gecko-dev/js/js2/utilities.h
2000-03-02 03:43:40 +00:00

620 lines
21 KiB
C++

// -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*-
//
// The contents of this file are subject to the Netscape Public
// License Version 1.1 (the "License"); you may not use this file
// except in compliance with the License. You may obtain a copy of
// the License at http://www.mozilla.org/NPL/
//
// Software distributed under the License is distributed on an "AS
// IS" basis, WITHOUT WARRANTY OF ANY KIND, either express oqr
// implied. See the License for the specific language governing
// rights and limitations under the License.
//
// The Original Code is the JavaScript 2 Prototype.
//
// The Initial Developer of the Original Code is Netscape
// Communications Corporation. Portions created by Netscape are
// Copyright (C) 1998 Netscape Communications Corporation. All
// Rights Reserved.
#ifndef utilities_h
#define utilities_h
#include "systemtypes.h"
#include <memory>
#include <new>
#include <string>
#include <iterator>
#include <iostream>
#ifndef _WIN32 // Microsoft Visual C++ 6.0 bug: standard identifiers should be in std namespace
using std::size_t;
using std::ptrdiff_t;
using std::strlen;
using std::strcpy;
#define STD std
#else
#define STD
#endif
using std::string;
using std::istream;
using std::ostream;
using std::auto_ptr;
namespace JavaScript {
//
// Assertions
//
#ifdef DEBUG
void Assert(const char *s, const char *file, int line);
#define ASSERT(_expr) ((_expr) ? (void)0 : JavaScript::Assert(#_expr, __FILE__, __LINE__))
#define NOT_REACHED(_reasonStr) JavaScript::Assert(_reasonStr, __FILE__, __LINE__)
#define DEBUG_ONLY(_stmt) _stmt
#else
#define ASSERT(expr)
#define NOT_REACHED(reasonStr)
#define DEBUG_ONLY(_stmt)
#endif
//
// Numerics
//
template<class N> N min(N v1, N v2) {return v1 <= v2 ? v1 : v2;}
template<class N> N max(N v1, N v2) {return v1 >= v2 ? v1 : v2;}
//
// Bit manipulation
//
#define JS_BIT(n) ((uint32)1 << (n))
#define JS_BITMASK(n) (JS_BIT(n) - 1)
uint ceilingLog2(uint32 n);
uint floorLog2(uint32 n);
//
// Unicode UTF-16 characters and strings
//
// A string of UTF-16 characters. Nulls are allowed just like any other character.
// The string is not null-terminated.
// Use wstring if char16 is wchar_t. Otherwise use basic_string<uint16>.
//
// Eventually we'll want to use a custom class better suited for JavaScript that generates less
// code bloat and separates the concepts of a fixed, read-only string from a mutable buffer that
// is expanding. For now, though, we use the standard basic_string.
typedef std::basic_string<char16> String;
typedef uint32 char16orEOF; // A type that can hold any char16 plus one special value: ueof.
const char16orEOF char16eof = static_cast<char16orEOF>(-1);
// If c is a char16, return it; if c is char16eof, return the character \uFFFF.
inline char16 char16orEOFToChar16(char16orEOF c) {return static_cast<char16>(c);}
// Special char16s
namespace uni {
const char16 null = '\0';
const char16 cr = '\r';
const char16 lf = '\n';
const char16 ls = 0x2028;
const char16 ps = 0x2029;
}
const uint16 firstFormatChar = 0x200C; // Lowest Unicode Cf character
inline char16 widen(char ch) {return static_cast<char16>(static_cast<uchar>(ch));}
#ifndef _WIN32
// Return a String containing the characters of the null-terminated C string cstr
// (without the trailing null).
inline String widenCString(const char *cstr)
{
size_t len = strlen(cstr);
const uchar *ucstr = reinterpret_cast<const uchar *>(cstr);
return String(ucstr, ucstr+len);
}
// Widen and append length characters starting at chars to the end of str.
inline void appendChars(String &str, const char *chars, size_t length)
{
const uchar *uchars = reinterpret_cast<const uchar *>(chars);
str.append(uchars, uchars + length);
}
#else // Microsoft VC6 bug: String constructor and append limited to char16 iterators
String widenCString(const char *cstr);
void appendChars(String &str, const char *chars, size_t length);
#endif
String &operator+=(String &str, const char *cstr);
String operator+(const String &str, const char *cstr);
String operator+(const char *cstr, const String &str);
inline String &operator+=(String &str, char c) {return str += widen(c);}
inline void clear(String &s) {s.resize(0);}
class CharInfo {
uint32 info; // Word from table a.
