//
// This helper converts a serialized CFG grammar header into an in-memory header
//
internal static ScriptRef[] LoadScriptRefs(StreamMarshaler streamHelper, CfgSerializedHeader pFH)
{
//
// Because in 64-bit code, pointers != sizeof(ULONG) we copy each member explicitly.
//
if (pFH.FormatId != CfgGrammar._SPGDF_ContextFree)
{
return(null);
}
//We know that in SAPI 5.0 grammar format pszWords follows header immediately.
if (pFH.pszWords < Marshal.SizeOf <CfgSerializedHeader>())
{
// Must be SAPI 6.0 or above to hold a .NET script
return(null);
}
// Get the symbols
StringBlob symbols = LoadStringBlob(streamHelper, pFH.pszSymbols, pFH.cchSymbols);
// Get the script refs
CfgScriptRef[] cfgScripts = Load <CfgScriptRef>(streamHelper, pFH.pScripts, pFH.cScripts);
// Convert the CFG script reference to ScriptRef
ScriptRef[] scripts = new ScriptRef[cfgScripts.Length];
for (int i = 0; i < cfgScripts.Length; i++)
{
CfgScriptRef cfgScript = cfgScripts[i];
scripts[i] = new ScriptRef(symbols[cfgScript._idRule], symbols[cfgScript._idMethod], cfgScript._method);
}
return(scripts);
}
internal static ScriptRef[] LoadIL(Stream stream)
{
using (StreamMarshaler streamMarshaler = new StreamMarshaler(stream))
{
CfgSerializedHeader cfgSerializedHeader = new CfgSerializedHeader();
streamMarshaler.ReadStream(cfgSerializedHeader);
return(LoadScriptRefs(streamMarshaler, cfgSerializedHeader));
}
}
private static void CheckValidCfgFormat(CfgSerializedHeader pFH, CfgHeader header, bool includeAllGrammarData)
{
if (pFH.pszWords < 100)
{
XmlParser.ThrowSrgsException(SRID.UnsupportedFormat);
}
int start = (int)pFH.pszWords;
CheckSetOffsets(pFH.pszWords, pFH.cchWords * 2, ref start, pFH.ulTotalSerializedSize);
CheckSetOffsets(pFH.pszSymbols, pFH.cchSymbols * 2, ref start, pFH.ulTotalSerializedSize);
if (pFH.cRules > 0)
{
CheckSetOffsets(pFH.pRules, pFH.cRules * Marshal.SizeOf(typeof(CfgRule)), ref start, pFH.ulTotalSerializedSize);
}
if (pFH.cArcs > 0)
{
CheckSetOffsets(pFH.pArcs, pFH.cArcs * Marshal.SizeOf(typeof(CfgArc)), ref start, pFH.ulTotalSerializedSize);
}
if (pFH.pWeights != 0)
{
CheckSetOffsets(pFH.pWeights, pFH.cArcs * Marshal.SizeOf(typeof(float)), ref start, pFH.ulTotalSerializedSize);
}
if (pFH.cTags > 0)
{
CheckSetOffsets(pFH.tags, pFH.cTags * Marshal.SizeOf(typeof(CfgSemanticTag)), ref start, pFH.ulTotalSerializedSize);
if (includeAllGrammarData)
{
for (int i = 0; i < header.tags.Length; i++)
{
int startArcIndex = (int)header.tags[i].StartArcIndex;
int endArcIndex = (int)header.tags[i].EndArcIndex;
int num = header.arcs.Length;
if (startArcIndex == 0 || startArcIndex >= num || endArcIndex == 0 || endArcIndex >= num || (header.tags[i].PropVariantType != 0 && header.tags[i].PropVariantType == VarEnum.VT_BSTR && header.tags[i].PropVariantType == VarEnum.VT_BOOL && header.tags[i].PropVariantType == VarEnum.VT_R8 && header.tags[i].PropVariantType == VarEnum.VT_I4))
{
XmlParser.ThrowSrgsException(SRID.UnsupportedFormat);
}
}
}
}
if (pFH.cScripts > 0)
{
CheckSetOffsets(pFH.pScripts, pFH.cScripts * Marshal.SizeOf(typeof(CfgScriptRef)), ref start, pFH.ulTotalSerializedSize);
}
if (pFH.cIL > 0)
{
CheckSetOffsets(pFH.pIL, pFH.cIL * Marshal.SizeOf(typeof(byte)), ref start, pFH.ulTotalSerializedSize);
}
if (pFH.cPDB > 0)
{
CheckSetOffsets(pFH.pPDB, pFH.cPDB * Marshal.SizeOf(typeof(byte)), ref start, pFH.ulTotalSerializedSize);
}
}
