Added formatters for libc++ (http://libcxx.llvm.org):

 std::string has a summary provider
 std::vector std::list and std::map have both a summary and a synthetic children provider
Given the usage of a custom namespace (std::__1::classname) for the implementation of libc++, we keep both libstdcpp and libc++ formatters enabled at the same time since that raises no conflicts and enabled for seamless transition between the two
The formatters for libc++ reside in a libcxx category, and are loaded from libcxx.py (to be found in examples/synthetic)

The formatters-stl test cases have been divided to be separate for libcxx and libstdcpp. This separation is necessary because
 (a) we need different compiler flags for libc++ than for libstdcpp
 (b) libc++ inlines a lot more than libstdcpp and some code changes were required to accommodate this difference

llvm-svn: 152570

1 files changed, 481 insertions, 0 deletions

diff --git a/lldb/examples/synthetic/libcxx.py b/lldb/examples/synthetic/libcxx.py
new file mode 100644
index 00000000000..5e405df451a
--- /dev/null
+++ b/lldb/examples/synthetic/libcxx.py
@@ -0,0 +1,481 @@
+import lldb
+
+# libcxx STL formatters for LLDB
+# These formatters are based upon the implementation of libc++ that
+# ships with current releases of OS X - They will not work for other implementations
+# of the standard C++ library - and they are bound to use the libc++-specific namespace
+
+# this could probably be made more efficient but since it only reads a handful of bytes at a time
+# we probably don't need to worry too much about this for the time being
+def make_string(F,L):
+	strval = ''
+	G = F.GetData().uint8
+	for X in range(L):
+		V = G[X]
+		if V == 0:
+			break
+		strval = strval + chr(V % 256)
+	return '"' + strval + '"'
+
+# if we ever care about big-endian, these two functions might need to change
+def is_short_string(value):
+	return True if (value & 1) == 0 else False
+def extract_short_size(value):
+	return ((value >> 1) % 256)
+
+# some of the members of libc++ std::string are anonymous or have internal names that convey
+# no external significance - we access them by index since this saves a name lookup that would add
+# no information for readers of the code, but when possible try to use meaningful variable names
+def stdstring_SummaryProvider(valobj,dict):
+	r = valobj.GetChildAtIndex(0)
+	B = r.GetChildAtIndex(0)
+	first = B.GetChildAtIndex(0)
+	D = first.GetChildAtIndex(0)
+	l = D.GetChildAtIndex(0)
+	s = D.GetChildAtIndex(1)
+	D20 = s.GetChildAtIndex(0)
+	size_mode = D20.GetChildAtIndex(0).GetValueAsUnsigned(0)
+	if is_short_string(size_mode):
+		size = extract_short_size(size_mode)
+		return make_string(s.GetChildAtIndex(1),size)
+	else:
+		data_ptr = l.GetChildAtIndex(2)
+		size_vo = l.GetChildAtIndex(1)
+		size = size_vo.GetValueAsUnsigned(0)+1 # the NULL terminator must be accounted for
+		if size <= 1: # should never be the case
+			return '""'
+		data = data_ptr.GetPointeeData(0,size)
+		error = lldb.SBError()
+		strval = data.GetString(error,0)
+		if error.Fail():
+			return '<error:' + error.GetCString() + '>'
+		else:
+			return '"' + strval + '"'
+
+class stdvector_SynthProvider:
+
+	def __init__(self, valobj, dict):
+		self.valobj = valobj;
+		self.update()
+
+	def num_children(self):
+		try:
+			start_val = self.start.GetValueAsUnsigned(0)
+			finish_val = self.finish.GetValueAsUnsigned(0)
+			# Before a vector has been constructed, it will contain bad values
+			# so we really need to be careful about the length we return since
+			# unitialized data can cause us to return a huge number. We need
+			# to also check for any of the start, finish or end of storage values
+			# being zero (NULL). If any are, then this vector has not been 
+			# initialized yet and we should return zero
+
+			# Make sure nothing is NULL
+			if start_val == 0 or finish_val == 0:
+				return 0
+			# Make sure start is less than finish
+			if start_val >= finish_val:
+				return 0
+
+			num_children = (finish_val-start_val)
+			if (num_children % self.data_size) != 0:
+				return 0
+			else:
+				num_children = num_children/self.data_size
+			return num_children
