Clear cyclic references in the parser and the emitter to avoid extra GC calls.

git-svn-id: http://svn.pyyaml.org/pyyaml/trunk@377 18f92427-320e-0410-9341-c67f048884a3

11 files changed, 130 insertions, 44 deletions

diff --git a/ext/_yaml.pyx b/ext/_yaml.pyx
index df19492..5158fb4 100644
--- a/ext/_yaml.pyx
+++ b/ext/_yaml.pyx
@@ -311,6 +311,9 @@ cdef class CParser:
         yaml_parser_delete(&self.parser)
         yaml_event_delete(&self.parsed_event)
 
+    def dispose(self):
+        pass
+
     cdef object _parser_error(self):
         if self.parser.error == YAML_MEMORY_ERROR:
             return MemoryError
@@ -993,6 +996,9 @@ cdef class CEmitter:
     def __dealloc__(self):
         yaml_emitter_delete(&self.emitter)
 
+    def dispose(self):
+        pass
+
     cdef object _emitter_error(self):
         if self.emitter.error == YAML_MEMORY_ERROR:
             return MemoryError
diff --git a/lib/yaml/__init__.py b/lib/yaml/__init__.py
index c0fd1f3..d414be8 100644
--- a/lib/yaml/__init__.py
+++ b/lib/yaml/__init__.py
@@ -21,16 +21,22 @@ def scan(stream, Loader=Loader):
     Scan a YAML stream and produce scanning tokens.
     """
     loader = Loader(stream)
-    while loader.check_token():
-        yield loader.get_token()
+    try:
+        while loader.check_token():
+            yield loader.get_token()
+    finally:
+        loader.dispose()
 
 def parse(stream, Loader=Loader):
     """
     Parse a YAML stream and produce parsing events.
     """
     loader = Loader(stream)
-    while loader.check_event():
-        yield loader.get_event()
+    try:
+        while loader.check_event():
+            yield loader.get_event()
+    finally:
+        loader.dispose()
 
 def compose(stream, Loader=Loader):
     """
@@ -38,7 +44,10 @@ def compose(stream, Loader=Loader):
     and produce the corresponding representation tree.
     """
     loader = Loader(stream)
-    return loader.get_single_node()
+    try:
+        return loader.get_single_node()
+    finally:
+        loader.dispose()
 
 def compose_all(stream, Loader=Loader):
     """
@@ -46,8 +55,11 @@ def compose_all(stream, Loader=Loader):
     and produce corresponding representation trees.
     """
     loader = Loader(stream)
-    while loader.check_node():
-        yield loader.get_node()
+    try:
+        while loader.check_node():
+            yield loader.get_node()
+    finally:
+        loader.dispose()
 
 def load(stream, Loader=Loader):
     """
@@ -55,7 +67,10 @@ def load(stream, Loader=Loader):
     and produce the corresponding Python object.
     """
     loader = Loader(stream)
-    return loader.get_single_data()
+    try:
+        return loader.get_single_data()
+    finally:
+        loader.dispose()
 
 def load_all(stream, Loader=Loader):
     """
@@ -63,8 +78,11 @@ def load_all(stream, Loader=Loader):
     and produce corresponding Python objects.
     """
     loader = Loader(stream)
-    while loader.check_data():
-        yield loader.get_data()
+    try:
+        while loader.check_data():
+            yield loader.get_data()
+    finally:
+        loader.dispose()
 
 def safe_load(stream):
     """
@@ -96,8 +114,11 @@ def emit(events, stream=None, Dumper=Dumper,
         getvalue = stream.getvalue
     dumper = Dumper(stream, canonical=canonical, indent=indent, width=width,
             allow_unicode=allow_unicode, line_break=line_break)
-    for event in events:
-        dumper.emit(event)
+    try:
+        for event in events:
+            dumper.emit(event)
+    finally:
+        dumper.dispose()
     if getvalue:
         return getvalue()
 
@@ -122,10 +143,13 @@ def serialize_all(nodes, stream=None, Dumper=Dumper,
             allow_unicode=allow_unicode, line_break=line_break,
             encoding=encoding, version=version, tags=tags,
             explicit_start=explicit_start, explicit_end=explicit_end)
-    dumper.open()
-    for node in nodes:
-        dumper.serialize(node)
-    dumper.close()
+    try:
+        dumper.open()
+        for node in nodes:
+            dumper.serialize(node)
+        dumper.close()
+    finally:
+        dumper.dispose()
     if getvalue:
         return getvalue()
 