internal static CfgHeader ConvertCfgHeader(StreamMarshaler streamHelper, bool includeAllGrammarData, bool loadSymbols, out CfgSerializedHeader cfgSerializedHeader)
{
cfgSerializedHeader = new CfgSerializedHeader(streamHelper.Stream);
CfgHeader cfgHeader = default(CfgHeader);
cfgHeader.FormatId = cfgSerializedHeader.FormatId;
cfgHeader.GrammarGUID = cfgSerializedHeader.GrammarGUID;
cfgHeader.langId = cfgSerializedHeader.LangID;
cfgHeader.pszGlobalTags = cfgSerializedHeader.pszSemanticInterpretationGlobals;
cfgHeader.cArcsInLargestState = cfgSerializedHeader.cArcsInLargestState;
cfgHeader.rules = Load <CfgRule>(streamHelper, cfgSerializedHeader.pRules, cfgSerializedHeader.cRules);
if (includeAllGrammarData || loadSymbols)
{
cfgHeader.pszSymbols = LoadStringBlob(streamHelper, cfgSerializedHeader.pszSymbols, cfgSerializedHeader.cchSymbols);
}
if (includeAllGrammarData)
{
cfgHeader.pszWords = LoadStringBlob(streamHelper, cfgSerializedHeader.pszWords, cfgSerializedHeader.cchWords);
cfgHeader.arcs = Load <CfgArc>(streamHelper, cfgSerializedHeader.pArcs, cfgSerializedHeader.cArcs);
cfgHeader.tags = Load <CfgSemanticTag>(streamHelper, cfgSerializedHeader.tags, cfgSerializedHeader.cTags);
cfgHeader.weights = Load <float>(streamHelper, cfgSerializedHeader.pWeights, cfgSerializedHeader.cArcs);
}
if (cfgSerializedHeader.pszWords < Marshal.SizeOf(typeof(CfgSerializedHeader)))
{
cfgHeader.ulRootRuleIndex = uint.MaxValue;
cfgHeader.GrammarOptions = GrammarOptions.KeyValuePairs;
cfgHeader.BasePath = null;
cfgHeader.GrammarMode = GrammarType.VoiceGrammar;
}
else
{
cfgHeader.ulRootRuleIndex = cfgSerializedHeader.ulRootRuleIndex;
cfgHeader.GrammarOptions = cfgSerializedHeader.GrammarOptions;
cfgHeader.GrammarMode = (GrammarType)cfgSerializedHeader.GrammarMode;
if (includeAllGrammarData)
{
cfgHeader.scripts = Load <CfgScriptRef>(streamHelper, cfgSerializedHeader.pScripts, cfgSerializedHeader.cScripts);
}
if (cfgSerializedHeader.cBasePath != 0)
{
streamHelper.Stream.Position = (int)cfgSerializedHeader.pRules + cfgHeader.rules.Length * Marshal.SizeOf(typeof(CfgRule));
cfgHeader.BasePath = streamHelper.ReadNullTerminatedString();
}
}
CheckValidCfgFormat(cfgSerializedHeader, cfgHeader, includeAllGrammarData);
return(cfgHeader);
}
internal static ScriptRef[] LoadScriptRefs(StreamMarshaler streamHelper, CfgSerializedHeader pFH)
{
if (pFH.FormatId != _SPGDF_ContextFree)
{
return(null);
}
if (pFH.pszWords < Marshal.SizeOf(typeof(CfgSerializedHeader)))
{
return(null);
}
StringBlob stringBlob = LoadStringBlob(streamHelper, pFH.pszSymbols, pFH.cchSymbols);
CfgScriptRef[] array = Load <CfgScriptRef>(streamHelper, pFH.pScripts, pFH.cScripts);
ScriptRef[] array2 = new ScriptRef[array.Length];
for (int i = 0; i < array.Length; i++)
{
CfgScriptRef cfgScriptRef = array[i];
array2[i] = new ScriptRef(stringBlob[cfgScriptRef._idRule], stringBlob[cfgScriptRef._idMethod], cfgScriptRef._method);
}
return(array2);
}
internal static bool LoadIL(Stream stream, out byte[] assemblyContent, out byte[] assemblyDebugSymbols, out ScriptRef[] scripts)
{
assemblyContent = (assemblyDebugSymbols = null);
scripts = null;
using (StreamMarshaler streamMarshaler = new StreamMarshaler(stream))
{
CfgSerializedHeader cfgSerializedHeader = new CfgSerializedHeader();
streamMarshaler.ReadStream(cfgSerializedHeader);
scripts = LoadScriptRefs(streamMarshaler, cfgSerializedHeader);
if (scripts == null)
{
return(false);
}
if (cfgSerializedHeader.cIL == 0)
{
return(false);