// Unicode character attribute lookup tables
static const uint8 x[];
static const uint8 y[];
static const uint32 a[];
public:
// Enumerated Unicode general category types
enum Type {
Unassigned = 0, // Cn
UppercaseLetter = 1, // Lu
LowercaseLetter = 2, // Ll
TitlecaseLetter = 3, // Lt
ModifierLetter = 4, // Lm
OtherLetter = 5, // Lo
NonSpacingMark = 6, // Mn
EnclosingMark = 7, // Me
CombiningSpacingMark = 8, // Mc
DecimalDigitNumber = 9, // Nd
LetterNumber = 10, // Nl
OtherNumber = 11, // No
SpaceSeparator = 12, // Zs
LineSeparator = 13, // Zl
ParagraphSeparator = 14, // Zp
Control = 15, // Cc
Format = 16, // Cf
PrivateUse = 18, // Co
Surrogate = 19, // Cs
DashPunctuation = 20, // Pd
StartPunctuation = 21, // Ps
EndPunctuation = 22, // Pe
ConnectorPunctuation = 23, // Pc
OtherPunctuation = 24, // Po
MathSymbol = 25, // Sm
CurrencySymbol = 26, // Sc
ModifierSymbol = 27, // Sk
OtherSymbol = 28 // So
};
enum Group {
NonIdGroup, // 0 May not be part of an identifier
FormatGroup, // 1 Format control
IdGroup, // 2 May start or continue a JS identifier (includes $ and _)
IdContinueGroup, // 3 May continue a JS identifier [(IdContinueGroup & -2) == IdGroup]
WhiteGroup, // 4 White space character (but not line break)
LineBreakGroup // 5 Line break character [(LineBreakGroup & -2) == WhiteGroup]
};
CharInfo() {}
CharInfo(char16 c): info(a[y[x[static_cast<uint16>(c)>>6]<<6 | c&0x3F]]) {}
CharInfo(const CharInfo &ci): info(ci.info) {}
friend Type cType(const CharInfo &ci) {return static_cast<Type>(ci.info & 0x1F);}
friend Group cGroup(const CharInfo &ci) {return static_cast<Group>(ci.info >> 16 & 7);}
friend bool isAlpha(const CharInfo &ci)
{
return ((1<<UppercaseLetter | 1<<LowercaseLetter | 1<<TitlecaseLetter | 1<<ModifierLetter | 1<<OtherLetter)
>> cType(ci) & 1) != 0;
}
friend bool isAlphanumeric(const CharInfo &ci)
{
return ((1<<UppercaseLetter | 1<<LowercaseLetter | 1<<TitlecaseLetter | 1<<ModifierLetter | 1<<OtherLetter |
1<<DecimalDigitNumber | 1<<LetterNumber)
>> cType(ci) & 1) != 0;
}
// Return true if this character can start a JavaScript identifier
friend bool isIdLeading(const CharInfo &ci) {return cGroup(ci) == IdGroup;}
// Return true if this character can continue a JavaScript identifier
friend bool isIdContinuing(const CharInfo &ci) {return (cGroup(ci) & -2) == IdGroup;}
// Return true if this character is a Unicode decimal digit (Nd) character
friend bool isDecimalDigit(const CharInfo &ci) {return cType(ci) == DecimalDigitNumber;}
// Return true if this character is a Unicode white space or line break character
friend bool isSpace(const CharInfo &ci) {return (cGroup(ci) & -2) == WhiteGroup;}
// Return true if this character is a Unicode line break character (LF, CR, LS, or PS)
friend bool isLineBreak(const CharInfo &ci) {return cGroup(ci) == LineBreakGroup;}
// Return true if this character is a Unicode format control character (Cf)
friend bool isFormat(const CharInfo &ci) {return cGroup(ci) == FormatGroup;}
friend bool isUpper(const CharInfo &ci) {return cType(ci) == UppercaseLetter;}
friend bool isLower(const CharInfo &ci) {return cType(ci) == LowercaseLetter;}
friend char16 toUpper(char16 c);
friend char16 toLower(char16 c);
};
inline bool isASCIIDecimalDigit(char16 c) {return c >= '0' && c <= '9';}
bool isASCIIHexDigit(char16 c, uint &digit);
const char16 *skipWhiteSpace(const char16 *str, const char16 *strEnd);
//
// Algorithms
//
// Assign zero to every element between first inclusive and last exclusive.
// This is equivalent ot fill(first, last, 0) but may be more efficient.
template<class For>
inline void zero(For first, For last)
{
while (first != last)
*first++ = 0;
}
// Assign zero to n elements starting at first.
// This is equivalent ot fill_n(first, n, 0) but may be more efficient.
template<class For, class Size>
inline void zero_n(For first, Size n)
{
while (n--)
*first++ = 0;
}
//
// Arenas
//
#ifndef _WIN32
// Pretend that obj points to a value of class T and call obj's destructor.
template<class T>
void classDestructor(void *obj)
{
static_cast<T *>(obj)->~T();
}
#else // Microsoft Visual C++ 6.0 bug workaround
template<class T>
struct DestructorHolder {
static void destroy(void *obj) {static_cast<T *>(obj)->~T();}
};
#endif
// An arena is a region of memory from which objects either derived from ArenaObject or allocated
// using a ArenaAllocator can be allocated. Deleting these objects individually runs the destructors,
// if any, but does not deallocate the memory. On the other hand, the entire arena can be deallocated
// as a whole.
//
// One may also allocate other objects in an arena by using the Arena specialization of the global
// operator new. However, be careful not to delete any such objects explicitly!
//
// Destructors can be registered for objects (or parts of objects) allocated in the arena. These
// destructors are called, in reverse order of being registered, at the time the arena is deallocated
// or cleared. When registering destructors for an object O be careful not to delete O manually because that
// would run its destructor twice.
class Arena {
struct Directory {
enum {maxNBlocks = 31};
Directory *next; // Next directory in linked list
uint nBlocks; // Number of blocks used in this directory
void *blocks[maxNBlocks]; // Pointers to data blocks; only the first nBlocks are valid
Directory(): nBlocks(0) {}
void clear();
};
struct DestructorEntry;
char *freeBegin; // Pointer to free bytes left in current block
char *freeEnd; // Pointer to end of free bytes left in current block
size_t blockSize; // Size of individual arena blocks
Directory *currentDirectory; // Directory in which the last block was allocated
Directory rootDirectory; // Initial directory; root of linked list of Directories
DestructorEntry *destructorEntries; // Linked list of destructor registrations, ordered from most to least recently registered
public:
explicit Arena(size_t blockSize = 1024);
private:
Arena(const Arena&); // No copy constructor
void operator=(const Arena&); // No assignment operator
public:
void clear();
~Arena() {clear();}
private:
void *newBlock(size_t size);
void newDestructorEntry(void (*destructor)(void *), void *object);
public:
void *allocate(size_t size);
// Ensure that object's destructor is called at the time the arena is deallocated or cleared.
// The destructors will be called in reverse order of being registered.
// registerDestructor might itself runs out of memory, in which case it immediately
// calls object's destructor before throwing bad_alloc.
#ifndef _WIN32
template<class T> void registerDestructor(T *object) {newDestructorEntry(&classDestructor<T>, object);}
#else
template<class T> void registerDestructor(T *object) {newDestructorEntry(&DestructorHolder<T>::destroy, object);}
#endif
};
// Objects derived from this class will be contained in the Arena passed to the new operator.
struct ArenaObject {
void *operator new(size_t size, Arena &arena) {return arena.allocate(size);}
void *operator new[](size_t size, Arena &arena) {return arena.allocate(size);}
#ifndef __MWERKS__ // Metrowerks 5.3 bug: These aren't supported yet
void operator delete(void *, Arena &) {}
void operator delete[](void *, Arena &) {}
#endif
private:
void operator delete(void *, size_t) {}
void operator delete[](void *) {}
};
// Objects allocated by passing this class to standard containers will be contained in the Arena
// passed to the ArenaAllocator's constructor.
