// Parse a block of C++ source code, adding any types found
public void AddFromDeclarationText(string text)
{
using StringReader lines = new StringReader(text);
var rgxForwardDecl = new Regex(@"(struct|union)\s+(\S+);");
var rgxTypedefAlias = new Regex(@"typedef\s+(?:(struct|union)\s+)?(\S+)\s+(\S+);");
var rgxTypedefFnPtr = new Regex(@"typedef\s+(?:struct\s+)?" + CppFnPtrType.Regex + ";");
var rgxDefinition = new Regex(@"^(typedef\s+)?(struct|union|enum)");
var rgxFieldFnPtr = new Regex(CppFnPtrType.Regex + @";");
var rgxField = new Regex(@"^(?:struct\s+|enum\s+)?(\S+?\s*\**)\s*((?:\S|\s*,\s*)+)(?:\s*:\s*([0-9]+))?;");
var rgxEnumValue = new Regex(@"^\s*([A-Za-z0-9_]+)(?:\s*=\s*(.+?))?,?\s*$");
var rgxIsConst = new Regex(@"\bconst\b");
var rgxStripKeywords = new Regex(@"\b(?:const|unsigned|volatile)\b");
var rgxCompressPtrs = new Regex(@"\*\s+\*");
var rgxArrayField = new Regex(@"(\S+?)\[([0-9]+)\]");
var rgxAlignment = new Regex(@"__attribute__\(\(aligned\(([0-9]+)\)\)\)\s+");
var rgxIsBitDirective = new Regex(@"#ifdef\s+IS_(32|64)BIT");
var rgxSingleLineComment = new Regex(@"/\*.*?\*/");
var currentType = new Stack <CppComplexType>();
bool falseIfBlock = false;
bool inComment = false;
bool inMethod = false;
bool inTypedef = false;
var nextEnumValue = 0ul;
string rawLine, line;
while ((rawLine = line = lines.ReadLine()) != null)
{
// Remove comments
if (line.Contains("//"))
{
line = line.Substring(0, line.IndexOf("//", StringComparison.Ordinal));
}
// End of multi-line comment?
if (line.Contains("*/") && inComment)
{
inComment = false;
line = line.Substring(line.IndexOf("*/", StringComparison.Ordinal) + 2);
}
if (inComment)
{
Debug.WriteLine($"[COMMENT ] {line}");
continue;
}
// Remove all single-line comments
line = rgxSingleLineComment.Replace(line, "");
// Start of multi-line comment?
if (line.Contains("/*") && !inComment)
{
inComment = true;
line = line.Substring(0, line.IndexOf("/*"));
}
// Ignore global variables
if (line.StartsWith("const ") && currentType.Count == 0)
{
Debug.WriteLine($"[GLOBAL ] {line}");
continue;
}
// Ignore methods
// Note: This is a very lazy way of processing early version IL2CPP headers
if (line != "}" && inMethod)
{
Debug.WriteLine($"[METHOD ] {line}");
continue;
}
if (line == "}" && inMethod)
{
inMethod = false;
Debug.WriteLine($"[METHOD END ] {line}");
continue;
}
if (line.StartsWith("static inline "))
{
inMethod = true;
Debug.WriteLine($"[METHOD START ] {line}");
continue;
}
// Remove keywords we don't care about
line = rgxStripKeywords.Replace(line, "");
// Remove whitespace in multiple indirections
line = rgxCompressPtrs.Replace(line, "**");
// Process __attribute((aligned(x)))
var alignment = 0;
var alignmentMatch = rgxAlignment.Match(line);
if (alignmentMatch.Success)
{
alignment = int.Parse(alignmentMatch.Groups[1].Captures[0].ToString());
line = rgxAlignment.Replace(line, "");
}
line = line.Trim();
// Ignore blank lines
if (line.Length == 0)
{
continue;
}
// Ignore defines
if (line.StartsWith("#define"))
{
continue;
}
// Process #ifdefs before anything else
// Doesn't handle nesting but we probably don't need to (use a Stack if we do)
var ifdef = rgxIsBitDirective.Match(line);
if (ifdef.Success)
{
var bits = int.Parse(ifdef.Groups[1].Captures[0].ToString());
if (bits != WordSize)
{
falseIfBlock = true;
