// Copyright (c) 2011 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include "net/base/net_util.h"
#include <unicode/regex.h>
#include <unicode/ucnv.h>
#include <unicode/uidna.h>
#include <unicode/ulocdata.h>
#include <unicode/uniset.h>
#include <unicode/uscript.h>
#include <unicode/uset.h>
#include <algorithm>
#include <iterator>
#include <map>
#include "build/build_config.h"
#if defined(OS_WIN)
#include <windows.h>
#include <winsock2.h>
#include <iphlpapi.h>
#pragma comment(lib, "iphlpapi.lib")
#elif defined(OS_POSIX)
#include <fcntl.h>
#ifndef ANDROID
#include <ifaddrs.h>
#endif
#include <netdb.h>
#include <net/if.h>
#include <netinet/in.h>
#if defined(__BIONIC__) && defined(ANDROID)
#include <netinet/in6.h>
#endif
#endif
#include "base/base64.h"
#include "base/basictypes.h"
#include "base/file_path.h"
#include "base/file_util.h"
#include "base/i18n/file_util_icu.h"
#include "base/i18n/icu_string_conversions.h"
#include "base/i18n/time_formatting.h"
#include "base/json/string_escape.h"
#include "base/logging.h"
#include "base/memory/singleton.h"
#include "base/message_loop.h"
#include "base/metrics/histogram.h"
#include "base/path_service.h"
#include "base/stl_util-inl.h"
#include "base/string_number_conversions.h"
#include "base/string_piece.h"
#include "base/string_split.h"
#include "base/string_tokenizer.h"
#include "base/string_util.h"
#include "base/stringprintf.h"
#include "base/synchronization/lock.h"
#include "base/sys_string_conversions.h"
#include "base/time.h"
#include "base/utf_offset_string_conversions.h"
#include "base/utf_string_conversions.h"
#include "googleurl/src/gurl.h"
#include "googleurl/src/url_canon.h"
#include "googleurl/src/url_canon_ip.h"
#include "googleurl/src/url_parse.h"
#include "grit/net_resources.h"
#include "net/base/dns_util.h"
#include "net/base/escape.h"
#include "net/base/net_module.h"
#if defined(OS_WIN)
#include "net/base/winsock_init.h"
#endif
#include "unicode/datefmt.h"
using base::Time;
namespace net {
namespace {
// what we prepend to get a file URL
static const FilePath::CharType kFileURLPrefix[] =
FILE_PATH_LITERAL("file:///");
// The general list of blocked ports. Will be blocked unless a specific
// protocol overrides it. (Ex: ftp can use ports 20 and 21)
static const int kRestrictedPorts[] = {
1, // tcpmux
7, // echo
9, // discard
11, // systat
13, // daytime
15, // netstat
17, // qotd
19, // chargen
20, // ftp data
21, // ftp access
22, // ssh
23, // telnet
25, // smtp
37, // time
42, // name
43, // nicname
53, // domain
77, // priv-rjs
79, // finger
87, // ttylink
95, // supdup
101, // hostriame
102, // iso-tsap
103, // gppitnp
104, // acr-nema
109, // pop2
110, // pop3
111, // sunrpc
113, // auth
115, // sftp
117, // uucp-path
119, // nntp
123, // NTP
135, // loc-srv /epmap
139, // netbios
143, // imap2
179, // BGP
389, // ldap
465, // smtp+ssl
512, // print / exec
513, // login
514, // shell
515, // printer
526, // tempo
530, // courier
531, // chat
532, // netnews
540, // uucp
556, // remotefs
563, // nntp+ssl
587, // stmp?
601, // ??
636, // ldap+ssl
993, // ldap+ssl
995, // pop3+ssl
2049, // nfs
3659, // apple-sasl / PasswordServer
4045, // lockd
6000, // X11
6665, // Alternate IRC [Apple addition]
6666, // Alternate IRC [Apple addition]
6667, // Standard IRC [Apple addition]
6668, // Alternate IRC [Apple addition]
6669, // Alternate IRC [Apple addition]
0xFFFF, // Used to block all invalid port numbers (see
// third_party/WebKit/Source/WebCore/platform/KURLGoogle.cpp, port())
};
// FTP overrides the following restricted ports.
static const int kAllowedFtpPorts[] = {
21, // ftp data
22, // ssh
};
template<typename STR>
STR GetSpecificHeaderT(const STR& headers, const STR& name) {
// We want to grab the Value from the "Key: Value" pairs in the headers,
// which should look like this (no leading spaces, \n-separated) (we format
// them this way in url_request_inet.cc):
// HTTP/1.1 200 OK\n
// ETag: "6d0b8-947-24f35ec0"\n
// Content-Length: 2375\n
// Content-Type: text/html; charset=UTF-8\n
// Last-Modified: Sun, 03 Sep 2006 04:34:43 GMT\n
if (headers.empty())
return STR();
STR match;
match.push_back('\n');
match.append(name);
match.push_back(':');
typename STR::const_iterator begin =
search(headers.begin(), headers.end(), match.begin(), match.end(),
base::CaseInsensitiveCompareASCII<typename STR::value_type>());
if (begin == headers.end())
return STR();
begin += match.length();
typename STR::const_iterator end = find(begin, headers.end(), '\n');
STR ret;
TrimWhitespace(STR(begin, end), TRIM_ALL, &ret);
return ret;
}
// Similar to Base64Decode. Decodes a Q-encoded string to a sequence
// of bytes. If input is invalid, return false.
bool QPDecode(const std::string& input, std::string* output) {
std::string temp;
temp.reserve(input.size());
std::string::const_iterator it = input.begin();
while (it != input.end()) {
if (*it == '_') {
temp.push_back(' ');
} else if (*it == '=') {
if (input.end() - it < 3) {
return false;
}
if (IsHexDigit(static_cast<unsigned char>(*(it + 1))) &&
IsHexDigit(static_cast<unsigned char>(*(it + 2)))) {
unsigned char ch = HexDigitToInt(*(it + 1)) * 16 +
HexDigitToInt(*(it + 2));
temp.push_back(static_cast<char>(ch));
++it;
++it;
} else {
return false;
}
} else if (0x20 < *it && *it < 0x7F) {
// In a Q-encoded word, only printable ASCII characters
// represent themselves. Besides, space, '=', '_' and '?' are
// not allowed, but they're already filtered out.
DCHECK(*it != 0x3D && *it != 0x5F && *it != 0x3F);
temp.push_back(*it);
} else {
return false;
}
++it;
}
output->swap(temp);
return true;
}
enum RFC2047EncodingType {Q_ENCODING, B_ENCODING};
bool DecodeBQEncoding(const std::string& part, RFC2047EncodingType enc_type,
const std::string& charset, std::string* output) {
std::string decoded;
if (enc_type == B_ENCODING) {
if (!base::Base64Decode(part, &decoded)) {
return false;
}
} else {
if (!QPDecode(part, &decoded)) {
return false;
}
}
UErrorCode err = U_ZERO_ERROR;
UConverter* converter(ucnv_open(charset.c_str(), &err));
if (U_FAILURE(err)) {
return false;
}
// A single byte in a legacy encoding can be expanded to 3 bytes in UTF-8.
// A 'two-byte character' in a legacy encoding can be expanded to 4 bytes
// in UTF-8. Therefore, the expansion ratio is 3 at most.
int length = static_cast<int>(decoded.length());
char* buf = WriteInto(output, length * 3);
length = ucnv_toAlgorithmic(UCNV_UTF8, converter, buf, length * 3,
decoded.data(), length, &err);
ucnv_close(converter);
if (U_FAILURE(err)) {
return false;
}
output->resize(length);
return true;
}
bool DecodeWord(const std::string& encoded_word,
const std::string& referrer_charset,
bool* is_rfc2047,
std::string* output) {
*is_rfc2047 = false;
output->clear();
if (encoded_word.empty())
return true;
if (!IsStringASCII(encoded_word)) {
// Try UTF-8, referrer_charset and the native OS default charset in turn.
if (IsStringUTF8(encoded_word)) {
*output = encoded_word;
} else {
std::wstring wide_output;
if (!referrer_charset.empty() &&
base::CodepageToWide(encoded_word, referrer_charset.c_str(),
base::OnStringConversionError::FAIL,
&wide_output)) {
*output = WideToUTF8(wide_output);
} else {
*output = WideToUTF8(base::SysNativeMBToWide(encoded_word));
}
}
return true;
}
// RFC 2047 : one of encoding methods supported by Firefox and relatively
// widely used by web servers.
// =?charset?<E>?<encoded string>?= where '<E>' is either 'B' or 'Q'.
// We don't care about the length restriction (72 bytes) because
// many web servers generate encoded words longer than the limit.
std::string tmp;
*is_rfc2047 = true;
int part_index = 0;
std::string charset;
StringTokenizer t(encoded_word, "?");
RFC2047EncodingType enc_type = Q_ENCODING;
while (*is_rfc2047 && t.GetNext()) {
std::string part = t.token();
switch (part_index) {
case 0:
if (part != "=") {
*is_rfc2047 = false;
break;
}
++part_index;
break;
case 1:
// Do we need charset validity check here?
charset = part;
++part_index;
break;
case 2:
if (part.size() > 1 ||
part.find_first_of("bBqQ") == std::string::npos) {
*is_rfc2047 = false;
break;
}
if (part[0] == 'b' || part[0] == 'B') {
enc_type = B_ENCODING;
}
++part_index;
break;
case 3:
*is_rfc2047 = DecodeBQEncoding(part, enc_type, charset, &tmp);
if (!*is_rfc2047) {
// Last minute failure. Invalid B/Q encoding. Rather than
// passing it through, return now.
return false;
}
++part_index;
break;
case 4:
if (part != "=") {
// Another last minute failure !
// Likely to be a case of two encoded-words in a row or
// an encoded word followed by a non-encoded word. We can be
// generous, but it does not help much in terms of compatibility,
// I believe. Return immediately.