@@ -160,10 +184,13 @@ def dump_all(documents, stream=None, Dumper=Dumper,
             allow_unicode=allow_unicode, line_break=line_break,
             encoding=encoding, version=version, tags=tags,
             explicit_start=explicit_start, explicit_end=explicit_end)
-    dumper.open()
-    for data in documents:
-        dumper.represent(data)
-    dumper.close()
+    try:
+        dumper.open()
+        for data in documents:
+            dumper.represent(data)
+        dumper.close()
+    finally:
+        dumper.dispose()
     if getvalue:
         return getvalue()
 
diff --git a/lib/yaml/constructor.py b/lib/yaml/constructor.py
index 420c434..081d78a 100644
--- a/lib/yaml/constructor.py
+++ b/lib/yaml/constructor.py
@@ -58,11 +58,11 @@ class BaseConstructor(object):
         return data
 
     def construct_object(self, node, deep=False):
+        if node in self.constructed_objects:
+            return self.constructed_objects[node]
         if deep:
             old_deep = self.deep_construct
             self.deep_construct = True
-        if node in self.constructed_objects:
-            return self.constructed_objects[node]
         if node in self.recursive_objects:
             raise ConstructorError(None, None,
                     "found unconstructable recursive node", node.start_mark)
diff --git a/lib/yaml/emitter.py b/lib/yaml/emitter.py
index 4cb2c8a..e5bcdcc 100644
--- a/lib/yaml/emitter.py
+++ b/lib/yaml/emitter.py
@@ -103,6 +103,11 @@ class Emitter(object):
         self.analysis = None
         self.style = None
 
+    def dispose(self):
+        # Reset the state attributes (to clear self-references)
+        self.states = []
+        self.state = None
+
     def emit(self, event):
         self.events.append(event)
         while not self.need_more_events():
diff --git a/lib/yaml/parser.py b/lib/yaml/parser.py
index b6a7416..f9e3057 100644
--- a/lib/yaml/parser.py
+++ b/lib/yaml/parser.py
@@ -86,6 +86,11 @@ class Parser(object):
         self.marks = []
         self.state = self.parse_stream_start
 
+    def dispose(self):
+        # Reset the state attributes (to clear self-references)
+        self.states = []
+        self.state = None
+
     def check_event(self, *choices):
         # Check the type of the next event.
         if self.current_event is None:
diff --git a/lib3/yaml/__init__.py b/lib3/yaml/__init__.py
index cd2ea5a..2d49ac8 100644
--- a/lib3/yaml/__init__.py
+++ b/lib3/yaml/__init__.py
@@ -22,16 +22,22 @@ def scan(stream, Loader=Loader):
     Scan a YAML stream and produce scanning tokens.
     """
     loader = Loader(stream)
-    while loader.check_token():
-        yield loader.get_token()
+    try:
+        while loader.check_token():
+            yield loader.get_token()
+    finally:
+        loader.dispose()
 
 def parse(stream, Loader=Loader):
     """
     Parse a YAML stream and produce parsing events.
     """
     loader = Loader(stream)
-    while loader.check_event():
-        yield loader.get_event()
+    try:
+        while loader.check_event():
+            yield loader.get_event()
+    finally:
+        loader.dispose()
 
 def compose(stream, Loader=Loader):
     """
@@ -39,7 +45,10 @@ def compose(stream, Loader=Loader):
     and produce the corresponding representation tree.
     """
     loader = Loader(stream)
-    return loader.get_single_node()
+    try:
+        return loader.get_single_node()
+    finally:
+        loader.dispose()
 
 def compose_all(stream, Loader=Loader):
     """
@@ -47,8 +56,11 @@ def compose_all(stream, Loader=Loader):
     and produce corresponding representation trees.
     """
     loader = Loader(stream)
-    while loader.check_node():
-        yield loader.get_node()
+    try:
+        while loader.check_node():
+            yield loader.get_node()
+    finally:
+        loader.dispose()
 
 def load(stream, Loader=Loader):
     """
@@ -56,7 +68,10 @@ def load(stream, Loader=Loader):
     and produce the corresponding Python object.
     """
     loader = Loader(stream)
-    return loader.get_single_data()
+    try:
+        return loader.get_single_data()
+    finally:
+        loader.dispose()
 
 def load_all(stream, Loader=Loader):
     """
@@ -64,8 +79,11 @@ def load_all(stream, Loader=Loader):
     and produce corresponding Python objects.
     """
     loader = Loader(stream)
-    while loader.check_data():
-        yield loader.get_data()
+    try:
+        while loader.check_data():
+            yield loader.get_data()
+    finally:
+        loader.dispose()
 
 def safe_load(stream):
     """
@@ -96,8 +114,11 @@ def emit(events, stream=None, Dumper=Dumper,
         getvalue = stream.getvalue
     dumper = Dumper(stream, canonical=canonical, indent=indent, width=width,
             allow_unicode=allow_unicode, line_break=line_break)
-    for event in events:
-        dumper.emit(event)
+    try:
+        for event in events:
+            dumper.emit(event)
+    finally:
+        dumper.dispose()
     if getvalue:
         return getvalue()
 
