|
/////////////////////////////////////////////////////////////////////////////
//
// File: SelectionWordBreaker.cs
//
// Description: Word breaker used for TextSelection's auto-word selection and
// ctl-arrow navigation.
//
/////////////////////////////////////////////////////////////////////////////
namespace System.Windows.Documents
{
using MS.Win32;
using MS.Internal; // Invariant
// Word breaker used for TextSelection's auto-word selection and ctl-arrow
// navigation.
//
// Unicode code points are broken down into classes, and with several exceptions
// word breaks are defined as locations where the classes of two consequative
// code points differ.
//
// This code is based on RichEdit's WB_MOVEWORDLEFT/RIGHT implementation.
// It supports east-asian and european scripts, but not south-east asian
// scripts such as Thai, Khmer, or Lao.
internal static class SelectionWordBreaker
{
//------------------------------------------------------
//
// Internal Methods
//
//------------------------------------------------------
#region Internal Methods
// Returns true if position points to a word break in the supplied
// char array. position is an inter-character offset -- 0 points
// to the space preceeding the first char, 1 points between the
// first and second char, etc.
//
// insideWordDirection specifies whether we're looking for a word start
// or word end. If insideWordDirection == LogicalDirection.Forward, then
// text = "abc def", position = 4 will return true, but if the direction is
// backward, no word boundary will be found (looking backward position is
// at the edge of whitespace, not a word).
//
// This method requires at least MinContextLength chars ahead of and
// following position to give accurate results, but no more.
internal static bool IsAtWordBoundary(char[] text, int position, LogicalDirection insideWordDirection)
{
CharClass[] classes = GetClasses(text);
// If the inside text is blank, it's not a word boundary.
if (insideWordDirection == LogicalDirection.Backward)
{
if (position == text.Length)
{
return true;
}
if (position == 0 || IsWhiteSpace(text[position - 1], classes[position - 1]))
{
return false;
}
}
else
{
if (position == 0)
{
return true;
}
if (position == text.Length || IsWhiteSpace(text[position], classes[position]))
{
return false;
}
}
UInt16[] charType3 = new UInt16[2];
SafeNativeMethods.GetStringTypeEx(0 /* ignored */, SafeNativeMethods.CT_CTYPE3, new char[] { text[position - 1], text[position] }, 2, charType3);
// Otherwise we're at a word boundary if the classes of the surrounding text differ.
return IsWordBoundary(text[position - 1], text[position]) ||
(
!IsSameClass(charType3[0], classes[position - 1], charType3[1], classes[position]) &&
!IsMidLetter(text, position - 1, classes) &&
!IsMidLetter(text, position, classes)
);
}
#endregion Internal Methods
//------------------------------------------------------
//
// Internal Properties
//
//------------------------------------------------------
#region Internal Properties
// The minimum char count required to give accurate word breaking
// results.
//
// This value specifies the count in each direction, so in general
// calls to IsAtWordBoundary will require at least MinContextLength*2
// chars surrounding the test position.
internal static int MinContextLength
{
get
{
return 2;
}
}
#endregion Internal Properties
//------------------------------------------------------
//
// Private Methods
//
//------------------------------------------------------
#region Private Methods
// Returns true if the position between a pair of consequative chars is
// always a word break.
private static bool IsWordBoundary(char previousChar, char followingChar)
{
bool isWordBoundary = false;
if (followingChar == CarriageReturnChar)
{
// xxCR
isWordBoundary = true;
}
return isWordBoundary;
}
// Returns true if the char specified by index is a MidLetter as defined
// by the Unicode Standard Annex #29. (Actually we use a subset of all
// possible MidLetter values.)
//
// MidLetters are exceptions to the rule that consequative characters
// with different classes are word breaks.
private static bool IsMidLetter(char []text, int index, CharClass []classes)
{
Invariant.Assert(text.Length == classes.Length);
return (text[index] == ApostropheChar || text[index] == RightSingleQuotationChar || text[index] == SoftHyphenChar) &&
(index > 0 && index + 1 < classes.Length) &&
((classes[index - 1] == CharClass.Alphanumeric && classes[index + 1] == CharClass.Alphanumeric) ||
(text[index] == QuotationMarkChar && IsHebrew(text[index - 1]) && IsHebrew(text[index + 1])));
}
// Returns true if the specified C3 type matches an east-asian code point.
private static bool IsIdeographicCharType(UInt16 charType3)
{
return (charType3 & (SafeNativeMethods.C3_KATAKANA | SafeNativeMethods.C3_HIRAGANA | SafeNativeMethods.C3_IDEOGRAPH)) != 0;
}
// Return true if two chars are in the same class.
// Ideographic chars are a special case -- each is considered to be
// unique except for several exceptions in japanese.
private static bool IsSameClass(UInt16 preceedingType3, CharClass preceedingClass,
UInt16 followingType3, CharClass followingClass)
{
const UInt16 IdeographicKanaTypes = SafeNativeMethods.C3_HALFWIDTH | SafeNativeMethods.C3_FULLWIDTH | SafeNativeMethods.C3_KATAKANA | SafeNativeMethods.C3_HIRAGANA;
const UInt16 IdeographicTypes = IdeographicKanaTypes | SafeNativeMethods.C3_IDEOGRAPH;
bool isSameClass;
// Assume just one of the two chars is ideographic, in which case
// the chars are not in the same class.
isSameClass = false;
if (IsIdeographicCharType(preceedingType3) && IsIdeographicCharType(followingType3))
{
// Both chars are ideographic.
