Move Bin and SymInfo into their own files

tools/isledecomp/isledecomp/__init__.py

@@ -1,3 +1,5 @@
+from .bin import *
 from .dir import *
-from .utils import *
 from .parser import *
+from .syminfo import *
+from .utils import *

tools/isledecomp/isledecomp/bin.py

@@ -0,0 +1,47 @@
+import struct
+
+
+# Declare a class that can automatically convert virtual executable addresses
+# to file addresses
+class Bin:
+    def __init__(self, filename, logger):
+        self.logger = logger
+        self.logger.debug('Parsing headers of "%s"... ', filename)
+        self.filename = filename
+        self.file = None
+        self.imagebase = None
+        self.textvirt = None
+        self.textraw = None
+
+    def __enter__(self):
+        self.logger.debug(f"Bin {self.filename} Enter")
+        self.file = open(self.filename, "rb")
+
+        # HACK: Strictly, we should be parsing the header, but we know where
+        #      everything is in these two files so we just jump straight there
+
+        # Read ImageBase
+        self.file.seek(0xB4)
+        (self.imagebase,) = struct.unpack("<i", self.file.read(4))
+
+        # Read .text VirtualAddress
+        self.file.seek(0x184)
+        (self.textvirt,) = struct.unpack("<i", self.file.read(4))
+
+        # Read .text PointerToRawData
+        self.file.seek(0x18C)
+        (self.textraw,) = struct.unpack("<i", self.file.read(4))
+        self.logger.debug("... Parsing finished")
+        return self
+
+    def __exit__(self, exc_type, exc_value, exc_traceback):
+        self.logger.debug(f"Bin {self.filename} Exit")
+        if self.file:
+            self.file.close()
+
+    def get_addr(self, virt):
+        return virt - self.imagebase - self.textvirt + self.textraw
+
+    def read(self, offset, size):
+        self.file.seek(self.get_addr(offset))
+        return self.file.read(size)

tools/isledecomp/isledecomp/syminfo.py

@@ -0,0 +1,138 @@
+import os
+import subprocess
+from .utils import get_file_in_script_dir
+
+
+class RecompiledInfo:
+    addr = None
+    size = None
+    name = None
+    start = None
+
+
+# Declare a class that parses the output of cvdump for fast access later
+class SymInfo:
+    funcs = {}
+    lines = {}
+    names = {}
+
+    def __init__(self, pdb, sym_recompfile, sym_logger, sym_wine_path_converter=None):
+        self.logger = sym_logger
+        call = [get_file_in_script_dir("cvdump.exe"), "-l", "-s"]
+
+        if sym_wine_path_converter:
+            # Run cvdump through wine and convert path to Windows-friendly wine path
+            call.insert(0, "wine")
+            call.append(sym_wine_path_converter.get_wine_path(pdb))
+        else:
+            call.append(pdb)
+
+        self.logger.info("Parsing %s ...", pdb)
+        self.logger.debug("Command = %s", call)
+        line_dump = subprocess.check_output(call).decode("utf-8").split("\r\n")
+
+        current_section = None
+
+        self.logger.debug("Parsing output of cvdump.exe ...")
+
+        for i, line in enumerate(line_dump):
+            if line.startswith("***"):
+                current_section = line[4:]
+
+            if current_section == "SYMBOLS" and "S_GPROC32" in line:
+                sym_addr = int(line[26:34], 16)
+
+                info = RecompiledInfo()
+                info.addr = (
+                    sym_addr + sym_recompfile.imagebase + sym_recompfile.textvirt
+                )
+
+                use_dbg_offs = False
+                if use_dbg_offs:
+                    debug_offs = line_dump[i + 2]
+                    debug_start = int(debug_offs[22:30], 16)
+                    debug_end = int(debug_offs[43:], 16)
+
+                    info.start = debug_start
+                    info.size = debug_end - debug_start
+                else:
+                    info.start = 0
+                    info.size = int(line[41:49], 16)
+
+                info.name = line[77:]
+
+                self.names[info.name] = info
+                self.funcs[sym_addr] = info
+            elif (
+                current_section == "LINES"
+                and line.startswith("  ")
+                and not line.startswith("   ")
+            ):
+                sourcepath = line.split()[0]
+
+                if sym_wine_path_converter:
+                    # Convert filename to Unix path for file compare
+                    sourcepath = sym_wine_path_converter.get_unix_path(sourcepath)
+
+                if sourcepath not in self.lines:
+                    self.lines[sourcepath] = {}
+
+                j = i + 2
+                while True:
+                    ll = line_dump[j].split()
+                    if len(ll) == 0:
+                        break
+
+                    k = 0
+                    while k < len(ll):
+                        linenum = int(ll[k + 0])
+                        address = int(ll[k + 1], 16)
+                        if linenum not in self.lines[sourcepath]:
+                            self.lines[sourcepath][linenum] = address
+                        k += 2
+
+                    j += 1
+
+        self.logger.debug("... Parsing output of cvdump.exe finished")
+
+    def get_recompiled_address(self, filename, line):
+        recompiled_addr = None
+
+        self.logger.debug("Looking for %s:%s", filename, line)
+        filename_basename = os.path.basename(filename).lower()
+
+        for fn in self.lines:
+            # Sometimes a PDB is compiled with a relative path while we always have
+            # an absolute path. Therefore we must
+            try:
+                if os.path.basename(
+                    fn
+                ).lower() == filename_basename and os.path.samefile(fn, filename):
+                    filename = fn
+                    break
+            except FileNotFoundError:
+                continue
+
+        if filename in self.lines and line in self.lines[filename]:
+            recompiled_addr = self.lines[filename][line]
+
+            if recompiled_addr in self.funcs:
+                return self.funcs[recompiled_addr]
+            self.logger.error(
+                "Failed to find function symbol with address: %x", recompiled_addr
+            )
+            return None
+        self.logger.error(
+            "Failed to find function symbol with filename and line: %s:%s",
+            filename,
+            line,
+        )
+        return None
+
+    def get_recompiled_address_from_name(self, name):
+        self.logger.debug("Looking for %s", name)
+
+        if name in self.names:
+            return self.names[name]
+        self.logger.error("Failed to find function symbol with name: %s", name)
+        return None