*is_rfc2047 = false;
return false;
}
++part_index;
break;
default:
*is_rfc2047 = false;
return false;
}
}
if (*is_rfc2047) {
if (*(encoded_word.end() - 1) == '=') {
output->swap(tmp);
return true;
}
// encoded_word ending prematurelly with '?' or extra '?'
*is_rfc2047 = false;
return false;
}
// We're not handling 'especial' characters quoted with '\', but
// it should be Ok because we're not an email client but a
// web browser.
// What IE6/7 does: %-escaped UTF-8.
tmp = UnescapeURLComponent(encoded_word, UnescapeRule::SPACES);
if (IsStringUTF8(tmp)) {
output->swap(tmp);
return true;
// We can try either the OS default charset or 'origin charset' here,
// As far as I can tell, IE does not support it. However, I've seen
// web servers emit %-escaped string in a legacy encoding (usually
// origin charset).
// TODO(jungshik) : Test IE further and consider adding a fallback here.
}
return false;
}
bool DecodeParamValue(const std::string& input,
const std::string& referrer_charset,
std::string* output) {
std::string tmp;
// Tokenize with whitespace characters.
StringTokenizer t(input, " \t\n\r");
t.set_options(StringTokenizer::RETURN_DELIMS);
bool is_previous_token_rfc2047 = true;
while (t.GetNext()) {
if (t.token_is_delim()) {
// If the previous non-delimeter token is not RFC2047-encoded,
// put in a space in its place. Otheriwse, skip over it.
if (!is_previous_token_rfc2047) {
tmp.push_back(' ');
}
continue;
}
// We don't support a single multibyte character split into
// adjacent encoded words. Some broken mail clients emit headers
// with that problem, but most web servers usually encode a filename
// in a single encoded-word. Firefox/Thunderbird do not support
// it, either.
std::string decoded;
if (!DecodeWord(t.token(), referrer_charset, &is_previous_token_rfc2047,
&decoded))
return false;
tmp.append(decoded);
}
output->swap(tmp);
return true;
}
// TODO(mpcomplete): This is a quick and dirty implementation for now. I'm
// sure this doesn't properly handle all (most?) cases.
template<typename STR>
STR GetHeaderParamValueT(const STR& header, const STR& param_name,
QuoteRule::Type quote_rule) {
// This assumes args are formatted exactly like "bla; arg1=value; arg2=value".
typename STR::const_iterator param_begin =
search(header.begin(), header.end(), param_name.begin(), param_name.end(),
base::CaseInsensitiveCompareASCII<typename STR::value_type>());
if (param_begin == header.end())
return STR();
param_begin += param_name.length();
STR whitespace;
whitespace.push_back(' ');
whitespace.push_back('\t');
const typename STR::size_type equals_offset =
header.find_first_not_of(whitespace, param_begin - header.begin());
if (equals_offset == STR::npos || header.at(equals_offset) != '=')
return STR();
param_begin = header.begin() + equals_offset + 1;
if (param_begin == header.end())
return STR();
typename STR::const_iterator param_end;
if (*param_begin == '"' && quote_rule == QuoteRule::REMOVE_OUTER_QUOTES) {
++param_begin; // skip past the quote.
param_end = find(param_begin, header.end(), '"');
// If the closing quote is missing, we will treat the rest of the
// string as the parameter. We can't set |param_end| to the
// location of the separator (';'), since the separator is
// technically quoted. See: http://crbug.com/58840
} else {
param_end = find(param_begin+1, header.end(), ';');
}
return STR(param_begin, param_end);
}
// Does some simple normalization of scripts so we can allow certain scripts
// to exist together.
// TODO(brettw) bug 880223: we should allow some other languages to be
// oombined such as Chinese and Latin. We will probably need a more
// complicated system of language pairs to have more fine-grained control.
UScriptCode NormalizeScript(UScriptCode code) {
switch (code) {
case USCRIPT_KATAKANA:
case USCRIPT_HIRAGANA:
case USCRIPT_KATAKANA_OR_HIRAGANA:
case USCRIPT_HANGUL: // This one is arguable.
return USCRIPT_HAN;
default:
return code;
}
}
bool IsIDNComponentInSingleScript(const char16* str, int str_len) {
UScriptCode first_script = USCRIPT_INVALID_CODE;
bool is_first = true;
int i = 0;
while (i < str_len) {
unsigned code_point;
U16_NEXT(str, i, str_len, code_point);
UErrorCode err = U_ZERO_ERROR;
UScriptCode cur_script = uscript_getScript(code_point, &err);
if (err != U_ZERO_ERROR)
return false; // Report mixed on error.
cur_script = NormalizeScript(cur_script);
// TODO(brettw) We may have to check for USCRIPT_INHERENT as well.
if (is_first && cur_script != USCRIPT_COMMON) {
first_script = cur_script;
is_first = false;
} else {
if (cur_script != USCRIPT_COMMON && cur_script != first_script)
return false;
}
}
return true;
}
// Check if the script of a language can be 'safely' mixed with
// Latin letters in the ASCII range.
bool IsCompatibleWithASCIILetters(const std::string& lang) {
// For now, just list Chinese, Japanese and Korean (positive list).
// An alternative is negative-listing (languages using Greek and
// Cyrillic letters), but it can be more dangerous.
return !lang.substr(0, 2).compare("zh") ||
!lang.substr(0, 2).compare("ja") ||
!lang.substr(0, 2).compare("ko");
}
typedef std::map<std::string, icu::UnicodeSet*> LangToExemplarSetMap;
class LangToExemplarSet {
public:
static LangToExemplarSet* GetInstance() {
return Singleton<LangToExemplarSet>::get();
}
private:
LangToExemplarSetMap map;
LangToExemplarSet() { }
~LangToExemplarSet() {
STLDeleteContainerPairSecondPointers(map.begin(), map.end());
}
friend class Singleton<LangToExemplarSet>;
friend struct DefaultSingletonTraits<LangToExemplarSet>;
friend bool GetExemplarSetForLang(const std::string&, icu::UnicodeSet**);
friend void SetExemplarSetForLang(const std::string&, icu::UnicodeSet*);
DISALLOW_COPY_AND_ASSIGN(LangToExemplarSet);
};
bool GetExemplarSetForLang(const std::string& lang,
icu::UnicodeSet** lang_set) {
const LangToExemplarSetMap& map = LangToExemplarSet::GetInstance()->map;
LangToExemplarSetMap::const_iterator pos = map.find(lang);
if (pos != map.end()) {
*lang_set = pos->second;
return true;
}
return false;
}
void SetExemplarSetForLang(const std::string& lang,
icu::UnicodeSet* lang_set) {
LangToExemplarSetMap& map = LangToExemplarSet::GetInstance()->map;
map.insert(std::make_pair(lang, lang_set));
}
static base::Lock lang_set_lock;
// Returns true if all the characters in component_characters are used by
// the language |lang|.
bool IsComponentCoveredByLang(const icu::UnicodeSet& component_characters,
const std::string& lang) {
static const icu::UnicodeSet kASCIILetters(0x61, 0x7a); // [a-z]
icu::UnicodeSet* lang_set;
// We're called from both the UI thread and the history thread.
{
base::AutoLock lock(lang_set_lock);
if (!GetExemplarSetForLang(lang, &lang_set)) {
UErrorCode status = U_ZERO_ERROR;
ULocaleData* uld = ulocdata_open(lang.c_str(), &status);
// TODO(jungshik) Turn this check on when the ICU data file is
// rebuilt with the minimal subset of locale data for languages
// to which Chrome is not localized but which we offer in the list
// of languages selectable for Accept-Languages. With the rebuilt ICU
// data, ulocdata_open never should fall back to the default locale.
// (issue 2078)
// DCHECK(U_SUCCESS(status) && status != U_USING_DEFAULT_WARNING);
if (U_SUCCESS(status) && status != U_USING_DEFAULT_WARNING) {
lang_set = reinterpret_cast<icu::UnicodeSet *>(
ulocdata_getExemplarSet(uld, NULL, 0,
ULOCDATA_ES_STANDARD, &status));
// If |lang| is compatible with ASCII Latin letters, add them.
if (IsCompatibleWithASCIILetters(lang))
lang_set->addAll(kASCIILetters);
} else {
lang_set = new icu::UnicodeSet(1, 0);
}
lang_set->freeze();
SetExemplarSetForLang(lang, lang_set);
ulocdata_close(uld);
}
}
return !lang_set->isEmpty() && lang_set->containsAll(component_characters);
}
// Returns true if the given Unicode host component is safe to display to the
// user.
bool IsIDNComponentSafe(const char16* str,
int str_len,
const std::wstring& languages) {
// Most common cases (non-IDN) do not reach here so that we don't
// need a fast return path.
// TODO(jungshik) : Check if there's any character inappropriate
// (although allowed) for domain names.
// See http://www.unicode.org/reports/tr39/#IDN_Security_Profiles and
// http://www.unicode.org/reports/tr39/data/xidmodifications.txt
// For now, we borrow the list from Mozilla and tweaked it slightly.
// (e.g. Characters like U+00A0, U+3000, U+3002 are omitted because
// they're gonna be canonicalized to U+0020 and full stop before
// reaching here.)
