bug: not all true types can use winansiencoding. need to check cmaps

PDFWriter/AbstractWrittenFont.cpp

@@ -36,9 +36,10 @@
 
 using namespace PDFHummus;
 
-AbstractWrittenFont::AbstractWrittenFont(ObjectsContext* inObjectsContext)
+AbstractWrittenFont::AbstractWrittenFont(ObjectsContext* inObjectsContext, FreeTypeFaceWrapper* inFontInfo)
 {
 	mObjectsContext = inObjectsContext;
+	mFontInfo = inFontInfo;
 	mCIDRepresentation = NULL;
 	mANSIRepresentation = NULL;
 }

PDFWriter/AbstractWrittenFont.h

@@ -32,7 +32,7 @@ class PDFParser;
 class AbstractWrittenFont : public IWrittenFont
 {
 public:
-	AbstractWrittenFont(ObjectsContext* inObjectsContext);
+	AbstractWrittenFont(ObjectsContext* inObjectsContext, FreeTypeFaceWrapper* inFontInfo);
 	virtual ~AbstractWrittenFont(void);
 
 	virtual void AppendGlyphs(const GlyphUnicodeMappingList& inGlyphsList,
@@ -47,6 +47,7 @@ class AbstractWrittenFont : public IWrittenFont
 	WrittenFontRepresentation* mCIDRepresentation;
 	WrittenFontRepresentation* mANSIRepresentation;
 	ObjectsContext* mObjectsContext;
+	FreeTypeFaceWrapper* mFontInfo;
 
 	PDFHummus::EStatusCode WriteStateInDictionary(ObjectsContext* inStateWriter,DictionaryContext* inDerivedObjectDictionary);
 	PDFHummus::EStatusCode WriteStateAfterDictionary(ObjectsContext* inStateWriter);

PDFWriter/FreeTypeFaceWrapper.cpp

@@ -655,11 +655,11 @@ IWrittenFont* FreeTypeFaceWrapper::CreateWrittenFontObject(ObjectsContext* inObj
 			if(FT_Get_CID_Is_Internally_CID_Keyed(mFace,&isCID) != 0)
 				isCID = false;	
 
-			result = new WrittenFontCFF(inObjectsContext,isCID != 0, inFontIsToBeEmbedded); // CFF fonts should know if font is to be embedded, as the embedding code involves re-encoding of glyphs
+			result = new WrittenFontCFF(inObjectsContext, this,isCID != 0, inFontIsToBeEmbedded); // CFF fonts should know if font is to be embedded, as the embedding code involves re-encoding of glyphs
 		}
 		else if(strcmp(fontFormat,scTrueType) == 0)
 		{
-			result = new WrittenFontTrueType(inObjectsContext);
+			result = new WrittenFontTrueType(inObjectsContext, this);
 		}
 		else
 		{

PDFWriter/IWrittenFont.h

@@ -61,7 +61,7 @@ class IWrittenFont
 	/*
 		Write a font definition using the glyphs appended.
 	*/
-	virtual PDFHummus::EStatusCode WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo,bool inEmbedFont) = 0;
+	virtual PDFHummus::EStatusCode WriteFontDefinition(bool inEmbedFont) = 0;
 
 	// state read and write
 	virtual PDFHummus::EStatusCode WriteState(ObjectsContext* inStateWriter,ObjectIDType inObjectID) = 0;

PDFWriter/PDFUsedFont.cpp

@@ -122,7 +122,7 @@ EStatusCode PDFUsedFont::WriteFontDefinition()
     if(!mWrittenFont)
         return eSuccess;
     else
-        return mWrittenFont->WriteFontDefinition(mFaceWrapper, mEmbedFont);
+        return mWrittenFont->WriteFontDefinition(mEmbedFont);
 }
 
 EStatusCode PDFUsedFont::WriteState(ObjectsContext* inStateWriter,ObjectIDType inObjectID)

PDFWriter/WrittenFontCFF.cpp

@@ -34,7 +34,7 @@
 
 using namespace PDFHummus;
 
-WrittenFontCFF::WrittenFontCFF(ObjectsContext* inObjectsContext,bool inIsCID, bool inFontWillBeEmbedded):AbstractWrittenFont(inObjectsContext)
+WrittenFontCFF::WrittenFontCFF(ObjectsContext* inObjectsContext, FreeTypeFaceWrapper* inFontInfo, bool inIsCID, bool inFontWillBeEmbedded):AbstractWrittenFont(inObjectsContext, inFontInfo)
 {
 	mAvailablePositionsCount = 255;
 	mFreeList.push_back(UCharAndUChar(1,255)); 
@@ -161,7 +161,7 @@ unsigned char WrittenFontCFF::AllocateFromFreeList(unsigned int inGlyph)
 	return result;
 }
 