UInt16 typeDelta = (UInt16)((preceedingType3 & IdeographicTypes) ^ (followingType3 & IdeographicTypes));
// Only a few japanese ideographic chars are considered the same class.
isSameClass = (preceedingType3 & IdeographicKanaTypes) != 0 &&
(typeDelta == 0 ||
typeDelta == SafeNativeMethods.C3_FULLWIDTH ||
typeDelta == SafeNativeMethods.C3_HIRAGANA ||
typeDelta == (SafeNativeMethods.C3_FULLWIDTH | SafeNativeMethods.C3_HIRAGANA));
}
else if (!IsIdeographicCharType(preceedingType3) && !IsIdeographicCharType(followingType3))
{
// Neither char is ideographic.
isSameClass = (preceedingClass & CharClass.WBF_CLASS) == (followingClass & CharClass.WBF_CLASS);
}
return isSameClass;
}
// Returns true is the specified char is whitespace.
private static bool IsWhiteSpace(char ch, CharClass charClass)
{
return (charClass & CharClass.WBF_CLASS) == CharClass.Blank && ch != ObjectReplacementChar;
}
// Computes the character classes for each char of an array of text.
private static CharClass[] GetClasses(char[] text)
{
CharClass[] classes = new CharClass[text.Length];
for (int i = 0; i < text.Length; i++)
{
CharClass classification;
char ch = text[i];
if (ch < 0x0100)
{
classification = (CharClass)_latinClasses[ch];
}
else if (IsKorean(ch))
{
classification = CharClass.Alphanumeric;
}
else if (IsThai(ch))
{
classification = CharClass.Alphanumeric;
}
else if (ch == ObjectReplacementChar)
{
classification = CharClass.Blank | CharClass.WBF_BREAKAFTER;
}
else
{
UInt16[] charType1 = new UInt16[1];
SafeNativeMethods.GetStringTypeEx(0 /* ignored */, SafeNativeMethods.CT_CTYPE1, new char[] { ch }, 1, charType1);
if ((charType1[0] & SafeNativeMethods.C1_SPACE) != 0)
{
if ((charType1[0] & SafeNativeMethods.C1_BLANK) != 0)
{
classification = CharClass.Blank | CharClass.WBF_ISWHITE;
}
else
{
classification = CharClass.WhiteSpace | CharClass.WBF_ISWHITE;
}
}
else if ((charType1[0] & SafeNativeMethods.C1_PUNCT) != 0 && !IsDiacriticOrKashida(ch))
{
classification = CharClass.Punctuation;
}
else
{
classification = CharClass.Alphanumeric;
}
}
classes[i] = classification;
}
return classes;
}
// Returns true if a char is a non-spacing diacritic or kashida.
private static bool IsDiacriticOrKashida(char ch)
{
UInt16 []charType3 = new UInt16[1];
SafeNativeMethods.GetStringTypeEx(0 /* ignored */, SafeNativeMethods.CT_CTYPE3, new char[] { ch }, 1, charType3);
return (charType3[0] & (SafeNativeMethods.C3_DIACRITIC | SafeNativeMethods.C3_NONSPACING | SafeNativeMethods.C3_VOWELMARK | SafeNativeMethods.C3_KASHIDA)) != 0;
}
// Returns true if a character falls within a specified code point range.
private static bool IsInRange(uint lower, char ch, uint upper)
{
return (lower <= (uint)ch && (uint)ch <= upper);
}
// Returns true if the specified char is a Korean char.
private static bool IsKorean(char ch)
{
return IsInRange(0xac00, ch, 0xd7ff);
}
// Returns true if the specified char is a Thai char.
private static bool IsThai(char ch)
{
return IsInRange(0x0e00, ch, 0x0e7f);
}
// Returns true if the specified char is a Hebrew char.
private static bool IsHebrew(char ch)
{
return IsInRange(0x05d0, ch, 0x05f2);
}
#endregion Private Methods
//------------------------------------------------------
//
// Private Fields
//
//------------------------------------------------------
#region Private Fields
// Unicode line feed char.
const char LineFeedChar = (char)0x000a;
// Unicode carriage return char.
const char CarriageReturnChar = (char)0x000d;
// Unicode quotation mark char.
const char QuotationMarkChar = (char)0x0022;
// Unicode apostrophe char.
const char ApostropheChar = (char)0x0027;
// Unicode soft hyphen char.
const char SoftHyphenChar = (char)0x00ad;
// Unicode right single quotation char.
const char RightSingleQuotationChar = (char)0x2019;
// Unicode object replacement char.
private const char ObjectReplacementChar = (char)0xfffc;
// A sub-set of the GetStringTypeEx C1 char classifications.
[Flags]
private enum CharClass : byte
{
// Low-order nibble is classification.
Alphanumeric = 0,
Punctuation = 1,
Blank = 2,
WhiteSpace = 4,
// High-order nibble holds attributes (matching rich edit's documented WBF flags).
WBF_CLASS = 0xf, // Mask for low order nibble.
WBF_ISWHITE = 0x10, // Whitespace char.
WBF_BREAKAFTER = 0x40, // Break char.
}
// Character classifications for u+0000 - u+00ff.
static readonly byte []_latinClasses = new byte[] {
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, //0x00
0x00, 0x13, 0x14, 0x14, 0x14, 0x14, 0x00, 0x00, //0x08
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x10
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x18
0x32, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, //0x20
0x01, 0x01, 0x01, 0x01, 0x01, 0x41, 0x01, 0x01, //0x28
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x30
0x00, 0x00, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, //0x38
0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x40
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x48
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x50
0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x01, 0x01, //0x58
0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x60
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x68
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x70
0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x01, 0x00, //0x78
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x80
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x88
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x90
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0x98
0x12, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, //0xA0
0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, //0xA8
0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, //0xB0
0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, //0xB8
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0xC0
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0xC8
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, //0xD0
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0xD8
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0xE0
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0xE8
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0xF0
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00};//0xF8
#endregion Private Fields
}
}
|