// The original list is available at
// http://kb.mozillazine.org/Network.IDN.blacklist_chars and
// at http://mxr.mozilla.org/seamonkey/source/modules/libpref/src/init/all.js#703
UErrorCode status = U_ZERO_ERROR;
#ifdef U_WCHAR_IS_UTF16
icu::UnicodeSet dangerous_characters(icu::UnicodeString(
L"[[\\ \u00bc\u00bd\u01c3\u0337\u0338"
L"\u05c3\u05f4\u06d4\u0702\u115f\u1160][\u2000-\u200b]"
L"[\u2024\u2027\u2028\u2029\u2039\u203a\u2044\u205f]"
L"[\u2154-\u2156][\u2159-\u215b][\u215f\u2215\u23ae"
L"\u29f6\u29f8\u2afb\u2afd][\u2ff0-\u2ffb][\u3014"
L"\u3015\u3033\u3164\u321d\u321e\u33ae\u33af\u33c6\u33df\ufe14"
L"\ufe15\ufe3f\ufe5d\ufe5e\ufeff\uff0e\uff06\uff61\uffa0\ufff9]"
L"[\ufffa-\ufffd]]"), status);
DCHECK(U_SUCCESS(status));
icu::RegexMatcher dangerous_patterns(icu::UnicodeString(
// Lone katakana no, so, or n
L"[^\\p{Katakana}][\u30ce\u30f3\u30bd][^\\p{Katakana}]"
// Repeating Japanese accent characters
L"|[\u3099\u309a\u309b\u309c][\u3099\u309a\u309b\u309c]"),
0, status);
#else
icu::UnicodeSet dangerous_characters(icu::UnicodeString(
"[[\\u0020\\u00bc\\u00bd\\u01c3\\u0337\\u0338"
"\\u05c3\\u05f4\\u06d4\\u0702\\u115f\\u1160][\\u2000-\\u200b]"
"[\\u2024\\u2027\\u2028\\u2029\\u2039\\u203a\\u2044\\u205f]"
"[\\u2154-\\u2156][\\u2159-\\u215b][\\u215f\\u2215\\u23ae"
"\\u29f6\\u29f8\\u2afb\\u2afd][\\u2ff0-\\u2ffb][\\u3014"
"\\u3015\\u3033\\u3164\\u321d\\u321e\\u33ae\\u33af\\u33c6\\u33df\\ufe14"
"\\ufe15\\ufe3f\\ufe5d\\ufe5e\\ufeff\\uff0e\\uff06\\uff61\\uffa0\\ufff9]"
"[\\ufffa-\\ufffd]]", -1, US_INV), status);
DCHECK(U_SUCCESS(status));
icu::RegexMatcher dangerous_patterns(icu::UnicodeString(
// Lone katakana no, so, or n
"[^\\p{Katakana}][\\u30ce\\u30f3\u30bd][^\\p{Katakana}]"
// Repeating Japanese accent characters
"|[\\u3099\\u309a\\u309b\\u309c][\\u3099\\u309a\\u309b\\u309c]"),
0, status);
#endif
DCHECK(U_SUCCESS(status));
icu::UnicodeSet component_characters;
icu::UnicodeString component_string(str, str_len);
component_characters.addAll(component_string);
if (dangerous_characters.containsSome(component_characters))
return false;
DCHECK(U_SUCCESS(status));
dangerous_patterns.reset(component_string);
if (dangerous_patterns.find())
return false;
// If the language list is empty, the result is completely determined
// by whether a component is a single script or not. This will block
// even "safe" script mixing cases like <Chinese, Latin-ASCII> that are
// allowed with |languages| (while it blocks Chinese + Latin letters with
// an accent as should be the case), but we want to err on the safe side
// when |languages| is empty.
if (languages.empty())
return IsIDNComponentInSingleScript(str, str_len);
// |common_characters| is made up of ASCII numbers, hyphen, plus and
// underscore that are used across scripts and allowed in domain names.
// (sync'd with characters allowed in url_canon_host with square
// brackets excluded.) See kHostCharLookup[] array in url_canon_host.cc.
icu::UnicodeSet common_characters(UNICODE_STRING_SIMPLE("[[0-9]\\-_+\\ ]"),
status);
DCHECK(U_SUCCESS(status));
// Subtract common characters because they're always allowed so that
// we just have to check if a language-specific set contains
// the remainder.
component_characters.removeAll(common_characters);
std::string languages_list(WideToASCII(languages));
StringTokenizer t(languages_list, ",");
while (t.GetNext()) {
if (IsComponentCoveredByLang(component_characters, t.token()))
return true;
}
return false;
}
// Converts one component of a host (between dots) to IDN if safe. The result
// will be APPENDED to the given output string and will be the same as the input
// if it is not IDN or the IDN is unsafe to display. Returns whether any
// conversion was performed.
bool IDNToUnicodeOneComponent(const char16* comp,
size_t comp_len,
const std::wstring& languages,
string16* out) {
DCHECK(out);
if (comp_len == 0)
return false;
// Only transform if the input can be an IDN component.
static const char16 kIdnPrefix[] = {'x', 'n', '-', '-'};
if ((comp_len > arraysize(kIdnPrefix)) &&
!memcmp(comp, kIdnPrefix, arraysize(kIdnPrefix) * sizeof(char16))) {
// Repeatedly expand the output string until it's big enough. It looks like
// ICU will return the required size of the buffer, but that's not
// documented, so we'll just grow by 2x. This should be rare and is not on a
// critical path.
size_t original_length = out->length();
for (int extra_space = 64; ; extra_space *= 2) {
UErrorCode status = U_ZERO_ERROR;
out->resize(out->length() + extra_space);
int output_chars = uidna_IDNToUnicode(comp,
static_cast<int32_t>(comp_len), &(*out)[original_length], extra_space,
UIDNA_DEFAULT, NULL, &status);
if (status == U_ZERO_ERROR) {
// Converted successfully.
out->resize(original_length + output_chars);
if (IsIDNComponentSafe(out->data() + original_length, output_chars,
languages))
return true;
}
if (status != U_BUFFER_OVERFLOW_ERROR)
break;
}
// Failed, revert back to original string.
out->resize(original_length);
}
// We get here with no IDN or on error, in which case we just append the
// literal input.
out->append(comp, comp_len);
return false;
}
struct SubtractFromOffset {
explicit SubtractFromOffset(size_t amount)
: amount(amount) {}
void operator()(size_t& offset) {
if (offset != std::wstring::npos) {
if (offset >= amount)
offset -= amount;
else
offset = std::wstring::npos;
}
}
size_t amount;
};
struct AddToOffset {
explicit AddToOffset(size_t amount)
: amount(amount) {}
void operator()(size_t& offset) {
if (offset != std::wstring::npos)
offset += amount;
}
size_t amount;
};
std::vector<size_t> OffsetsIntoSection(
std::vector<size_t>* offsets_for_adjustment,
size_t section_begin) {
std::vector<size_t> offsets_into_section;
if (offsets_for_adjustment) {
std::transform(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
std::back_inserter(offsets_into_section),
ClampComponentOffset(section_begin));
std::for_each(offsets_into_section.begin(), offsets_into_section.end(),
SubtractFromOffset(section_begin));
}
return offsets_into_section;
}
void ApplySectionAdjustments(const std::vector<size_t>& offsets_into_section,
std::vector<size_t>* offsets_for_adjustment,
size_t old_section_len,
size_t new_section_len,
size_t section_begin) {
if (offsets_for_adjustment) {
DCHECK_EQ(offsets_for_adjustment->size(), offsets_into_section.size());
std::vector<size_t>::const_iterator host_offsets_iter =
offsets_into_section.begin();
for (std::vector<size_t>::iterator offsets_iter =
offsets_for_adjustment->begin();
offsets_iter != offsets_for_adjustment->end();
++offsets_iter, ++host_offsets_iter) {
size_t offset = *offsets_iter;
if (offset == std::wstring::npos || offset < section_begin) {
// The offset is before the host section so leave it as is.
continue;
}
if (offset >= section_begin + old_section_len) {
// The offset is after the host section so adjust by host length delta.
offset += new_section_len - old_section_len;
} else if (*host_offsets_iter != std::wstring::npos) {
// The offset is within the host and valid so adjust by the host
// reformatting offsets results.
offset = section_begin + *host_offsets_iter;
} else {
// The offset is invalid.
offset = std::wstring::npos;
}
*offsets_iter = offset;
}
}
}
// If |component| is valid, its begin is incremented by |delta|.
void AdjustComponent(int delta, url_parse::Component* component) {
if (!component->is_valid())
return;
DCHECK(delta >= 0 || component->begin >= -delta);
component->begin += delta;
}
// Adjusts all the components of |parsed| by |delta|, except for the scheme.
void AdjustComponents(int delta, url_parse::Parsed* parsed) {
AdjustComponent(delta, &(parsed->username));
AdjustComponent(delta, &(parsed->password));
AdjustComponent(delta, &(parsed->host));
AdjustComponent(delta, &(parsed->port));
AdjustComponent(delta, &(parsed->path));
AdjustComponent(delta, &(parsed->query));
AdjustComponent(delta, &(parsed->ref));
}
std::wstring FormatUrlInternal(const GURL& url,
const std::wstring& languages,
FormatUrlTypes format_types,
UnescapeRule::Type unescape_rules,
url_parse::Parsed* new_parsed,
size_t* prefix_end,
std::vector<size_t>* offsets_for_adjustment);
// Helper for FormatUrl()/FormatUrlInternal().
std::wstring FormatViewSourceUrl(const GURL& url,
const std::wstring& languages,
FormatUrlTypes format_types,
UnescapeRule::Type unescape_rules,
url_parse::Parsed* new_parsed,
size_t* prefix_end,
std::vector<size_t>* offsets_for_adjustment) {
DCHECK(new_parsed);
DCHECK(offsets_for_adjustment);
const wchar_t* const kWideViewSource = L"view-source:";
const size_t kViewSourceLengthPlus1 = 12;
std::vector<size_t> saved_offsets(*offsets_for_adjustment);
GURL real_url(url.possibly_invalid_spec().substr(kViewSourceLengthPlus1));
// Clamp the offsets to the source area.
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
SubtractFromOffset(kViewSourceLengthPlus1));
std::wstring result = FormatUrlInternal(real_url, languages, format_types,
unescape_rules, new_parsed, prefix_end, offsets_for_adjustment);
result.insert(0, kWideViewSource);
// Adjust position values.
if (new_parsed->scheme.is_nonempty()) {
// Assume "view-source:real-scheme" as a scheme.
new_parsed->scheme.len += kViewSourceLengthPlus1;
} else {
new_parsed->scheme.begin = 0;
new_parsed->scheme.len = kViewSourceLengthPlus1 - 1;
}
AdjustComponents(kViewSourceLengthPlus1, new_parsed);
if (prefix_end)
*prefix_end += kViewSourceLengthPlus1;
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
AddToOffset(kViewSourceLengthPlus1));
// Restore all offsets which were not affected by FormatUrlInternal.
