/*
- * Copyright 2014 Facebook, Inc.
+ * Copyright 2011-present Facebook, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* 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>
-
-#ifdef _GLIBCXX_SYMVER
-#include <ext/hash_set>
-#include <ext/hash_map>
-#endif
-
-#include <unordered_set>
#include <unordered_map>
+#include <unordered_set>
+#include <vector>
-#include "folly/Conv.h"
-#include "folly/Demangle.h"
-#include "folly/FBString.h"
-#include "folly/FBVector.h"
-#include "folly/Portability.h"
-#include "folly/Range.h"
-#include "folly/ScopeGuard.h"
+#include <boost/regex/pending/unicode_iterator.hpp>
+#include <boost/type_traits.hpp>
+
+#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>
+#include <folly/Traits.h>
// Compatibility function, to make sure toStdString(s) can be called
// to convert a std::string or fbstring variable s into type std::string
* 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 signiture.
- */
-void stringPrintf(std::string* out, const char* fmt, ...)
- __attribute__ ((format (printf, 2, 3)));
+/* Similar to stringPrintf, with different signature. */
+void stringPrintf(std::string* out, FOLLY_PRINTF_FORMAT const char* fmt, ...)
+ FOLLY_PRINTF_FORMAT_ATTR(2, 3);
-std::string& stringAppendf(std::string* output, const char* format, ...)
- __attribute__ ((format (printf, 2, 3)));
+std::string& stringAppendf(std::string* output,
+ FOLLY_PRINTF_FORMAT const char* format, ...)
+ FOLLY_PRINTF_FORMAT_ATTR(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
* 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;
}
* 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
* 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 {
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.
*
*/
fbstring errnoStr(int err);
-/**
- * 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());
-}
-
-inline fbstring exceptionStr(std::exception_ptr ep) {
- try {
- std::rethrow_exception(ep);
- } catch (const std::exception& e) {
- return exceptionStr(e);
- } catch (...) {
- return "<unknown exception>";
- }
-}
-
/*
* Split a string into a list of tokens by delimiter.
*
* 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);
/*
* Split a string into a fixed number of string pieces and/or numeric types
- * by delimiter. Any numeric type that folly::to<> can convert to from a
- * string piece is supported as a target. Returns 'true' if the fields were
- * all successfully populated.
+ * 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:
*
* 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>
-using IsSplitTargetType = std::integral_constant<bool,
- std::is_arithmetic<T>::value ||
- std::is_same<T, StringPiece>::value>;
-
-template<bool exact = true,
- 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);
+namespace detail {
+template <typename Void, typename OutputType>
+struct IsConvertible : std::false_type {};
+
+template <typename OutputType>
+struct IsConvertible<
+ void_t<decltype(parseTo(StringPiece{}, std::declval<OutputType&>()))>,
+ OutputType> : std::true_type {};
+} // namespace detail
+template <typename OutputType>
+struct IsConvertible : detail::IsConvertible<void, OutputType> {};
+
+template <bool exact = true, class Delim, class... OutputTypes>
+typename std::enable_if<
+ StrictConjunction<IsConvertible<OutputTypes>...>::value &&
+ sizeof...(OutputTypes) >= 1,
+ bool>::type
+split(const Delim& delimiter, StringPiece input, OutputTypes&... outputs);
/*
* Join list of tokens.
return output;
}
-} // namespace folly
+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;
+}
-// Hash functions to make std::string usable with e.g. hash_map
-//
-// Handle interaction with different C++ standard libraries, which
-// expect these types to be in different namespaces.
-namespace std {
+/**
+ * Returns a subpiece with all whitespace removed from the front of @sp.
+ * Whitespace means any of [' ', '\n', '\r', '\t'].
+ */
+StringPiece ltrimWhitespace(StringPiece sp);
-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());
- }
-};
+/**
+ * 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));
}
-#if defined(_GLIBCXX_SYMVER) && !defined(__BIONIC__)
-namespace __gnu_cxx {
+/**
+ * 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);
+}
-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());
- }
-};
+/**
+ * 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());
}
-#endif
-// Hook into boost's type traits
-namespace boost {
-template <class T>
-struct has_nothrow_constructor<folly::basic_fbstring<T> > : true_type {
- enum { value = true };
+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>