tools/isledecomp/isledecomp/utils.py

@@ -1,5 +1,5 @@
-import struct
+import os
-
+import sys
 import colorama
 
 
@@ -24,57 +24,8 @@ def print_diff(udiff, plain):
     return has_diff
 
 
-# Declare a class that can automatically convert virtual executable addresses
+def get_file_in_script_dir(fn):
-# to file addresses
+    return os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])), fn)
-class Bin:
-    def __init__(self, filename, logger):
-        self.logger = logger
-        self.logger.debug('Parsing headers of "%s"... ', filename)
-        self.filename = filename
-        self.file = None
-        self.imagebase = None
-        self.textvirt = None
-        self.textraw = None
-
-    def __enter__(self):
-        self.logger.debug(f"Bin {self.filename} Enter")
-        self.file = open(self.filename, "rb")
-
-        # HACK: Strictly, we should be parsing the header, but we know where
-        #      everything is in these two files so we just jump straight there
-
-        # Read ImageBase
-        self.file.seek(0xB4)
-        (self.imagebase,) = struct.unpack("<i", self.file.read(4))
-
-        # Read .text VirtualAddress
-        self.file.seek(0x184)
-        (self.textvirt,) = struct.unpack("<i", self.file.read(4))
-
-        # Read .text PointerToRawData
-        self.file.seek(0x18C)
-        (self.textraw,) = struct.unpack("<i", self.file.read(4))
-        self.logger.debug("... Parsing finished")
-        return self
-
-    def __exit__(self, exc_type, exc_value, exc_traceback):
-        self.logger.debug(f"Bin {self.filename} Exit")
-        if self.file:
-            self.file.close()
-
-    def get_addr(self, virt):
-        return virt - self.imagebase - self.textvirt + self.textraw
-
-    def read(self, offset, size):
-        self.file.seek(self.get_addr(offset))
-        return self.file.read(size)
-
-
-class RecompiledInfo:
-    addr = None
-    size = None
-    name = None
-    start = None
 
 
 class OffsetPlaceholderGenerator:

tools/reccmp/reccmp.py

@@ -7,15 +7,14 @@
 import logging
 import os
 import re
-import subprocess
-import sys
 
 from isledecomp import (
     Bin,
     find_code_blocks,
+    get_file_in_script_dir,
     OffsetPlaceholderGenerator,
     print_diff,
-    RecompiledInfo,
+    SymInfo,
     walk_source_dir,
     WinePathConverter,
 )
@@ -47,138 +46,6 @@
 WORDS = re.compile(r"\w+")
 