DCHECK_EQ(saved_offsets.size(), offsets_for_adjustment->size());
for (size_t i = 0; i < saved_offsets.size(); ++i) {
if (saved_offsets[i] < kViewSourceLengthPlus1)
(*offsets_for_adjustment)[i] = saved_offsets[i];
}
return result;
}
// Appends the substring |in_component| inside of the URL |spec| to |output|,
// and the resulting range will be filled into |out_component|. |unescape_rules|
// defines how to clean the URL for human readability. |offsets_for_adjustment|
// is an array of offsets into |output| each of which will be adjusted based on
// how it maps to the component being converted; if it is less than
// output->length(), it will be untouched, and if it is greater than
// output->length() + in_component.len it will be adjusted by the difference in
// lengths between the input and output components. Otherwise it points into
// the component being converted, and is adjusted to point to the same logical
// place in |output|. |offsets_for_adjustment| may not be NULL.
void AppendFormattedComponent(const std::string& spec,
const url_parse::Component& in_component,
UnescapeRule::Type unescape_rules,
std::wstring* output,
url_parse::Component* out_component,
std::vector<size_t>* offsets_for_adjustment) {
DCHECK(output);
DCHECK(offsets_for_adjustment);
if (in_component.is_nonempty()) {
size_t component_begin = output->length();
out_component->begin = static_cast<int>(component_begin);
// Compose a list of offsets within the component area.
std::vector<size_t> offsets_into_component =
OffsetsIntoSection(offsets_for_adjustment, component_begin);
if (unescape_rules == UnescapeRule::NONE) {
output->append(UTF8ToWideAndAdjustOffsets(
spec.substr(in_component.begin, in_component.len),
&offsets_into_component));
} else {
output->append(UTF16ToWideHack(
UnescapeAndDecodeUTF8URLComponentWithOffsets(
spec.substr(in_component.begin, in_component.len), unescape_rules,
&offsets_into_component)));
}
size_t new_component_len = output->length() - component_begin;
out_component->len = static_cast<int>(new_component_len);
// Apply offset adjustments.
size_t old_component_len = static_cast<size_t>(in_component.len);
ApplySectionAdjustments(offsets_into_component, offsets_for_adjustment,
old_component_len, new_component_len, component_begin);
} else {
out_component->reset();
}
}
// TODO(viettrungluu): This is really the old-fashioned version, made internal.
// I need to really convert |FormatUrl()|.
std::wstring FormatUrlInternal(const GURL& url,
const std::wstring& languages,
FormatUrlTypes format_types,
UnescapeRule::Type unescape_rules,
url_parse::Parsed* new_parsed,
size_t* prefix_end,
std::vector<size_t>* offsets_for_adjustment) {
url_parse::Parsed parsed_temp;
if (!new_parsed)
new_parsed = &parsed_temp;
else
*new_parsed = url_parse::Parsed();
std::vector<size_t> offsets_temp;
if (!offsets_for_adjustment)
offsets_for_adjustment = &offsets_temp;
std::wstring url_string;
// Check for empty URLs or 0 available text width.
if (url.is_empty()) {
if (prefix_end)
*prefix_end = 0;
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
LimitOffset<std::wstring>(0));
return url_string;
}
// Special handling for view-source:. Don't use chrome::kViewSourceScheme
// because this library shouldn't depend on chrome.
const char* const kViewSource = "view-source";
// Reject "view-source:view-source:..." to avoid deep recursion.
const char* const kViewSourceTwice = "view-source:view-source:";
if (url.SchemeIs(kViewSource) &&
!StartsWithASCII(url.possibly_invalid_spec(), kViewSourceTwice, false)) {
return FormatViewSourceUrl(url, languages, format_types,
unescape_rules, new_parsed, prefix_end, offsets_for_adjustment);
}
// We handle both valid and invalid URLs (this will give us the spec
// regardless of validity).
const std::string& spec = url.possibly_invalid_spec();
const url_parse::Parsed& parsed = url.parsed_for_possibly_invalid_spec();
size_t spec_length = spec.length();
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
LimitOffset<std::wstring>(spec_length));
// Copy everything before the username (the scheme and the separators.)
// These are ASCII.
url_string.insert(url_string.end(), spec.begin(),
spec.begin() + parsed.CountCharactersBefore(url_parse::Parsed::USERNAME,
true));
const wchar_t kHTTP[] = L"http://";
const char kFTP[] = "ftp.";
// URLFixerUpper::FixupURL() treats "ftp.foo.com" as ftp://ftp.foo.com. This
// means that if we trim "http://" off a URL whose host starts with "ftp." and
// the user inputs this into any field subject to fixup (which is basically
// all input fields), the meaning would be changed. (In fact, often the
// formatted URL is directly pre-filled into an input field.) For this reason
// we avoid stripping "http://" in this case.
bool omit_http =
(format_types & kFormatUrlOmitHTTP) && (url_string == kHTTP) &&
(url.host().compare(0, arraysize(kFTP) - 1, kFTP) != 0);
new_parsed->scheme = parsed.scheme;
if ((format_types & kFormatUrlOmitUsernamePassword) != 0) {
// Remove the username and password fields. We don't want to display those
// to the user since they can be used for attacks,
// e.g. "http://google.com:search@evil.ru/"
new_parsed->username.reset();
new_parsed->password.reset();
// Update the offsets based on removed username and/or password.
if (!offsets_for_adjustment->empty() &&
(parsed.username.is_nonempty() || parsed.password.is_nonempty())) {
AdjustOffset::Adjustments adjustments;
if (parsed.username.is_nonempty() && parsed.password.is_nonempty()) {
// The seeming off-by-one and off-by-two in these first two lines are to
// account for the ':' after the username and '@' after the password.
adjustments.push_back(AdjustOffset::Adjustment(
static_cast<size_t>(parsed.username.begin),
static_cast<size_t>(parsed.username.len + parsed.password.len +
2), 0));
} else {
const url_parse::Component* nonempty_component =
parsed.username.is_nonempty() ? &parsed.username : &parsed.password;
// The seeming off-by-one in below is to account for the '@' after the
// username/password.
adjustments.push_back(AdjustOffset::Adjustment(
static_cast<size_t>(nonempty_component->begin),
static_cast<size_t>(nonempty_component->len + 1), 0));
}
// Make offset adjustment.
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
AdjustOffset(adjustments));
}
} else {
AppendFormattedComponent(spec, parsed.username, unescape_rules, &url_string,
&new_parsed->username, offsets_for_adjustment);
if (parsed.password.is_valid())
url_string.push_back(':');
AppendFormattedComponent(spec, parsed.password, unescape_rules, &url_string,
&new_parsed->password, offsets_for_adjustment);
if (parsed.username.is_valid() || parsed.password.is_valid())
url_string.push_back('@');
}
if (prefix_end)
*prefix_end = static_cast<size_t>(url_string.length());
AppendFormattedHostWithOffsets(url, languages, &url_string, new_parsed,
offsets_for_adjustment);
// Port.
if (parsed.port.is_nonempty()) {
url_string.push_back(':');
new_parsed->port.begin = url_string.length();
url_string.insert(url_string.end(),
spec.begin() + parsed.port.begin,
spec.begin() + parsed.port.end());
new_parsed->port.len = url_string.length() - new_parsed->port.begin;
} else {
new_parsed->port.reset();
}
// Path and query both get the same general unescape & convert treatment.
if (!(format_types & kFormatUrlOmitTrailingSlashOnBareHostname) ||
!CanStripTrailingSlash(url)) {
AppendFormattedComponent(spec, parsed.path, unescape_rules, &url_string,
&new_parsed->path, offsets_for_adjustment);
}
if (parsed.query.is_valid())
url_string.push_back('?');
AppendFormattedComponent(spec, parsed.query, unescape_rules, &url_string,
&new_parsed->query, offsets_for_adjustment);
// Reference is stored in valid, unescaped UTF-8, so we can just convert.
if (parsed.ref.is_valid()) {
url_string.push_back('#');
size_t ref_begin = url_string.length();
new_parsed->ref.begin = static_cast<int>(ref_begin);
// Compose a list of offsets within the section.
std::vector<size_t> offsets_into_ref =
OffsetsIntoSection(offsets_for_adjustment, ref_begin);
if (parsed.ref.len > 0) {
url_string.append(UTF8ToWideAndAdjustOffsets(spec.substr(parsed.ref.begin,
parsed.ref.len),
&offsets_into_ref));
}
size_t old_ref_len = static_cast<size_t>(parsed.ref.len);
size_t new_ref_len = url_string.length() - new_parsed->ref.begin;
new_parsed->ref.len = static_cast<int>(new_ref_len);
// Apply offset adjustments.
ApplySectionAdjustments(offsets_into_ref, offsets_for_adjustment,
old_ref_len, new_ref_len, ref_begin);
}
// If we need to strip out http do it after the fact. This way we don't need
// to worry about how offset_for_adjustment is interpreted.
const size_t kHTTPSize = arraysize(kHTTP) - 1;
if (omit_http && !url_string.compare(0, kHTTPSize, kHTTP)) {
url_string = url_string.substr(kHTTPSize);
AdjustOffset::Adjustments adjustments;
adjustments.push_back(AdjustOffset::Adjustment(0, kHTTPSize, 0));
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
AdjustOffset(adjustments));
if (prefix_end)
*prefix_end -= kHTTPSize;
// Adjust new_parsed.