-EStatusCode WrittenFontCFF::WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo,bool inEmbedFont)
+EStatusCode WrittenFontCFF::WriteFontDefinition(bool inEmbedFont)
 {
 	EStatusCode status = PDFHummus::eSuccess;
 	do
@@ -170,7 +170,7 @@ EStatusCode WrittenFontCFF::WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo,
 		{
 			CFFANSIFontWriter fontWriter;
 
-			status = fontWriter.WriteFont(inFontInfo, mANSIRepresentation, mObjectsContext, inEmbedFont);
+			status = fontWriter.WriteFont(*mFontInfo, mANSIRepresentation, mObjectsContext, inEmbedFont);
 			if(status != PDFHummus::eSuccess)
 			{
 				TRACE_LOG("WrittenFontCFF::WriteFontDefinition, Failed to write Ansi font definition");
@@ -184,7 +184,7 @@ EStatusCode WrittenFontCFF::WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo,
 			CIDFontWriter fontWriter;
 			CFFDescendentFontWriter descendentFontWriter;
 
-			status = fontWriter.WriteFont(inFontInfo, mCIDRepresentation, mObjectsContext, &descendentFontWriter, inEmbedFont);
+			status = fontWriter.WriteFont(*mFontInfo, mCIDRepresentation, mObjectsContext, &descendentFontWriter, inEmbedFont);
 			if(status != PDFHummus::eSuccess)
 			{
 				TRACE_LOG("WrittenFontCFF::WriteFontDefinition, Failed to write CID font definition");

PDFWriter/WrittenFontCFF.h

@@ -33,11 +33,11 @@ typedef std::list<UCharAndUChar> UCharAndUCharList;
 class WrittenFontCFF : public AbstractWrittenFont
 {
 public:
-	WrittenFontCFF(ObjectsContext* inObjectsContext,bool inIsCID, bool inFontWillBeEmbedded);
+	WrittenFontCFF(ObjectsContext* inObjectsContext, FreeTypeFaceWrapper* inFontInfo, bool inIsCID, bool inFontWillBeEmbedded);
 	virtual ~WrittenFontCFF(void);
 
 
-	virtual PDFHummus::EStatusCode WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo, bool inEmbedFont);
+	virtual PDFHummus::EStatusCode WriteFontDefinition(bool inEmbedFont);
 
 	virtual PDFHummus::EStatusCode WriteState(ObjectsContext* inStateWriter,ObjectIDType inObjectId);
 	virtual PDFHummus::EStatusCode ReadState(PDFParser* inStateReader,ObjectIDType inObjectID);

PDFWriter/WrittenFontTrueType.cpp

@@ -29,11 +29,33 @@
 #include "PDFObjectCast.h"
 #include "PDFParser.h"
 #include "PDFDictionary.h"
+#include "FreeTypeFaceWrapper.h"
+
+#include <ft2build.h>
+#include FT_FREETYPE_H
 
 using namespace PDFHummus;
 
-WrittenFontTrueType::WrittenFontTrueType(ObjectsContext* inObjectsContext):AbstractWrittenFont(inObjectsContext)
+bool FontHasCmapsForWinAnsiEncoding(FT_Face font) {
+	// See PDF Reference 5.5.5 Character Encoding, Encodings for TrueType Fonts. When is mapping from glyph name is possible, and so non CID encoding can be built - only if font contains either of 
+	// two possible cmaps: win unicode bmp or macintosh roman. so check if got either of those.
+	for(FT_Int i = 0; i < font->num_charmaps; ++i) {
+		FT_CharMap charmap = font->charmaps[i];
+
+		if (charmap->platform_id == 3 && charmap->encoding_id == 1)
+			return true; // Windows Unicode BMP
+
+		if (charmap->platform_id == 1 && charmap->encoding_id == 0)
+			return true; // Macintosh Roman
+
+	}
+
+	return false;
+}
+
+WrittenFontTrueType::WrittenFontTrueType(ObjectsContext* inObjectsContext, FreeTypeFaceWrapper* inFontInfo):AbstractWrittenFont(inObjectsContext, inFontInfo)
 {
+	fontSupportsWinAnsiEncoding = FontHasCmapsForWinAnsiEncoding(*inFontInfo);
 }
 
 WrittenFontTrueType::~WrittenFontTrueType(void)
@@ -43,18 +65,19 @@ WrittenFontTrueType::~WrittenFontTrueType(void)
 /*
 here's what i'm deciding on:
 1. Can encoding if/f all text codes are available through WinAnsiEncoding.  
-[maybe should also make sure that the font has the relevant cmaps?! Or maybe I'm just assuming that...]
+[maybe should also make sure that the font has the relevant cmaps?! Or maybe I'm just assuming that...] [That's what FontHasCmapsForWinAnsiEncoding and fontSupportsWinAnsiEncoding are for]
 2. While encoding use WinAnsiEncoding values, of course. This will necasserily work
 3. While writing the font description simply write the WinAnsiEncoding glyph name, and pray.*/
 
