import os, sys, re # the list only for debugging. The real list, used in the real OpenCV build, is specified in CMakeLists.txt opencv_hdr_list = [ "../../core/include/opencv2/core/core.hpp", "../../ml/include/opencv2/ml/ml.hpp", "../../imgproc/include/opencv2/imgproc/imgproc.hpp", "../../calib3d/include/opencv2/calib3d/calib3d.hpp", "../../features2d/include/opencv2/features2d/features2d.hpp", "../../video/include/opencv2/video/tracking.hpp", "../../video/include/opencv2/video/background_segm.hpp", "../../objdetect/include/opencv2/objdetect/objdetect.hpp", "../../highgui/include/opencv2/highgui/highgui.hpp", "opencv_extra_api.hpp", ] """ Each declaration is [funcname, return_value_type /* in C, not in Python */, , ], where each element of is 4-element list itself: [argtype, argname, default_value /* or "" if none */, ] where the list of modifiers is yet another nested list of strings (currently recognized are "/O" for output argument, "/S" for static (i.e. class) methods and "/A value" for the plain C arrays with counters) """ class CppHeaderParser(object): def __init__(self): self.BLOCK_TYPE = 0 self.BLOCK_NAME = 1 self.PROCESS_FLAG = 2 self.PUBLIC_SECTION = 3 self.CLASS_DECL = 4 def batch_replace(self, s, pairs): for before, after in pairs: s = s.replace(before, after) return s def get_macro_arg(self, arg_str, npos): npos2 = npos3 = arg_str.find("(", npos) if npos2 < 0: print "Error: no arguments for the macro at %d" % (self.lineno,) sys.exit(-1) balance = 1 while 1: t, npos3 = self.find_next_token(arg_str, ['(', ')'], npos3+1) if npos3 < 0: print "Error: no matching ')' in the macro call at %d" % (self.lineno,) sys.exit(-1) if t == '(': balance += 1 if t == ')': balance -= 1 if balance == 0: break return arg_str[npos2+1:npos3].strip(), npos3 def parse_arg(self, arg_str, argno): """ Parses [arg_name] Returns arg_type, arg_name, modlist, argno, where modlist is the list of wrapper-related modifiers (such as "output argument", "has counter", ...) and argno is the new index of an anonymous argument. That is, if no arg_str is just an argument type without argument name, the argument name is set to "arg" + str(argno), and then argno is incremented. """ modlist = [] # pass 0: extracts the modifiers if "CV_OUT" in arg_str: modlist.append("/O") arg_str = arg_str.replace("CV_OUT", "") if "CV_IN_OUT" in arg_str: modlist.append("/IO") arg_str = arg_str.replace("CV_IN_OUT", "") isarray = False npos = arg_str.find("CV_CARRAY") if npos >= 0: isarray = True macro_arg, npos3 = self.get_macro_arg(arg_str, npos) modlist.append("/A " + macro_arg) arg_str = arg_str[:npos] + arg_str[npos3+1:] npos = arg_str.find("CV_CUSTOM_CARRAY") if npos >= 0: isarray = True macro_arg, npos3 = self.get_macro_arg(arg_str, npos) modlist.append("/CA " + macro_arg) arg_str = arg_str[:npos] + arg_str[npos3+1:] arg_str = arg_str.strip() word_start = 0 word_list = [] npos = -1 #print self.lineno, ":\t", arg_str # pass 1: split argument type into tokens while 1: npos += 1 t, npos = self.find_next_token(arg_str, [" ", "&", "*", "<", ">", ","], npos) w = arg_str[word_start:npos].strip() if w == "operator": word_list.append("operator " + arg_str[npos:].strip()) break if w not in ["", "const"]: word_list.append(w) if t not in ["", " ", "&"]: word_list.append(t) if not t: break word_start = npos+1 npos = word_start - 1 arg_type = "" arg_name = "" angle_stack = [] #print self.lineno, ":\t", word_list # pass 2: decrypt the list wi = -1 prev_w = "" for w in word_list: wi += 1 if w == "*": if prev_w == "char" and not isarray: arg_type = arg_type[:-len("char")] + "c_string" else: arg_type += w continue elif w == "<": arg_type += "_" angle_stack.append(0) elif w == "," or w == '>': if not angle_stack: print "Error at %d: argument contains ',' or '>' not within template arguments" % (self.lineno,) sys.exit(-1) if w == ",": arg_type += "_and_" elif w == ">": if angle_stack[0] == 0: print "Error at %d: template has no arguments" % (self.lineno,) sys.exit(-1) if angle_stack[0] > 1: arg_type += "_end_" angle_stack[-1:] = [] elif angle_stack: arg_type += w angle_stack[-1] += 1 elif arg_type and arg_type != "~": arg_name = " ".join(word_list[wi:]) break else: arg_type += w prev_w = w counter_str = "" add_star = False if ("[" in arg_name) and not ("operator" in arg_str): #print arg_str p1 = arg_name.find("[") p2 = arg_name.find("]",p1+1) if p2 < 0: print "Error at %d: no closing ]" % (self.lineno,) sys.exit(-1) counter_str = arg_name[p1+1:p2].strip() if counter_str == "": counter_str = "?" if not isarray: modlist.append("/A " + counter_str.strip()) arg_name = arg_name[:p1] add_star = True if not arg_name: if arg_type.startswith("operator"): arg_type, arg_name = "", arg_type else: arg_name = "arg" + str(argno) argno += 1 while arg_type.endswith("_end_"): arg_type = arg_type[:-len("_end_")] if add_star: arg_type += "*" arg_type = self.batch_replace(arg_type, [("std::", ""), ("cv::", "")]) return arg_type, arg_name, modlist, argno def parse_enum(self, decl_str): l = decl_str ll = l.split(",") prev_val = "" prev_val_delta = -1 decl = [] for pair in ll: pv = pair.split("=") if len(pv) == 1: prev_val_delta += 1 val = "" if prev_val: val = prev_val + "+" val += str(prev_val_delta) else: prev_val_delta = 0 prev_val = val = pv[1].strip() decl.append(["const " + self.get_dotted_name(pv[0].strip()), val, [], []]) return decl def parse_class_decl(self, decl_str): """ Parses class/struct declaration start in the form: {class|struct} [CV_EXPORTS] [: public [, ...]] Returns class_name1, """ l = decl_str modlist = [] if "CV_EXPORTS_W_MAP" in l: l = l.replace("CV_EXPORTS_W_MAP", "") modlist.append("/Map") if "CV_EXPORTS_W_SIMPLE" in l: l = l.replace("CV_EXPORTS_W_SIMPLE", "") modlist.append("/Simple") npos = l.find("CV_EXPORTS_AS") if npos >= 0: macro_arg, npos3 = self.get_macro_arg(l, npos) modlist.append("=" + macro_arg) l = l[:npos] + l[npos3+1:] l = self.batch_replace(l, [("CV_EXPORTS_W", ""), ("CV_EXPORTS", ""), ("public ", " "), ("::", ".")]).strip() ll = re.split(r'\s*[,:]?\s*', l) ll = [le for le in ll if le] classname = ll[1] bases = ll[2:] return classname, bases, modlist def parse_func_decl(self, decl_str): """ Parses the function or method declaration in the form: [([CV_EXPORTS] ) | CVAPI(rettype)] [~] ( [=] [, [=] ...]) [const] {; | } Returns the function declaration entry: [, , , ] (see above) """ if not (("CV_EXPORTS_AS" in decl_str) or ("CV_EXPORTS_W" in decl_str) or \ ("CV_WRAP" in decl_str) or ("CV_WRAP_AS" in decl_str)): return [] top = self.block_stack[-1] func_modlist = [] npos = decl_str.find("CV_EXPORTS_AS") if npos >= 0: arg, npos3 = self.get_macro_arg(decl_str, npos) func_modlist.append("="+arg) decl_str = decl_str[:npos] + decl_str[npos3+1:] npos = decl_str.find("CV_WRAP_AS") if npos >= 0: arg, npos3 = self.get_macro_arg(decl_str, npos) func_modlist.append("="+arg) decl_str = decl_str[:npos] + decl_str[npos3+1:] # filter off some common prefixes, which are meaningless for Python wrappers. # note that we do not strip "static" prefix, which does matter; # it means class methods, not instance methods decl_str = self.batch_replace(decl_str, [("virtual", ""), ("static inline", ""), ("inline", ""),\ ("CV_EXPORTS_W", ""), ("CV_EXPORTS", ""), ("CV_WRAP ", " "), ("static CV_INLINE", ""), ("CV_INLINE", "")]).strip() static_method = False context = top[0] if decl_str.startswith("static") and (context == "class" or context == "struct"): decl_str = decl_str[len("static"):].lstrip() static_method = True args_begin = decl_str.find("(") if decl_str.startswith("CVAPI"): rtype_end = decl_str.find(")", args_begin+1) if rtype_end < 0: print "Error at %d. no terminating ) in CVAPI() macro: %s" % (self.lineno, decl_str) sys.exit(-1) decl_str = decl_str[args_begin+1:rtype_end] + " " + decl_str[rtype_end+1:] args_begin = decl_str.find("(") if args_begin < 0: print "Error at %d: no args in '%s'" % (self.lineno, decl_str) sys.exit(-1) decl_start = decl_str[:args_begin].strip() # handle operator () case if decl_start.endswith("operator"): args_begin = decl_str.find("(", args_begin+1) if args_begin < 0: print "Error at %d: no args in '%s'" % (self.lineno, decl_str) sys.exit(-1) decl_start = decl_str[:args_begin].strip() rettype, funcname, modlist, argno = self.parse_arg(decl_start, -1) if argno >= 0: classname = top[1] if rettype == classname or rettype == "~" + classname: rettype, funcname = "", rettype else: print "Error at %d. the function/method name is missing: '%s'" % (self.lineno, decl_start) sys.exit(-1) if ("::" in funcname) or funcname.startswith("~"): # if there is :: in function name (and this is in the header file), # it means, this is inline implementation of a class method. # Thus the function has been already declared within the class and we skip this repeated # declaration. # Also, skip the destructors, as they are always wrapped return [] funcname = self.get_dotted_name(funcname) arg_start = args_begin+1 npos = arg_start-1 balance = 1 angle_balance = 0 # scan the argument list; handle nested parentheses args_decls = [] args = [] argno = 1 while balance > 0: npos += 1 t, npos = self.find_next_token(decl_str, ["(", ")", ",", "<", ">"], npos) if not t: print "Error: no closing ')' at %d" % (self.lineno,) print decl_str print decl_str[arg_start:] sys.exit(-1) if t == "<": angle_balance += 1 if t == ">": angle_balance -= 1 if t == "(": balance += 1 if t == ")": balance -= 1 if (t == "," and balance == 1 and angle_balance == 0) or balance == 0: # process next function argument a = decl_str[arg_start:npos].strip() #print "arg = ", a arg_start = npos+1 if a: eqpos = a.find("=") defval = "" modlist = [] if eqpos >= 0: defval = a[eqpos+1:].strip() else: eqpos = a.find("CV_DEFAULT") if eqpos >= 0: defval, pos3 = self.get_macro_arg(a, eqpos) else: eqpos = a.find("CV_WRAP_DEFAULT") if eqpos >= 0: defval, pos3 = self.get_macro_arg(a, eqpos) if defval == "NULL": defval = "0" if eqpos >= 0: a = a[:eqpos].strip() arg_type, arg_name, modlist, argno = self.parse_arg(a, argno) if arg_type == "InputArray" or arg_type == "InputOutputArray": arg_type = "Mat" elif arg_type == "OutputArray": arg_type = "Mat" modlist.append("/O") elif arg_type == "InputArrayOfArrays" or arg_type == "InputOutputArrayOfArrays": arg_type = "vector_Mat" elif arg_type == "OutputArrayOfArrays": arg_type = "vector_Mat" modlist.append("/O") defval = self.batch_replace(defval, [("InputArrayOfArrays", "vector"), ("InputOutputArrayOfArrays", "vector"), ("OutputArrayOfArrays", "vector"), ("InputArray", "Mat"), ("InputOutputArray", "Mat"), ("OutputArray", "Mat")]).strip() args.append([arg_type, arg_name, defval, modlist]) npos = arg_start-1 