apibuild.py

libxml,在UNIX/LINUX下非常重要的一个库,为XML相关应用提供方便.目前上载的是最新版本,若要取得最新版本,请参考里面的readme.

#!/usr/bin/python -u
#
# This is the API builder, it parses the C sources and build the
# API formal description in XML.
#
# See Copyright for the status of this software.
#
# [e-mail]
#
import os, sys
import string
import glob

debug=0
#debugsym='ignorableWhitespaceSAXFunc'
debugsym=None

#
# C parser analysis code
#
ignored_files = {
  "trio": "too many non standard macros",
  "trio.c": "too many non standard macros",
  "trionan.c": "too many non standard macros",
  "triostr.c": "too many non standard macros",
  "acconfig.h": "generated portability layer",
  "config.h": "generated portability layer",
  "libxml.h": "internal only",
  "testOOM.c": "out of memory tester",
  "testOOMlib.h": "out of memory tester",
  "testOOMlib.c": "out of memory tester",
  "rngparser.c": "not yet integrated",
  "rngparser.h": "not yet integrated",
  "elfgcchack.h": "not a normal header",
  "testapi.c": "generated regression tests",
  "tst.c": "not part of the library",
  "testdso.c": "test for dynamid shared libraries",
}

ignored_words = {
  "WINAPI": (0, "Windows keyword"),
  "LIBXML_DLL_IMPORT": (0, "Special macro to flag external keywords"),
  "XMLPUBVAR": (0, "Special macro for extern vars for win32"),
  "XSLTPUBVAR": (0, "Special macro for extern vars for win32"),
  "EXSLTPUBVAR": (0, "Special macro for extern vars for win32"),
  "XMLPUBFUN": (0, "Special macro for extern funcs for win32"),
  "XSLTPUBFUN": (0, "Special macro for extern funcs for win32"),
  "EXSLTPUBFUN": (0, "Special macro for extern funcs for win32"),
  "XMLCALL": (0, "Special macro for win32 calls"),
  "XSLTCALL": (0, "Special macro for win32 calls"),
  "EXSLTCALL": (0, "Special macro for win32 calls"),
  "__declspec": (3, "Windows keyword"),
  "__stdcall": (0, "Windows keyword"),
  "ATTRIBUTE_UNUSED": (0, "macro keyword"),
  "LIBEXSLT_PUBLIC": (0, "macro keyword"),
  "X_IN_Y": (5, "macro function builder"),
}

def escape(raw):
    raw = string.replace(raw, '&', '&')
    raw = string.replace(raw, '<', '&lt;')
    raw = string.replace(raw, '>', '&gt;')
    raw = string.replace(raw, "'", '&apos;')
    raw = string.replace(raw, '"', '&quot;')
    return raw

def uniq(items):
    d = {}
    for item in items:
        d[item]=1
    return d.keys()

class identifier:
    def __init__(self, name, header=None, module=None, type=None, lineno = 0,
                 info=None, extra=None, conditionals = None):
        self.name = name
	self.header = header
	self.module = module
	self.type = type
	self.info = info
	self.extra = extra
	self.lineno = lineno
	self.static = 0
	if conditionals == None or len(conditionals) == 0:
	    self.conditionals = None
	else:
	    self.conditionals = conditionals[:]
	if self.name == debugsym:
	    print "=> define %s : %s" % (debugsym, (module, type, info,
	                                 extra, conditionals))

    def __repr__(self):
        r = "%s %s:" % (self.type, self.name)
	if self.static:
	    r = r + " static"
	if self.module != None:
	    r = r + " from %s" % (self.module)
	if self.info != None:
	    r = r + " " +  `self.info`
	if self.extra != None:
	    r = r + " " + `self.extra`
	if self.conditionals != None:
	    r = r + " " + `self.conditionals`
	return r


    def set_header(self, header):
        self.header = header
    def set_module(self, module):
        self.module = module
    def set_type(self, type):
        self.type = type
    def set_info(self, info):
        self.info = info
    def set_extra(self, extra):
        self.extra = extra
    def set_lineno(self, lineno):
        self.lineno = lineno
    def set_static(self, static):
        self.static = static
    def set_conditionals(self, conditionals):
	if conditionals == None or len(conditionals) == 0:
	    self.conditionals = None
	else:
	    self.conditionals = conditionals[:]

