X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=folly%2FString-inl.h;h=27b02a614f96f4fffe949cd6ccdfc6bc19a279c0;hb=4ebfdff303924da5636e512ef7268612cba7c803;hp=02d376d82f06e02fd4b64b30f6758705091cc0ca;hpb=f3f96c69481d5dd078e3563b13bf05903dba6517;p=folly.git

diff --git a/folly/String-inl.h b/folly/String-inl.h
index 02d376d8..27b02a61 100644
--- a/folly/String-inl.h
+++ b/folly/String-inl.h
@@ -1,5 +1,5 @@
 /*
- * Copyright 2012 Facebook, Inc.
+ * Copyright 2014 Facebook, Inc.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -149,6 +149,95 @@ void cUnescape(StringPiece str, String& out, bool strict) {
   out.append(&*last, p - last);
 }
 
+namespace detail {
+// Map from character code to escape mode:
+// 0 = pass through
+// 1 = unused
+// 2 = pass through in PATH mode
+// 3 = space, replace with '+' in QUERY mode
+// 4 = percent-encode
+extern const unsigned char uriEscapeTable[];
+}  // namespace detail
+
+template <class String>
+void uriEscape(StringPiece str, String& out, UriEscapeMode mode) {
+  static const char hexValues[] = "0123456789abcdef";
+  char esc[3];
+  esc[0] = '%';
+  // Preallocate assuming that 25% of the input string will be escaped
+  out.reserve(out.size() + str.size() + 3 * (str.size() / 4));
+  auto p = str.begin();
+  auto last = p;  // last regular character
+  // We advance over runs of passthrough characters and copy them in one go;
+  // this is faster than calling push_back repeatedly.
+  unsigned char minEncode = static_cast<unsigned char>(mode);
+  while (p != str.end()) {
+    char c = *p;
+    unsigned char v = static_cast<unsigned char>(c);
+    unsigned char discriminator = detail::uriEscapeTable[v];
+    if (LIKELY(discriminator <= minEncode)) {
+      ++p;
+    } else if (mode == UriEscapeMode::QUERY && discriminator == 3) {
+      out.append(&*last, p - last);
+      out.push_back('+');
+      ++p;
+      last = p;
+    } else {
+      out.append(&*last, p - last);
+      esc[1] = hexValues[v >> 4];
+      esc[2] = hexValues[v & 0x0f];
+      out.append(esc, 3);
+      ++p;
+      last = p;
+    }
+  }
+  out.append(&*last, p - last);
+}
+
+template <class String>
+void uriUnescape(StringPiece str, String& out, UriEscapeMode mode) {
+  out.reserve(out.size() + str.size());
+  auto p = str.begin();
+  auto last = p;
+  // We advance over runs of passthrough characters and copy them in one go;
+  // this is faster than calling push_back repeatedly.
+  while (p != str.end()) {
+    char c = *p;
+    unsigned char v = static_cast<unsigned char>(v);
+    switch (c) {
+    case '%':
+      {
+        if (UNLIKELY(std::distance(p, str.end()) < 3)) {
+          throw std::invalid_argument("incomplete percent encode sequence");
+        }
+        auto h1 = detail::hexTable[static_cast<unsigned char>(p[1])];
+        auto h2 = detail::hexTable[static_cast<unsigned char>(p[2])];
+        if (UNLIKELY(h1 == 16 || h2 == 16)) {
+          throw std::invalid_argument("invalid percent encode sequence");
+        }
+        out.append(&*last, p - last);
+        out.push_back((h1 << 4) | h2);
+        p += 3;
+        last = p;
+        break;
+      }
+    case '+':
+      if (mode == UriEscapeMode::QUERY) {
+        out.append(&*last, p - last);
+        out.push_back(' ');
+        ++p;
+        last = p;
+        break;
+      }
+      // else fallthrough
+    default:
+      ++p;
+      break;
+    }
+  }
+  out.append(&*last, p - last);
+}
+
 namespace detail {
 
 /*
@@ -211,7 +300,7 @@ template<> struct OutputConverter<StringPiece> {
 template<class OutStringT, class DelimT, class OutputIterator>
 void internalSplit(DelimT delim, StringPiece sp, OutputIterator out,
     bool ignoreEmpty) {
-  assert(sp.start() != nullptr);
+  assert(sp.empty() || sp.start() != nullptr);
 
   const char* s = sp.start();
   const size_t strSize = sp.size();
@@ -231,9 +320,9 @@ 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));
@@ -246,9 +335,8 @@ 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));
   }
 }
@@ -258,6 +346,50 @@ template<class String> StringPiece prepareDelim(const String& s) {
 }
 inline char prepareDelim(char c) { return c; }
 