 
-def get_file_in_script_dir(fn):
-    return os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])), fn)
-
-
-# Declare a class that parses the output of cvdump for fast access later
-class SymInfo:
-    funcs = {}
-    lines = {}
-    names = {}
-
-    def __init__(self, pdb, sym_recompfile, sym_logger, sym_wine_path_converter=None):
-        self.logger = sym_logger
-        call = [get_file_in_script_dir("cvdump.exe"), "-l", "-s"]
-
-        if sym_wine_path_converter:
-            # Run cvdump through wine and convert path to Windows-friendly wine path
-            call.insert(0, "wine")
-            call.append(sym_wine_path_converter.get_wine_path(pdb))
-        else:
-            call.append(pdb)
-
-        self.logger.info("Parsing %s ...", pdb)
-        self.logger.debug("Command = %s", call)
-        line_dump = subprocess.check_output(call).decode("utf-8").split("\r\n")
-
-        current_section = None
-
-        self.logger.debug("Parsing output of cvdump.exe ...")
-
-        for i, line in enumerate(line_dump):
-            if line.startswith("***"):
-                current_section = line[4:]
-
-            if current_section == "SYMBOLS" and "S_GPROC32" in line:
-                sym_addr = int(line[26:34], 16)
-
-                info = RecompiledInfo()
-                info.addr = (
-                    sym_addr + sym_recompfile.imagebase + sym_recompfile.textvirt
-                )
-
-                use_dbg_offs = False
-                if use_dbg_offs:
-                    debug_offs = line_dump[i + 2]
-                    debug_start = int(debug_offs[22:30], 16)
-                    debug_end = int(debug_offs[43:], 16)
-
-                    info.start = debug_start
-                    info.size = debug_end - debug_start
-                else:
-                    info.start = 0
-                    info.size = int(line[41:49], 16)
-
-                info.name = line[77:]
-
-                self.names[info.name] = info
-                self.funcs[sym_addr] = info
-            elif (
-                current_section == "LINES"
-                and line.startswith("  ")
-                and not line.startswith("   ")
-            ):
-                sourcepath = line.split()[0]
-
-                if sym_wine_path_converter:
-                    # Convert filename to Unix path for file compare
-                    sourcepath = sym_wine_path_converter.get_unix_path(sourcepath)
-
-                if sourcepath not in self.lines:
-                    self.lines[sourcepath] = {}
-
-                j = i + 2
-                while True:
-                    ll = line_dump[j].split()
-                    if len(ll) == 0:
-                        break
-
-                    k = 0
-                    while k < len(ll):
-                        linenum = int(ll[k + 0])
-                        address = int(ll[k + 1], 16)
-                        if linenum not in self.lines[sourcepath]:
-                            self.lines[sourcepath][linenum] = address
-                        k += 2
-
-                    j += 1
-
-        self.logger.debug("... Parsing output of cvdump.exe finished")
-
-    def get_recompiled_address(self, filename, line):
-        recompiled_addr = None
-
-        self.logger.debug("Looking for %s:%s", filename, line)
-        filename_basename = os.path.basename(filename).lower()
-
-        for fn in self.lines:
-            # Sometimes a PDB is compiled with a relative path while we always have
-            # an absolute path. Therefore we must
-            try:
-                if os.path.basename(
-                    fn
-                ).lower() == filename_basename and os.path.samefile(fn, filename):
-                    filename = fn
-                    break
-            except FileNotFoundError:
-                continue
-
-        if filename in self.lines and line in self.lines[filename]:
-            recompiled_addr = self.lines[filename][line]
-
-            if recompiled_addr in self.funcs:
-                return self.funcs[recompiled_addr]
-            self.logger.error(
-                "Failed to find function symbol with address: %x", recompiled_addr
-            )
-            return None
-        self.logger.error(
-            "Failed to find function symbol with filename and line: %s:%s",
-            filename,
-            line,
-        )
-        return None
-
-    def get_recompiled_address_from_name(self, name):
-        self.logger.debug("Looking for %s", name)
-
-        if name in self.names:
-            return self.names[name]
-        self.logger.error("Failed to find function symbol with name: %s", name)
-        return None
-
-
 def sanitize(file, placeholder_generator, mnemonic, op_str):
     op_str_is_number = False
     try: