | Delete

$this->state = self::STATE_VALUE;

[421] Fix | Delete

}

[422] Fix | Delete

[423] Fix | Delete

/**

[424] Fix | Delete

* Parse an escaped character within quotes

[425] Fix | Delete

* @return void

[426] Fix | Delete

*/

[427] Fix | Delete

protected function quote_escaped()

[428] Fix | Delete

{

[429] Fix | Delete

$this->value .= $this->data[$this->position];

[430] Fix | Delete

$this->position++;

[431] Fix | Delete

$this->state = self::STATE_QUOTE;

[432] Fix | Delete

}

[433] Fix | Delete

[434] Fix | Delete

/**

[435] Fix | Delete

* Parse the body

[436] Fix | Delete

* @return void

[437] Fix | Delete

*/

[438] Fix | Delete

protected function body()

[439] Fix | Delete

{

[440] Fix | Delete

$this->body = substr($this->data, $this->position);

[441] Fix | Delete

if (!empty($this->headers['transfer-encoding'])) {

[442] Fix | Delete

unset($this->headers['transfer-encoding']);

[443] Fix | Delete

$this->state = self::STATE_CHUNKED;

[444] Fix | Delete

} else {

[445] Fix | Delete

$this->state = self::STATE_EMIT;

[446] Fix | Delete

}

[447] Fix | Delete

}

[448] Fix | Delete

[449] Fix | Delete

/**

[450] Fix | Delete

* Parsed a "Transfer-Encoding: chunked" body

[451] Fix | Delete

* @return void

[452] Fix | Delete

*/

[453] Fix | Delete

protected function chunked()

[454] Fix | Delete

{

[455] Fix | Delete

if (!preg_match('/^([0-9a-f]+)[^\r\n]*\r\n/i', trim($this->body))) {

[456] Fix | Delete

$this->state = self::STATE_EMIT;

[457] Fix | Delete

return;

[458] Fix | Delete

}

[459] Fix | Delete

[460] Fix | Delete

$decoded = '';

[461] Fix | Delete

$encoded = $this->body;

[462] Fix | Delete

[463] Fix | Delete

while (true) {

[464] Fix | Delete

$is_chunked = (bool) preg_match('/^([0-9a-f]+)[^\r\n]*\r\n/i', $encoded, $matches);

[465] Fix | Delete

if (!$is_chunked) {

[466] Fix | Delete

// Looks like it's not chunked after all

[467] Fix | Delete

$this->state = self::STATE_EMIT;

[468] Fix | Delete

return;

[469] Fix | Delete

}

[470] Fix | Delete

[471] Fix | Delete

$length = hexdec(trim($matches[1]));

[472] Fix | Delete

// For PHPStan: this will only be float when larger than PHP_INT_MAX.

[473] Fix | Delete

// But even on 32-bit systems, it would mean 2GiB chunk, which sounds unlikely.

[474] Fix | Delete

\assert(\is_int($length), "Length needs to be shorter than PHP_INT_MAX");

[475] Fix | Delete

if ($length === 0) {

[476] Fix | Delete

// Ignore trailer headers

[477] Fix | Delete

$this->state = self::STATE_EMIT;

[478] Fix | Delete

$this->body = $decoded;

[479] Fix | Delete

return;

[480] Fix | Delete

}

[481] Fix | Delete

[482] Fix | Delete

$chunk_length = strlen($matches[0]);

[483] Fix | Delete

$decoded .= substr($encoded, $chunk_length, $length);

[484] Fix | Delete

$encoded = substr($encoded, $chunk_length + $length + 2);

[485] Fix | Delete

[486] Fix | Delete

// BC for PHP < 8.0: substr() can return bool instead of string

[487] Fix | Delete

$encoded = ($encoded === false) ? '' : $encoded;

[488] Fix | Delete

[489] Fix | Delete

if (trim($encoded) === '0' || empty($encoded)) {

[490] Fix | Delete

$this->state = self::STATE_EMIT;

[491] Fix | Delete

$this->body = $decoded;

[492] Fix | Delete

return;

[493] Fix | Delete

}

[494] Fix | Delete

}

[495] Fix | Delete

}

[496] Fix | Delete

[497] Fix | Delete

/**

[498] Fix | Delete

* Prepare headers (take care of proxies headers)

[499] Fix | Delete