TTF sbit parser

# TTF sbit parser
# supports EBLC and EBDT table

# [email protected]
# Byounyoung Lee

import struct
import StringIO
import collections
import sys


def sUnpack( sData, size):
    return unpack( sData.read(size), size)

def unpack( data, size):
    if size == 1:
        return struct.unpack(">B", data)[0]
    elif size == 2:
        return struct.unpack(">H", data)[0]
    elif size == 4:
        return struct.unpack(">I", data)[0]
    else:
        print "ERROR : unpack size"
        return ""

class SbitParser:
    def __init__(self, ttfFilename ):
        self.tables = {}
        self.tableNames = ["EBLC", "EBDT" ]   # EBLC should be parsed before EBDT

        self.ttfFilename = ttfFilename
        self.ttf = open(ttfFilename, "rb").read()

        for tableName in self.tableNames:
            self.parseTables( tableName)
        return


    def parseTables(self,tableName):
        # table header
        idx = self.ttf.find(tableName)

        tName = self.ttf[idx:idx+4]

        tChecksum = self.ttf[idx+4:idx+8]
        tOffset = self.ttf[idx+8:idx+12]
        tSize = self.ttf[idx+12:idx+16]

        tChecksum = struct.unpack(">I", tChecksum)[0]
        tOffset = struct.unpack(">I", tOffset)[0]
        tSize  = struct.unpack(">I", tSize)[0]

        tData = self.ttf[tOffset:tOffset+tSize]
        self.tables[tableName] = {}

        self.tables[tableName]["checksum"] = tChecksum
        self.tables[tableName]["offset"] = tOffset
        self.tables[tableName]["size"] = tSize
        self.tables[tableName]["data"] = tData

        sData = StringIO.StringIO()
        sData.write(tData)
        sData.seek(0)

        d = getattr(self, "parse%s" % tName)(sData) # call each parse function
        self.tables[tName] = d

        self.dumpTable(d)
        return

    def dumpTable( self, d, depth = 1):
        indent = " " * depth * 4
        for key, value in d.iteritems():
            if type(value) == type(1): # integer
                print indent, "%s : 0x%x" % (key, value)

            elif type(value) == type([]): # list
                for i, k in enumerate(value):
                    print indent, "[-] %dth %s " % (i,key)
                    self.dumpTable(k, depth+1)
            else:
                binStr = " ".join(["%02x" % ord(c) for c in value])
                print indent, "%s : %s" % (key, binStr)
        return

    def __hBigGlyphMetrics(self, sData, d):
        d["height"] = sUnpack( sData, 1)
        d["width"] = sUnpack( sData, 1)
        d["horiBearingX"] = sUnpack( sData, 1)
        d["horiBearingY"] = sUnpack( sData, 1)
        d["horiAdvance"] = sUnpack( sData, 1)
        d["vertBearingX"] = sUnpack( sData, 1)
        d["vertBearingY"] = sUnpack( sData, 1)
        d["vertAdvance"] = sUnpack( sData, 1)
        return d

    def __hSmallGlyphMetrics(self, sData, d):
        d["height"] = sUnpack( sData, 1)
        d["width"] = sUnpack( sData, 1)
        d["BearingX"] = sUnpack( sData, 1)
        d["BearingY"] = sUnpack( sData, 1)
        d["Advance"] = sUnpack( sData, 1)
        return d


    def parseEBDT( self,sData):
        print "[*] parseEBDT"
        ebdt = collections.OrderedDict()

        ebdt["version"] = sUnpack(sData, 4)

        ebdt["metrics"] = []
        for bTable in self.tables["EBLC"]["bitmapSizeTable"]:
            for i in range(len(bTable["subArray"])):
                imageFormat = bTable["subArray"][i]["imageFormat"]

                imageDataOffset = bTable["subArray"][i]["imageDataOffset"]

                # now only supports 1 component
                arrayOffset0 = bTable["subArray"][i]["offsetArray0"]
                arrayOffset1 = bTable["subArray"][i]["offsetArray1"]

                offset = imageDataOffset + arrayOffset0

                pos = sData.tell()
                sData.seek(offset, 0)

                m = collections.OrderedDict()
                if imageFormat == 1:
                    self.__hSmallGlyphMetrics(sData, m)
                    imageSize = imageDataOffset + arrayOffset1 - sData.tell()
                    m["imageData"] = sData.read(imageSize)

                elif imageFormat == 8:
                    self.__hSmallGlyphMetrics(sData, m)
                    m["pad"] = sUnpack( sData, 1)
                    m["numComponents"] = sUnpack( sData, 2)
                    m["ComponentArray"] = []
                    for i in range(m["numComponents"]):
                        c = collections.OrderedDict()
                        c["glyphCode"] = sUnpack( sData, 2)
                        c["xOffset"] = sUnpack( sData, 1)
                        c["yOffset"] = sUnpack( sData, 1)
                        m["ComponentArray"].append( c)

                sData.seek(pos, 0)
                ebdt["metrics"].append( m)

        return ebdt

    def parseEBLC( self, sData):
        print "[*] parseEBLC"
        eblc = collections.OrderedDict()
        eblc["version"] = sUnpack( sData, 4)
        eblc["numSizes"] = sUnpack( sData, 4)

        eblc["bitmapSizeTable"] = []
        for i in range(eblc["numSizes"]):
            t = collections.OrderedDict()
            t["indexSubTableArrayOffset"] = sUnpack( sData, 4)
            t["indexTablesSize"] = sUnpack( sData, 4)
            t["numberOfIndexSubTables"] = sUnpack( sData, 4)
            t["colorRef"] = sUnpack( sData, 4)
            t["hori"] = sData.read(12) # hori
            t["vert"] = sData.read(12) # vert
            t["startGlyphIndex"] = sUnpack( sData, 2)
            t["endGlyphIndex"] = sUnpack( sData, 2)
            t["ppemX"] = sUnpack( sData, 1)
            t["ppemY"] = sUnpack( sData, 1)
            t["bitDepth"] = sUnpack( sData, 1)
            t["flags"] = sUnpack( sData, 1)

            #subtable
            oldPos = sData.tell()
            sData.seek(t["indexSubTableArrayOffset"], 0)

            t["subArray"] = []
            for j in range(t["numberOfIndexSubTables"]):
                s = collections.OrderedDict()
                s["firstGlyphIndex"] = sUnpack( sData, 2)
                s["lastGlyphIndex"] = sUnpack( sData, 2)
                s["additionalOffsetToIndexSubtable"] = sUnpack( sData, 4)

                oldPos2 = sData.tell()
                newPos = t["indexSubTableArrayOffset"]
                newPos += s["additionalOffsetToIndexSubtable"]

                sData.seek(newPos, 0)

                s["indexFormat"] = sUnpack( sData, 2)
                s["imageFormat"] = sUnpack( sData, 2)
                s["imageDataOffset"] = sUnpack( sData, 4)

                assert s["indexFormat"] == 3, "only supports indexFormat == 3"

                sizeOfArray = (s["lastGlyphIndex"] - s["firstGlyphIndex"] + 1) + 1

                for k in range(sizeOfArray):
                    s["offsetArray%d"%k] = sUnpack( sData, 2)

                sData.seek(oldPos2)

                t["subArray"].append(s)

            sData.seek(oldPos, 0)
            eblc["bitmapSizeTable"].append(t)

        return eblc

if __name__ == '__main__':
    ttfFilename = sys.argv[1]
    t = SbitParser(ttfFilename )