Merge pull request #59 from praetorian20/perf_improvements

Performance improvements - cache attributes and avoid runtime checks (#56)

pygccxml/declarations/class_declaration.py

@@ -67,32 +67,29 @@ def __init__(self, related_class=None, access=None, is_virtual=False):
             assert(access in ACCESS_TYPES.ALL)
         self._access = access
         self._is_virtual = is_virtual
+        self._declaration_path = None
+        self._declaration_path_hash = None
 
     def __eq__(self, other):
         if not isinstance(other, hierarchy_info_t):
             return False
-        return declaration_utils.declaration_path(self.related_class) == \
-            declaration_utils.declaration_path(other.related_class) \
-            and self.access == other.access \
-            and self.is_virtual == other.is_virtual
+        return (self.declaration_path_hash ==
+                other.declaration_path_hash) \
+            and self._declaration_path == other._declaration_path \
+            and self._access == other._access \
+            and self._is_virtual == other._is_virtual
 
     def __hash__(self):
-        return hash(declaration_utils.declaration_path(self.related_class))
+        return self.declaration_path_hash
 
     def __ne__(self, other):
         return not self.__eq__(other)
 
     def __lt__(self, other):
         if not isinstance(other, self.__class__):
             return self.__class__.__name__ < other.__class__.__name__
-        return (
-            declaration_utils.declaration_path(self.related_class),
-            self.access,
-            self.is_virtual) < (
-            declaration_utils.declaration_path(
-                other.related_class),
-            other.access,
-            self.is_virtual)
+        return (self.declaration_path, self.access, self.is_virtual) < \
+            (other.declaration_path, other.access, other.is_virtual)
 
     @property
     def related_class(self):
@@ -104,6 +101,8 @@ def related_class(self, new_related_class):
         if new_related_class:
             assert(isinstance(new_related_class, class_t))
         self._related_class = new_related_class
+        self._declaration_path = None
+        self._declaration_path_hash = None
 
     @property
     def access(self):
@@ -135,6 +134,19 @@ def is_virtual(self):
     def is_virtual(self, new_is_virtual):
         self._is_virtual = new_is_virtual
 
+    @property
+    def declaration_path(self):
+        if self._declaration_path is None:
+            self._declaration_path = declaration_utils.declaration_path(
+                self.related_class)
+        return self._declaration_path
+
+    @property
+    def declaration_path_hash(self):
+        if self._declaration_path_hash is None:
+            self._declaration_path_hash = hash(tuple(self.declaration_path))
+        return self._declaration_path_hash
+
 
 class class_declaration_t(declaration.declaration_t):
 

pygccxml/parser/scanner.py

@@ -154,6 +154,11 @@ def __init__(self, xml_file, decl_factory, config, *args):
         self.__mangled_suffix = ' *INTERNAL* '
         self.__mangled_suffix_len = len(self.__mangled_suffix)
 
+        self.__name_attrs_to_skip = []
+        self.__read_location = \
+            lambda decl, attrs, to_skip: self.__read_location_bootstrap(
+                self, decl, attrs, to_skip)
+
     def read(self):
         xml.sax.parse(self.xml_file, self)
 
@@ -223,7 +228,7 @@ def startElement(self, name, attrs):
                 self.__update_membership(attrs)
                 self.__declarations[element_id] = obj
                 if not isinstance(obj, declarations.namespace_t):
-                    self.__read_location(obj, attrs)
+                    self.__read_location(obj, attrs, self.__name_attrs_to_skip)
                 if isinstance(obj, declarations.class_t):
                     self.__read_bases(obj, attrs)
                 if isinstance(obj, declarations.typedef_t):
@@ -260,16 +265,19 @@ def endElement(self, name):
             self.__inst = None
 
     @staticmethod
-    def __read_location(decl, attrs):
+    def __read_location_bootstrap(inst, decl, attrs, to_skip):
+        """ This function monkey patches the __read_location function to either
+        __read_location_gccxml or __read_location_castxml depending on the
+        xml generator in use
+        """
 
-        to_skip = []
         if "CastXML" in utils.xml_generator:
             # These fields are generated by clang, and have no location.
             # Just set an empty location for them. Gccxml does not have
             # this problem.
             # bug #19: gp_offset, fp_offset, overflow_arg_area, reg_save_area
             # bug #32: isa, flags, str and length were added in llvm 3.9
-            to_skip = [
+            inst.__name_attrs_to_skip = [
                 "gp_offset",
                 "fp_offset",
                 "overflow_arg_area",
@@ -279,7 +287,19 @@ def __read_location(decl, attrs):
                 "str",
                 "length"
             ]
+            inst.__read_location = inst.__read_location_castxml
+        else:
+            inst.__read_location = inst.__read_location_gccxml
+        return inst.__read_location(decl, attrs, inst.__name_attrs_to_skip)
 
+    @staticmethod
+    def __read_location_gccxml(decl, attrs, to_skip):
+        decl.location = declarations.location_t(
+            file_name=attrs[XML_AN_FILE],
+            line=int(attrs[XML_AN_LINE]))
+
+    @staticmethod
+    def __read_location_castxml(decl, attrs, to_skip):
         if "name" in attrs and attrs["name"] in to_skip:
             decl.location = declarations.location_t('', -1)
         else: