Merge pull request #125 from Decompollaborate/develop

1.15.0

pyproject.toml

@@ -4,7 +4,7 @@
 [project]
 name = "spimdisasm"
 # Version should be synced with spimdisasm/__init__.py
-version = "1.14.3"
+version = "1.15.0"
 description = "MIPS disassembler"
 # license = "MIT"
 readme = "README.md"

spimdisasm/__init__.py

@@ -5,17 +5,17 @@
 
 from __future__ import annotations
 
-__version_info__ = (1, 14, 3)
+__version_info__ = (1, 15, 0)
 __version__ = ".".join(map(str, __version_info__))
 __author__ = "Decompollaborate"
 
-from . import common
-from . import elf32
-from . import mips
+from . import common as common
+from . import elf32 as elf32
+from . import mips as mips
 
 # Front-end scripts
-from . import frontendCommon
-from . import disasmdis
-from . import rspDisasm
-from . import elfObjDisasm
-from . import singleFileDisasm
+from . import frontendCommon as frontendCommon
+from . import disasmdis as disasmdis
+from . import rspDisasm as rspDisasm
+from . import elfObjDisasm as elfObjDisasm
+from . import singleFileDisasm as singleFileDisasm

spimdisasm/common/ContextSymbols.py

@@ -89,7 +89,7 @@ def getAllTypes(self) -> set[str]:
 
 
 gKnownTypes: set[str] = {
-    "asciz", "char", "char*"
+    "asciz", "char", "char*", "String", "Char"
 }
 
 for kind in gAccessKinds.values():
@@ -122,6 +122,8 @@ class ContextSymbol:
 
     isMaybeString: bool = False
 
+    isMaybePascalString: bool = False
+
     referenceCounter: int = 0
     "How much this symbol is referenced by something else"
 
@@ -205,6 +207,11 @@ def hasNoType(self) -> bool:
         currentType = self.getTypeSpecial()
         return (currentType is None or currentType == "") and self.accessType is None
 
+    def hasOnlyAutodetectedType(self) -> bool:
+        if self.userDeclaredType is not None and self.userDeclaredType != "":
+            return False
+        return (self.autodetectedType is not None and self.autodetectedType != "") or self.accessType is not None
+
 
     def isTrustableFunction(self, rsp: bool=False) -> bool:
         """Checks if the function symbol should be trusted based on the current disassembler settings"""
@@ -263,16 +270,53 @@ def isShort(self) -> bool:
     def isString(self) -> bool:
         currentType = self.getTypeSpecial()
 
+        if self.sectionType == FileSectionType.Rodata:
+            stringGuesserLevel = GlobalConfig.RODATA_STRING_GUESSER_LEVEL
+        else:
+            stringGuesserLevel = GlobalConfig.DATA_STRING_GUESSER_LEVEL
+
         if currentType in {"char", "char*", "asciz"}:
             return True
         if not self.isMaybeString:
             return False
-        if not GlobalConfig.STRING_GUESSER:
+
+        if stringGuesserLevel < 1:
             return False
-        if self.hasNoType(): # no type information, let's try to guess
+
+        if self.hasNoType():
+            # no type information, let's try to guess
+            return True
+
+        if self.hasOnlyAutodetectedType():
+            if stringGuesserLevel >= 4:
+                # There's autodetected type information, but we are going to ignore it and try to guess
+                return True
+        return False
+
+    def isPascalString(self) -> bool:
+        currentType = self.getTypeSpecial()
+
+        if self.sectionType == FileSectionType.Rodata:
+            stringGuesserLevel = GlobalConfig.PASCAL_RODATA_STRING_GUESSER_LEVEL
+        else:
+            stringGuesserLevel = GlobalConfig.PASCAL_DATA_STRING_GUESSER_LEVEL
+
+        if currentType in {"String", "Char"}:
             return True
-        if GlobalConfig.AGGRESSIVE_STRING_GUESSER:
+        if not self.isMaybePascalString:
+            return False
+
+        if stringGuesserLevel < 1:
+            return False
+
+        if self.hasNoType():
+            # no type information, let's try to guess
             return True
+
+        if self.hasOnlyAutodetectedType():
+            if stringGuesserLevel >= 4:
+                # There's autodetected type information, but we are going to ignore it and try to guess
+                return True
         return False
 
     def isFloat(self) -> bool:
@@ -519,7 +563,7 @@ def getCsvHeader() -> str:
         output += "autodetectedSize,"
         output += "getSize,getVrom,sectionType,"
 
-        output += "isDefined,isUserDeclared,isAutogenerated,isMaybeString,"
+        output += "isDefined,isUserDeclared,isAutogenerated,isMaybeString,isMaybePascalString,"
         output += "referenceCounter,overlayCategory,unknownSegment,"
         output += "isGot,isGotGlobal,isGotLocal,gotIndex,"
         output += "firstLoAccess,isAutogeneratedPad,isElfNotype"
@@ -544,7 +588,7 @@ def toCsv(self) -> str:
         else:
             output += f"0x{self.autodetectedSize:X},"
         output += f"0x{self.getSize():X},0x{self.getVrom():X},{self.sectionType.toStr()},"
-        output += f"{self.isDefined},{self.isUserDeclared},{self.isAutogenerated},{self.isMaybeString},"
+        output += f"{self.isDefined},{self.isUserDeclared},{self.isAutogenerated},{self.isMaybeString},{self.isMaybePascalString},"
         output += f"{self.referenceCounter},{self.overlayCategory},{self.unknownSegment},"
         output += f"{self.isGot},{self.isGotGlobal},{self.isGotLocal},{self.gotIndex},"
         output += f"{self.firstLoAccess},{self.isAutogeneratedPad()},{self.isElfNotype}"