template<class T>
class ArenaAllocator {
Arena &arena;
public:
typedef T value_type;
typedef size_t size_type;
typedef ptrdiff_t difference_type;
typedef T *pointer;
typedef const T *const_pointer;
typedef T &reference;
typedef const T &const_reference;
static pointer address(reference r) {return &r;}
static const_pointer address(const_reference r) {return &r;}
ArenaAllocator(Arena &arena): arena(arena) {}
template<class U> ArenaAllocator(const ArenaAllocator<U> &u): arena(u.arena) {}
pointer allocate(size_type n, const void *hint = 0) {return static_cast<pointer>(arena.allocate(n*sizeof(T)));}
static void deallocate(pointer, size_type) {}
static void construct(pointer p, const T &val) {new(p) T(val);}
static void destroy(pointer p) {p->~T();}
static size_type max_size() {return std::numeric_limits<size_type>::max() / sizeof(T);}
template<class U> struct rebind {typedef ArenaAllocator<U> other;};
};
String *newArenaString(Arena &arena);
String *newArenaString(Arena &arena, const String &str);
//
// Array auto_ptr's
//
// An ArrayAutoPtr holds a pointer to an array initialized by new T[x].
// A regular auto_ptr cannot be used here because it deletes its pointer using
// delete rather than delete[].
// An appropriate operator[] is also provided.
template <typename T>
class ArrayAutoPtr {
T *ptr;
public:
explicit ArrayAutoPtr(T *p = 0): ptr(p) {}
ArrayAutoPtr(ArrayAutoPtr &a): ptr(a.ptr) {a.ptr = 0;}
ArrayAutoPtr &operator=(ArrayAutoPtr &a) {reset(a.release());}
~ArrayAutoPtr() {delete[] ptr;}
T &operator*() const {return *ptr;}
T &operator->() const {return *ptr;}
template<class N> T &operator[](N i) const {return ptr[i];}
T *get() const {return ptr;}
T *release() {T *p = ptr; ptr = 0; return p;}
void reset(T *p = 0) {delete[] ptr; ptr = p;}
};
typedef ArrayAutoPtr<char> CharAutoPtr;
//
// Growable arrays
//
// private
template <typename T>
class ProtoArrayBuffer {
protected:
T *buffer;
int32 length;
int32 bufferSize;
void append(const T *elts, int32 nElts, T *cache);
};
// private
template <typename T>
void ProtoArrayBuffer<T>::append(const T *elts, int32 nElts, T *cache)
{
assert(nElts >= 0);
int32 newLength = length + nElts;
if (newLength > bufferSize) {
// Allocate a new buffer and copy the current buffer's contents there.
int32 newBufferSize = newLength + bufferSize;
auto_ptr<T> newBuffer = new T[newBufferSize];
T *p = buffer;
T *pLimit = old + length;
T *q = newBuffer.get();
while (p != pLimit)
*q++ = *p++;
if (buffer != cache)
delete buffer;
buffer = newBuffer.release();
bufferSize = newBufferSize;
}
length = newLength;
}
// An ArrayBuffer represents an array of elements of type T. The ArrayBuffer contains
// storage for a fixed size array of cacheSize elements; if this size is exceeded, the
// ArrayBuffer allocates the array from the heap.
// Use append to append nElts elements to the end of the ArrayBuffer.
template <typename T, int32 cacheSize>
class ArrayBuffer: public ProtoArrayBuffer<T> {
T cache[cacheSize];
public:
ArrayBuffer() {buffer = &cache; length = cacheSize; bufferSize = cacheSize;}
~ArrayBuffer() {if (buffer != &cache) delete buffer;}
int32 size() const {return length;}
T *front() const {return buffer;}
void append(const T *elts, int32 nElts) {ProtoArrayBuffer<T>::append(elts, nElts, cache);}
};
//
// Linked Lists
//
// In some cases it is desirable to manipulate ordinary C-style linked lists as though
// they were STL-like sequences. These classes define STL forward iterators that walk
// through singly-linked lists of objects threaded through fields named 'next'. The type
// parameter E must be a class that has a member named 'next' whose type is E* or const E*.
