llvm-project/lldb/scripts/Python/modify-python-lldb.py

#
# modify-python-lldb.py
#
# This script modifies the lldb module (which was automatically generated via
# running swig) to support iteration and/or equality operations for certain lldb
# objects, implements truth value testing for certain lldb objects, and adds a
# global variable 'debugger_unique_id' which is initialized to 0.
#
# As a cleanup step, it also removes the 'residues' from the autodoc features of
# swig.  For an example, take a look at SBTarget.h header file, where we take
# advantage of the already existing doxygen C++-docblock and make it the Python
# docstring for the same method.  The 'residues' in this context include the
# '#endif', the '#ifdef SWIG', the c comment marker, the trailing blank (SPC's)
# line, and the doxygen comment start marker.
#
# In addition to the 'residues' removal during the cleanup step, it also
# transforms the 'char' data type (which was actually 'char *' but the 'autodoc'
# feature of swig removes ' *' from it) into 'str' (as a Python str type).
#
# It also calls SBDebugger.Initialize() to initialize the lldb debugger
# subsystem.
#

import sys, re, StringIO

if len (sys.argv) != 2:
    output_name = "./lldb.py"
else:
    output_name = sys.argv[1] + "/lldb.py"

# print "output_name is '" + output_name + "'"

#
# Residues to be removed.
#
c_endif_swig = "#endif"
c_ifdef_swig = "#ifdef SWIG"
c_comment_marker = "//------------"
# The pattern for recognizing the doxygen comment block line.
doxygen_comment_start = re.compile("^\s*(/// ?)")
# The demarcation point for turning on/off residue removal state.
# When bracketed by the lines, the CLEANUP_DOCSTRING state (see below) is ON.
toggle_docstring_cleanup_line = '        """'

def char_to_str_xform(line):
    """This transforms the 'char', i.e, 'char *' to 'str', Python string."""
    line = line.replace(' char', ' str')
    line = line.replace('char ', 'str ')
    # Special case handling of 'char **argv' and 'char **envp'.
    line = line.replace('str argv', 'list argv')
    line = line.replace('str envp', 'list envp')
    return line

#
# The one-liner docstring also needs char_to_str transformation, btw.
#
TWO_SPACES = ' ' * 2
EIGHT_SPACES = ' ' * 8
one_liner_docstring_pattern = re.compile('^(%s|%s)""".*"""$' % (TWO_SPACES, EIGHT_SPACES))

#
# lldb_helpers and lldb_iter() should appear before our first SB* class definition.
#
lldb_helpers = '''
# ==================================
# Helper function for SBModule class
# ==================================
def in_range(symbol, section):
    """Test whether a symbol is within the range of a section."""
    symSA = symbol.GetStartAddress().GetFileAddress()
    symEA = symbol.GetEndAddress().GetFileAddress()
    secSA = section.GetFileAddress()
    secEA = secSA + section.GetByteSize()

    if symEA != LLDB_INVALID_ADDRESS:
        if secSA <= symSA and symEA <= secEA:
            return True
        else:
            return False
    else:
        if secSA <= symSA and symSA < secEA:
            return True
        else:
            return False
'''

lldb_iter_def = '''
# ===================================
# Iterator for lldb container objects
# ===================================
def lldb_iter(obj, getsize, getelem):
    """A generator adaptor to support iteration for lldb container objects."""
    size = getattr(obj, getsize)
    elem = getattr(obj, getelem)
    for i in range(size()):
        yield elem(i)

# ==============================================================================
# The modify-python-lldb.py script is responsible for post-processing this SWIG-
# generated lldb.py module.  It is responsible for adding the above lldb_iter()
# function definition as well as the supports, in the following, for iteration
# protocol: __iter__, rich comparison methods: __eq__ and __ne__, truth value
# testing (and built-in operation bool()): __nonzero__, and built-in function
# len(): __len__.
# ==============================================================================
'''

#
# linked_list_iter() is a special purpose iterator to treat the SBValue as the
# head of a list data structure, where you specify the child member name which
# points to the next item on the list and you specify the end-of-list function
# which takes an SBValue and returns True if EOL is reached and False if not.
#
linked_list_iter_def = '''
    def __eol_test__(val):
        """Default function for end of list test takes an SBValue object.