DCHECK(new_parsed->scheme.is_valid());
int delta = -(new_parsed->scheme.len + 3); // +3 for ://.
new_parsed->scheme.reset();
AdjustComponents(delta, new_parsed);
}
return url_string;
}
char* do_strdup(const char* src) {
#if defined(OS_WIN)
return _strdup(src);
#else
return strdup(src);
#endif
}
} // namespace
const FormatUrlType kFormatUrlOmitNothing = 0;
const FormatUrlType kFormatUrlOmitUsernamePassword = 1 << 0;
const FormatUrlType kFormatUrlOmitHTTP = 1 << 1;
const FormatUrlType kFormatUrlOmitTrailingSlashOnBareHostname = 1 << 2;
const FormatUrlType kFormatUrlOmitAll = kFormatUrlOmitUsernamePassword |
kFormatUrlOmitHTTP | kFormatUrlOmitTrailingSlashOnBareHostname;
// TODO(viettrungluu): We don't want non-POD globals; change this.
std::multiset<int> explicitly_allowed_ports;
GURL FilePathToFileURL(const FilePath& path) {
// Produce a URL like "file:///C:/foo" for a regular file, or
// "file://///server/path" for UNC. The URL canonicalizer will fix up the
// latter case to be the canonical UNC form: "file://server/path"
FilePath::StringType url_string(kFileURLPrefix);
url_string.append(path.value());
// Now do replacement of some characters. Since we assume the input is a
// literal filename, anything the URL parser might consider special should
// be escaped here.
// must be the first substitution since others will introduce percents as the
// escape character
ReplaceSubstringsAfterOffset(&url_string, 0,
FILE_PATH_LITERAL("%"), FILE_PATH_LITERAL("%25"));
// semicolon is supposed to be some kind of separator according to RFC 2396
ReplaceSubstringsAfterOffset(&url_string, 0,
FILE_PATH_LITERAL(";"), FILE_PATH_LITERAL("%3B"));
ReplaceSubstringsAfterOffset(&url_string, 0,
FILE_PATH_LITERAL("#"), FILE_PATH_LITERAL("%23"));
#if defined(OS_POSIX)
ReplaceSubstringsAfterOffset(&url_string, 0,
FILE_PATH_LITERAL("\\"), FILE_PATH_LITERAL("%5C"));
#endif
return GURL(url_string);
}
std::wstring GetSpecificHeader(const std::wstring& headers,
const std::wstring& name) {
return GetSpecificHeaderT(headers, name);
}
std::string GetSpecificHeader(const std::string& headers,
const std::string& name) {
return GetSpecificHeaderT(headers, name);
}
bool DecodeCharset(const std::string& input,
std::string* decoded_charset,
std::string* value) {
StringTokenizer t(input, "'");
t.set_options(StringTokenizer::RETURN_DELIMS);
std::string temp_charset;
std::string temp_value;
int numDelimsSeen = 0;
while (t.GetNext()) {
if (t.token_is_delim()) {
++numDelimsSeen;
continue;
} else {
switch (numDelimsSeen) {
case 0:
temp_charset = t.token();
break;
case 1:
// Language is ignored.
break;
case 2:
temp_value = t.token();
break;
default:
return false;
}
}
}
if (numDelimsSeen != 2)
return false;
if (temp_charset.empty() || temp_value.empty())
return false;
decoded_charset->swap(temp_charset);
value->swap(temp_value);
return true;
}
std::string GetFileNameFromCD(const std::string& header,
const std::string& referrer_charset) {
std::string decoded;
std::string param_value = GetHeaderParamValue(header, "filename*",
QuoteRule::KEEP_OUTER_QUOTES);
if (!param_value.empty()) {
if (param_value.find('"') == std::string::npos) {
std::string charset;
std::string value;
if (DecodeCharset(param_value, &charset, &value)) {
// RFC 5987 value should be ASCII-only.
if (!IsStringASCII(value))
return std::string();
std::string tmp = UnescapeURLComponent(
value,
UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS);
if (base::ConvertToUtf8AndNormalize(tmp, charset, &decoded))
return decoded;
}
}
}
param_value = GetHeaderParamValue(header, "filename",
QuoteRule::REMOVE_OUTER_QUOTES);
if (param_value.empty()) {
// Some servers use 'name' parameter.
param_value = GetHeaderParamValue(header, "name",
QuoteRule::REMOVE_OUTER_QUOTES);
}
if (param_value.empty())
return std::string();
if (DecodeParamValue(param_value, referrer_charset, &decoded))
return decoded;
return std::string();
}
std::wstring GetHeaderParamValue(const std::wstring& field,
const std::wstring& param_name,
QuoteRule::Type quote_rule) {
return GetHeaderParamValueT(field, param_name, quote_rule);
}
std::string GetHeaderParamValue(const std::string& field,
const std::string& param_name,
QuoteRule::Type quote_rule) {
return GetHeaderParamValueT(field, param_name, quote_rule);
}
// TODO(brettw) bug 734373: check the scripts for each host component and
// don't un-IDN-ize if there is more than one. Alternatively, only IDN for
// scripts that the user has installed. For now, just put the entire
// path through IDN. Maybe this feature can be implemented in ICU itself?
//
// We may want to skip this step in the case of file URLs to allow unicode
// UNC hostnames regardless of encodings.
std::wstring IDNToUnicodeWithOffsets(
const char* host,
size_t host_len,
const std::wstring& languages,
std::vector<size_t>* offsets_for_adjustment) {
// Convert the ASCII input to a wide string for ICU.
string16 input16;
input16.reserve(host_len);
input16.insert(input16.end(), host, host + host_len);
// Do each component of the host separately, since we enforce script matching
// on a per-component basis.
AdjustOffset::Adjustments adjustments;
string16 out16;
for (size_t component_start = 0, component_end;
component_start < input16.length();
component_start = component_end + 1) {
// Find the end of the component.
component_end = input16.find('.', component_start);
if (component_end == string16::npos)
component_end = input16.length(); // For getting the last component.
size_t component_length = component_end - component_start;
size_t new_component_start = out16.length();
bool converted_idn = false;
if (component_end > component_start) {
// Add the substring that we just found.
converted_idn = IDNToUnicodeOneComponent(input16.data() + component_start,
component_length, languages, &out16);
}
size_t new_component_length = out16.length() - new_component_start;
if (converted_idn && offsets_for_adjustment) {
adjustments.push_back(AdjustOffset::Adjustment(
component_start, component_length, new_component_length));
}
// Need to add the dot we just found (if we found one).
if (component_end < input16.length())
out16.push_back('.');
}
// Make offset adjustment.
if (offsets_for_adjustment && !adjustments.empty()) {
std::for_each(offsets_for_adjustment->begin(),
offsets_for_adjustment->end(),
AdjustOffset(adjustments));
}
return UTF16ToWideAndAdjustOffsets(out16, offsets_for_adjustment);
}
std::wstring IDNToUnicode(const char* host,
size_t host_len,
const std::wstring& languages,
size_t* offset_for_adjustment) {
std::vector<size_t> offsets;
if (offset_for_adjustment)
offsets.push_back(*offset_for_adjustment);
std::wstring result =
IDNToUnicodeWithOffsets(host, host_len, languages, &offsets);
if (offset_for_adjustment)
*offset_for_adjustment = offsets[0];
return result;
}
std::string CanonicalizeHost(const std::string& host,
url_canon::CanonHostInfo* host_info) {
// Try to canonicalize the host.
const url_parse::Component raw_host_component(
0, static_cast<int>(host.length()));
std::string canon_host;
url_canon::StdStringCanonOutput canon_host_output(&canon_host);
url_canon::CanonicalizeHostVerbose(host.c_str(), raw_host_component,
&canon_host_output, host_info);
if (host_info->out_host.is_nonempty() &&
host_info->family != url_canon::CanonHostInfo::BROKEN) {
// Success! Assert that there's no extra garbage.
canon_host_output.Complete();
DCHECK_EQ(host_info->out_host.len, static_cast<int>(canon_host.length()));
} else {
// Empty host, or canonicalization failed. We'll return empty.
canon_host.clear();
}
return canon_host;
}
std::string CanonicalizeHost(const std::wstring& host,
url_canon::CanonHostInfo* host_info) {
std::string converted_host;
WideToUTF8(host.c_str(), host.length(), &converted_host);
return CanonicalizeHost(converted_host, host_info);
}
std::string GetDirectoryListingHeader(const string16& title) {
static const base::StringPiece header(
NetModule::GetResource(IDR_DIR_HEADER_HTML));
// This can be null in unit tests.
DLOG_IF(WARNING, header.empty()) <<
"Missing resource: directory listing header";
std::string result;
if (!header.empty())
result.assign(header.data(), header.size());
result.append("<script>start(");
base::JsonDoubleQuote(title, true, &result);
result.append(");</script>\n");
return result;
}
inline bool IsHostCharAlpha(char c) {
// We can just check lowercase because uppercase characters have already been
// normalized.
return (c >= 'a') && (c <= 'z');
}
inline bool IsHostCharDigit(char c) {
return (c >= '0') && (c <= '9');
}
bool IsCanonicalizedHostCompliant(const std::string& host,
const std::string& desired_tld) {
if (host.empty())
return false;
bool in_component = false;
bool most_recent_component_started_alpha = false;
bool last_char_was_hyphen_or_underscore = false;
for (std::string::const_iterator i(host.begin()); i != host.end(); ++i) {
const char c = *i;
if (!in_component) {
most_recent_component_started_alpha = IsHostCharAlpha(c);
if (!most_recent_component_started_alpha && !IsHostCharDigit(c))
return false;
in_component = true;
} else {
if (c == '.') {
if (last_char_was_hyphen_or_underscore)
return false;
in_component = false;
} else if (IsHostCharAlpha(c) || IsHostCharDigit(c)) {
last_char_was_hyphen_or_underscore = false;
} else if ((c == '-') || (c == '_')) {
last_char_was_hyphen_or_underscore = true;
} else {
return false;
}
}
}
return most_recent_component_started_alpha ||
(!desired_tld.empty() && IsHostCharAlpha(desired_tld[0]));
}
std::string GetDirectoryListingEntry(const string16& name,
const std::string& raw_bytes,
bool is_dir,
int64 size,
Time modified) {
std::string result;
result.append("<script>addRow(");
base::JsonDoubleQuote(name, true, &result);
result.append(",");
if (raw_bytes.empty()) {
base::JsonDoubleQuote(EscapePath(UTF16ToUTF8(name)),
true, &result);
} else {
base::JsonDoubleQuote(EscapePath(raw_bytes), true, &result);
}
if (is_dir) {
result.append(",1,");
} else {
result.append(",0,");
}
base::JsonDoubleQuote(
FormatBytes(size, GetByteDisplayUnits(size), true),
true,
&result);
result.append(",");
string16 modified_str;
// |modified| can be NULL in FTP listings.
if (!modified.is_null()) {
modified_str = base::TimeFormatShortDateAndTime(modified);
}
base::JsonDoubleQuote(modified_str, true, &result);
result.append(");</script>\n");
return result;
}
string16 StripWWW(const string16& text) {
const string16 www(ASCIIToUTF16("www."));
return StartsWith(text, www, true) ? text.substr(www.length()) : text;
}
string16 GetSuggestedFilename(const GURL& url,
const std::string& content_disposition,
const std::string& referrer_charset,
const string16& default_name) {
// TODO: this function to be updated to match the httpbis recommendations.