    def get_name(self):
        return self.name
    def get_header(self):
        return self.module
    def get_module(self):
        return self.module
    def get_type(self):
        return self.type
    def get_info(self):
        return self.info
    def get_lineno(self):
        return self.lineno
    def get_extra(self):
        return self.extra
    def get_static(self):
        return self.static
    def get_conditionals(self):
        return self.conditionals

    def update(self, header, module, type = None, info = None, extra=None,
               conditionals=None):
	if self.name == debugsym:
	    print "=> update %s : %s" % (debugsym, (module, type, info,
	                                 extra, conditionals))
        if header != None and self.header == None:
	    self.set_header(module)
        if module != None and (self.module == None or self.header == self.module):
	    self.set_module(module)
        if type != None and self.type == None:
	    self.set_type(type)
        if info != None:
	    self.set_info(info)
        if extra != None:
	    self.set_extra(extra)
        if conditionals != None:
	    self.set_conditionals(conditionals)

class index:
    def __init__(self, name = "noname"):
        self.name = name
        self.identifiers = {}
        self.functions = {}
	self.variables = {}
	self.includes = {}
	self.structs = {}
	self.enums = {}
	self.typedefs = {}
	self.macros = {}
	self.references = {}
	self.info = {}

    def add_ref(self, name, header, module, static, type, lineno, info=None, extra=None, conditionals = None):
        if name[0:2] == '__':
	    return None
        d = None
        try:
	   d = self.identifiers[name]
	   d.update(header, module, type, lineno, info, extra, conditionals)
	except:
	   d = identifier(name, header, module, type, lineno, info, extra, conditionals)
	   self.identifiers[name] = d

	if d != None and static == 1:
	    d.set_static(1)

	if d != None and name != None and type != None:
	    self.references[name] = d

	if name == debugsym:
	    print "New ref: %s" % (d)

	return d

    def add(self, name, header, module, static, type, lineno, info=None, extra=None, conditionals = None):
        if name[0:2] == '__':
	    return None
        d = None
        try:
	   d = self.identifiers[name]
	   d.update(header, module, type, lineno, info, extra, conditionals)
	except:
	   d = identifier(name, header, module, type, lineno, info, extra, conditionals)
	   self.identifiers[name] = d

	if d != None and static == 1:
	    d.set_static(1)

	if d != None and name != None and type != None:
	    if type == "function":
	        self.functions[name] = d
	    elif type == "functype":
	        self.functions[name] = d
	    elif type == "variable":
	        self.variables[name] = d
	    elif type == "include":
	        self.includes[name] = d
	    elif type == "struct":
	        self.structs[name] = d
	    elif type == "enum":
	        self.enums[name] = d
	    elif type == "typedef":
	        self.typedefs[name] = d
	    elif type == "macro":
	        self.macros[name] = d
	    else:
	        print "Unable to register type ", type

	if name == debugsym:
	    print "New symbol: %s" % (d)