        Return True if val is invalid or it corresponds to a null pointer.
        Otherwise, return False.
        """
        if not val or val.GetValueAsUnsigned() == 0:
            return True
        else:
            return False

    # ==================================================
    # Iterator for lldb.SBValue treated as a linked list
    # ==================================================
    def linked_list_iter(self, next_item_name, end_of_list_test=__eol_test__):
        """Generator adaptor to support iteration for SBValue as a linked list.

        linked_list_iter() is a special purpose iterator to treat the SBValue as
        the head of a list data structure, where you specify the child member
        name which points to the next item on the list and you specify the
        end-of-list test function which takes an SBValue for an item and returns
        True if EOL is reached and False if not.

        linked_list_iter() also detects infinite loop and bails out early.

        The end_of_list_test arg, if omitted, defaults to the __eol_test__
        function above.

        For example,

        # Get Frame #0.
        ...

        # Get variable 'task_head'.
        task_head = frame0.FindVariable('task_head')
        ...

        for t in task_head.linked_list_iter('next'):
            print t
        """
        if end_of_list_test(self):
            return
        item = self
        visited = set()
        try:
            while not end_of_list_test(item) and not item.GetValueAsUnsigned() in visited:
                visited.add(item.GetValueAsUnsigned())
                yield item
                # Prepare for the next iteration.
                item = item.GetChildMemberWithName(next_item_name)
        except:
            # Exception occurred.  Stop the generator.
            pass

        return
'''

# This supports the iteration protocol.
iter_def = "    def __iter__(self): return lldb_iter(self, '%s', '%s')"
module_iter = "    def module_iter(self): return lldb_iter(self, '%s', '%s')"
breakpoint_iter = "    def breakpoint_iter(self): return lldb_iter(self, '%s', '%s')"
watchpoint_iter = "    def watchpoint_iter(self): return lldb_iter(self, '%s', '%s')"
section_iter = "    def section_iter(self): return lldb_iter(self, '%s', '%s')"

# Called to implement the built-in function len().
# Eligible objects are those containers with unambiguous iteration support.
len_def = "    def __len__(self): return self.%s()"

# This supports the rich comparison methods of __eq__ and __ne__.
eq_def = "    def __eq__(self, other): return isinstance(other, %s) and %s"
ne_def = "    def __ne__(self, other): return not self.__eq__(other)"

# Called to implement truth value testing and the built-in operation bool();
# should return False or True, or their integer equivalents 0 or 1.
# Delegate to self.IsValid() if it is defined for the current lldb object.
nonzero_def = "    def __nonzero__(self): return self.IsValid()"

# A convenience iterator for SBSymbol!
symbol_in_section_iter_def = '''
    def symbol_in_section_iter(self, section):
        """Given a module and its contained section, returns an iterator on the
        symbols within the section."""
        for sym in self:
            if in_range(sym, section):
                yield sym
'''

#
# This dictionary defines a mapping from classname to (getsize, getelem) tuple.
#
d = { 'SBBreakpoint':  ('GetNumLocations',   'GetLocationAtIndex'),
      'SBCompileUnit': ('GetNumLineEntries', 'GetLineEntryAtIndex'),
      'SBDebugger':    ('GetNumTargets',     'GetTargetAtIndex'),
      'SBModule':      ('GetNumSymbols',     'GetSymbolAtIndex'),
      'SBProcess':     ('GetNumThreads',     'GetThreadAtIndex'),
      'SBSection':     ('GetNumSubSections', 'GetSubSectionAtIndex'),
      'SBThread':      ('GetNumFrames',      'GetFrameAtIndex'),

      'SBInstructionList':   ('GetSize', 'GetInstructionAtIndex'),
      'SBStringList':        ('GetSize', 'GetStringAtIndex',),
      'SBSymbolContextList': ('GetSize', 'GetContextAtIndex'),
      'SBTypeList':          ('GetSize', 'GetTypeAtIndex'),
      'SBValueList':         ('GetSize', 'GetValueAtIndex'),

      'SBType':  ('GetNumberChildren', 'GetChildAtIndex'),
      'SBValue': ('GetNumChildren',    'GetChildAtIndex'),

