X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;ds=sidebyside;f=folly%2FString-inl.h;h=dbfefa010fb6f97ac530ce280b74f086996b7ec0;hb=79e5fd2fa293cf03269d4b7bba9ed7a31dda1cd8;hp=a2179209e4e240e6d83bf81f6da3550b6a3d342f;hpb=52ba96edd198f7985f0315b034f84c6dd29234c5;p=folly.git diff --git a/folly/String-inl.h b/folly/String-inl.h index a2179209..dbfefa01 100644 --- a/folly/String-inl.h +++ b/folly/String-inl.h @@ -1,5 +1,5 @@ /* - * Copyright 2014 Facebook, Inc. + * Copyright 2017 Facebook, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -14,13 +14,12 @@ * limitations under the License. */ -#ifndef FOLLY_STRING_INL_H_ -#define FOLLY_STRING_INL_H_ +#pragma once #include #include -#ifndef FOLLY_BASE_STRING_H_ +#ifndef FOLLY_STRING_H_ #error This file may only be included from String.h #endif @@ -51,7 +50,7 @@ void cEscape(StringPiece str, String& out) { if (e == 'P') { // printable ++p; } else if (e == 'O') { // octal - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); esc[1] = '0' + ((v >> 6) & 7); esc[2] = '0' + ((v >> 3) & 7); esc[3] = '0' + (v & 7); @@ -59,14 +58,14 @@ void cEscape(StringPiece str, String& out) { ++p; last = p; } else { // special 1-character escape - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); esc[1] = e; out.append(esc, 2); ++p; last = p; } } - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); } namespace detail { @@ -178,12 +177,12 @@ void uriEscape(StringPiece str, String& out, UriEscapeMode mode) { if (LIKELY(discriminator <= minEncode)) { ++p; } else if (mode == UriEscapeMode::QUERY && discriminator == 3) { - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); out.push_back('+'); ++p; last = p; } else { - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); esc[1] = hexValues[v >> 4]; esc[2] = hexValues[v & 0x0f]; out.append(esc, 3); @@ -191,7 +190,7 @@ void uriEscape(StringPiece str, String& out, UriEscapeMode mode) { last = p; } } - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); } template @@ -203,7 +202,6 @@ void uriUnescape(StringPiece str, String& out, UriEscapeMode mode) { // this is faster than calling push_back repeatedly. while (p != str.end()) { char c = *p; - unsigned char v = static_cast(v); switch (c) { case '%': { @@ -215,7 +213,7 @@ void uriUnescape(StringPiece str, String& out, UriEscapeMode mode) { if (UNLIKELY(h1 == 16 || h2 == 16)) { throw std::invalid_argument("invalid percent encode sequence"); } - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); out.push_back((h1 << 4) | h2); p += 3; last = p; @@ -223,7 +221,7 @@ void uriUnescape(StringPiece str, String& out, UriEscapeMode mode) { } case '+': if (mode == UriEscapeMode::QUERY) { - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); out.push_back(' '); ++p; last = p; @@ -235,7 +233,7 @@ void uriUnescape(StringPiece str, String& out, UriEscapeMode mode) { break; } } - out.append(&*last, p - last); + out.append(&*last, size_t(p - last)); } namespace detail { @@ -265,29 +263,6 @@ inline char delimFront(StringPiece s) { return *s.start(); } -/* - * These output conversion templates allow us to support multiple - * output string types, even when we are using an arbitrary - * OutputIterator. - */ -template struct OutputConverter {}; - -template<> struct OutputConverter { - std::string operator()(StringPiece sp) const { - return sp.toString(); - } -}; - -template<> struct OutputConverter { - fbstring operator()(StringPiece sp) const { - return sp.toFbstring(); - } -}; - -template<> struct OutputConverter { - StringPiece operator()(StringPiece sp) const { return sp; } -}; - /* * Shared implementation for all the split() overloads. * @@ -306,11 +281,9 @@ void internalSplit(DelimT delim, StringPiece sp, OutputIterator out, const size_t strSize = sp.size(); const size_t dSize = delimSize(delim); - OutputConverter conv; - if (dSize > strSize || dSize == 0) { if (!ignoreEmpty || strSize > 0) { - *out++ = conv(sp); + *out++ = to(sp); } return; } @@ -320,12 +293,12 @@ void internalSplit(DelimT delim, StringPiece sp, OutputIterator out, ignoreEmpty); } - int tokenStartPos = 0; - int tokenSize = 0; - for (int i = 0; i <= strSize - dSize; ++i) { + size_t tokenStartPos = 0; + size_t tokenSize = 0; + for (size_t i = 0; i <= strSize - dSize; ++i) { if (atDelim(&s[i], delim)) { if (!ignoreEmpty || tokenSize > 0) { - *out++ = conv(StringPiece(&s[tokenStartPos], tokenSize)); + *out++ = to(sp.subpiece(tokenStartPos, tokenSize)); } tokenStartPos = i + dSize; @@ -335,10 +308,9 @@ void internalSplit(DelimT delim, StringPiece sp, OutputIterator out, ++tokenSize; } } - + tokenSize = strSize - tokenStartPos; if (!ignoreEmpty || tokenSize > 0) { - tokenSize = strSize - tokenStartPos; - *out++ = conv(StringPiece(&s[tokenStartPos], tokenSize)); + *out++ = to(sp.subpiece(tokenStartPos, tokenSize)); } } @@ -347,25 +319,25 @@ template StringPiece prepareDelim(const String& s) { } inline char prepareDelim(char c) { return c; } -template -bool splitFixed(const Delim& delimiter, - StringPiece input, - StringPiece& out) { +template +bool splitFixed(const Delim& delimiter, StringPiece input, OutputType& output) { + static_assert( + exact || std::is_same::value || + IsSomeString::value, + "split() requires that the last argument be a string type"); if (exact && UNLIKELY(std::string::npos != input.find(delimiter))) { return false; } - out = input; + output = folly::to(input); return true; } -template -bool splitFixed(const Delim& delimiter, - StringPiece input, - StringPiece& outHead, - StringPieces&... outTail) { +template +bool splitFixed( + const Delim& delimiter, + StringPiece input, + OutputType& outHead, + OutputTypes&... outTail) { size_t cut = input.find(delimiter); if (UNLIKELY(cut == std::string::npos)) { return false; @@ -374,7 +346,7 @@ bool splitFixed(const Delim& delimiter, StringPiece tail(input.begin() + cut + detail::delimSize(delimiter), input.end()); if (LIKELY(splitFixed(delimiter, tail, outTail...))) { - outHead = head; + outHead = folly::to(head); return true; } return false; @@ -421,25 +393,31 @@ void splitTo(const Delim& delimiter, ignoreEmpty); } -template -bool split(const Delim& delimiter, - StringPiece input, - StringPiece& outHead, - StringPieces&... outTail) { +template +typename std::enable_if< + AllConvertible::value && sizeof...(OutputTypes) >= 1, + bool>::type +split(const Delim& delimiter, StringPiece input, OutputTypes&... outputs) { return detail::splitFixed( - detail::prepareDelim(delimiter), - input, - outHead, - outTail...); + detail::prepareDelim(delimiter), input, outputs...); } namespace detail { +/* + * If a type can have its string size determined cheaply, we can more + * efficiently append it in a loop (see internalJoinAppend). Note that the + * struct need not conform to the std::string api completely (ex. does not need + * to implement append()). + */ +template struct IsSizableString { + enum { value = IsSomeString::value + || std::is_same::value }; +}; + template -struct IsStringContainerIterator : - IsSomeString::value_type> { +struct IsSizableStringContainerIterator : + IsSizableString::value_type> { }; template @@ -460,7 +438,7 @@ void internalJoinAppend(Delim delimiter, } template -typename std::enable_if::value>::type +typename std::enable_if::value>::type internalJoin(Delim delimiter, Iterator begin, Iterator end, @@ -480,7 +458,8 @@ internalJoin(Delim delimiter, } template -typename std::enable_if::value>::type +typename +std::enable_if::value>::type internalJoin(Delim delimiter, Iterator begin, Iterator end, @@ -543,8 +522,8 @@ void backslashify(const String1& input, String2& output, bool hex_style) { template void humanify(const String1& input, String2& output) { - int numUnprintable = 0; - int numPrintablePrefix = 0; + size_t numUnprintable = 0; + size_t numPrintablePrefix = 0; for (unsigned char c : input) { if (c < 0x20 || c > 0x7e || c == '\\') { ++numUnprintable; @@ -591,9 +570,9 @@ bool hexlify(const InputString& input, OutputString& output, if (!append_output) output.clear(); static char hexValues[] = "0123456789abcdef"; - int j = output.size(); + auto j = output.size(); output.resize(2 * input.size() + output.size()); - for (int i = 0; i < input.size(); ++i) { + for (size_t i = 0; i < input.size(); ++i) { int ch = input[i]; output[j++] = hexValues[(ch >> 4) & 0xf]; output[j++] = hexValues[ch & 0xf]; @@ -608,17 +587,12 @@ bool unhexlify(const InputString& input, OutputString& output) { } output.resize(input.size() / 2); int j = 0; - auto unhex = [](char c) -> int { - return c >= '0' && c <= '9' ? c - '0' : - c >= 'A' && c <= 'F' ? c - 'A' + 10 : - c >= 'a' && c <= 'f' ? c - 'a' + 10 : - -1; - }; - - for (int i = 0; i < input.size(); i += 2) { - int highBits = unhex(input[i]); - int lowBits = unhex(input[i + 1]); - if (highBits < 0 || lowBits < 0) { + + for (size_t i = 0; i < input.size(); i += 2) { + int highBits = detail::hexTable[static_cast(input[i])]; + int lowBits = detail::hexTable[static_cast(input[i + 1])]; + if ((highBits | lowBits) & 0x10) { + // One of the characters wasn't a hex digit return false; } output[j++] = (highBits << 4) + lowBits; @@ -646,5 +620,3 @@ void hexDump(const void* ptr, size_t size, OutIt out) { } } // namespace folly - -#endif /* FOLLY_STRING_INL_H_ */