pmacs3/ctags.py

#!/usr/bin/python
#
# by Erik Osheim

import os, re, sets, sys

entries = {}
packages = {}
classes = {}
class_methods = {}

class_re    = re.compile('^(.*?)\t(.*?)\t(.*?)\tc$')
function_re = re.compile('^(.*?)\t(.*?)\t(.*?)\tf$')
method_re   = re.compile('^([^\t]+)\t([^\t]+)\t([^\t]+)\tm\tclass:([^\t]+)(?:\t.*)?$')

class_supers_re = re.compile('^\/\^ *class +[_A-Za-z][_A-Za-z0-9]* *\((.*?)\) *: *\$\/;\"$')
def_args_re     = re.compile('^\/\^ *def +[_A-Za-z][_A-Za-z0-9]* *\((.*?)\) *: *\$\/;\"$')
find_args_re    = re.compile('[\*_a-zA-Z][\*_a-zA-Z0-9]*(?:=(?:[^,\'" ]+|"(?:\\.|[^\\"])*"|\'(?:\\.|[^\\"])*\'))?')

base_objects = sets.Set(['object', 'list', 'dict'])

def is_fully_qualified(s):
    return s in base_objects or '.' in s

def parse_entry(line):
    m = class_re.match(line)
    if m:
        return ClassEntry(m.group(1), m.group(2), m.group(3))
    m = function_re.match(line)
    if m:
        return FunctionEntry(m.group(1), m.group(2), m.group(3))
    m = method_re.match(line)
    if m:
        return MethodEntry(m.group(1), m.group(2), m.group(3), m.group(4))
    raise Exception, "Oh no: %s" % line

class Entry:
    type = 'generic'
    def __init__(self, symbol, path):
        self.symbol = symbol
        self.path = path
    def __repr__(self):
        return '<%s %s.%s>' % (self.type.title(), self.package(), self.symbol)
    def package(self):
        return self.path[:-3].replace('/', '.')
    def fullname(self):
        return '%s.%s' % (self.package(), self.symbol)
    def prototype(self):
        return self.fullname()
    def dump(self):
        return '%s %s' % (self.type, self.prototype())
class ClassEntry(Entry):
    type = 'class'
    def __init__(self, symbol, path, match):
        Entry.__init__(self, symbol, path)
        m = class_supers_re.match(match)
        self.match = match
        self.supers = []
        if m:
            self.supers = [x.strip() for x in m.group(1).split(',')]
        for i in range(0, len(self.supers)):
            if not is_fully_qualified(self.supers[i]):
                self.supers[i] = '%s.%s' % (self.package(), self.supers[i])
    def prototype(self):
        return '%s(%s)' % (self.fullname(), ', '.join(self.supers))
class FunctionEntry(Entry):
    type = 'function'
    def __init__(self, symbol, path, match):
        Entry.__init__(self, symbol, path)
        m = def_args_re.match(match)
        self.match = match
        self.args = []
        if m:
            self.args = re.findall(find_args_re, m.group(1))
    def prototype(self):
        return '%s(%s)' % (self.fullname(), ', '.join(self.args))
class MethodEntry(FunctionEntry):
    type = 'method'
    def __init__(self, symbol, path, match, parent):
        FunctionEntry.__init__(self, symbol, path, match)
        self.parent = parent
        if is_fully_qualified(parent):
            self.parent = parent
        else:
            self.parent = '%s.%s' % (self.package(), parent)
    def fullname(self):
        return '%s.%s' % (self.parent, self.symbol)

def process_tagfile(path):
    global entries, classes, packages
    f = open(path, 'r')
    data = f.read()
    f.close()
    process_data(data)

def process_direct():
    (stdin, stdout, stderr) = os.popen3("exuberant-ctags -L - -f -")
    for base in ('.'):
        for root, dirs, files in os.walk(base):
            if 'CVS' in dirs:
                dirs.remove('CVS')
            for name in files:
                if name.endswith('.py'):
                    if base != '.':
                        path = os.path.join(root, name)
                    else:
                        path = name
                    stdin.write('%s\n' % path)
                    stdin.flush()
    stdin.close()
    data = stdout.read()
    stdout.close()
    stderr.close()
    process_data(data)

def process_data(data):
    global entries, classes, packages
    # process the ctags output data
    for l in data.split('\n'):
        if not l:
            continue
        elif l.startswith('!'):
            continue
        else:
            e = parse_entry(l)
            entries[e.fullname()] = e
    
            package = e.package()
            if e.type == 'method':
                p = e.parent
                if not is_fully_qualified(p):
                    p = '%s.%s' % (package, p)
                classes.setdefault(p, {})
                classes[p][e.symbol] = e
            else:
                packages.setdefault(package, {})
                packages[package][e.symbol] = e

# this returns the methods available in the class
def get_methods_for_class(c):
    global class_methods

    cn = c.fullname()

    # if we haven't determined this classes methods yet, then let's do it
    if cn not in class_methods:
        classes_seen = sets.Set()
        methods_seen = sets.Set()
        class_methods[cn] = []

        # create a queue of classes to process... this solves the ordering
        # problem for class inheritance...i.e.:
        #     class Shape
        #     class Rectangle(Shape)
        #     class Rhombus(Shape)
        #     class Square(Rectangle, Rhombus)
        # 1. [Square]             --> process Square    --> [Rectangle, Rhombus]
        # 2. [Rectangle, Rhombus] --> process Rectangle --> [Rhombus, Shape]
        # 3. [Rhombus, Shape]     --> process Rhombus   --> [Shape, Shape]
        # 4. [Shape, Shape]       --> process Shape     --> [Shape]
        # 5. [Shape]              --> already processed Shape, skipping
        to_process = [c]
        while to_process:
            e = to_process.pop(0)
            fn = e.fullname()
    
            # if we've seen this class already, then skip it
            if fn in classes_seen:
                continue
    
            # mark that we've seen this class; if we don't know about it's
            # methods, then let's just skip it.
            classes_seen.add(fn)
            if fn not in classes:
                continue
    
            # for each method in the class, add it to our list if it's new
            for msymbol in classes[fn]:
                if msymbol not in methods_seen:
                    class_methods[cn].append(classes[fn][msymbol])
                    methods_seen.add(msymbol)
    
            # for each parent of this class, append it to the end of the queue
            # if we know about it
            for sfn in e.supers:
                if sfn in entries:
                    to_process.append(entries[sfn])

    return class_methods[cn]

if __name__ == "__main__":
    if len(sys.argv[1:]) == 0:
        process_direct()
    else:
        process_tagfile(sys.argv[1])

    # exit here to get good timing data
    #sys.exit(0)

    # for each package, print out the classes and functions in that package
    for p in packages:
        print 'package %s' % p
        for es in packages[p]:
            e = packages[p][es]
            print '    %s %s' % (e.type, e.prototype())
            fn = e.fullname()
            # for each class, print out the methods that class provides (either
            # implemented directly or inherited from a super class)
            if e.type == 'class':
                for m in get_methods_for_class(e):
                    # status determines whether the class is being inherited,
                    # or implemented directly
                    if fn != m.parent:
                        status = '*'
                    else:
                        status = ' '
                    print '      %s %s' % (status, m.dump())
        print ''