@@ -121,10 +142,13 @@ def serialize_all(nodes, stream=None, Dumper=Dumper,
             allow_unicode=allow_unicode, line_break=line_break,
             encoding=encoding, version=version, tags=tags,
             explicit_start=explicit_start, explicit_end=explicit_end)
-    dumper.open()
-    for node in nodes:
-        dumper.serialize(node)
-    dumper.close()
+    try:
+        dumper.open()
+        for node in nodes:
+            dumper.serialize(node)
+        dumper.close()
+    finally:
+        dumper.dispose()
     if getvalue:
         return getvalue()
 
@@ -158,10 +182,13 @@ def dump_all(documents, stream=None, Dumper=Dumper,
             allow_unicode=allow_unicode, line_break=line_break,
             encoding=encoding, version=version, tags=tags,
             explicit_start=explicit_start, explicit_end=explicit_end)
-    dumper.open()
-    for data in documents:
-        dumper.represent(data)
-    dumper.close()
+    try:
+        dumper.open()
+        for data in documents:
+            dumper.represent(data)
+        dumper.close()
+    finally:
+        dumper.dispose()
     if getvalue:
         return getvalue()
 
diff --git a/lib3/yaml/constructor.py b/lib3/yaml/constructor.py
index bd25b79..981543a 100644
--- a/lib3/yaml/constructor.py
+++ b/lib3/yaml/constructor.py
@@ -51,11 +51,11 @@ class BaseConstructor:
         return data
 
     def construct_object(self, node, deep=False):
+        if node in self.constructed_objects:
+            return self.constructed_objects[node]
         if deep:
             old_deep = self.deep_construct
             self.deep_construct = True
-        if node in self.constructed_objects:
-            return self.constructed_objects[node]
         if node in self.recursive_objects:
             raise ConstructorError(None, None,
                     "found unconstructable recursive node", node.start_mark)
diff --git a/lib3/yaml/emitter.py b/lib3/yaml/emitter.py
index 8947d04..34cb145 100644
--- a/lib3/yaml/emitter.py
+++ b/lib3/yaml/emitter.py
@@ -103,6 +103,11 @@ class Emitter:
         self.analysis = None
         self.style = None
 
+    def dispose(self):
+        # Reset the state attributes (to clear self-references)
+        self.states = []
+        self.state = None
+
     def emit(self, event):
         self.events.append(event)
         while not self.need_more_events():
diff --git a/lib3/yaml/parser.py b/lib3/yaml/parser.py
index c147dd6..13a5995 100644
--- a/lib3/yaml/parser.py
+++ b/lib3/yaml/parser.py
@@ -86,6 +86,11 @@ class Parser:
         self.marks = []
         self.state = self.parse_stream_start
 
+    def dispose(self):
+        # Reset the state attributes (to clear self-references)
+        self.states = []
+        self.state = None
+
     def check_event(self, *choices):
         # Check the type of the next event.
         if self.current_event is None:
diff --git a/tests/lib/canonical.py b/tests/lib/canonical.py
index 41d111a..020e6db 100644
--- a/tests/lib/canonical.py
+++ b/tests/lib/canonical.py
@@ -211,6 +211,9 @@ class CanonicalParser:
         self.events = []
         self.parsed = False
 
+    def dispose(self):
+        pass
+
     # stream: STREAM-START document* STREAM-END
     def parse_stream(self):
         self.get_token(yaml.StreamStartToken)
diff --git a/tests/lib3/canonical.py b/tests/lib3/canonical.py
index e04477b..a8b4e3a 100644
--- a/tests/lib3/canonical.py
+++ b/tests/lib3/canonical.py
@@ -212,6 +212,9 @@ class CanonicalParser:
         self.events = []
         self.parsed = False
 
+    def dispose(self):
+        pass
+
     # stream: STREAM-START document* STREAM-END
     def parse_stream(self):
         self.get_token(yaml.StreamStartToken)