}
Debug.WriteLine($"[IF ] {line}");
continue;
}
// Other #ifdef
if (line.StartsWith("#ifdef") || line.StartsWith("#if "))
{
falseIfBlock = true;
Debug.WriteLine($"[IF ] {line}");
continue;
}
if (line.StartsWith("#ifndef"))
{
falseIfBlock = false;
Debug.WriteLine($"[IF ] {line}");
continue;
}
if (line == "#else")
{
falseIfBlock = !falseIfBlock;
Debug.WriteLine($"[ELSE ] {line}");
continue;
}
if (line == "#endif")
{
falseIfBlock = false;
Debug.WriteLine($"[ENDIF ] {line}");
continue;
}
if (falseIfBlock)
{
Debug.WriteLine($"[FALSE ] {line}");
continue;
}
// Forward declaration
// <struct|union> <external-type>;
var externDecl = rgxForwardDecl.Match(line);
if (externDecl.Success)
{
var complexType = complexTypeMap[externDecl.Groups[1].Captures[0].ToString()];
var declType = externDecl.Groups[2].Captures[0].ToString();
switch (complexType)
{
case ComplexValueType.Struct: Struct(declType); break;
case ComplexValueType.Union: Union(declType); break;
}
Debug.WriteLine($"[FORWARD DECL ] {line}");
continue;
}
// Typedef alias
// typedef <struct|union> <existing-type> <alias>
var typedef = rgxTypedefAlias.Match(line);
if (typedef.Success)
{
//var complexType = complexTypeMap[typedef.Groups[1].Captures[0].ToString()];
var existingType = typedef.Groups[2].Captures[0].ToString();
var alias = typedef.Groups[3].Captures[0].ToString();
// Sometimes we might get multiple forward declarations for the same type
// Potential multiple indirection
var type = GetType(existingType);
// C++ allows the same typedef to be defined more than once
TypedefAliases.TryAdd(alias, type);
var pointers = line.Count(c => c == '*');
Debug.WriteLine($"[TYPEDEF {(pointers > 0 ? "PTR" : "VAL")} ] {line} -- Adding typedef from {type.Name} to {alias}");
continue;
}
// Function pointer alias
// typedef <retType> (*<alias>)(<args>);
typedef = rgxTypedefFnPtr.Match(line);
if (typedef.Success)
{
var alias = typedef.Groups[2].Captures[0].ToString();
var fnPtrType = CppFnPtrType.FromSignature(this, line);
fnPtrType.Group = currentGroup;
TypedefAliases.Add(alias, fnPtrType);
Debug.WriteLine($"[TYPEDEF FNPTR] {line} -- Adding method pointer typedef to {alias}");
continue;
}
// Start of struct/union/enum
// [typedef] <struct|union|enum> [optional-tag-name]
var definition = rgxDefinition.Match(line);
if (definition.Success && line.IndexOf(";", StringComparison.Ordinal) == -1 && currentType.Count == 0)
{
// Must have a name if not a typedef, might have a name if it is
var split = line.Split(' ');
if (split[0] == "typedef")
{
split = split.Skip(1).ToArray();
}
var name = split.Length > 1 && split[1] != "{" ? split[1] : "";
currentType.Push(complexTypeMap[split[0]] switch {
ComplexValueType.Struct => Struct(name),
ComplexValueType.Union => Union(name),
ComplexValueType.Enum => NewDefaultEnum(name),
_ => throw new InvalidOperationException("Unknown complex type")
});
// Parse a block of C++ source code, adding any types found
public void AddFromDeclarationText(string text)
{
using StringReader lines = new StringReader(text);
var rgxExternDecl = new Regex(@"struct (\S+);");
var rgxTypedefForwardDecl = new Regex(@"typedef struct (\S+) (\S+);");
var rgxTypedefFnPtr = new Regex(@"typedef\s+(?:struct )?" + CppFnPtrType.Regex + ";");
var rgxTypedef = new Regex(@"typedef (\S+?)\s*\**\s*(\S+);");