-bool WrittenFontTrueType::AddToANSIRepresentation(	const GlyphUnicodeMappingList& inGlyphsList,
-													UShortList& outEncodedCharacters)
-{
+bool WrittenFontTrueType::AddANSICandidates(const GlyphUnicodeMappingList& inGlyphsList, UShortList& ioCandidates) {
+	if(!fontSupportsWinAnsiEncoding) {
+		return false;
+	}
+
 	// i'm totally relying on the text here, which is fine till i'll do ligatures, in which case
 	// i'll need to make something different out of the text.
 	// as you can see this has little to do with glyphs (mainly cause i can't use FreeType to map the glyphs
 	// back to the rleevant unicode values...but no need anyways...that's why i carry the text).
-	UShortList candidates;
 	BoolAndByte encodingResult(true,0);
 	WinAnsiEncoding winAnsiEncoding;
 	GlyphUnicodeMappingList::const_iterator it = inGlyphsList.begin(); 
@@ -81,11 +104,20 @@ bool WrittenFontTrueType::AddToANSIRepresentation(	const GlyphUnicodeMappingList
 		{
 			encodingResult = winAnsiEncoding.Encode(it->mUnicodeValues.front());
 			if(encodingResult.first)
-				candidates.push_back(encodingResult.second);
+				ioCandidates.push_back(encodingResult.second);
 		}
 	}
 
-	if(encodingResult.first)
+	return encodingResult.first;
+}
+
+bool WrittenFontTrueType::AddToANSIRepresentation(const GlyphUnicodeMappingList& inGlyphsList, UShortList& outEncodedCharacters)
+{
+	UShortList candidates;
+
+	bool result = AddANSICandidates(inGlyphsList, candidates);
+
+	if(result)
 	{
 		// for the first time, add also 0,0 mapping
 		if(mANSIRepresentation->mGlyphIDToEncodedChar.size() == 0)
@@ -104,11 +136,11 @@ bool WrittenFontTrueType::AddToANSIRepresentation(	const GlyphUnicodeMappingList
 		outEncodedCharacters = candidates;
 	}
 
-	return encodingResult.first;
+	return result;
 }
 
 
-EStatusCode WrittenFontTrueType::WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo,bool inEmbedFont)
+EStatusCode WrittenFontTrueType::WriteFontDefinition(bool inEmbedFont)
 {
 	EStatusCode status = PDFHummus::eSuccess;
 	do
@@ -117,7 +149,7 @@ EStatusCode WrittenFontTrueType::WriteFontDefinition(FreeTypeFaceWrapper& inFont
 		{
 			TrueTypeANSIFontWriter fontWriter;
 
-			status = fontWriter.WriteFont(inFontInfo, mANSIRepresentation, mObjectsContext, inEmbedFont);
+			status = fontWriter.WriteFont(*mFontInfo, mANSIRepresentation, mObjectsContext, inEmbedFont);
 			if(status != PDFHummus::eSuccess)
 			{
 				TRACE_LOG("WrittenFontTrueType::WriteFontDefinition, Failed to write Ansi font definition");
@@ -131,7 +163,7 @@ EStatusCode WrittenFontTrueType::WriteFontDefinition(FreeTypeFaceWrapper& inFont
 			CIDFontWriter fontWriter;
 			TrueTypeDescendentFontWriter descendentFontWriter;
 
-			status = fontWriter.WriteFont(inFontInfo, mCIDRepresentation, mObjectsContext, &descendentFontWriter, inEmbedFont);
+			status = fontWriter.WriteFont(*mFontInfo, mCIDRepresentation, mObjectsContext, &descendentFontWriter, inEmbedFont);
 			if(status != PDFHummus::eSuccess)
 			{
 				TRACE_LOG("WrittenFontTrueType::WriteFontDefinition, Failed to write CID font definition");
@@ -150,46 +182,20 @@ bool WrittenFontTrueType::AddToANSIRepresentation(	const GlyphUnicodeMappingList
 	UShortListList candidatesList;
 	UShortList candidates;
 	BoolAndByte encodingResult(true,0);
-	WinAnsiEncoding winAnsiEncoding;
 	GlyphUnicodeMappingListList::const_iterator itList = inGlyphsList.begin(); 
-	GlyphUnicodeMappingList::const_iterator it; 
+	bool result = true;
 
