X-Git-Url: http://plrg.eecs.uci.edu/git/?p=folly.git;a=blobdiff_plain;f=folly%2FString.h;h=750863b986bbe52b4812aaa0ed8f84cc89052779;hp=604c0883ff4a31359b753747a54ecd2521a02056;hb=fe9d8cadb9bba9ddc7ea151f779458b8dc1904cd;hpb=2afd1f1acb7b864fc0e5f2791c5250c1fbc82396

diff --git a/folly/String.h b/folly/String.h
index 604c0883..750863b9 100644
--- a/folly/String.h
+++ b/folly/String.h
@@ -1,5 +1,5 @@
 /*
- * Copyright 2012 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.
@@ -14,22 +14,26 @@
  * limitations under the License.
  */
 
-#ifndef FOLLY_BASE_STRING_H_
-#define FOLLY_BASE_STRING_H_
+#pragma once
+#define FOLLY_STRING_H_
 
+#include <cstdarg>
+#include <exception>
 #include <string>
-#include <boost/type_traits.hpp>
+#include <unordered_map>
+#include <unordered_set>
+#include <vector>
 
-#ifdef __GNUC__
-# include <ext/hash_set>
-# include <ext/hash_map>
-#endif
+#include <boost/regex/pending/unicode_iterator.hpp>
+#include <boost/type_traits.hpp>
 
-#include "folly/Conv.h"
-#include "folly/FBString.h"
-#include "folly/FBVector.h"
-#include "folly/Range.h"
-#include "folly/ScopeGuard.h"
+#include <folly/Conv.h>
+#include <folly/ExceptionString.h>
+#include <folly/FBString.h>
+#include <folly/FBVector.h>
+#include <folly/Portability.h>
+#include <folly/Range.h>
+#include <folly/ScopeGuard.h>
 
 // Compatibility function, to make sure toStdString(s) can be called
 // to convert a std::string or fbstring variable s into type std::string
@@ -111,22 +115,82 @@ String cUnescape(StringPiece str, bool strict = true) {
   return out;
 }
 
+/**
+ * URI-escape a string.  Appends the result to the output string.
+ *
+ * Alphanumeric characters and other characters marked as "unreserved" in RFC
+ * 3986 ( -_.~ ) are left unchanged.  In PATH mode, the forward slash (/) is
+ * also left unchanged.  In QUERY mode, spaces are replaced by '+'.  All other
+ * characters are percent-encoded.
+ */
+enum class UriEscapeMode : unsigned char {
+  // The values are meaningful, see generate_escape_tables.py
+  ALL = 0,
+  QUERY = 1,
+  PATH = 2
+};
+template <class String>
+void uriEscape(StringPiece str,
+               String& out,
+               UriEscapeMode mode = UriEscapeMode::ALL);
+
+/**
+ * Similar to uriEscape above, but returns the escaped string.
+ */
+template <class String>
+String uriEscape(StringPiece str, UriEscapeMode mode = UriEscapeMode::ALL) {
+  String out;
+  uriEscape(str, out, mode);
+  return out;
+}
+
+/**
+ * URI-unescape a string.  Appends the result to the output string.
+ *
+ * In QUERY mode, '+' are replaced by space.  %XX sequences are decoded if
+ * XX is a valid hex sequence, otherwise we throw invalid_argument.
+ */
+template <class String>
+void uriUnescape(StringPiece str,
+                 String& out,
+                 UriEscapeMode mode = UriEscapeMode::ALL);
+
+/**
+ * Similar to uriUnescape above, but returns the unescaped string.
+ */
+template <class String>
+String uriUnescape(StringPiece str, UriEscapeMode mode = UriEscapeMode::ALL) {
+  String out;
+  uriUnescape(str, out, mode);
+  return out;
+}
+
 /**
  * stringPrintf is much like printf but deposits its result into a
  * string. Two signatures are supported: the first simply returns the
  * resulting string, and the second appends the produced characters to
  * the specified string and returns a reference to it.
  */
-std::string stringPrintf(const char* format, ...)
-  __attribute__ ((format (printf, 1, 2)));
+std::string stringPrintf(FOLLY_PRINTF_FORMAT const char* format, ...)
+  FOLLY_PRINTF_FORMAT_ATTR(1, 2);
+
+/* Similar to stringPrintf, with different signature. */
+void stringPrintf(std::string* out, FOLLY_PRINTF_FORMAT const char* fmt, ...)
+  FOLLY_PRINTF_FORMAT_ATTR(2, 3);
 