var rgxFieldFnPtr = new Regex(CppFnPtrType.Regex + @";");
var rgxField = new Regex(@"^(?:struct |enum )?(\S+?)\s*\**\s*((?:\S|\s*,\s*)+)(?:\s*:\s*([0-9]+))?;");
var rgxEnumValue = new Regex(@"^\s*([A-Za-z0-9_]+)(?:\s*=\s*(.+?))?,?\s*$");
var rgxStripKeywords = new Regex(@"\b(?:const|unsigned|volatile)\b");
var rgxCompressPtrs = new Regex(@"\*\s+\*");
var rgxArrayField = new Regex(@"(\S+?)\[([0-9]+)\]");
var rgxAlignment = new Regex(@"__attribute__\(\(aligned\(([0-9]+)\)\)\)");
var rgxIsBitDirective = new Regex(@"#ifdef\s+IS_(32|64)BIT");
var rgxSingleLineComment = new Regex(@"/\*.*?\*/");
var currentType = new Stack <CppComplexType>();
bool falseIfBlock = false;
bool inComment = false;
bool inMethod = false;
var nextEnumValue = 0ul;
string line;
while ((line = lines.ReadLine()) != null)
{
// Remove comments
if (line.Contains("//"))
{
line = line.Substring(0, line.IndexOf("//", StringComparison.Ordinal));
}
// End of multi-line comment?
if (line.Contains("*/") && inComment)
{
inComment = false;
line = line.Substring(line.IndexOf("*/", StringComparison.Ordinal) + 2);
}
if (inComment)
{
Debug.WriteLine($"[COMMENT ] {line}");
continue;
}
// Remove all single-line comments
line = rgxSingleLineComment.Replace(line, "");
// Start of multi-line comment?
if (line.Contains("/*") && !inComment)
{
inComment = true;
line = line.Substring(0, line.IndexOf("/*"));
}
// Ignore global variables
if (line.StartsWith("const ") && currentType.Count == 0)
{
Debug.WriteLine($"[GLOBAL ] {line}");
continue;
}
// Ignore methods
// Note: This is a very lazy way of processing early version IL2CPP headers
if (line != "}" && inMethod)
{
Debug.WriteLine($"[METHOD ] {line}");
continue;
}
if (line == "}" && inMethod)
{
inMethod = false;
Debug.WriteLine($"[METHOD END ] {line}");
continue;
}
if (line.StartsWith("static inline "))
{
inMethod = true;
Debug.WriteLine($"[METHOD START ] {line}");
continue;
}
// Remove keywords we don't care about
line = rgxStripKeywords.Replace(line, "");
// Remove whitespace in multiple indirections
line = rgxCompressPtrs.Replace(line, "**");
// Process __attribute((aligned(x)))
var alignment = 0;
var alignmentMatch = rgxAlignment.Match(line);
if (alignmentMatch.Success)
{
alignment = int.Parse(alignmentMatch.Groups[1].Captures[0].ToString());
line = rgxAlignment.Replace(line, "");
}
line = line.Trim();
// Ignore blank lines
if (line.Length == 0)
{
continue;
}
// Process #ifs before anything else
// Doesn't handle nesting but we probably don't need to (use a Stack if we do)
var ifdef = rgxIsBitDirective.Match(line);
if (ifdef.Success)
{
var bits = int.Parse(ifdef.Groups[1].Captures[0].ToString());
if (bits != WordSize)
{
falseIfBlock = true;
}
Debug.WriteLine($"[IF ] {line}");
continue;
}
if (line == "#else")
{
falseIfBlock = !falseIfBlock;
Debug.WriteLine($"[ELSE ] {line}");
continue;
}
if (line == "#endif")
{
falseIfBlock = false;
Debug.WriteLine($"[ENDIF ] {line}");
continue;
}
if (falseIfBlock)
{
Debug.WriteLine($"[FALSE ] {line}");
continue;
}
// External declaration
// struct <external-type>;
// NOTE: Unfortunately we're not going to ever know the size of this type
var externDecl = rgxExternDecl.Match(line);
if (externDecl.Success)
{
var declType = externDecl.Groups[1].Captures[0].ToString();
Types.Add(declType, CppType.NewStruct(declType));
