diff options
author | Abseil Team <absl-team@google.com> | 2020-01-09T17·58-0800 |
---|---|---|
committer | Derek Mauro <dmauro@google.com> | 2020-01-09T19·19-0500 |
commit | b3aaac8a37c467a1125c794196caa90d0957bdc3 (patch) | |
tree | 054402e6c20470cf6d4150e58cf194737e122f3a /absl/strings/escaping.cc | |
parent | 63ee2f8877915a3565c29707dba8fe4d7822596a (diff) |
Export of internal Abseil changes
-- 9beb68204986a015c9cb065b9fae4f9a8879a788 by Abseil Team <absl-team@google.com>: Move Base64EscapeInternal and CalculateBase64EscapedLenInternal to an internal header. PiperOrigin-RevId: 288917378 -- 90acfbe03b3f9f6de3ffa49c39343dfaa2c5d38c by Greg Falcon <gfalcon@google.com>: Update macos CI script to support the ALTERNATE_OPTIONS environment variable. PiperOrigin-RevId: 288913564 -- f1572e870678cdcda6b48cb39780d1ad984e4c1b by Derek Mauro <dmauro@google.com>: Makes absl::NullSafeStringView constexpr Fixes https://github.com/abseil/abseil-cpp/issues/583 PiperOrigin-RevId: 288906940 -- d28a8471e32c10caa64bfffe6d6d4d0a8d144013 by Abseil Team <absl-team@google.com>: absl::GetFlag is lock free for small trivially copyable types. PiperOrigin-RevId: 288768172 -- 2643b8ed1a1dc836b38ab9e46538a1af129ffd67 by Gennadiy Rozental <rogeeff@google.com>: Eliminate call to callback from flag initialization. We do not need to have this invocation inside FlagImpl::Init since SetCallback performs invocation anyways. Calling InitCallback from inside of Init complicates separation of value initialization from data guard initialization, which is about to happen. PiperOrigin-RevId: 288732526 -- 22caa880b7a4cb6da34e16a2e064a473c99e880b by Abseil Team <absl-team@google.com>: Fix the documentation on how to create a null string_view. PiperOrigin-RevId: 288727968 -- 10727c5cadc561837141176f4c9b9717cec9233a by Greg Falcon <gfalcon@google.com>: Change CI scripts for gcc to use the ALTERNATE_OPTIONS file as well. PiperOrigin-RevId: 288718855 -- 5d1e2dd6c7fb12af8aa4337a0f61872f5f0c5992 by Greg Falcon <gfalcon@google.com>: Add an option for using inline namespaces in Abseil. PiperOrigin-RevId: 288614491 GitOrigin-RevId: 9beb68204986a015c9cb065b9fae4f9a8879a788 Change-Id: If9acd46301e3df8cb231b4c16f7ed651bf4fb3c3
Diffstat (limited to 'absl/strings/escaping.cc')
-rw-r--r-- | absl/strings/escaping.cc | 188 |
1 files changed, 13 insertions, 175 deletions
diff --git a/absl/strings/escaping.cc b/absl/strings/escaping.cc index d2fcd9c111bc..7adc1b657168 100644 --- a/absl/strings/escaping.cc +++ b/absl/strings/escaping.cc @@ -26,6 +26,7 @@ #include "absl/base/internal/raw_logging.h" #include "absl/base/internal/unaligned_access.h" #include "absl/strings/internal/char_map.h" +#include "absl/strings/internal/escaping.h" #include "absl/strings/internal/resize_uninitialized.h" #include "absl/strings/internal/utf8.h" #include "absl/strings/str_cat.h" @@ -764,177 +765,10 @@ constexpr signed char kUnWebSafeBase64[] = { }; /* clang-format on */ -size_t CalculateBase64EscapedLenInternal(size_t input_len, bool do_padding) { - // Base64 encodes three bytes of input at a time. If the input is not - // divisible by three, we pad as appropriate. - // - // (from https://tools.ietf.org/html/rfc3548) - // Special processing is performed if fewer than 24 bits are available - // at the end of the data being encoded. A full encoding quantum is - // always completed at the end of a quantity. When fewer than 24 input - // bits are available in an input group, zero bits are added (on the - // right) to form an integral number of 6-bit groups. Padding at the - // end of the data is performed using the '=' character. Since all base - // 64 input is an integral number of octets, only the following cases - // can arise: - - // Base64 encodes each three bytes of input into four bytes of output. - size_t len = (input_len / 3) * 4; - - if (input_len % 3 == 0) { - // (from https://tools.ietf.org/html/rfc3548) - // (1) the final quantum of encoding input is an integral multiple of 24 - // bits; here, the final unit of encoded output will be an integral - // multiple of 4 characters with no "=" padding, - } else if (input_len % 3 == 1) { - // (from https://tools.ietf.org/html/rfc3548) - // (2) the final quantum of encoding input is exactly 8 bits; here, the - // final unit of encoded output will be two characters followed by two - // "=" padding characters, or - len += 2; - if (do_padding) { - len += 2; - } - } else { // (input_len % 3 == 2) - // (from https://tools.ietf.org/html/rfc3548) - // (3) the final quantum of encoding input is exactly 16 bits; here, the - // final unit of encoded output will be three characters followed by one - // "=" padding character. - len += 3; - if (do_padding) { - len += 1; - } - } - - assert(len >= input_len); // make sure we didn't overflow - return len; -} - -size_t Base64EscapeInternal(const unsigned char* src, size_t szsrc, char* dest, - size_t szdest, const char* base64, - bool do_padding) { - static const char kPad64 = '='; - - if (szsrc * 4 > szdest * 3) return 0; - - char* cur_dest = dest; - const