+		except:
+			return 0;
+
+	def get_child_index(self,name):
+		try:
+			return int(name.lstrip('[').rstrip(']'))
+		except:
+			return -1
+
+	def get_child_at_index(self,index):
+		if index < 0:
+			return None;
+		if index >= self.num_children():
+			return None;
+		try:
+			offset = index * self.data_size
+			return self.start.CreateChildAtOffset('['+str(index)+']',offset,self.data_type)
+		except:
+			return None
+
+	def update(self):
+		try:
+			self.start = self.valobj.GetChildMemberWithName('__begin_')
+			self.finish = self.valobj.GetChildMemberWithName('__end_')
+			# the purpose of this field is unclear, but it is the only field whose type is clearly T* for a vector<T>
+			# if this ends up not being correct, we can use the APIs to get at template arguments
+			data_type_finder = self.valobj.GetChildMemberWithName('__end_cap_').GetChildMemberWithName('__first_')
+			self.data_type = data_type_finder.GetType().GetPointeeType()
+			self.data_size = self.data_type.GetByteSize()
+		except:
+			pass
+
+def stdvector_SummaryProvider(valobj,dict):
+	prov = stdvector_SynthProvider(valobj,None)
+	return 'size=' + str(prov.num_children())
+
+class stdlist_entry:
+
+	def __init__(self,entry):
+		self.entry = entry
+
+	def _next_impl(self):
+		return stdlist_entry(self.entry.GetChildMemberWithName('__next_'))
+
+	def _prev_impl(self):
+		return stdlist_entry(self.entry.GetChildMemberWithName('__prev_'))
+
+	def _value_impl(self):
+		return self.entry.GetValueAsUnsigned(0)
+
+	def _isnull_impl(self):
+		return self._value_impl() == 0
+
+	def _sbvalue_impl(self):
+		return self.entry
+
+	next = property(_next_impl,None)
+	value = property(_value_impl,None)
+	is_null = property(_isnull_impl,None)
+	sbvalue = property(_sbvalue_impl,None)
+
+class stdlist_iterator:
+
+	def increment_node(self,node):
+		if node.is_null:
+			return None
+		return node.next
+
+	def __init__(self,node):
+		self.node = stdlist_entry(node) # we convert the SBValue to an internal node object on entry
+
+	def value(self):
+		return self.node.sbvalue # and return the SBValue back on exit
+
+	def next(self):
+		node = self.increment_node(self.node)
+		if node != None and node.sbvalue.IsValid() and not(node.is_null):
+			self.node = node
+			return self.value()
+		else:
+			return None
+
+	def advance(self,N):
+		if N < 0:
+			return None
+		if N == 0:
+			return self.value()
+		if N == 1:
+			return self.next()
+		while N > 0:
+			self.next()
+			N = N - 1
+		return self.value()
+
+
+class stdlist_SynthProvider:
+	def __init__(self, valobj, dict):
+		self.valobj = valobj
+		self.update()
+
+	def next_node(self,node):
+		return node.GetChildMemberWithName('__next_')
+
+	def value(self,node):
+		return node.GetValueAsUnsigned()
+
+	# Floyd's cyle-finding algorithm
+	# try to detect if this list has a loop
+	def has_loop(self):
+		slow = stdlist_entry(self.head)
+		fast1 = stdlist_entry(self.head)
+		fast2 = stdlist_entry(self.head)
+		while slow.next.value != self.node_address:
+			slow_value = slow.value
+			fast1 = fast2.next
+			fast2 = fast1.next
+			if fast1.value == slow_value or fast2.value == slow_value:
+				return True
+			slow = slow.next
+		return False
+
+	def num_children(self):
+		if self.count == None:
+			self.count = self.num_children_impl()
+		return self.count
+
+	def num_children_impl(self):
+		try:
+			next_val = self.head.GetValueAsUnsigned(0)
+			prev_val = self.tail.GetValueAsUnsigned(0)
+			# After a std::list has been initialized, both next and prev will be non-NULL
+			if next_val == 0 or prev_val == 0:
+				return 0
+			if next_val == self.node_address:
+				return 0
+			if next_val == prev_val:
+				return 1
+			if self.has_loop():
+				return 0
+			size = 2
+			current = stdlist_entry(self.head)
+			while current.next.value != self.node_address:
+				size = size + 1
+				current = current.next
+			return (size - 1)
+		except:
+			return 0;
+
+	def get_child_index(self,name):
+		try:
+			return int(name.lstrip('[').rstrip(']'))
+		except:
+			return -1
+
+	def get_child_at_index(self,index):
+		if index < 0:
+			return None;
+		if index >= self.num_children():
+			return None;