// Talk to abarth for the latest news.
// We don't translate this fallback string, "download". If localization is
// needed, the caller should provide localized fallback default_name.
static const char* kFinalFallbackName = "download";
// about: and data: URLs don't have file names, but esp. data: URLs may
// contain parts that look like ones (i.e., contain a slash).
// Therefore we don't attempt to divine a file name out of them.
if (url.SchemeIs("about") || url.SchemeIs("data")) {
return default_name.empty() ? ASCIIToUTF16(kFinalFallbackName)
: default_name;
}
std::string filename = GetFileNameFromCD(content_disposition,
referrer_charset);
if (!filename.empty()) {
// Replace any path information the server may have sent, by changing
// path separators with underscores.
ReplaceSubstringsAfterOffset(&filename, 0, "/", "_");
ReplaceSubstringsAfterOffset(&filename, 0, "\\", "_");
// Next, remove "." from the beginning and end of the file name to avoid
// tricks with hidden files, "..", and "."
TrimString(filename, ".", &filename);
}
if (filename.empty()) {
if (url.is_valid()) {
const std::string unescaped_url_filename = UnescapeURLComponent(
url.ExtractFileName(),
UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS);
// The URL's path should be escaped UTF-8, but may not be.
std::string decoded_filename = unescaped_url_filename;
if (!IsStringASCII(decoded_filename)) {
bool ignore;
// TODO(jshin): this is probably not robust enough. To be sure, we
// need encoding detection.
DecodeWord(unescaped_url_filename, referrer_charset, &ignore,
&decoded_filename);
}
filename = decoded_filename;
}
}
#if defined(OS_WIN)
{ // Handle CreateFile() stripping trailing dots and spaces on filenames
// http://support.microsoft.com/kb/115827
std::string::size_type pos = filename.find_last_not_of(" .");
if (pos == std::string::npos)
filename.resize(0);
else
filename.resize(++pos);
}
#endif
// Trim '.' once more.
TrimString(filename, ".", &filename);
// If there's no filename or it gets trimed to be empty, use
// the URL hostname or default_name
if (filename.empty()) {
if (!default_name.empty()) {
return default_name;
} else if (url.is_valid()) {
// Some schemes (e.g. file) do not have a hostname. Even though it's
// not likely to reach here, let's hardcode the last fallback name.
// TODO(jungshik) : Decode a 'punycoded' IDN hostname. (bug 1264451)
filename = url.host().empty() ? kFinalFallbackName : url.host();
} else {
NOTREACHED();
}
}
#if defined(OS_WIN)
string16 path = UTF8ToUTF16(filename);
file_util::ReplaceIllegalCharactersInPath(&path, '-');
return path;
#else
std::string path = filename;
file_util::ReplaceIllegalCharactersInPath(&path, '-');
return UTF8ToUTF16(path);
#endif
}
bool IsPortAllowedByDefault(int port) {
int array_size = arraysize(kRestrictedPorts);
for (int i = 0; i < array_size; i++) {
if (kRestrictedPorts[i] == port) {
return false;
}
}
return true;
}
bool IsPortAllowedByFtp(int port) {
int array_size = arraysize(kAllowedFtpPorts);
for (int i = 0; i < array_size; i++) {
if (kAllowedFtpPorts[i] == port) {
return true;
}
}
// Port not explicitly allowed by FTP, so return the default restrictions.
return IsPortAllowedByDefault(port);
}
bool IsPortAllowedByOverride(int port) {
if (explicitly_allowed_ports.empty())
return false;
return explicitly_allowed_ports.count(port) > 0;
}
int SetNonBlocking(int fd) {
#if defined(OS_WIN)
unsigned long no_block = 1;
return ioctlsocket(fd, FIONBIO, &no_block);
#elif defined(OS_POSIX)
int flags = fcntl(fd, F_GETFL, 0);
if (-1 == flags)
return flags;
return fcntl(fd, F_SETFL, flags | O_NONBLOCK);
#endif
}
bool ParseHostAndPort(std::string::const_iterator host_and_port_begin,
std::string::const_iterator host_and_port_end,
std::string* host,
int* port) {
if (host_and_port_begin >= host_and_port_end)
return false;
// When using url_parse, we use char*.
const char* auth_begin = &(*host_and_port_begin);
int auth_len = host_and_port_end - host_and_port_begin;
url_parse::Component auth_component(0, auth_len);
url_parse::Component username_component;
url_parse::Component password_component;
url_parse::Component hostname_component;
url_parse::Component port_component;
url_parse::ParseAuthority(auth_begin, auth_component, &username_component,
&password_component, &hostname_component, &port_component);
// There shouldn't be a username/password.
if (username_component.is_valid() || password_component.is_valid())
return false;
if (!hostname_component.is_nonempty())
return false; // Failed parsing.
int parsed_port_number = -1;
if (port_component.is_nonempty()) {
parsed_port_number = url_parse::ParsePort(auth_begin, port_component);
// If parsing failed, port_number will be either PORT_INVALID or
// PORT_UNSPECIFIED, both of which are negative.
if (parsed_port_number < 0)
return false; // Failed parsing the port number.
}
if (port_component.len == 0)
return false; // Reject inputs like "foo:"
// Pass results back to caller.
host->assign(auth_begin + hostname_component.begin, hostname_component.len);
*port = parsed_port_number;
return true; // Success.
}
bool ParseHostAndPort(const std::string& host_and_port,
std::string* host,
int* port) {
return ParseHostAndPort(
host_and_port.begin(), host_and_port.end(), host, port);
}
std::string GetHostAndPort(const GURL& url) {
// For IPv6 literals, GURL::host() already includes the brackets so it is
// safe to just append a colon.
return base::StringPrintf("%s:%d", url.host().c_str(),
url.EffectiveIntPort());
}
std::string GetHostAndOptionalPort(const GURL& url) {
// For IPv6 literals, GURL::host() already includes the brackets
// so it is safe to just append a colon.
if (url.has_port())
return base::StringPrintf("%s:%s", url.host().c_str(), url.port().c_str());
return url.host();
}
std::string NetAddressToString(const struct addrinfo* net_address) {
return NetAddressToString(net_address->ai_addr, net_address->ai_addrlen);
}
std::string NetAddressToString(const struct sockaddr* net_address,
socklen_t address_len) {
#if defined(OS_WIN)
EnsureWinsockInit();
#endif
// This buffer is large enough to fit the biggest IPv6 string.
char buffer[INET6_ADDRSTRLEN];
int result = getnameinfo(net_address, address_len, buffer, sizeof(buffer),
NULL, 0, NI_NUMERICHOST);
if (result != 0) {
DVLOG(1) << "getnameinfo() failed with " << result << ": "
<< gai_strerror(result);
buffer[0] = '\0';
}
return std::string(buffer);
}
std::string NetAddressToStringWithPort(const struct addrinfo* net_address) {
return NetAddressToStringWithPort(
net_address->ai_addr, net_address->ai_addrlen);
}
std::string NetAddressToStringWithPort(const struct sockaddr* net_address,
socklen_t address_len) {
std::string ip_address_string = NetAddressToString(net_address, address_len);
if (ip_address_string.empty())
return std::string(); // Failed.
int port = GetPortFromSockaddr(net_address, address_len);
if (ip_address_string.find(':') != std::string::npos) {
// Surround with square brackets to avoid ambiguity.
return base::StringPrintf("[%s]:%d", ip_address_string.c_str(), port);
}
return base::StringPrintf("%s:%d", ip_address_string.c_str(), port);
}
std::string GetHostName() {
#if defined(OS_WIN)
EnsureWinsockInit();
#endif
// Host names are limited to 255 bytes.
char buffer[256];
int result = gethostname(buffer, sizeof(buffer));
if (result != 0) {
DVLOG(1) << "gethostname() failed with " << result;
buffer[0] = '\0';
}
return std::string(buffer);
}
void GetIdentityFromURL(const GURL& url,
string16* username,
string16* password) {
UnescapeRule::Type flags =
UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS;
*username = UnescapeAndDecodeUTF8URLComponent(url.username(), flags, NULL);
*password = UnescapeAndDecodeUTF8URLComponent(url.password(), flags, NULL);
}
std::string GetHostOrSpecFromURL(const GURL& url) {
return url.has_host() ? TrimEndingDot(url.host()) : url.spec();
}
void AppendFormattedHostWithOffsets(
const GURL& url,
const std::wstring& languages,
std::wstring* output,
url_parse::Parsed* new_parsed,
std::vector<size_t>* offsets_for_adjustment) {
DCHECK(output);
const url_parse::Component& host =
url.parsed_for_possibly_invalid_spec().host;
if (host.is_nonempty()) {
// Handle possible IDN in the host name.
size_t host_begin = output->length();
if (new_parsed)
new_parsed->host.begin = static_cast<int>(host_begin);
size_t old_host_len = static_cast<size_t>(host.len);
// Compose a list of offsets within the host area.
std::vector<size_t> offsets_into_host =
OffsetsIntoSection(offsets_for_adjustment, host_begin);
const std::string& spec = url.possibly_invalid_spec();
DCHECK(host.begin >= 0 &&
((spec.length() == 0 && host.begin == 0) ||
host.begin < static_cast<int>(spec.length())));
output->append(IDNToUnicodeWithOffsets(&spec[host.begin], old_host_len,
languages, &offsets_into_host));
size_t new_host_len = output->length() - host_begin;
if (new_parsed)
new_parsed->host.len = static_cast<int>(new_host_len);
// Apply offset adjustments.