Debug.WriteLine($"[EXTERN DECL ] {line}");
continue;
}
// Forward declaration
// typedef struct <struct-type> <alias>
var typedef = rgxTypedefForwardDecl.Match(line);
if (typedef.Success)
{
var alias = typedef.Groups[2].Captures[0].ToString();
var declType = typedef.Groups[1].Captures[0].ToString();
// Sometimes we might get multiple forward declarations for the same type
if (!Types.ContainsKey(declType))
{
Types.Add(declType, CppType.NewStruct(declType));
}
// Sometimes the alias might be the same name as the type (this is usually the case)
if (!Types.ContainsKey(alias))
{
Types.Add(alias, Types[declType].AsAlias(alias));
}
Debug.WriteLine($"[FORWARD DECL ] {line}");
continue;
}
// Function pointer
// typedef <retType> (*<alias>)(<args>);
typedef = rgxTypedefFnPtr.Match(line);
if (typedef.Success)
{
var alias = typedef.Groups[2].Captures[0].ToString();
var fnPtrType = CppFnPtrType.FromSignature(this, line);
fnPtrType.Name = alias;
Types.Add(alias, fnPtrType);
Debug.WriteLine($"[TYPEDEF FNPTR] {line} -- Adding method pointer typedef to {alias}");
continue;
}
// Alias
// typedef <targetType>[*..] <alias>;
typedef = rgxTypedef.Match(line);
if (typedef.Success)
{
var alias = typedef.Groups[2].Captures[0].ToString();
var existingType = typedef.Groups[1].Captures[0].ToString();
// Potential multiple indirection
var type = Types[existingType];
var pointers = line.Count(c => c == '*');
for (int i = 0; i < pointers; i++)
{
type = type.AsPointer(WordSize);
}
Types.Add(alias, type.AsAlias(alias));
Debug.WriteLine($"[TYPEDEF {(pointers > 0? "PTR":"VAL")} ] {line} -- Adding typedef from {type.Name} to {alias}");
continue;
}
// Start of struct
// typedef struct <optional-type-name>
if ((line.StartsWith("typedef struct") || line.StartsWith("struct ")) && line.IndexOf(";", StringComparison.Ordinal) == -1 &&
currentType.Count == 0)
{
currentType.Push(CppType.NewStruct());
if (line.StartsWith("struct "))
{
currentType.Peek().Name = line.Split(' ')[1];
}
Debug.WriteLine($"\n[STRUCT START ] {line}");
continue;
}
// Start of union
// typedef union <optional-type-name>
if (line.StartsWith("typedef union") && line.IndexOf(";", StringComparison.Ordinal) == -1)
{
currentType.Push(CppType.NewUnion());
Debug.WriteLine($"\n[UNION START ] {line}");
continue;
}
// Start of enum
// typedef enum <optional-type-name>
if (line.StartsWith("typedef enum") && line.IndexOf(";", StringComparison.Ordinal) == -1)
{
currentType.Push(NewDefaultEnum());
nextEnumValue = 0;
Debug.WriteLine($"\n[ENUM START ] {line}");
continue;
}
// Nested complex field
// struct <optional-type-name>
// union <optional-type-name>
var words = line.Split(' ');
if ((words[0] == "union" || words[0] == "struct") && words.Length <= 2)
{
currentType.Push(words[0] == "struct" ? CppType.NewStruct() : CppType.NewUnion());
Debug.WriteLine($"[FIELD START ] {line}");
continue;
}
// End of already named struct
if (line == "};" && currentType.Count == 1)
{
var ct = currentType.Pop();
if (!Types.ContainsKey(ct.Name))
{
Types.Add(ct.Name, ct);
}
else
{
((CppComplexType)Types[ct.Name]).Fields = ct.Fields;
}
Debug.WriteLine($"[STRUCT END ] {line} -- {ct.Name}\n");
continue;
}
// End of complex field, complex type or enum
// end of [typedef] struct/union/enum
if (line.StartsWith("}") && line.EndsWith(";"))
{
var name = line[1..^ 1].Trim();