+		try:
+			current = stdlist_iterator(self.head)
+			current = current.advance(index)
+			# we do not return __value_ because then all our children would be named __value_
+			# we need to make a copy of __value__ with the right name - unfortunate
+			obj = current.GetChildMemberWithName('__value_')
+			obj_data = obj.GetData()
+			return self.valobj.CreateValueFromData('[' + str(index) + ']',obj_data,self.data_type)
+		except:
+			return None
+
+	def extract_type(self):
+		list_type = self.valobj.GetType().GetUnqualifiedType()
+		if list_type.GetNumberOfTemplateArguments() > 0:
+			data_type = list_type.GetTemplateArgumentType(0)
+		else:
+			data_type = None
+		return data_type
+
+	def update(self):
+		try:
+			impl = self.valobj.GetChildMemberWithName('__end_')
+			self.node_address = self.valobj.AddressOf().GetValueAsUnsigned(0)
+			self.head = impl.GetChildMemberWithName('__next_')
+			self.tail = impl.GetChildMemberWithName('__prev_')
+			self.data_type = self.extract_type()
+			self.data_size = self.data_type.GetByteSize()
+			self.count = None
+		except:
+			pass
+
+def stdlist_SummaryProvider(valobj,dict):
+	prov = stdlist_SynthProvider(valobj,None)
+	return 'size=' + str(prov.num_children())
+
+# a tree node - this class makes the syntax in the actual iterator nicer to read and maintain
+class stdmap_iterator_node:
+	def _left_impl(self):
+		return stdmap_iterator_node(self.node.GetChildMemberWithName("__left_"))
+
+	def _right_impl(self):
+		return stdmap_iterator_node(self.node.GetChildMemberWithName("__right_"))
+
+	def _parent_impl(self):
+		return stdmap_iterator_node(self.node.GetChildMemberWithName("__parent_"))
+
+	def _value_impl(self):
+		return self.node.GetValueAsUnsigned(0)
+
+	def _sbvalue_impl(self):
+		return self.node
+
+	def _null_impl(self):
+		return self.value == 0
+
+	def __init__(self,node):
+		self.node = node
+
+	left = property(_left_impl,None)
+	right = property(_right_impl,None)
+	parent = property(_parent_impl,None)
+	value = property(_value_impl,None)
+	is_null = property(_null_impl,None)
+	sbvalue = property(_sbvalue_impl,None)
+
+# a Python implementation of the tree iterator used by libc++
+class stdmap_iterator:
+
+	def tree_min(self,x):
+		if x.is_null:
+			return None
+		while (not x.left.is_null):
+			x = x.left
+		return x
+
+	def tree_max(self,x):
+		if x.is_null:
+			return None
+		while (not x.right.is_null):
+			x =  x.right
+		return x
+
+	def tree_is_left_child(self,x):
+		if x.is_null:
+			return None
+		return True if x.value == x.parent.left.value else False
+
+	def increment_node(self,node):
+		if node.is_null:
+			return None
+		if not node.right.is_null:
+			return self.tree_min(node.right)
+		while (not self.tree_is_left_child(node)):
+			node = node.parent
+		return node.parent
+
+	def __init__(self,node):
+		self.node = stdmap_iterator_node(node) # we convert the SBValue to an internal node object on entry
+
+	def value(self):
+		return self.node.sbvalue # and return the SBValue back on exit
+
+	def next(self):
+		node = self.increment_node(self.node)
+		if node != None and node.sbvalue.IsValid() and not(node.is_null):
+			self.node = node
+			return self.value()
+		else:
+			return None
+
+	def advance(self,N):
+		if N < 0:
+			return None
+		if N == 0:
+			return self.value()
+		if N == 1:
+			return self.next()
+		while N > 0:
+			self.next()
+			N = N - 1
+		return self.value()
+
+class stdmap_SynthProvider:
+
+	def __init__(self, valobj, dict):
+		self.valobj = valobj;
+		self.pointer_size = self.valobj.GetProcess().GetAddressByteSize()
+		self.update()
+
+	def update(self):
+		try:
+			self.tree = self.valobj.GetChildMemberWithName('__tree_')
+			self.root_node = self.tree.GetChildMemberWithName('__begin_node_')
+			# this data is either lazily-calculated, or cannot be inferred at this moment
+			# we still need to mark it as None, meaning "please set me ASAP"
+			self.data_type = None
+			self.data_size = None
+			self.skip_size = None
+			self.count = None
+		except:
+			pass
+
+	def num_children(self):
+		if self.count == None:
+			self.count = self.num_children_impl()
+		return self.count
+