ApplySectionAdjustments(offsets_into_host, offsets_for_adjustment,
old_host_len, new_host_len, host_begin);
} else if (new_parsed) {
new_parsed->host.reset();
}
}
void AppendFormattedHost(const GURL& url,
const std::wstring& languages,
std::wstring* output,
url_parse::Parsed* new_parsed,
size_t* offset_for_adjustment) {
std::vector<size_t> offsets;
if (offset_for_adjustment)
offsets.push_back(*offset_for_adjustment);
AppendFormattedHostWithOffsets(url, languages, output, new_parsed, &offsets);
if (offset_for_adjustment)
*offset_for_adjustment = offsets[0];
}
// TODO(viettrungluu): convert the wstring |FormatUrlInternal()|.
string16 FormatUrlWithOffsets(const GURL& url,
const std::string& languages,
FormatUrlTypes format_types,
UnescapeRule::Type unescape_rules,
url_parse::Parsed* new_parsed,
size_t* prefix_end,
std::vector<size_t>* offsets_for_adjustment) {
return WideToUTF16Hack(
FormatUrlInternal(url, ASCIIToWide(languages), format_types,
unescape_rules, new_parsed, prefix_end,
offsets_for_adjustment));
}
string16 FormatUrl(const GURL& url,
const std::string& languages,
FormatUrlTypes format_types,
UnescapeRule::Type unescape_rules,
url_parse::Parsed* new_parsed,
size_t* prefix_end,
size_t* offset_for_adjustment) {
std::vector<size_t> offsets;
if (offset_for_adjustment)
offsets.push_back(*offset_for_adjustment);
string16 result = WideToUTF16Hack(
FormatUrlInternal(url, ASCIIToWide(languages), format_types,
unescape_rules, new_parsed, prefix_end, &offsets));
if (offset_for_adjustment)
*offset_for_adjustment = offsets[0];
return result;
}
bool CanStripTrailingSlash(const GURL& url) {
// Omit the path only for standard, non-file URLs with nothing but "/" after
// the hostname.
return url.IsStandard() && !url.SchemeIsFile() && !url.has_query() &&
!url.has_ref() && url.path() == "/";
}
GURL SimplifyUrlForRequest(const GURL& url) {
DCHECK(url.is_valid());
GURL::Replacements replacements;
replacements.ClearUsername();
replacements.ClearPassword();
replacements.ClearRef();
return url.ReplaceComponents(replacements);
}
// Specifies a comma separated list of port numbers that should be accepted
// despite bans. If the string is invalid no allowed ports are stored.
void SetExplicitlyAllowedPorts(const std::string& allowed_ports) {
if (allowed_ports.empty())
return;
std::multiset<int> ports;
size_t last = 0;
size_t size = allowed_ports.size();
// The comma delimiter.
const std::string::value_type kComma = ',';
// Overflow is still possible for evil user inputs.
for (size_t i = 0; i <= size; ++i) {
// The string should be composed of only digits and commas.
if (i != size && !IsAsciiDigit(allowed_ports[i]) &&
(allowed_ports[i] != kComma))
return;
if (i == size || allowed_ports[i] == kComma) {
if (i > last) {
int port;
base::StringToInt(allowed_ports.begin() + last,
allowed_ports.begin() + i,
&port);
ports.insert(port);
}
last = i + 1;
}
}
explicitly_allowed_ports = ports;
}
ScopedPortException::ScopedPortException(int port) : port_(port) {
explicitly_allowed_ports.insert(port);
}
ScopedPortException::~ScopedPortException() {
std::multiset<int>::iterator it = explicitly_allowed_ports.find(port_);
if (it != explicitly_allowed_ports.end())
explicitly_allowed_ports.erase(it);
else
NOTREACHED();
}
enum IPv6SupportStatus {
IPV6_CANNOT_CREATE_SOCKETS,
IPV6_CAN_CREATE_SOCKETS,
IPV6_GETIFADDRS_FAILED,
IPV6_GLOBAL_ADDRESS_MISSING,
IPV6_GLOBAL_ADDRESS_PRESENT,
IPV6_INTERFACE_ARRAY_TOO_SHORT,
IPV6_SUPPORT_MAX // Bounding values for enumeration.
};
static void IPv6SupportResults(IPv6SupportStatus result) {
static bool run_once = false;
if (!run_once) {
run_once = true;
UMA_HISTOGRAM_ENUMERATION("Net.IPv6Status", result, IPV6_SUPPORT_MAX);
} else {
UMA_HISTOGRAM_ENUMERATION("Net.IPv6Status_retest", result,
IPV6_SUPPORT_MAX);
}
}
// TODO(jar): The following is a simple estimate of IPv6 support. We may need
// to do a test resolution, and a test connection, to REALLY verify support.
// static
bool IPv6Supported() {
#ifdef ANDROID
// Android does not have the ifaddrs.h header
return false;
#elif defined(OS_POSIX)
int test_socket = socket(AF_INET6, SOCK_STREAM, 0);
if (test_socket == -1) {
IPv6SupportResults(IPV6_CANNOT_CREATE_SOCKETS);
return false;
}
close(test_socket);
// Check to see if any interface has a IPv6 address.
struct ifaddrs* interface_addr = NULL;
int rv = getifaddrs(&interface_addr);
if (rv != 0) {
IPv6SupportResults(IPV6_GETIFADDRS_FAILED);
return true; // Don't yet block IPv6.
}
bool found_ipv6 = false;
for (struct ifaddrs* interface = interface_addr;
interface != NULL;
interface = interface->ifa_next) {
if (!(IFF_UP & interface->ifa_flags))
continue;
if (IFF_LOOPBACK & interface->ifa_flags)
continue;
struct sockaddr* addr = interface->ifa_addr;
if (!addr)
continue;
if (addr->sa_family != AF_INET6)
continue;
// Safe cast since this is AF_INET6.
struct sockaddr_in6* addr_in6 =
reinterpret_cast<struct sockaddr_in6*>(addr);
struct in6_addr* sin6_addr = &addr_in6->sin6_addr;
if (IN6_IS_ADDR_LOOPBACK(sin6_addr) || IN6_IS_ADDR_LINKLOCAL(sin6_addr))
continue;
found_ipv6 = true;
break;
}
freeifaddrs(interface_addr);
if (!found_ipv6) {
IPv6SupportResults(IPV6_GLOBAL_ADDRESS_MISSING);
return false;
}
IPv6SupportResults(IPV6_GLOBAL_ADDRESS_PRESENT);
return true;
#elif defined(OS_WIN)
EnsureWinsockInit();
SOCKET test_socket = socket(AF_INET6, SOCK_STREAM, 0);
if (test_socket == INVALID_SOCKET) {
IPv6SupportResults(IPV6_CANNOT_CREATE_SOCKETS);
return false;
}
closesocket(test_socket);
// Check to see if any interface has a IPv6 address.
// The GetAdaptersAddresses MSDN page recommends using a size of 15000 to
// avoid reallocation.
ULONG adapters_size = 15000;
scoped_ptr_malloc<IP_ADAPTER_ADDRESSES> adapters;
ULONG error;
int num_tries = 0;
do {
adapters.reset(
reinterpret_cast<PIP_ADAPTER_ADDRESSES>(malloc(adapters_size)));
// Return only unicast addresses.
error = GetAdaptersAddresses(AF_UNSPEC,
GAA_FLAG_SKIP_ANYCAST |
GAA_FLAG_SKIP_MULTICAST |
GAA_FLAG_SKIP_DNS_SERVER |
GAA_FLAG_SKIP_FRIENDLY_NAME,
NULL, adapters.get(), &adapters_size);
num_tries++;
} while (error == ERROR_BUFFER_OVERFLOW && num_tries <= 3);
if (error == ERROR_NO_DATA) {
IPv6SupportResults(IPV6_GLOBAL_ADDRESS_MISSING);
return false;
}
if (error != ERROR_SUCCESS) {
IPv6SupportResults(IPV6_GETIFADDRS_FAILED);
return true; // Don't yet block IPv6.
}
PIP_ADAPTER_ADDRESSES adapter;
for (adapter = adapters.get(); adapter; adapter = adapter->Next) {
if (adapter->OperStatus != IfOperStatusUp)
continue;
if (adapter->IfType == IF_TYPE_SOFTWARE_LOOPBACK)
continue;
PIP_ADAPTER_UNICAST_ADDRESS unicast_address;
for (unicast_address = adapter->FirstUnicastAddress;
unicast_address;
unicast_address = unicast_address->Next) {
if (unicast_address->Address.lpSockaddr->sa_family != AF_INET6)
continue;
// Safe cast since this is AF_INET6.
struct sockaddr_in6* addr_in6 = reinterpret_cast<struct sockaddr_in6*>(
unicast_address->Address.lpSockaddr);
struct in6_addr* sin6_addr = &addr_in6->sin6_addr;
if (IN6_IS_ADDR_LOOPBACK(sin6_addr) || IN6_IS_ADDR_LINKLOCAL(sin6_addr))
continue;
IPv6SupportResults(IPV6_GLOBAL_ADDRESS_PRESENT);
return true;
}
}
IPv6SupportResults(IPV6_GLOBAL_ADDRESS_MISSING);
return false;
#else
NOTIMPLEMENTED();
return true;
#endif // defined(various platforms)
}
bool HaveOnlyLoopbackAddresses() {
#if defined(ANDROID)
// Android has no <ifaddrs.h>
return false;
#elif defined(OS_POSIX)
struct ifaddrs* interface_addr = NULL;
int rv = getifaddrs(&interface_addr);
if (rv != 0) {
DVLOG(1) << "getifaddrs() failed with errno = " << errno;
return false;
}
bool result = true;
for (struct ifaddrs* interface = interface_addr;
interface != NULL;
interface = interface->ifa_next) {
if (!(IFF_UP & interface->ifa_flags))
continue;
if (IFF_LOOPBACK & interface->ifa_flags)
continue;
const struct sockaddr* addr = interface->ifa_addr;
if (!addr)
continue;
if (addr->sa_family == AF_INET6) {
// Safe cast since this is AF_INET6.
const struct sockaddr_in6* addr_in6 =
reinterpret_cast<const struct sockaddr_in6*>(addr);
const struct in6_addr* sin6_addr = &addr_in6->sin6_addr;
if (IN6_IS_ADDR_LOOPBACK(sin6_addr) || IN6_IS_ADDR_LINKLOCAL(sin6_addr))
continue;
}
if (addr->sa_family != AF_INET6 && addr->sa_family != AF_INET)
continue;
result = false;
break;
}
freeifaddrs(interface_addr);
return result;
#elif defined(OS_WIN)
// TODO(wtc): implement with the GetAdaptersAddresses function.