	return d

    def merge(self, idx):
        for id in idx.functions.keys():
              #
              # macro might be used to override functions or variables
              # definitions
              #
	     if self.macros.has_key(id):
	         del self.macros[id]
	     if self.functions.has_key(id):
	         print "function %s from %s redeclared in %s" % (
		    id, self.functions[id].header, idx.functions[id].header)
	     else:
	         self.functions[id] = idx.functions[id]
		 self.identifiers[id] = idx.functions[id]
        for id in idx.variables.keys():
              #
              # macro might be used to override functions or variables
              # definitions
              #
	     if self.macros.has_key(id):
	         del self.macros[id]
	     if self.variables.has_key(id):
	         print "variable %s from %s redeclared in %s" % (
		    id, self.variables[id].header, idx.variables[id].header)
	     else:
	         self.variables[id] = idx.variables[id]
		 self.identifiers[id] = idx.variables[id]
        for id in idx.structs.keys():
	     if self.structs.has_key(id):
	         print "struct %s from %s redeclared in %s" % (
		    id, self.structs[id].header, idx.structs[id].header)
	     else:
	         self.structs[id] = idx.structs[id]
		 self.identifiers[id] = idx.structs[id]
        for id in idx.typedefs.keys():
	     if self.typedefs.has_key(id):
	         print "typedef %s from %s redeclared in %s" % (
		    id, self.typedefs[id].header, idx.typedefs[id].header)
	     else:
	         self.typedefs[id] = idx.typedefs[id]
		 self.identifiers[id] = idx.typedefs[id]
        for id in idx.macros.keys():
              #
              # macro might be used to override functions or variables
              # definitions
              #
             if self.variables.has_key(id):
                 continue
             if self.functions.has_key(id):
                 continue
             if self.enums.has_key(id):
                 continue
	     if self.macros.has_key(id):
	         print "macro %s from %s redeclared in %s" % (
		    id, self.macros[id].header, idx.macros[id].header)
	     else:
	         self.macros[id] = idx.macros[id]
		 self.identifiers[id] = idx.macros[id]
        for id in idx.enums.keys():
	     if self.enums.has_key(id):
	         print "enum %s from %s redeclared in %s" % (
		    id, self.enums[id].header, idx.enums[id].header)
	     else:
	         self.enums[id] = idx.enums[id]
		 self.identifiers[id] = idx.enums[id]

    def merge_public(self, idx):
        for id in idx.functions.keys():
	     if self.functions.has_key(id):
	         # check that function condition agrees with header
	         if idx.functions[id].conditionals != \
		    self.functions[id].conditionals:
		     print "Header condition differs from Function for %s:" \
		        % id
		     print "  H: %s" % self.functions[id].conditionals
		     print "  C: %s" % idx.functions[id].conditionals
	         up = idx.functions[id]
	         self.functions[id].update(None, up.module, up.type, up.info, up.extra)
	 #     else:
	 #         print "Function %s from %s is not declared in headers" % (
	 #	        id, idx.functions[id].module)
	 # TODO: do the same for variables.

    def analyze_dict(self, type, dict):
        count = 0
	public = 0
        for name in dict.keys():
	    id = dict[name]
	    count = count + 1
	    if id.static == 0:
	        public = public + 1
        if count != public:
	    print "  %d %s , %d public" % (count, type, public)
	elif count != 0:
	    print "  %d public %s" % (count, type)


    def analyze(self):
	self.analyze_dict("functions", self.functions)
	self.analyze_dict("variables", self.variables)
	self.analyze_dict("structs", self.structs)
	self.analyze_dict("typedefs", self.typedefs)
	self.analyze_dict("macros", self.macros)
         
class CLexer:
    """A lexer for the C language, tokenize the input by reading and
       analyzing it line by line"""
    def __init__(self, input):
        self.input = input
	self.tokens = []
	self.line = ""
	self.lineno = 0

    def getline(self):
        line = ''
	while line == '':
	    line = self.input.readline()
	    if not line:
		return None
	    self.lineno = self.lineno + 1
	    line = string.lstrip(line)
	    line = string.rstrip(line)
	    if line == '':
	        continue
	    while line[-1] == '\\':
	        line = line[:-1]
		n = self.input.readline()
		self.lineno = self.lineno + 1
		n = string.lstrip(n)
		n = string.rstrip(n)
		if not n:
		    break
		else:
		    line = line + n
        return line
	 
    def getlineno(self):
        return self.lineno

    def push(self, token):
        self.tokens.insert(0, token);

    def debug(self):
        print "Last token: ", self.last
	print "Token queue: ", self.tokens
	print "Line %d end: " % (self.lineno), self.line

    def token(self):
        while self.tokens == []:
	    if self.line == "":
		line = self.getline()
	    else:
	        line = self.line
		self.line = ""
	    if line == None:
	        return None

	    if line[0] == '#':
	        self.tokens = map((lambda x: ('preproc', x)),
		                  string.split(line))
		break;
	    l = len(line)
	    if line[0] == '"' or line[0] == "'":
	        end = line[0]
	        line = line[1:]
		found = 0
		tok = ""
		while found == 0:
		    i = 0
		    l = len(line)
		    while i < l:
			if line[i] == end:
			    self.line = line[i+1:]