      # SBTarget needs special processing, see below.
      'SBTarget': {'module':     ('GetNumModules', 'GetModuleAtIndex'),
                   'breakpoint': ('GetNumBreakpoints', 'GetBreakpointAtIndex'),
                   'watchpoint': ('GetNumWatchpoints', 'GetWatchpointAtIndex')
                   },

      # SBModule has an additional section_iter(), see below.
      'SBModule-section': ('GetNumSections', 'GetSectionAtIndex'),
      # As well as symbol_in_section_iter().
      'SBModule-symbol-in-section': symbol_in_section_iter_def
      }

#
# This dictionary defines a mapping from classname to equality method name(s).
#
e = { 'SBAddress':            ['GetFileAddress', 'GetModule'],
      'SBBreakpoint':         ['GetID'],
      'SBWatchpoint':         ['GetID'],
      'SBFileSpec':           ['GetFilename', 'GetDirectory'],
      'SBModule':             ['GetFileSpec', 'GetUUIDString'],
      'SBType':               ['GetByteSize', 'GetName']
      }

def list_to_frag(list):
    """Transform a list to equality program fragment.

    For example, ['GetID'] is transformed to 'self.GetID() == other.GetID()',
    and ['GetFilename', 'GetDirectory'] to 'self.GetFilename() == other.GetFilename()
    and self.GetDirectory() == other.GetDirectory()'.
    """
    if not list:
        raise Exception("list should be non-empty")
    frag = StringIO.StringIO()
    for i in range(len(list)):
        if i > 0:
            frag.write(" and ")
        frag.write("self.{0}() == other.{0}()".format(list[i]))
    return frag.getvalue()

class NewContent(StringIO.StringIO):
    """Simple facade to keep track of the previous line to be committed."""
    def __init__(self):
        StringIO.StringIO.__init__(self)
        self.prev_line = None
    def add_line(self, a_line):
        """Add a line to the content, if there is a previous line, commit it."""
        if self.prev_line != None:
            print >> self, self.prev_line
        self.prev_line = a_line
    def del_line(self):
        """Forget about the previous line, do not commit it."""
        self.prev_line = None
    def del_blank_line(self):
        """Forget about the previous line if it is a blank line."""
        if self.prev_line != None and not self.prev_line.strip():
            self.prev_line = None
    def finish(self):
        """Call this when you're finished with populating content."""
        if self.prev_line != None:
            print >> self, self.prev_line
        self.prev_line = None

# The new content will have the iteration protocol defined for our lldb objects.
new_content = NewContent()

with open(output_name, 'r') as f_in:
    content = f_in.read()

# The pattern for recognizing the beginning of an SB class definition.
class_pattern = re.compile("^class (SB.*)\(_object\):$")

# The pattern for recognizing the beginning of the __init__ method definition.
init_pattern = re.compile("^    def __init__\(self.*\):")

# The pattern for recognizing the beginning of the IsValid method definition.
isvalid_pattern = re.compile("^    def IsValid\(")

# These define the states of our finite state machine.
NORMAL = 0
DEFINING_ITERATOR = 1
DEFINING_EQUALITY = 2
CLEANUP_DOCSTRING = 4

# The lldb_iter_def only needs to be inserted once.
lldb_iter_defined = False;

# Our FSM begins its life in the NORMAL state, and transitions to the
# DEFINING_ITERATOR and/or DEFINING_EQUALITY state whenever it encounters the
# beginning of certain class definitions, see dictionaries 'd' and 'e' above.
#
# Note that the two states DEFINING_ITERATOR and DEFINING_EQUALITY are
# orthogonal in that our FSM can be in one, the other, or both states at the
# same time.  During such time, the FSM is eagerly searching for the __init__
# method definition in order to insert the appropriate method(s) into the lldb
# module.
#
# The state CLEANUP_DOCSTRING can be entered from either the NORMAL or the
# DEFINING_ITERATOR/EQUALITY states.  While in this state, the FSM is fixing/
# cleaning the Python docstrings generated by the swig docstring features.
#
# The FSM, in all possible states, also checks the current input for IsValid()
# definition, and inserts a __nonzero__() method definition to implement truth
# value testing and the built-in operation bool().
state = NORMAL
for line in content.splitlines():
    # Handle the state transition into CLEANUP_DOCSTRING state as it is possible
    # to enter this state from either NORMAL or DEFINING_ITERATOR/EQUALITY.
    #
    # If '        """' is the sole line, prepare to transition to the
    # CLEANUP_DOCSTRING state or out of it.
    if line == toggle_docstring_cleanup_line:
        if state & CLEANUP_DOCSTRING:
            # Special handling of the trailing blank line right before the '"""'
            # end docstring marker.
            new_content.del_blank_line()
            state ^= CLEANUP_DOCSTRING
        else:
            state |= CLEANUP_DOCSTRING

