url_canon_ip.cc | Explore in Territory

// Copyright 2013 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#ifdef UNSAFE_BUFFERS_BUILD
// TODO(crbug.com/40285824): Remove this and convert code to safer constructs.
#pragma allow_unsafe_buffers
#endif

// This is a copy of url/url_canon_ip.cc circa 2023. It should be used only by
// components/feedback/redaction_tool/.
// We need a copy because the components/feedback/redaction_tool source code is
// shared into ChromeOS and needs to have no dependencies outside of base/.

#include "components/feedback/redaction_tool/url_canon_ip.h"

#include <stdint.h>
#include <stdlib.h>

#include <limits>

#include "base/check.h"
#include "components/feedback/redaction_tool/url_canon_internal.h"

namespace redaction_internal {

namespace {

// Converts one of the character types that represent a numerical base to the
// corresponding base.
int BaseForType(SharedCharTypes type) { … }

// Converts an IPv4 component to a 32-bit number, while checking for overflow.
//
// Possible return values:
// - IPV4    - The number was valid, and did not overflow.
// - BROKEN  - The input was numeric, but too large for a 32-bit field.
// - NEUTRAL - Input was not numeric.
//
// The input is assumed to be ASCII. The components are assumed to be non-empty.
template <typename CHAR>
CanonHostInfo::Family IPv4ComponentToNumber(const CHAR* spec,
                                            const Component& component,
                                            uint32_t* number) { … }

// See declaration of IPv4AddressToNumber for documentation.
template <typename CHAR, typename UCHAR>
CanonHostInfo::Family DoIPv4AddressToNumber(const CHAR* spec,
                                            Component host,
                                            unsigned char address[4],
                                            int* num_ipv4_components) { … }

// Helper class that describes the main components of an IPv6 input string.
// See the following examples to understand how it breaks up an input string:
//
// [Example 1]: input = "[::aa:bb]"
//  ==> num_hex_components = 2
//  ==> hex_components[0] = Component(3,2) "aa"
//  ==> hex_components[1] = Component(6,2) "bb"
//  ==> index_of_contraction = 0
//  ==> ipv4_component = Component(0, -1)
//
// [Example 2]: input = "[1:2::3:4:5]"
//  ==> num_hex_components = 5
//  ==> hex_components[0] = Component(1,1) "1"
//  ==> hex_components[1] = Component(3,1) "2"
//  ==> hex_components[2] = Component(6,1) "3"
//  ==> hex_components[3] = Component(8,1) "4"
//  ==> hex_components[4] = Component(10,1) "5"
//  ==> index_of_contraction = 2
//  ==> ipv4_component = Component(0, -1)
//
// [Example 3]: input = "[::ffff:192.168.0.1]"
//  ==> num_hex_components = 1
//  ==> hex_components[0] = Component(3,4) "ffff"
//  ==> index_of_contraction = 0
//  ==> ipv4_component = Component(8, 11) "192.168.0.1"
//
// [Example 4]: input = "[1::]"
//  ==> num_hex_components = 1
//  ==> hex_components[0] = Component(1,1) "1"
//  ==> index_of_contraction = 1
//  ==> ipv4_component = Component(0, -1)
//
// [Example 5]: input = "[::192.168.0.1]"
//  ==> num_hex_components = 0
//  ==> index_of_contraction = 0
//  ==> ipv4_component = Component(8, 11) "192.168.0.1"
//
struct IPv6Parsed { … };

// Parse the IPv6 input string. If parsing succeeded returns true and fills
// |parsed| with the information. If parsing failed (because the input is
// invalid) returns false.
template <typename CHAR, typename UCHAR>
bool DoParseIPv6(const CHAR* spec, const Component& host, IPv6Parsed* parsed) { … }

// Verifies the parsed IPv6 information, checking that the various components
// add up to the right number of bits (hex components are 16 bits, while
// embedded IPv4 formats are 32 bits, and contractions are placeholders for
// 16 or more bits). Returns true if sizes match up, false otherwise. On
// success writes the length of the contraction (if any) to
// |out_num_bytes_of_contraction|.
bool CheckIPv6ComponentsSize(const IPv6Parsed& parsed,
                             int* out_num_bytes_of_contraction) { … }

// Converts a hex component into a number. This cannot fail since the caller has
// already verified that each character in the string was a hex digit, and
// that there were no more than 4 characters.
template <typename CHAR>
uint16_t IPv6HexComponentToNumber(const CHAR* spec,
                                  const Component& component) { … }

// Converts an IPv6 address to a 128-bit number (network byte order), returning
// true on success. False means that the input was not a valid IPv6 address.
template <typename CHAR, typename UCHAR>
bool DoIPv6AddressToNumber(const CHAR* spec,
                           const Component& host,
                           unsigned char address[16]) { … }

// Searches for the longest sequence of zeros in |address|, and writes the
// range into |contraction_range|. The run of zeros must be at least 16 bits,
// and if there is a tie the first is chosen.
void ChooseIPv6ContractionRange(const unsigned char address[16],
                                Component* contraction_range) { … }

}  // namespace

void AppendIPv4Address(const unsigned char address[4], CanonOutput* output) { … }

void AppendIPv6Address(const unsigned char address[16], CanonOutput* output) { … }

CanonHostInfo::Family IPv4AddressToNumber(const char* spec,
                                          const Component& host,
                                          unsigned char address[4],
                                          int* num_ipv4_components) { … }

CanonHostInfo::Family IPv4AddressToNumber(const char16_t* spec,
                                          const Component& host,
                                          unsigned char address[4],
                                          int* num_ipv4_components) { … }

bool IPv6AddressToNumber(const char* spec,
                         const Component& host,
                         unsigned char address[16]) { … }

bool IPv6AddressToNumber(const char16_t* spec,
                         const Component& host,
                         unsigned char address[16]) { … }

}  // namespace redaction_internal
chromium/components/feedback/redaction_tool/url_canon_ip.cc