unsigned char* cur_src = src; - - char* const limit_dest = dest + szdest; - const unsigned char* const limit_src = src + szsrc; - - // Three bytes of data encodes to four characters of cyphertext. - // So we can pump through three-byte chunks atomically. - if (szsrc >= 3) { // "limit_src - 3" is UB if szsrc < 3. - while (cur_src < limit_src - 3) { // While we have >= 32 bits. - uint32_t in = absl::big_endian::Load32(cur_src) >> 8; - - cur_dest[0] = base64[in >> 18]; - in &= 0x3FFFF; - cur_dest[1] = base64[in >> 12]; - in &= 0xFFF; - cur_dest[2] = base64[in >> 6]; - in &= 0x3F; - cur_dest[3] = base64[in]; - - cur_dest += 4; - cur_src += 3; - } - } - // To save time, we didn't update szdest or szsrc in the loop. So do it now. - szdest = limit_dest - cur_dest; - szsrc = limit_src - cur_src; - - /* now deal with the tail (<=3 bytes) */ - switch (szsrc) { - case 0: - // Nothing left; nothing more to do. - break; - case 1: { - // One byte left: this encodes to two characters, and (optionally) - // two pad characters to round out the four-character cypherblock. - if (szdest < 2) return 0; - uint32_t in = cur_src[0]; - cur_dest[0] = base64[in >> 2]; - in &= 0x3; - cur_dest[1] = base64[in << 4]; - cur_dest += 2; - szdest -= 2; - if (do_padding) { - if (szdest < 2) return 0; - cur_dest[0] = kPad64; - cur_dest[1] = kPad64; - cur_dest += 2; - szdest -= 2; - } - break; - } - case 2: { - // Two bytes left: this encodes to three characters, and (optionally) - // one pad character to round out the four-character cypherblock. - if (szdest < 3) return 0; - uint32_t in = absl::big_endian::Load16(cur_src); - cur_dest[0] = base64[in >> 10]; - in &= 0x3FF; - cur_dest[1] = base64[in >> 4]; - in &= 0x00F; - cur_dest[2] = base64[in << 2]; - cur_dest += 3; - szdest -= 3; - if (do_padding) { - if (szdest < 1) return 0; - cur_dest[0] = kPad64; - cur_dest += 1; - szdest -= 1; - } - break; - } - case 3: { - // Three bytes left: same as in the big loop above. We can't do this in - // the loop because the loop above always reads 4 bytes, and the fourth - // byte is past the end of the input. - if (szdest < 4) return 0; - uint32_t in = (cur_src[0] << 16) + absl::big_endian::Load16(cur_src + 1); - cur_dest[0] = base64[in >> 18]; - in &= 0x3FFFF; - cur_dest[1] = base64[in >> 12]; - in &= 0xFFF; - cur_dest[2] = base64[in >> 6]; - in &= 0x3F; - cur_dest[3] = base64[in]; - cur_dest += 4; - szdest -= 4; - break; - } - default: - // Should not be reached: blocks of 4 bytes are handled - // in the while loop before this switch statement. - ABSL_RAW_LOG(FATAL, "Logic problem? szsrc = %zu", szsrc); - break; - } - return (cur_dest - dest); -} - -constexpr char kBase64Chars[] = - "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; - constexpr char kWebSafeBase64Chars[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_"; template <typename String> -void Base64EscapeInternal(const unsigned char* src, size_t szsrc, String* dest, - bool do_padding, const char* base64_chars) { - const size_t calc_escaped_size = - CalculateBase64EscapedLenInternal(szsrc, do_padding); - strings_internal::STLStringResizeUninitialized(dest, calc_escaped_size); - - const size_t escaped_len = Base64EscapeInternal( - src, szsrc, &(*dest)[0], dest->size(), base64_chars, do_padding); - assert(calc_escaped_size == escaped_len); - dest->erase(escaped_len); -} - -template <typename String> bool Base64UnescapeInternal(const char* src, size_t slen, String* dest, const signed char* unbase64) { // Determine the size of the output std::string. Base64 encodes every 3 bytes into @@ -1068,26 +902,30 @@ bool WebSafeBase64Unescape(absl::string_view src, std::string* dest) { } void Base64Escape(absl::string_view src, std::string* dest) { - Base64EscapeInternal(reinterpret_cast<const unsigned char*>(src.data()), - src.size(), dest, true, kBase64Chars); + strings_internal::Base64EscapeInternal( + reinterpret_cast<const unsigned char*>(src.data()), src.size(), dest, + true, strings_internal::kBase64Chars); } void WebSafeBase64Escape(absl::string_view src, std::string* dest) { - Base64EscapeInternal(reinterpret_cast<const unsigned char*>(src.data()), - src.size(), dest, false, kWebSafeBase64Chars); + strings_internal::Base64EscapeInternal( + reinterpret_cast<const unsigned char*>(src.data()), src.size(), dest, + false, kWebSafeBase64Chars); } std::string Base64Escape(absl::string_view src) { std::string dest; - Base64EscapeInternal(reinterpret_cast<const unsigned char*>(src.data()), - src.size(), &dest, true, kBase64Chars); + strings_internal::Base64EscapeInternal( + reinterpret_cast<const unsigned char*>(src.data()), src.size(), &dest, + true, strings_internal::kBase64Chars); return dest; } std::string WebSafeBase64Escape(absl::string_view src) { std::string dest; - Base64EscapeInternal(reinterpret_cast<const unsigned char*>(src.data()), - src.size(), &dest, false, kWebSafeBase64Chars); + strings_internal::Base64EscapeInternal( + reinterpret_cast<const unsigned char*>(src.data()), src.size(), &dest, + false, kWebSafeBase64Chars); return dest; } |