-	for(; itList != inGlyphsList.end() && encodingResult.first; ++itList)
+	for(; itList != inGlyphsList.end() && result; ++itList)
 	{
-		it = itList->begin();
-		for(; it != itList->end() && encodingResult.first; ++it)
-		{
-			// don't bother with characters of more or less than one unicode
-			if(it->mUnicodeValues.size() != 1)
-			{
-				encodingResult.first = false;
-			}
-			else if(0x2022 == it->mUnicodeValues.front())
-			{
-				// From the reference:
-				// In WinAnsiEncoding, all unused codes greater than 40 map to the bullet character. 
-				// However, only code 225 is specifically assigned to the bullet character; other codes are subject to future reassignment.
-
-				// now i don't know if it's related or not...but acrobat isn't happy when i'm using winansi with bullet. and text coming after that bullet may be
-				// corrupted.
-				// so i'm forcing CID if i hit bullet till i know better.
-				encodingResult.first = false;
-			}
-			else
-			{
-				encodingResult = winAnsiEncoding.Encode(it->mUnicodeValues.front());
-				if(encodingResult.first)
-					candidates.push_back(encodingResult.second);
-			}
-		}
-		if(encodingResult.first)
+		result = AddANSICandidates(*itList, candidates);
+		if(result)
 		{
 			candidatesList.push_back(candidates);
 			candidates.clear();
 		}
 	}
 
-	if(encodingResult.first)
+	if(result)
 	{
 		// for the first time, add also 0,0 mapping
 		if(mANSIRepresentation->mGlyphIDToEncodedChar.size() == 0)
@@ -216,7 +222,7 @@ bool WrittenFontTrueType::AddToANSIRepresentation(	const GlyphUnicodeMappingList
 		outEncodedCharacters = candidatesList;
 	}
 
-	return encodingResult.first;	
+	return result;	
 }
 
 EStatusCode WrittenFontTrueType::WriteState(ObjectsContext* inStateWriter,ObjectIDType inObjectID)
@@ -246,9 +252,9 @@ EStatusCode WrittenFontTrueType::ReadState(PDFParser* inStateReader,ObjectIDType
 	return AbstractWrittenFont::ReadStateFromObject(inStateReader,writtenFontState.GetPtr());
 }
 
-unsigned short WrittenFontTrueType::EncodeCIDGlyph(unsigned int inGlyphId) {
-	// Gal 26/8/2017: Most of the times, the glyph IDs are CIDs. this is to retain a few requirements of True type fonts, and the case of fonts when they are not embedded.
-	// However, when CFF fonts are embedded, the matching code actually recreates a font from just the subset, and renumbers them based on the order
-	// of them joining the font. Hence, we need a slight difference for this case, and an override is provided
-	return (unsigned short)inGlyphId;
+unsigned short WrittenFontTrueType::EncodeCIDGlyph(unsigned int inGlyphId) {
+	// Gal 26/8/2017: Most of the times, the glyph IDs are CIDs. this is to retain a few requirements of True type fonts, and the case of fonts when they are not embedded.
+	// However, when CFF fonts are embedded, the matching code actually recreates a font from just the subset, and renumbers them based on the order
+	// of them joining the font. Hence, we need a slight difference for this case, and an override is provided
+	return (unsigned short)inGlyphId;
 }

PDFWriter/WrittenFontTrueType.h

@@ -24,16 +24,18 @@
 class WrittenFontTrueType : public AbstractWrittenFont
 {
 public:
-	WrittenFontTrueType(ObjectsContext* inObjectsContext);
+	WrittenFontTrueType(ObjectsContext* inObjectsContext, FreeTypeFaceWrapper* inFontInfo);
 	~WrittenFontTrueType(void);
 
-	virtual PDFHummus::EStatusCode WriteFontDefinition(FreeTypeFaceWrapper& inFontInfo,bool inEmbedFont);
+	virtual PDFHummus::EStatusCode WriteFontDefinition(bool inEmbedFont);
 
 	virtual PDFHummus::EStatusCode WriteState(ObjectsContext* inStateWriter,ObjectIDType inObjectId);
 	virtual PDFHummus::EStatusCode ReadState(PDFParser* inStateReader,ObjectIDType inObjectID);
 
 
 private:
+    bool fontSupportsWinAnsiEncoding;
+
 	virtual bool AddToANSIRepresentation(	const GlyphUnicodeMappingList& inGlyphsList,
 											UShortList& outEncodedCharacters);
 
@@ -44,4 +46,8 @@ class WrittenFontTrueType : public AbstractWrittenFont
 
 	virtual unsigned short EncodeCIDGlyph(unsigned int inGlyphId);
 
+    bool AddANSICandidates(const GlyphUnicodeMappingList& inGlyphsList, UShortList& ioCandidates);
+
+
+
 };

PDFWriterTesting/CMakeLists.txt

@@ -67,6 +67,7 @@ create_test_sourcelist (Tests
   TIFFImageTest.cpp
   TiffSpecialsTest.cpp
   TimerTest.cpp
+  TrueTypeAnsiWriteBug.cpp
   TrueTypeTest.cpp
   TTCTest.cpp
   Type1Test.cpp