template <class E>
class ListIterator: public std::iterator<std::forward_iterator_tag, E> {
E *element;
public:
ListIterator() {}
explicit ListIterator(E *e): element(e) {}
E &operator*() const {return *element;}
E *operator->() const {return element;}
ListIterator &operator++() {element = element->next; return *this;}
ListIterator operator++(int) {ListIterator i(*this); element = element->next; return i;}
friend bool operator==(const ListIterator &i, const ListIterator &j) {return i.element == j.element;}
friend bool operator!=(const ListIterator &i, const ListIterator &j) {return i.element != j.element;}
};
template <class E>
#ifndef _WIN32 // Microsoft VC6 bug: std::iterator should support five template arguments
class ConstListIterator: public std::iterator<std::forward_iterator_tag, E, ptrdiff_t, const E*, const E&> {
#else
class ConstListIterator: public std::iterator<std::forward_iterator_tag, E, ptrdiff_t> {
#endif
const E *element;
public:
ConstListIterator() {}
ConstListIterator(const ListIterator<E> &i): element(&*i) {}
explicit ConstListIterator(const E *e): element(e) {}
const E &operator*() const {return *element;}
const E *operator->() const {return element;}
ConstListIterator &operator++() {element = element->next; return *this;}
ConstListIterator operator++(int) {ConstListIterator i(*this); element = element->next; return i;}
friend bool operator==(const ConstListIterator &i, const ConstListIterator &j) {return i.element == j.element;}
friend bool operator!=(const ConstListIterator &i, const ConstListIterator &j) {return i.element != j.element;}
};
//
// C++ I/O
//
// A class to remember the format of an ostream so that a function may modify it internally
// without changing it for the caller.
class SaveFormat {
#ifndef __GNUC__ // The GCC libraries don't support ios_base yet.
ostream &o;
std::ios_base::fmtflags flags;
char fill;
#endif
public:
explicit SaveFormat(ostream &out);
~SaveFormat();
};
void showChar(ostream &out, char16 ch);
template<class In>
void showString(ostream &out, In begin, In end)
{
while (begin != end)
showChar(out, *begin++);
}
void showString(ostream &out, const String &str);
//
// Source File Positions
//
// A FileOffset holds the raw, zero-based offset of a position in the source input.
// This offset is designed of easy indexing and depends on the format of the input.
// If the input is a String or array of char16, then this is merely a character index.
// If the input is utf-8, then this is a byte offset (which is very different from
// a character offset in this case!).
typedef uint32 FileOffset;
// A SourcePosition describes a character position in a source file or eval string.
struct SourcePosition {
FileOffset lineFileOffset; // Byte or character offset of start of source line relative to source file
uint32 lineNum; // One-based source line number
uint32 charPos; // Zero-based character offset of target character relative to the beginning of its source line
};
//
// Exceptions
//
// A JavaScript exception (other than out-of-memory, for which we use the standard C++
// exception bad_alloc).
struct Exception {
enum Kind {
SyntaxError
};
Kind kind; // The exception's kind
String message; // The detailed message
String sourceFile; // A description of the source code that caused the error
SourcePosition position; // Position of first character in token that caused the error
String sourceLine; // The text of the source line
Exception(Kind kind, const String &message): kind(kind), message(message) {position.lineNum = 0;}
Exception(Kind kind, const String &message, const String &sourceFile, SourcePosition &position, const String &sourceLine):
kind(kind), message(message), sourceFile(sourceFile), position(position), sourceLine(sourceLine) {}
const char *kindString() const;
String fullMessage() const;
};
}
inline void *operator new(size_t size, JavaScript::Arena &arena) {return arena.allocate(size);}
#ifndef _WIN32 // Microsoft Visual C++ 6.0 bug: new and new[] aren't distinguished
inline void *operator new[](size_t size, JavaScript::Arena &arena) {return arena.allocate(size);}
#endif
#ifndef __MWERKS__ // Metrowerks 5.3 bug: These aren't supported yet
// Global delete operators. These are only called in the rare cases that a constructor throws an exception
// and has to undo an operator new. An explicit delete statement will never invoke these.
inline void operator delete(void *, JavaScript::Arena &) {}
#ifndef _WIN32 // Microsoft Visual C++ 6.0 bug: new and new[] aren't distinguished
inline void operator delete[](void *, JavaScript::Arena &) {}
#endif
#endif
#endif