Load treebuilder and treewalker "submodules" only when needed (I'm expecting this could save some initialization time for applications using html5lib, though I have no figure and/or benchmark to prove that).

tbroyer · tbroyer · commit d25f3c614eec · 2007-05-29T14:25:37.000Z
Note to others: feel free to revert this changeset if my assumption proves wrong ;-)

--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40631
diff --git a/src/treebuilders/__init__.py b/src/treebuilders/__init__.py
@@ -33,14 +33,7 @@
 import os.path
 __path__.append(os.path.dirname(__path__[0]))
 
-import dom
-import simpletree
-import etree
-try:
-    import soup as beautifulsoup
-except:
-    pass
-
+treeBuilderCache = {}
 
 def getTreeBuilder(treeType, implementation=None, **kwargs):
     """Get a TreeBuilder class for various types of tree with built-in support
@@ -60,8 +53,15 @@ def getTreeBuilder(treeType, implementation=None, **kwargs):
                       implementing the tree type e.g. xml.etree.ElementTree or
                       lxml.etree."""
     
-    treeType = treeType.lower()
-    if treeType in ("dom", "simpletree", "beautifulsoup"):
-        return globals()[treeType].TreeBuilder
-    elif treeType == "etree":
-        return etree.getETreeModule(implementation, **kwargs).TreeBuilder
+    treeType = treeType.lower()
+    if treeType not in treeBuilderCache:
+        if treeType in ("dom", "simpletree"):
+            mod = __import__(treeType, globals())
+            treeBuilderCache[treeType] = mod.TreeBuilder
+        elif treeType == "beautifulsoup":
+            import soup
+            treeBuilderCache[treeType] = soup.TreeBuilder
+        elif treeType == "etree":
+            import etree
+            treeBuilderCache[treeType] = etree.getETreeModule(implementation, **kwargs).TreeBuilder
+    return treeBuilderCache.get(treeType)
diff --git a/src/treewalkers/__init__.py b/src/treewalkers/__init__.py
@@ -11,15 +11,7 @@
 import os.path
 __path__.append(os.path.dirname(__path__[0]))
 
-import dom
-import simpletree
-import etree
-import pulldom
-try:
-    import soup as beautifulsoup
-except:
-    pass
-
+treeWalkerCache = {}
 
 def getTreeWalker(treeType, implementation=None, **kwargs):
     """Get a TreeWalker class for various types of tree with built-in support
@@ -41,7 +33,14 @@ def getTreeWalker(treeType, implementation=None, **kwargs):
                       lxml.etree."""
 
     treeType = treeType.lower()
-    if treeType in ("dom", "pulldom", "simpletree", "beautifulsoup"):
-        return globals()[treeType].TreeWalker
-    elif treeType == "etree":
-        return etree.getETreeModule(implementation, **kwargs).TreeWalker
+    if treeType not in treeWalkerCache:
+        if treeType in ("dom", "pulldom", "simpletree"):
+            mod = __import__(treeType, globals())
+            treeWalkerCache[treeType] = mod.TreeWalker
+        elif treeType == "beautifulsoup":
+            import soup
+            treeWalkerCache[treeType] = soup.TreeWalker
+        elif treeType == "etree":
+            import etree
+            treeWalkerCache[treeType] = etree.getETreeModule(implementation, **kwargs).TreeWalker
+    return treeWalkerCache.get(treeType)
diff --git a/tests/test_serializer.py b/tests/test_serializer.py
@@ -23,18 +23,19 @@ def load(f):
 
 import html5parser
 import serializer
-import treewalkers
+from treewalkers._base import TreeWalker
 #END RELEASE
 
 #RELEASE add
 #import html5lib
-#from html5lib import html5parser, serializer, treewalkers
+#from html5lib import html5parser, serializer
+#from html5lib.treewalkers._base import TreeWalker
 #END RELEASE
 
 #Run the serialize error checks
 checkSerializeErrors = False
 
-class TestCase(unittest.TestCase, treewalkers._base.TreeWalker):
+class TestCase(unittest.TestCase, TreeWalker):
     def addTest(cls, name, expected, input, description, options):
         func = lambda self: self.mockTest(expected, input, options)
         func.__doc__ = "\t".join([description, str(input), str(options)])