-/** Similar to stringPrintf, with different signiture.
-  */
-void stringPrintf(std::string* out, const char* fmt, ...)
-  __attribute__ ((format (printf, 2, 3)));
+std::string& stringAppendf(std::string* output,
+                          FOLLY_PRINTF_FORMAT const char* format, ...)
+  FOLLY_PRINTF_FORMAT_ATTR(2, 3);
 
-std::string& stringAppendf(std::string* output, const char* format, ...)
-  __attribute__ ((format (printf, 2, 3)));
+/**
+ * Similar to stringPrintf, but accepts a va_list argument.
+ *
+ * As with vsnprintf() itself, the value of ap is undefined after the call.
+ * These functions do not call va_end() on ap.
+ */
+std::string stringVPrintf(const char* format, va_list ap);
+void stringVPrintf(std::string* out, const char* format, va_list ap);
+std::string& stringVAppendf(std::string* out, const char* format, va_list ap);
 
 /**
  * Backslashify a string, that is, replace non-printable characters
@@ -147,12 +211,15 @@ std::string& stringAppendf(std::string* output, const char* format, ...)
  * C++, use cEscape instead.  This function is for display purposes
  * only.
  */
-template <class String1, class String2>
-void backslashify(const String1& input, String2& output, bool hex_style=false);
-
-template <class String>
-String backslashify(const String& input, bool hex_style=false) {
-  String output;
+template <class OutputString>
+void backslashify(
+    folly::StringPiece input,
+    OutputString& output,
+    bool hex_style = false);
+
+template <class OutputString = std::string>
+OutputString backslashify(StringPiece input, bool hex_style = false) {
+  OutputString output;
   backslashify(input, output, hex_style);
   return output;
 }
@@ -185,17 +252,43 @@ String humanify(const String& input) {
  * If append_output is true, append data to the output rather than
  * replace it.
  */
-template<class InputString, class OutputString>
+template <class InputString, class OutputString>
 bool hexlify(const InputString& input, OutputString& output,
              bool append=false);
 
+template <class OutputString = std::string>
+OutputString hexlify(ByteRange input) {
+  OutputString output;
+  if (!hexlify(input, output)) {
+    // hexlify() currently always returns true, so this can't really happen
+    throw std::runtime_error("hexlify failed");
+  }
+  return output;
+}
+
+template <class OutputString = std::string>
+OutputString hexlify(StringPiece input) {
+  return hexlify<OutputString>(ByteRange{input});
+}
+
 /**
  * Same functionality as Python's binascii.unhexlify.  Returns true
  * on successful conversion.
  */
-template<class InputString, class OutputString>
+template <class InputString, class OutputString>
 bool unhexlify(const InputString& input, OutputString& output);
 
+template <class OutputString = std::string>
+OutputString unhexlify(StringPiece input) {
+  OutputString output;
+  if (!unhexlify(input, output)) {
+    // unhexlify() fails if the input has non-hexidecimal characters,
+    // or if it doesn't consist of a whole number of bytes
+    throw std::domain_error("unhexlify() called with non-hex input");
+  }
+  return output;
+}
+
 /*
  * A pretty-printer for numbers that appends suffixes of units of the
  * given type.  It prints 4 sig-figs of value with the most
@@ -212,7 +305,8 @@ bool unhexlify(const InputString& input, OutputString& output);
  *   PRETTY_UNITS_METRIC - k, M, G, etc (goes up by 10^3 = 1000 each time)
  *   PRETTY_UNITS_BINARY - k, M, G, etc (goes up by 2^10 = 1024 each time)
  *   PRETTY_UNITS_BINARY_IEC - Ki, Mi, Gi, etc
- *
+ *   PRETTY_SI           - full SI metric prefixes from yocto to Yotta
+ *                         http://en.wikipedia.org/wiki/Metric_prefix
  * @author Mark Rabkin <mrabkin@fb.com>
  */
 enum PrettyType {
@@ -228,11 +322,38 @@ enum PrettyType {
   PRETTY_UNITS_BINARY,
   PRETTY_UNITS_BINARY_IEC,
 
-  PRETTY_NUM_TYPES
+  PRETTY_SI,
+  PRETTY_NUM_TYPES,
 };
 
 std::string prettyPrint(double val, PrettyType, bool addSpace = true);
 
+/**
+ * This utility converts StringPiece in pretty format (look above) to double,
+ * with progress information. Alters the  StringPiece parameter
+ * to get rid of the already-parsed characters.
+ * Expects string in form <floating point number> {space}* [<suffix>]
+ * If string is not in correct format, utility finds longest valid prefix and
+ * if there at least one, returns double value based on that prefix and
+ * modifies string to what is left after parsing. Throws and std::range_error
+ * exception if there is no correct parse.
+ * Examples(for PRETTY_UNITS_METRIC):
+ * '10M' => 10 000 000
+ * '10 M' => 10 000 000
+ * '10' => 10
+ * '10 Mx' => 10 000 000, prettyString == "x"
+ * 'abc' => throws std::range_error
+ */
+double prettyToDouble(folly::StringPiece *const prettyString,
+                      const PrettyType type);
+
+/*
+ * Same as prettyToDouble(folly::StringPiece*, PrettyType), but
+ * expects whole string to be correctly parseable. Throws std::range_error
+ * otherwise
+ */
+double prettyToDouble(folly::StringPiece prettyString, const PrettyType type);
+
 /**
  * Write a hex dump of size bytes starting at ptr to out.
  *
@@ -262,29 +383,6 @@ std::string hexDump(const void* ptr, size_t size);
  */
 fbstring errnoStr(int err);
 
-/**
- * Return the demangled (prettyfied) version of a C++ type.
- *
- * This function tries to produce a human-readable type, but the type name will
- * be returned unchanged in case of error or if demangling isn't supported on
- * your system.
- *
- * Use for debugging -- do not rely on demangle() returning anything useful.
- *
- * This function may allocate memory (and therefore throw).
- */
-fbstring demangle(const char* name);
-inline fbstring demangle(const std::type_info& type) {
-  return demangle(type.name());
-}
-
-/**
- * Debug string for an exception: include type and what().
- */
-inline fbstring exceptionStr(const std::exception& e) {
-  return folly::to<fbstring>(demangle(typeid(e)), ": ", e.what());
-}
-
 /*
  * Split a string into a list of tokens by delimiter.
  *
@@ -314,56 +412,229 @@ inline fbstring exceptionStr(const std::exception& e) {
  * or not (generating empty tokens).
  */
 
-template<class Delim, class String, class OutputType>
+template <class Delim, class String, class OutputType>
 void split(const Delim& delimiter,
            const String& input,
            std::vector<OutputType>& out,
-           bool ignoreEmpty = false);
+           const bool ignoreEmpty = false);
 
-template<class Delim, class String, class OutputType>
+template <class Delim, class String, class OutputType>
 void split(const Delim& delimiter,
            const String& input,
            folly::fbvector<OutputType>& out,
-           bool ignoreEmpty = false);
+           const bool ignoreEmpty = false);
 
-template<class OutputValueType, class Delim, class String,
-         class OutputIterator>
+template <
+    class OutputValueType,
+    class Delim,
+    class String,
+    class OutputIterator>
 void splitTo(const Delim& delimiter,
              const String& input,
              OutputIterator out,
-             bool ignoreEmpty = false);
+             const bool ignoreEmpty = false);
 