+	def num_children_impl(self):
+		try:
+			return self.valobj.GetChildMemberWithName('__tree_').GetChildMemberWithName('__pair3_').GetChildMemberWithName('__first_').GetValueAsUnsigned()
+		except:
+			return 0;
+
+	def get_data_type(self):
+		if self.data_type == None or self.data_size == None:
+			if self.num_children() == 0:
+				return False
+			deref = self.root_node.Dereference()
+			if not(deref.IsValid()):
+				return False
+			value = deref.GetChildMemberWithName('__value_')
+			if not(value.IsValid()):
+				return False
+			self.data_type = value.GetType()
+			self.data_size = self.data_type.GetByteSize()
+			self.skip_size = None
+			return True
+		else:
+			return True
+
+	def get_value_offset(self,node):
+		if self.skip_size == None:
+			node_type = node.GetType()
+			fields_count = node_type.GetNumberOfFields()
+			for i in range(fields_count):
+				field = node_type.GetFieldAtIndex(i)
+				if field.GetName() == '__value_':
+					self.skip_size = field.GetOffsetInBytes()
+					break
+		return (self.skip_size != None)
+
+	def get_child_index(self,name):
+		try:
+			return int(name.lstrip('[').rstrip(']'))
+		except:
+			return -1
+
+	def get_child_at_index(self,index):
+		if index < 0:
+			return None
+		if index >= self.num_children():
+			return None;
+		try:
+			iterator = stdmap_iterator(self.root_node)
+			# the debug info for libc++ std::map is such that __begin_node_ has a very nice and useful type
+			# out of which we can grab the information we need - every other node has a less informative
+			# type which omits all value information and only contains housekeeping information for the RB tree
+			# hence, we need to know if we are at a node != 0, so that we can still get at the data
+			need_to_skip = (index > 0)
+			current = iterator.advance(index)
+			if self.get_data_type():
+				if not(need_to_skip):
+					current = current.Dereference()
+					obj = current.GetChildMemberWithName('__value_')
+					obj_data = obj.GetData()
+					self.get_value_offset(current) # make sure we have a valid offset for the next items
+					# we do not return __value_ because then we would end up with a child named
+					# __value_ instead of [0]
+					return self.valobj.CreateValueFromData('[' + str(index) + ']',obj_data,self.data_type)
+				else:
+					# FIXME we need to have accessed item 0 before accessing any other item!
+					if self.skip_size == None:
+						return None
+					return current.CreateChildAtOffset('[' + str(index) + ']',self.skip_size,self.data_type)
+			else:
+				print "foo"
+				return None
+		except Exception as err:
+			print err
+			return None
+
+def stdmap_SummaryProvider(valobj,dict):
+	prov = stdmap_SynthProvider(valobj,None)
+	return 'size=' + str(prov.num_children())
+
+
+# we can use two different categories for old and new formatters - type names are different enough that we should make no confusion
+# talking with libc++ developer: "std::__1::class_name is set in stone until we decide to change the ABI. That shouldn't happen within a 5 year time frame"
+def __lldb_init_module(debugger,dict):
+	debugger.HandleCommand('type summary add -F libcxx.stdstring_SummaryProvider "std::__1::string" -w libcxx')
+	debugger.HandleCommand('type summary add -F libcxx.stdstring_SummaryProvider "std::__1::basic_string<char, class std::__1::char_traits<char>, class std::__1::allocator<char> >" -w libcxx')
+	debugger.HandleCommand('type synthetic add -l libcxx.stdvector_SynthProvider -x "^(std::__1::)vector<.+>$" -w libcxx')
+	debugger.HandleCommand('type summary add -F libcxx.stdvector_SummaryProvider -e -x "^(std::__1::)vector<.+>$" -w libcxx')
+	debugger.HandleCommand('type synthetic add -l libcxx.stdlist_SynthProvider -x "^(std::__1::)list<.+>$" -w libcxx')
+	debugger.HandleCommand('type summary add -F libcxx.stdlist_SummaryProvider -e -x "^(std::__1::)list<.+>$" -w libcxx')
+	debugger.HandleCommand('type synthetic add -l libcxx.stdmap_SynthProvider -x "^(std::__1::)map<.+> >$" -w libcxx')
+	debugger.HandleCommand('type summary add -F libcxx.stdmap_SummaryProvider -e -x "^(std::__1::)map<.+> >$" -w libcxx')
+	debugger.HandleCommand("type category enable libcxx")