问题
Is there any simple and compatible GDI or .NET accessible subsystem of Windows that will give glyph position characters. The task here is the combining symbols such as those in Arabic which sometimes have chains of multiple combining symbols stacking on top of each others such as Arabic Fatha + Arabic Letter Superscript Alef + Arabic Maddah Above. The trouble is that though the X positions can be determined precisely with GDI GetCharacterPlacement, the Y position calculations which are derived from OpenType or TrueType font tables and anchors and a complex set of rules, is not available. Ultimately, to generate a PDF with Arabic properly formatted, the Y positions are needed and precisely. Studying Microsoft Word 2013's save as PDF feature, it is clear they have a way of properly gathering this data as studying the PDF details shows that each character is displayed at its precise position including the combining symbols.
WPF may contain some functions to do this in the GlyphRun class property GlyphOffsets. DirectWrite has the IDWriteTextAnalyzer interface which method GetGlyphPlacements can return DWRITE_GLYPH_OFFSETs and many other complex script information. Looking at the GDI Display and Printer Drive functions, STROBJ_bEnumPositionsOnly seems to return a set of GLYPHPOS structures with this information. GDI certainly renders this correctly under all circumstances if you send the full text to render but not if you want to do it glyph by glyph.
IXpsOMGlyphs in the XPS Object Model allows for a GetGlyphIndices call returning a set of XPS_GLYPH_INDEX gives horizontalOffset and verticalOffset though this library is hardly appropriate.
In the end the only appropriate library looks to be Uniscribe which is complex to use but supported since Internet Explorer 5 and Windows 2000 as opposed to all the other discussions beyond GDI which are generally Vista and later or requiring special dependencies. ScriptItemize returns an array of SCRIPT_STRING_ANALYSIS which can be passed to ScriptShape then ScriptPlace returning an array o GOFFSETs. In fact Uniscribe will given information about word breaks, diacritics, directional flow and many other aspects of what is happening in a complex script. I just wanted to know if there is a simpler method or if this is the minimum required and exact appropriate for such a task since Uniscribe appears to be extremely difficult to use directly from .NET and reasonably would need a C++ wrapper since there are a great deal of structures and pointers.
Update and answer: Uniscribe will not work for PDF purposes as it uses integers in GDI device units thus the accuracy is greatly compromised. Probably why Microsoft Word 2013 finally supports native PDF conversion support because ultimately DirectWrite seems to be relied upon. As mentioned below I posted both code solutions in .NET as tips on CodeProject. DirectWrite seems to be the only answer beyond designing a custom font shaping and calculation engine.
回答1:
Sample Uniscribe code in .NET since it is not available on the web currently:
_
Public Structure GCP_RESULTS
Public StructSize As UInteger
_
Public OutString As String
Public Order As IntPtr
Public Dx As IntPtr
Public CaretPos As IntPtr
Public [Class] As IntPtr
Public Glyphs As IntPtr
Public GlyphCount As UInteger
Public MaxFit As Integer
End Structure
_
Public Structure SCRIPT_CONTROL
Public ScriptControlFlags As UInteger
End Structure
_
Public Structure SCRIPT_STATE
Public ScriptStateFlags As UShort
End Structure
_
Public Structure SCRIPT_ANALYSIS
Public ScriptAnalysisFlags As UShort
Public s As SCRIPT_STATE
End Structure
_
Public Structure SCRIPT_VISATTR
Public ScriptVisAttrFlags As UShort
End Structure
_
Public Structure SCRIPT_ITEM
Public iCharPos As Integer
Public a As SCRIPT_ANALYSIS
End Structure
_
Public Structure GOFFSET
Public du As Integer
Public dv As Integer
End Structure
_
Public Structure ABC
Public abcA As Integer
Public abcB As UInteger
Public abcC As Integer
End Structure
Public Const E_OUTOFMEMORY As Integer = &H8007000E
Public Const E_PENDING As Integer = &H8000000A
Public Const USP_E_SCRIPT_NOT_IN_FONT As Integer = &H80040200
_
Public Shared Function GetCharacterPlacement(hdc As IntPtr, lpString As String, nCount As Integer, nMaxExtent As Integer, ByRef lpResults As GCP_RESULTS, dwFlags As UInteger) As UInteger
End Function
_
Public Shared Function ScriptItemize( wcInChars As String, cInChars As Integer, cMaxItems As Integer, psControl As SCRIPT_CONTROL, psState As SCRIPT_STATE, pItems() As SCRIPT_ITEM, ByRef pcItems As Integer) As Integer