-} // namespace folly
+/*
+ * Split a string into a fixed number of string pieces and/or numeric types
+ * by delimiter. Conversions are supported for any type which folly:to<> can
+ * target, including all overloads of parseTo(). Returns 'true' if the fields
+ * were all successfully populated.  Returns 'false' if there were too few
+ * fields in the input, or too many fields if exact=true.  Casting exceptions
+ * will not be caught.
+ *
+ * Examples:
+ *
+ *  folly::StringPiece name, key, value;
+ *  if (folly::split('\t', line, name, key, value))
+ *    ...
+ *
+ *  folly::StringPiece name;
+ *  double value;
+ *  int id;
+ *  if (folly::split('\t', line, name, value, id))
+ *    ...
+ *
+ * The 'exact' template parameter specifies how the function behaves when too
+ * many fields are present in the input string. When 'exact' is set to its
+ * default value of 'true', a call to split will fail if the number of fields in
+ * the input string does not exactly match the number of output parameters
+ * passed. If 'exact' is overridden to 'false', all remaining fields will be
+ * stored, unsplit, in the last field, as shown below:
+ *
+ *  folly::StringPiece x, y.
+ *  if (folly::split<false>(':', "a:b:c", x, y))
+ *    assert(x == "a" && y == "b:c");
+ *
+ * Note that this will likely not work if the last field's target is of numeric
+ * type, in which case folly::to<> will throw an exception.
+ */
+template <class T, class Enable = void>
+struct IsConvertible {
+  enum { value = false };
+};
+
+template <class T>
+struct IsConvertible<
+    T,
+    decltype(static_cast<void>(
+        parseTo(std::declval<folly::StringPiece>(), std::declval<T&>())))> {
+  enum { value = true };
+};
 
-// Hash functions for string and fbstring usable with e.g. hash_map
-#ifdef __GNUC__
-namespace __gnu_cxx {
+template <class... Types>
+struct AllConvertible;
 
-template <class C>
-struct hash<folly::basic_fbstring<C> > : private hash<const C*> {
-  size_t operator()(const folly::basic_fbstring<C> & s) const {
-    return hash<const C*>::operator()(s.c_str());
-  }
+template <class Head, class... Tail>
+struct AllConvertible<Head, Tail...> {
+  enum { value = IsConvertible<Head>::value && AllConvertible<Tail...>::value };
 };
 
-template <class C>
-struct hash<std::basic_string<C> > : private hash<const C*> {
-  size_t operator()(const std::basic_string<C> & s) const {
-    return hash<const C*>::operator()(s.c_str());
-  }
+template <>
+struct AllConvertible<> {
+  enum { value = true };
 };
 
-} // namespace __gnu_cxx
-#endif
+static_assert(AllConvertible<float>::value, "");
+static_assert(AllConvertible<int>::value, "");
+static_assert(AllConvertible<bool>::value, "");
+static_assert(AllConvertible<int>::value, "");
+static_assert(!AllConvertible<std::vector<int>>::value, "");
 
-// Hook into boost's type traits
-namespace boost {
-template <class T>
-struct has_nothrow_constructor<folly::basic_fbstring<T> > : true_type {
-  enum { value = true };
+template <bool exact = true, class Delim, class... OutputTypes>
+typename std::enable_if<
+    AllConvertible<OutputTypes...>::value && sizeof...(OutputTypes) >= 1,
+    bool>::type
+split(const Delim& delimiter, StringPiece input, OutputTypes&... outputs);
+
+/*
+ * Join list of tokens.
+ *
+ * Stores a string representation of tokens in the same order with
+ * deliminer between each element.
+ */
+
+template <class Delim, class Iterator, class String>
+void join(const Delim& delimiter,
+          Iterator begin,
+          Iterator end,
+          String& output);
+
+template <class Delim, class Container, class String>
+void join(const Delim& delimiter,
+          const Container& container,
+          String& output) {
+  join(delimiter, container.begin(), container.end(), output);
+}
+
+template <class Delim, class Value, class String>
+void join(const Delim& delimiter,
+          const std::initializer_list<Value>& values,
+          String& output) {
+  join(delimiter, values.begin(), values.end(), output);
+}
+
+template <class Delim, class Container>
+std::string join(const Delim& delimiter,
+                 const Container& container) {
+  std::string output;
+  join(delimiter, container.begin(), container.end(), output);
+  return output;
+}
+
+template <class Delim, class Value>
+std::string join(const Delim& delimiter,
+                 const std::initializer_list<Value>& values) {
+  std::string output;
+  join(delimiter, values.begin(), values.end(), output);
+  return output;
+}
+
+template <
+    class Delim,
+    class Iterator,
+    typename std::enable_if<std::is_same<
+        typename std::iterator_traits<Iterator>::iterator_category,
+        std::random_access_iterator_tag>::value>::type* = nullptr>
+std::string join(const Delim& delimiter, Iterator begin, Iterator end) {
+  std::string output;
+  join(delimiter, begin, end, output);
+  return output;
+}
+
+/**
+ * Returns a subpiece with all whitespace removed from the front of @sp.
+ * Whitespace means any of [' ', '\n', '\r', '\t'].
+ */
+StringPiece ltrimWhitespace(StringPiece sp);
+
+/**
+ * Returns a subpiece with all whitespace removed from the back of @sp.
+ * Whitespace means any of [' ', '\n', '\r', '\t'].
+ */
+StringPiece rtrimWhitespace(StringPiece sp);
+
+/**
+ * Returns a subpiece with all whitespace removed from the back and front of @sp.
+ * Whitespace means any of [' ', '\n', '\r', '\t'].
+ */
+inline StringPiece trimWhitespace(StringPiece sp) {
+  return ltrimWhitespace(rtrimWhitespace(sp));
+}
+
+/**
+ * Returns a subpiece with all whitespace removed from the front of @sp.
+ * Whitespace means any of [' ', '\n', '\r', '\t'].
+ * DEPRECATED: @see ltrimWhitespace @see rtrimWhitespace
+ */
+inline StringPiece skipWhitespace(StringPiece sp) {
+  return ltrimWhitespace(sp);
+}
+
+/**
+ *  Strips the leading and the trailing whitespace-only lines. Then looks for
+ *  the least indented non-whitespace-only line and removes its amount of
+ *  leading whitespace from every line. Assumes leading whitespace is either all
+ *  spaces or all tabs.
+ *
+ *  Purpose: including a multiline string literal in source code, indented to
+ *  the level expected from context.
+ */
+std::string stripLeftMargin(std::string s);
+
+/**
+ * Fast, in-place lowercasing of ASCII alphabetic characters in strings.
+ * Leaves all other characters unchanged, including those with the 0x80
+ * bit set.
+ * @param str String to convert
+ * @param length Length of str, in bytes
+ */
+void toLowerAscii(char* str, size_t length);
+
+inline void toLowerAscii(MutableStringPiece str) {
+  toLowerAscii(str.begin(), str.size());
+}
+
+inline void toLowerAscii(std::string& str) {
+  // str[0] is legal also if the string is empty.
+  toLowerAscii(&str[0], str.size());
+}
+
+template <
+    class Iterator = const char*,
+    class Base = folly::Range<boost::u8_to_u32_iterator<Iterator>>>
+class UTF8Range : public Base {
+ public:
+  /* implicit */ UTF8Range(const folly::Range<Iterator> baseRange)
+      : Base(boost::u8_to_u32_iterator<Iterator>(
+                 baseRange.begin(), baseRange.begin(), baseRange.end()),
+             boost::u8_to_u32_iterator<Iterator>(
+                 baseRange.end(), baseRange.begin(), baseRange.end())) {}
+  /* implicit */ UTF8Range(const std::string& baseString)
+      : Base(folly::Range<Iterator>(baseString)) {}
 };
-} // namespace boost
 
-#include "folly/String-inl.h"
+using UTF8StringPiece = UTF8Range<const char*>;
+
+} // namespace folly
 
-#endif
+#include <folly/String-inl.h>