NOTIMPLEMENTED();
return false;
#else
NOTIMPLEMENTED();
return false;
#endif // defined(various platforms)
}
bool ParseIPLiteralToNumber(const std::string& ip_literal,
IPAddressNumber* ip_number) {
// |ip_literal| could be either a IPv4 or an IPv6 literal. If it contains
// a colon however, it must be an IPv6 address.
if (ip_literal.find(':') != std::string::npos) {
// GURL expects IPv6 hostnames to be surrounded with brackets.
std::string host_brackets = "[" + ip_literal + "]";
url_parse::Component host_comp(0, host_brackets.size());
// Try parsing the hostname as an IPv6 literal.
ip_number->resize(16); // 128 bits.
return url_canon::IPv6AddressToNumber(host_brackets.data(),
host_comp,
&(*ip_number)[0]);
}
// Otherwise the string is an IPv4 address.
ip_number->resize(4); // 32 bits.
url_parse::Component host_comp(0, ip_literal.size());
int num_components;
url_canon::CanonHostInfo::Family family = url_canon::IPv4AddressToNumber(
ip_literal.data(), host_comp, &(*ip_number)[0], &num_components);
return family == url_canon::CanonHostInfo::IPV4;
}
IPAddressNumber ConvertIPv4NumberToIPv6Number(
const IPAddressNumber& ipv4_number) {
DCHECK(ipv4_number.size() == 4);
// IPv4-mapped addresses are formed by:
// <80 bits of zeros> + <16 bits of ones> + <32-bit IPv4 address>.
IPAddressNumber ipv6_number;
ipv6_number.reserve(16);
ipv6_number.insert(ipv6_number.end(), 10, 0);
ipv6_number.push_back(0xFF);
ipv6_number.push_back(0xFF);
ipv6_number.insert(ipv6_number.end(), ipv4_number.begin(), ipv4_number.end());
return ipv6_number;
}
bool ParseCIDRBlock(const std::string& cidr_literal,
IPAddressNumber* ip_number,
size_t* prefix_length_in_bits) {
// We expect CIDR notation to match one of these two templates:
// <IPv4-literal> "/" <number of bits>
// <IPv6-literal> "/" <number of bits>
std::vector<std::string> parts;
base::SplitString(cidr_literal, '/', &parts);
if (parts.size() != 2)
return false;
// Parse the IP address.
if (!ParseIPLiteralToNumber(parts[0], ip_number))
return false;
// Parse the prefix length.
int number_of_bits = -1;
if (!base::StringToInt(parts[1], &number_of_bits))
return false;
// Make sure the prefix length is in a valid range.
if (number_of_bits < 0 ||
number_of_bits > static_cast<int>(ip_number->size() * 8))
return false;
*prefix_length_in_bits = static_cast<size_t>(number_of_bits);
return true;
}
bool IPNumberMatchesPrefix(const IPAddressNumber& ip_number,
const IPAddressNumber& ip_prefix,
size_t prefix_length_in_bits) {
// Both the input IP address and the prefix IP address should be
// either IPv4 or IPv6.
DCHECK(ip_number.size() == 4 || ip_number.size() == 16);
DCHECK(ip_prefix.size() == 4 || ip_prefix.size() == 16);
DCHECK_LE(prefix_length_in_bits, ip_prefix.size() * 8);
// In case we have an IPv6 / IPv4 mismatch, convert the IPv4 addresses to
// IPv6 addresses in order to do the comparison.
if (ip_number.size() != ip_prefix.size()) {
if (ip_number.size() == 4) {
return IPNumberMatchesPrefix(ConvertIPv4NumberToIPv6Number(ip_number),
ip_prefix, prefix_length_in_bits);
}
return IPNumberMatchesPrefix(ip_number,
ConvertIPv4NumberToIPv6Number(ip_prefix),
96 + prefix_length_in_bits);
}
// Otherwise we are comparing two IPv4 addresses, or two IPv6 addresses.
// Compare all the bytes that fall entirely within the prefix.
int num_entire_bytes_in_prefix = prefix_length_in_bits / 8;
for (int i = 0; i < num_entire_bytes_in_prefix; ++i) {
if (ip_number[i] != ip_prefix[i])
return false;
}
// In case the prefix was not a multiple of 8, there will be 1 byte
// which is only partially masked.
int remaining_bits = prefix_length_in_bits % 8;
if (remaining_bits != 0) {
unsigned char mask = 0xFF << (8 - remaining_bits);
int i = num_entire_bytes_in_prefix;
if ((ip_number[i] & mask) != (ip_prefix[i] & mask))
return false;
}
return true;
}
struct addrinfo* CreateCopyOfAddrinfo(const struct addrinfo* info,
bool recursive) {
DCHECK(info);
struct addrinfo* copy = new addrinfo;
// Copy all the fields (some of these are pointers, we will fix that next).
memcpy(copy, info, sizeof(addrinfo));
// ai_canonname is a NULL-terminated string.
if (info->ai_canonname) {
copy->ai_canonname = do_strdup(info->ai_canonname);
}
// ai_addr is a buffer of length ai_addrlen.
if (info->ai_addr) {
copy->ai_addr = reinterpret_cast<sockaddr *>(new char[info->ai_addrlen]);
memcpy(copy->ai_addr, info->ai_addr, info->ai_addrlen);
}
// Recursive copy.
if (recursive && info->ai_next)
copy->ai_next = CreateCopyOfAddrinfo(info->ai_next, recursive);
else
copy->ai_next = NULL;
return copy;
}
void FreeCopyOfAddrinfo(struct addrinfo* info) {
DCHECK(info);
if (info->ai_canonname)
free(info->ai_canonname); // Allocated by strdup.
if (info->ai_addr)
delete [] reinterpret_cast<char*>(info->ai_addr);
struct addrinfo* next = info->ai_next;
delete info;
// Recursive free.
if (next)
FreeCopyOfAddrinfo(next);
}
// Returns the port field of the sockaddr in |info|.
uint16* GetPortFieldFromAddrinfo(struct addrinfo* info) {
const struct addrinfo* const_info = info;
const uint16* port_field = GetPortFieldFromAddrinfo(const_info);
return const_cast<uint16*>(port_field);
}
const uint16* GetPortFieldFromAddrinfo(const struct addrinfo* info) {
DCHECK(info);
const struct sockaddr* address = info->ai_addr;
DCHECK(address);
DCHECK_EQ(info->ai_family, address->sa_family);
return GetPortFieldFromSockaddr(address, info->ai_addrlen);
}
int GetPortFromAddrinfo(const struct addrinfo* info) {
const uint16* port_field = GetPortFieldFromAddrinfo(info);
if (!port_field)
return -1;
return ntohs(*port_field);
}
const uint16* GetPortFieldFromSockaddr(const struct sockaddr* address,
socklen_t address_len) {
if (address->sa_family == AF_INET) {
DCHECK_LE(sizeof(sockaddr_in), static_cast<size_t>(address_len));
const struct sockaddr_in* sockaddr =
reinterpret_cast<const struct sockaddr_in*>(address);
return &sockaddr->sin_port;
} else if (address->sa_family == AF_INET6) {
DCHECK_LE(sizeof(sockaddr_in6), static_cast<size_t>(address_len));
const struct sockaddr_in6* sockaddr =
reinterpret_cast<const struct sockaddr_in6*>(address);
return &sockaddr->sin6_port;
} else {
NOTREACHED();
return NULL;
}
}
int GetPortFromSockaddr(const struct sockaddr* address, socklen_t address_len) {
const uint16* port_field = GetPortFieldFromSockaddr(address, address_len);
if (!port_field)
return -1;
return ntohs(*port_field);
}
bool IsLocalhost(const std::string& host) {
if (host == "localhost" ||
host == "localhost.localdomain" ||
host == "localhost6" ||
host == "localhost6.localdomain6")
return true;
IPAddressNumber ip_number;
if (ParseIPLiteralToNumber(host, &ip_number)) {
size_t size = ip_number.size();
switch (size) {
case kIPv4AddressSize: {
IPAddressNumber localhost_prefix;
localhost_prefix.push_back(127);
for (int i = 0; i < 3; ++i) {
localhost_prefix.push_back(0);
}
return IPNumberMatchesPrefix(ip_number, localhost_prefix, 8);
}
case kIPv6AddressSize: {
struct in6_addr sin6_addr;
memcpy(&sin6_addr, &ip_number[0], kIPv6AddressSize);
return !!IN6_IS_ADDR_LOOPBACK(&sin6_addr);
}
default:
NOTREACHED();
}
}
return false;
}
NetworkInterface::NetworkInterface() {
}
NetworkInterface::NetworkInterface(const std::string& name,
const IPAddressNumber& address)
: name(name), address(address) {
}
NetworkInterface::~NetworkInterface() {
}
ClampComponentOffset::ClampComponentOffset(size_t component_start)
: component_start(component_start) {}
size_t ClampComponentOffset::operator()(size_t offset) {
return (offset >= component_start) ?
offset : std::wstring::npos;
}
} // namespace net