}
assemblyContent = Load <byte>(streamMarshaler, cfgSerializedHeader.pIL, cfgSerializedHeader.cIL);
assemblyDebugSymbols = ((cfgSerializedHeader.cPDB > 0) ? Load <byte>(streamMarshaler, cfgSerializedHeader.pPDB, cfgSerializedHeader.cPDB) : null);
}
return(true);
}
private static void CheckValidCfgFormat(CfgSerializedHeader pFH, CfgHeader header, bool includeAllGrammarData)
{
//See backend commit method to understand the layout of cfg format
if (pFH.pszWords < SP_SPCFGSERIALIZEDHEADER_500)
{
XmlParser.ThrowSrgsException(SRID.UnsupportedFormat);
}
int ullStartOffset = (int)pFH.pszWords;
//Check the word offset
//See stringblob implementation. pFH.cchWords * sizeof(WCHAR) isn't exactly the serialized size, but it is close and must be less than the serialized size
CheckSetOffsets(pFH.pszWords, pFH.cchWords * Helpers._sizeOfChar, ref ullStartOffset, pFH.ulTotalSerializedSize);
//Check the symbol offset
//symbol is right after word
//pFH.pszSymbols is very close to pFH.pszWords + pFH.cchWords * sizeof(WCHAR)
CheckSetOffsets(pFH.pszSymbols, pFH.cchSymbols * Helpers._sizeOfChar, ref ullStartOffset, pFH.ulTotalSerializedSize);
//Check the rule offset
if (pFH.cRules > 0)
{
CheckSetOffsets(pFH.pRules, pFH.cRules * Marshal.SizeOf <CfgRule>(), ref ullStartOffset, pFH.ulTotalSerializedSize);
}
//Check the arc offset
if (pFH.cArcs > 0)
{
CheckSetOffsets(pFH.pArcs, pFH.cArcs * Marshal.SizeOf <CfgArc>(), ref ullStartOffset, pFH.ulTotalSerializedSize);
}
//Check the weight offset
if (pFH.pWeights > 0)
{
CheckSetOffsets(pFH.pWeights, pFH.cArcs * sizeof(float), ref ullStartOffset, pFH.ulTotalSerializedSize);
}
//Check the semantic tag offset
if (pFH.cTags > 0)
{
CheckSetOffsets(pFH.tags, pFH.cTags * Marshal.SizeOf <CfgSemanticTag>(), ref ullStartOffset, pFH.ulTotalSerializedSize);
if (includeAllGrammarData)
{
//Validate the SPCFGSEMANTICTAG array pointed to by tags
//We use header for easy array access
//The first arc is dummy, so the start and end arcindex for semantic tag won't be zero
for (int i = 0; i < header.tags.Length; i++)
{
int startArc = (int)header.tags[i].StartArcIndex;
int endArc = (int)header.tags[i].EndArcIndex;
int cArcs = header.arcs.Length;
#pragma warning disable 0618 // VarEnum is obsolete
if (startArc == 0 ||
startArc >= cArcs ||
endArc == 0 ||
endArc >= cArcs ||
(
header.tags[i].PropVariantType != VarEnum.VT_EMPTY &&
header.tags[i].PropVariantType != VarEnum.VT_BSTR &&
header.tags[i].PropVariantType != VarEnum.VT_BOOL &&
header.tags[i].PropVariantType != VarEnum.VT_R8 &&
header.tags[i].PropVariantType != VarEnum.VT_I4)
)
{
XmlParser.ThrowSrgsException(SRID.UnsupportedFormat);
}
#pragma warning restore 0618
}
}
}
//Check the offset for the scripts
if (pFH.cScripts > 0)
{
CheckSetOffsets(pFH.pScripts, pFH.cScripts * Marshal.SizeOf <CfgScriptRef>(), ref ullStartOffset, pFH.ulTotalSerializedSize);
}
if (pFH.cIL > 0)
{
CheckSetOffsets(pFH.pIL, pFH.cIL * sizeof(byte), ref ullStartOffset, pFH.ulTotalSerializedSize);
}
if (pFH.cPDB > 0)
{
CheckSetOffsets(pFH.pPDB, pFH.cPDB * sizeof(byte), ref ullStartOffset, pFH.ulTotalSerializedSize);
}
}
internal static CfgHeader ConvertCfgHeader(StreamMarshaler streamHelper, bool includeAllGrammarData, bool loadSymbols, out CfgSerializedHeader cfgSerializedHeader)
{
cfgSerializedHeader = new CfgSerializedHeader(streamHelper.Stream);
//
// Because in 64-bit code, pointers != sizeof(ULONG) we copy each member explicitly.