+template <class Dst>
+struct convertTo {
+  template <class Src>
+  static Dst from(const Src& src) { return folly::to<Dst>(src); }
+  static Dst from(const Dst& src) { return src; }
+};
+
+template<bool exact,
+         class Delim,
+         class OutputType>
+typename std::enable_if<IsSplitTargetType<OutputType>::value, bool>::type
+splitFixed(const Delim& delimiter,
+           StringPiece input,
+           OutputType& out) {
+  if (exact && UNLIKELY(std::string::npos != input.find(delimiter))) {
+    return false;
+  }
+  out = convertTo<OutputType>::from(input);
+  return true;
+}
+
+template<bool exact,
+         class Delim,
+         class OutputType,
+         class... OutputTypes>
+typename std::enable_if<IsSplitTargetType<OutputType>::value, bool>::type
+splitFixed(const Delim& delimiter,
+           StringPiece input,
+           OutputType& outHead,
+           OutputTypes&... outTail) {
+  size_t cut = input.find(delimiter);
+  if (UNLIKELY(cut == std::string::npos)) {
+    return false;
+  }
+  StringPiece head(input.begin(), input.begin() + cut);
+  StringPiece tail(input.begin() + cut + detail::delimSize(delimiter),
+                   input.end());
+  if (LIKELY(splitFixed<exact>(delimiter, tail, outTail...))) {
+    outHead = convertTo<OutputType>::from(head);
+    return true;
+  }
+  return false;
+}
+
 }
 
 //////////////////////////////////////////////////////////////////////
@@ -278,7 +410,7 @@ template<class Delim, class String, class OutputType>
 void split(const Delim& delimiter,
            const String& input,
            fbvector<OutputType>& out,
-           bool ignoreEmpty = false) {
+           bool ignoreEmpty) {
   detail::internalSplit<OutputType>(
     detail::prepareDelim(delimiter),
     StringPiece(input),
@@ -299,11 +431,38 @@ void splitTo(const Delim& delimiter,
     ignoreEmpty);
 }
 
+template<bool exact,
+         class Delim,
+         class OutputType,
+         class... OutputTypes>
+typename std::enable_if<IsSplitTargetType<OutputType>::value, bool>::type
+split(const Delim& delimiter,
+      StringPiece input,
+      OutputType& outHead,
+      OutputTypes&... outTail) {
+  return detail::splitFixed<exact>(
+    detail::prepareDelim(delimiter),
+    input,
+    outHead,
+    outTail...);
+}
+
 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 <class T> struct IsSizableString {
+  enum { value = IsSomeString<T>::value
+         || std::is_same<T, StringPiece>::value };
+};
+
 template <class Iterator>
-struct IsStringContainerIterator :
-  IsSomeString<typename std::iterator_traits<Iterator>::value_type> {
+struct IsSizableStringContainerIterator :
+  IsSizableString<typename std::iterator_traits<Iterator>::value_type> {
 };
 
 template <class Delim, class Iterator, class String>
@@ -312,6 +471,11 @@ void internalJoinAppend(Delim delimiter,
                         Iterator end,
                         String& output) {
   assert(begin != end);
+  if (std::is_same<Delim, StringPiece>::value &&
+      delimSize(delimiter) == 1) {
+    internalJoinAppend(delimFront(delimiter), begin, end, output);
+    return;
+  }
   toAppend(*begin, &output);
   while (++begin != end) {
     toAppend(delimiter, *begin, &output);
@@ -319,7 +483,7 @@ void internalJoinAppend(Delim delimiter,
 }
 
 template <class Delim, class Iterator, class String>
-typename std::enable_if<IsStringContainerIterator<Iterator>::value>::type
+typename std::enable_if<IsSizableStringContainerIterator<Iterator>::value>::type
 internalJoin(Delim delimiter,
              Iterator begin,
              Iterator end,
@@ -339,7 +503,8 @@ internalJoin(Delim delimiter,
 }
 
 template <class Delim, class Iterator, class String>
-typename std::enable_if<!IsStringContainerIterator<Iterator>::value>::type
+typename
+std::enable_if<!IsSizableStringContainerIterator<Iterator>::value>::type
 internalJoin(Delim delimiter,
              Iterator begin,
              Iterator end,
@@ -446,13 +611,13 @@ void humanify(const String1& input, String2& output) {
 
 template<class InputString, class OutputString>
 bool hexlify(const InputString& input, OutputString& output,
-             bool append_output=false) {
+             bool append_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];
@@ -474,7 +639,7 @@ bool unhexlify(const InputString& input, OutputString& output) {
            -1;
   };
 
-  for (int i = 0; i < input.size(); i += 2) {
+  for (size_t i = 0; i < input.size(); i += 2) {
     int highBits = unhex(input[i]);
     int lowBits = unhex(input[i + 1]);
     if (highBits < 0 || lowBits < 0) {
@@ -507,4 +672,3 @@ void hexDump(const void* ptr, size_t size, OutIt out) {
 }  // namespace folly
 
 #endif /* FOLLY_STRING_INL_H_ */
-