    if state == NORMAL:
        match = class_pattern.search(line)
        # Inserts lldb_helpers and the lldb_iter() definition before the first
        # class definition.
        if not lldb_iter_defined and match:
            new_content.add_line(lldb_helpers)
            new_content.add_line(lldb_iter_def)
            lldb_iter_defined = True

        # If we are at the beginning of the class definitions, prepare to
        # transition to the DEFINING_ITERATOR/DEFINING_EQUALITY state for the
        # right class names.
        if match:
            cls = match.group(1)
            if cls in d:
                # Adding support for iteration for the matched SB class.
                state |= DEFINING_ITERATOR
            if cls in e:
                # Adding support for eq and ne for the matched SB class.
                state |= DEFINING_EQUALITY

    if (state & DEFINING_ITERATOR) or (state & DEFINING_EQUALITY):
        match = init_pattern.search(line)
        if match:
            # We found the beginning of the __init__ method definition.
            # This is a good spot to insert the iter and/or eq-ne support.
            #
            # But note that SBTarget has three types of iterations.
            if cls == "SBTarget":
                new_content.add_line(module_iter % (d[cls]['module']))
                new_content.add_line(breakpoint_iter % (d[cls]['breakpoint']))
                new_content.add_line(watchpoint_iter % (d[cls]['watchpoint']))
            else:
                if (state & DEFINING_ITERATOR):
                    new_content.add_line(iter_def % d[cls])
                    new_content.add_line(len_def % d[cls][0])
                if (state & DEFINING_EQUALITY):
                    new_content.add_line(eq_def % (cls, list_to_frag(e[cls])))
                    new_content.add_line(ne_def)

            # SBModule has an extra SBSection iterator and symbol_in_section_iter()!
            if cls == "SBModule":
                new_content.add_line(section_iter % d[cls+'-section'])
                new_content.add_line(d[cls+'-symbol-in-section'])
            
            # This special purpose iterator is for SBValue only!!!
            if cls == "SBValue":
                new_content.add_line(linked_list_iter_def)

            # Next state will be NORMAL.
            state = NORMAL

    if (state & CLEANUP_DOCSTRING):
        # Cleanse the lldb.py of the autodoc'ed residues.
        if c_ifdef_swig in line or c_endif_swig in line:
            continue
        # As well as the comment marker line.
        if c_comment_marker in line:
            continue

        # Also remove the '\a ' and '\b 'substrings.
        line = line.replace('\a ', '')
        line = line.replace('\b ', '')
        # And the leading '///' substring.
        doxygen_comment_match = doxygen_comment_start.match(line)
        if doxygen_comment_match:
            line = line.replace(doxygen_comment_match.group(1), '', 1)

        line = char_to_str_xform(line)

        # Note that the transition out of CLEANUP_DOCSTRING is handled at the
        # beginning of this function already.

    # This deals with one-liner docstring, for example, SBThread.GetName:
    # """GetName(self) -> char""".
    if one_liner_docstring_pattern.match(line):
        line = char_to_str_xform(line)

    # Look for 'def IsValid(*args):', and once located, add implementation
    # of truth value testing for this object by delegation.
    if isvalid_pattern.search(line):
        new_content.add_line(nonzero_def)

    # Pass the original line of content to new_content.
    new_content.add_line(line)

# We are finished with recording new content.
new_content.finish()

with open(output_name, 'w') as f_out:
    f_out.write(new_content.getvalue())
    f_out.write("debugger_unique_id = 0\n")
    f_out.write("SBDebugger.Initialize()\n")