npos = decl_str.replace(" ", "").find("=0", npos) if npos >= 0: # skip pure virtual functions return [] if static_method: func_modlist.append("/S") return [funcname, rettype, func_modlist, args] def get_dotted_name(self, name): """ adds the dot-separated container class/namespace names to the bare function/class name, e.g. when we have namespace cv { class A { public: f(int); }; } the function will convert "A" to "cv.A" and "f" to "cv.A.f". """ if not self.block_stack: return name n = "" for b in self.block_stack: block_type, block_name = b[self.BLOCK_TYPE], b[self.BLOCK_NAME] if block_type in ["file", "enum"]: continue if block_type not in ["struct", "class", "namespace"]: print "Error at %d: there are non-valid entries in the current block stack " % (self.lineno, self.block_stack) sys.exit(-1) if block_name: n += block_name + "." return n + name def parse_stmt(self, stmt, end_token): """ parses the statement (ending with ';' or '}') or a block head (ending with '{') The function calls parse_class_decl or parse_func_decl when necessary. It returns , , , where the first 3 values only make sense for blocks (i.e. code blocks, namespaces, classes, enums and such) """ stack_top = self.block_stack[-1] context = stack_top[self.BLOCK_TYPE] stmt_type = "" if end_token == "{": stmt_type = "block" if context == "block": print "Error at %d: should not call parse_stmt inside blocks" % (self.lineno,) sys.exit(-1) if context == "class" or context == "struct": while 1: colon_pos = stmt.find(":") if colon_pos < 0: break w = stmt[:colon_pos].strip() if w in ["public", "protected", "private"]: if w == "public": stack_top[self.PUBLIC_SECTION] = True else: stack_top[self.PUBLIC_SECTION] = False stmt = stmt[colon_pos+1:].strip() break # do not process hidden class members and template classes/functions if not stack_top[self.PUBLIC_SECTION] or stmt.startswith("template"): return stmt_type, "", False, None if end_token == "{": if stmt.startswith("class") or stmt.startswith("struct"): stmt_type = stmt.split()[0] classname, bases, modlist = self.parse_class_decl(stmt) decl = [] if ("CV_EXPORTS_W" in stmt) or ("CV_EXPORTS_AS" in stmt): decl = [stmt_type + " " + self.get_dotted_name(classname), "", modlist, []] if bases: decl[1] = ": " + " ".join(bases) return stmt_type, classname, True, decl if stmt.startswith("enum"): return "enum", "", True, None if stmt.startswith("namespace"): stmt_list = stmt.split() return stmt_list[0], stmt_list[1], True, None if stmt.startswith("extern") and "\"C\"" in stmt: return "namespace", "", True, None if end_token == "}" and context == "enum": decl = self.parse_enum(stmt) return "enum", "", False, decl if end_token == ";" and stmt.startswith("typedef"): # TODO: handle typedef's more intelligently return stmt_type, "", False, None paren_pos = stmt.find("(") if paren_pos >= 0: # assume it's function or method declaration, # since we filtered off the other places where '(' can normally occur: # - code blocks # - function pointer typedef's decl = self.parse_func_decl(stmt) # we return parse_flag == False to prevent the parser to look inside function/method bodies # (except for tracking the nested blocks) return stmt_type, "", False, decl if (context == "struct" or context == "class") and end_token == ";" and stmt: # looks like it's member declaration; append the members to the class declaration class_decl = stack_top[self.CLASS_DECL] if ("CV_PROP" in stmt): # or (class_decl and ("/Map" in class_decl[2])): var_modlist = [] if "CV_PROP_RW" in stmt: var_modlist.append("/RW") stmt = self.batch_replace(stmt, [("CV_PROP_RW", ""), ("CV_PROP", "")]).strip() var_list = stmt.split(",") var_type, var_name1, modlist, argno = self.parse_arg(var_list[0], -1) var_list = [var_name1] + [i.strip() for i in var_list[1:]] for v in var_list: class_decl[3].append([var_type, v, "", var_modlist]) return stmt_type, "", False, None # something unknown return stmt_type, "", False, None def find_next_token(self, s, tlist, p=0): """ Finds the next token from the 'tlist' in the input 's', starting from position 'p'. Returns the first occured token and its position, or ("", len(s)) when no token is found """ token = "" tpos = len(s) for t in tlist: pos = s.find(t, p) if pos < 0: continue if pos < tpos: tpos = pos token = t return token, tpos def parse(self, hname): """ The main method. Parses the input file. Returns the list of declarations (that can be print using print_decls) """ decls = [] f = open(hname, "rt") linelist = list(f.readlines()) f.close() # states: SCAN = 0 # outside of a comment or preprocessor directive COMMENT = 1 # inside a multi-line comment DIRECTIVE = 2 # inside a multi-line preprocessor directive state = SCAN self.block_stack = [["file", hname, True, True, None]] block_head = "" self.lineno = 0 for l0 in linelist: self.lineno += 1 #print self.lineno l = l0.strip() if state == SCAN and l.startswith("#"): state = DIRECTIVE # fall through to the if state == DIRECTIVE check if state == DIRECTIVE: if not l.endswith("\\"): state = SCAN continue if state == COMMENT: pos = l.find("*/") if pos < 0: continue l = l[pos+2:] state = SCAN if state != SCAN: print "Error at %d: invlid state = %d" % (self.lineno, state) sys.exit(-1) while 1: token, pos = self.find_next_token(l, [";", "\"", "{", "}", "//", "/*"]) if not token: block_head += " " + l break if token == "//": block_head += " " + l[:pos] break if token == "/*": block_head += " " + l[:pos] pos = l.find("*/", pos+2) if pos < 0: state = COMMENT break l = l[pos+2:] continue if token == "\"": pos2 = pos + 1 while 1: t2, pos2 = self.find_next_token(l, ["\\", "\""], pos2) if t2 == "": print "Error at %d: no terminating '\"'" % (self.lineno,) sys.exit(-1) if t2 == "\"": break pos2 += 2 block_head += " " + l[:pos2+1] l = l[pos2+1:] continue stmt = (block_head + " " + l[:pos]).strip() stmt = " ".join(stmt.split()) # normalize the statement stack_top = self.block_stack[-1] decl = None if stack_top[self.PROCESS_FLAG]: # even if stack_top[PUBLIC_SECTION] is False, we still try to process the statement, # since it can start with "public:" stmt_type, name, parse_flag, decl = self.parse_stmt(stmt, token) if decl: if stmt_type == "enum": for d in decl: decls.append(d) else: decls.append(decl) else: stmt_type, name, parse_flag = "block", "", False if token == "{": if stmt_type == "class": public_section = False else: public_section = True self.block_stack.append([stmt_type, name, parse_flag, public_section, decl]) if token == "}": if not self.block_stack: print "Error at %d: the block stack is empty" % (self.lineno,) self.block_stack[-1:] = [] if pos+1 < len(l) and l[pos+1] == ';': pos += 1 block_head = "" l = l[pos+1:] return decls def print_decls(self, decls): """ Prints the list of declarations, retrieived by the parse() method """ for d in decls: print d[0], d[1], ";".join(d[2]) for a in d[3]: print " ", a[0], a[1], a[2], if a[3]: print "; ".join(a[3]) else: print if __name__ == '__main__': parser = CppHeaderParser() decls = [] for hname in opencv_hdr_list: decls += parser.parse(hname) parser.print_decls(decls) print len(decls)