End Function
_
Public Shared Function ScriptShape(hdc As IntPtr, ByRef psc As IntPtr, wcChars As String, cChars As Integer, cMaxGlyphs As Integer, ByRef psa As SCRIPT_ANALYSIS, wOutGlyphs() As UShort, wLogClust() As UShort, psva() As SCRIPT_VISATTR, ByRef cGlyphs As Integer) As Integer
End Function
_
Public Shared Function ScriptPlace(hdc As IntPtr, ByRef psc As IntPtr, wGlyphs() As UShort, cGlyphs As Integer, psva() As SCRIPT_VISATTR, ByRef psa As SCRIPT_ANALYSIS, iAdvance() As Integer, pGoffset() As GOFFSET, ByRef pABC As ABC) As Integer
End Function
_
Public Shared Function ScriptFreeCache(ByRef psc As IntPtr) As Integer
End Function
_
Public Shared Function GetDC(hWnd As IntPtr) As IntPtr
End Function
_
Public Shared Function ReleaseDC(hWnd As IntPtr, hdc As IntPtr) As Integer
End Function
_
Private Shared Function SelectObject(ByVal hdc As IntPtr, ByVal hObject As IntPtr) As IntPtr
End Function
Structure CharPosInfo
Public Index As Integer
Public Width As Integer
Public PriorWidth As Integer
Public X As Integer
Public Y As Integer
End Structure
Public Shared Function GetWordDiacriticPositions(Str As String, useFont As Font) As CharPosInfo()
Dim hdc As IntPtr
Dim CharPosInfos As New List(Of CharPosInfo)
hdc = GetDC(IntPtr.Zero) 'desktop device context
Dim oldFont As IntPtr = SelectObject(hdc, useFont.ToHfont())
Dim MaxItems As Integer = 16
Dim Control As New SCRIPT_CONTROL With {.ScriptControlFlags = 0}
Dim State As New SCRIPT_STATE With {.ScriptStateFlags = 1} '0 LTR, 1 RTL
Dim Items() As SCRIPT_ITEM = Nothing
Dim ItemCount As Integer
Dim Result As Integer
Do
ReDim Items(MaxItems - 1)
Result = ScriptItemize(Str, Str.Length, MaxItems, Control, State, Items, ItemCount)
If Result = 0 Then
ReDim Preserve Items(ItemCount) 'there is a dummy last item so adding one here
Exit Do
ElseIf Result = E_OUTOFMEMORY Then
End If
MaxItems *= 2
Loop While True
If Result = 0 Then
'last item is dummy item pointing to end of string
Dim Cache As IntPtr = IntPtr.Zero
For Count = 0 To ItemCount - 2
Dim Logs() As UShort = Nothing
Dim Glyphs() As UShort = Nothing
Dim VisAttrs() As SCRIPT_VISATTR = Nothing
ReDim Glyphs((Items(Count + 1).iCharPos - Items(Count).iCharPos) * 3 \ 2 + 16 - 1)
ReDim VisAttrs((Items(Count + 1).iCharPos - Items(Count).iCharPos) * 3 \ 2 + 16 - 1)
ReDim Logs(Items(Count + 1).iCharPos - Items(Count).iCharPos - 1)
Dim dc As IntPtr = IntPtr.Zero
Do
Dim GlyphsUsed As Integer
Result = ScriptShape(dc, Cache, Str.Substring(Items(Count).iCharPos), Items(Count + 1).iCharPos - Items(Count).iCharPos, Glyphs.Length, Items(Count).a, Glyphs, Logs, VisAttrs, GlyphsUsed)
If Result = 0 Then
ReDim Preserve Glyphs(GlyphsUsed - 1)
ReDim Preserve VisAttrs(GlyphsUsed - 1)
Exit Do
ElseIf Result = E_PENDING Then
dc = hdc
ElseIf Result = E_OUTOFMEMORY Then
ReDim Glyphs(Glyphs.Length * 2 - 1)
ReDim VisAttrs(VisAttrs.Length * 2 - 1)
ElseIf Result = USP_E_SCRIPT_NOT_IN_FONT Then
Else
End If
Loop While True
If Result = 0 Then
Dim Advances(Glyphs.Length - 1) As Integer
Dim Offsets(Glyphs.Length - 1) As GOFFSET
Dim abc As New ABC With {.abcA = 0, .abcB = 0, .abcC = 0}
dc = IntPtr.Zero
Do
Result = ScriptPlace(dc, Cache, Glyphs, Glyphs.Length, VisAttrs, Items(Count).a, Advances, Offsets, abc)
If Result E_PENDING Then Exit Do
dc = hdc
Loop While True
If Result = 0 Then
Dim LastPriorWidth As Integer = 0
Dim RunStart As Integer = 0
For CharCount = 0 To Logs.Length - 1
Dim PriorWidth As Integer = 0
Dim RunCount As Integer = 0
For ResCount As Integer = Logs(CharCount) To If(CharCount = Logs.Length - 1, 0, Logs(CharCount + 1)) Step -1
'fDiacritic or fZeroWidth
If (VisAttrs(ResCount).ScriptVisAttrFlags And (32 Or 64)) 0 Then
CharPosInfos.Add(New CharPosInfo With {.Index = RunStart + RunCount, .PriorWidth = LastPriorWidth, .Width = Advances(ResCount), .X = Offsets(ResCount).du, .Y = Offsets(ResCount).dv})
End If
If CharCount = Logs.Length - 1 OrElse Logs(CharCount) Logs(CharCount + 1) Then
PriorWidth += Advances(ResCount)
RunCount += 1
End If
Next
LastPriorWidth += PriorWidth
If CharCount = Logs.Length - 1 OrElse Logs(CharCount) Logs(CharCount + 1) Then
RunStart = CharCount + 1
End If
Next
End If
End If
Next
ScriptFreeCache(Cache)
End If
SelectObject(hdc, oldFont)
ReleaseDC(IntPtr.Zero, hdc)
Return CharPosInfos.ToArray()
End Function
来源：https://stackoverflow.com/questions/27024074/calculating-the-positions-of-glyphs-in-windows