//
CfgHeader header = new();
header.FormatId = cfgSerializedHeader.FormatId;
header.GrammarGUID = cfgSerializedHeader.GrammarGUID;
header.langId = cfgSerializedHeader.LangID;
header.pszGlobalTags = cfgSerializedHeader.pszSemanticInterpretationGlobals;
header.cArcsInLargestState = cfgSerializedHeader.cArcsInLargestState;
// read all the common fields
header.rules = Load <CfgRule>(streamHelper, cfgSerializedHeader.pRules, cfgSerializedHeader.cRules);
if (includeAllGrammarData || loadSymbols)
{
header.pszSymbols = LoadStringBlob(streamHelper, cfgSerializedHeader.pszSymbols, cfgSerializedHeader.cchSymbols);
}
if (includeAllGrammarData)
{
header.pszWords = LoadStringBlob(streamHelper, cfgSerializedHeader.pszWords, cfgSerializedHeader.cchWords);
header.arcs = Load <CfgArc>(streamHelper, cfgSerializedHeader.pArcs, cfgSerializedHeader.cArcs);
header.tags = Load <CfgSemanticTag>(streamHelper, cfgSerializedHeader.tags, cfgSerializedHeader.cTags);
header.weights = Load <float>(streamHelper, cfgSerializedHeader.pWeights, cfgSerializedHeader.cArcs);
}
//We know that in SAPI 5.0 grammar format pszWords follows header immediately.
if (cfgSerializedHeader.pszWords < Marshal.SizeOf <CfgSerializedHeader>())
{
//This is SAPI 5.0 and SAPI 5.1 grammar format
header.ulRootRuleIndex = 0xFFFFFFFF;
header.GrammarOptions = GrammarOptions.KeyValuePairs;
header.BasePath = null;
header.GrammarMode = GrammarType.VoiceGrammar;
}
else
{
//This is SAPI 5.2 and beyond grammar format
header.ulRootRuleIndex = cfgSerializedHeader.ulRootRuleIndex;
header.GrammarOptions = cfgSerializedHeader.GrammarOptions;
header.GrammarMode = (GrammarType)cfgSerializedHeader.GrammarMode;
if (includeAllGrammarData)
{
header.scripts = Load <CfgScriptRef>(streamHelper, cfgSerializedHeader.pScripts, cfgSerializedHeader.cScripts);
}
// The BasePath string is written after the rules - no offset is provided
// Get the chars and build the string
if (cfgSerializedHeader.cBasePath > 0)
{
streamHelper.Stream.Position = (int)cfgSerializedHeader.pRules + (header.rules.Length * Marshal.SizeOf <CfgRule>());
header.BasePath = streamHelper.ReadNullTerminatedString();
}
}
// Check the content - should be valid for both SAPI 5.0 and SAPI 5.2 grammars
CheckValidCfgFormat(cfgSerializedHeader, header, includeAllGrammarData);
return(header);
}
//
// This helper converts a serialized CFG grammar header into an in-memory header
//
internal static CfgHeader ConvertCfgHeader(StreamMarshaler streamHelper)
{
CfgSerializedHeader cfgSerializedHeader = null;
return(ConvertCfgHeader(streamHelper, true, true, out cfgSerializedHeader));
}
