WIP

mingw_ldd/mingw_ldd.py

@@ -24,6 +24,7 @@
 # THE SOFTWARE.
 
 import argparse
+import concurrent.futures
 import os
 import pefile
 import sys
@@ -45,36 +46,66 @@ def get_dependency(pe_data):
     return deps
 
 
+def _find_dll(dll, dlls, dll_lookup_dirs, arch):
+    # (found, new_dll_path?, new_pe_data)
+    dll_lower = dll.lower()
+    if dll_lower in dlls:
+        return (True, None, None)
+    for dir in dll_lookup_dirs:
+        dll_path = find_file_ignore_case(dir, dll_lower)
+        if not dll_path:
+            continue
+        new_pe_data = pefile.PE(dll_path)
+        if new_pe_data.FILE_HEADER.Machine == arch:
+            return (True, dll_path, new_pe_data)
+    return (False, None, None)
+
+
 def dep_tree(pe, dll_lookup_dirs):
     dlls = {}
     deps = {}
-    dlls_blacklist = []
     pe_data = pefile.PE(pe)
     arch = pe_data.FILE_HEADER.Machine
 
+    def find_dll(dll):
+        # (found, new_dll_path?, new_pe_data)
+        dll_lower = dll.lower()
+        if dll_lower in dlls:
+            return (True, None, None)
+        for dir in dll_lookup_dirs:
+            dll_path = find_file_ignore_case(dir, dll_lower)
+            if not dll_path:
+                continue
+            new_pe_data = pefile.PE(dll_path)
+            if new_pe_data.FILE_HEADER.Machine == arch:
+                return (True, dll_path, new_pe_data)
+        return (False, None, None)
+
     def dep_tree_impl(pe, pe_data):
         pe = os.path.abspath(pe)
         if pe in deps:
             return
         deps[pe] = []
-        for dll in get_dependency(pe_data):
+        dll_to_future_result = []
+        with concurrent.futures.ProcessPoolExecutor() as executor:
+            future_to_dll = {executor.submit(_find_dll, dll, dlls, dll_lookup_dirs, arch): dll for dll in get_dependency(pe_data)}
+            dll_to_future_result = [(future_to_dll[future], future.result()) for future in concurrent.futures.as_completed(future_to_dll)]
+        for (dll, (found_dll, new_dll_path, _)) in dll_to_future_result:
+            deps[pe].append(dll)
             dll_lower = dll.lower()
-            if dll_lower in dlls:
-                deps[pe].append(dll)
+            # it's a cached dll, nothing to do
+            if found_dll and not new_dll_path:
                 continue
-            dlls[dll_lower] = 'not found'
-            for dir in dll_lookup_dirs:
-                dll_path = find_file_ignore_case(dir, dll)
-                if not dll_path or dll_path in dlls_blacklist:
-                    continue
-                new_pe_data = pefile.PE(dll_path)
-                if new_pe_data.FILE_HEADER.Machine == arch:
-                    dlls[dll_lower] = dll_path
-                    dep_tree_impl(dll_path, new_pe_data)
-                    break
-                else:
-                    dlls_blacklist.append(dll_path)
-            deps[pe].append(dll)
+            # it's a new dll, cache it
+            elif found_dll and new_dll_path:
+                dlls[dll_lower] = new_dll_path
+            # dll wasn't found, cache that it wasn't found
+            elif not found_dll:
+                dlls[dll_lower] = 'not found'
+        # recursively process newly found dlls
+        for (dll, (found_dll, new_dll_path, new_pe_data)) in dll_to_future_result:
+            if found_dll and new_dll_path:
+                dep_tree_impl(new_dll_path, new_pe_data)
 
     dep_tree_impl(pe, pe_data)
     return (dlls, deps)