Added new script for checking correctness of documentation

doc/check_docs2.py

+import os, sys, glob, re
+
+sys.path.append("../modules/python/src2/")
+sys.path.append("../modules/java/")
+
+import hdr_parser as hp
+import rst_parser as rp
+
+rp.show_warnings = False
+rp.show_errors = False
+
+DOCUMENTED_MARKER = "verified"
+
+ERROR_001_NOTACLASS        = 1
+ERROR_002_NOTASTRUCT       = 2
+ERROR_003_INCORRECTBASE    = 3
+ERROR_004_MISSEDNAMESPACE  = 4
+ERROR_005_MISSINGPYFUNC    = 5
+ERROR_006_INVALIDPYOLDDOC  = 6
+ERROR_007_INVALIDPYDOC     = 7
+ERROR_008_CFUNCISNOTGLOBAL = 8
+ERROR_009_OVERLOADNOTFOUND = 9
+
+do_python_crosscheck = True
+errors_disabled = [ERROR_004_MISSEDNAMESPACE]
+
+if do_python_crosscheck:
+    try:
+        import cv2
+    except ImportError:
+        print "Could not load cv2"
+        do_python_crosscheck = False
+
+def get_cv2_object(name):
+    if name.startswith("cv2."):
+        name = name[4:]
+    if name.startswith("cv."):
+        name = name[3:]
+    if name == "Algorithm":
+        return cv2.Algorithm__create("Feature2D.ORB"), name
+    elif name == "FeatureDetector":
+        return cv2.FeatureDetector_create("ORB"), name
+    elif name == "DescriptorExtractor":
+        return cv2.DescriptorExtractor_create("ORB"), name
+    elif name == "BackgroundSubtractor":
+        return cv2.BackgroundSubtractorMOG(), name
+    elif name == "StatModel":
+        return cv2.KNearest(), name
+    else:
+        return getattr(cv2, name)(), name
+
+def compareSignatures(f, s):
+    # function names
+    if f[0] != s[0]:
+        return False, "name mismatch"
+    # return type
+    stype = (s[1] or "void")
+    ftype = f[1]
+    if stype.startswith("cv::"):
+        stype = stype[4:]
+    if ftype and ftype.startswith("cv::"):
+        ftype = ftype[4:]
+    if ftype and ftype != stype:
+        return False, "return type mismatch"
+    if ("\C" in f[2]) ^ ("\C" in s[2]):
+        return False, "const qulifier mismatch"
+    if len(f[3]) != len(s[3]):
+        return False, "different number of arguments"
+    for idx, arg in enumerate(zip(f[3], s[3])):
+        farg = arg[0]
+        sarg = arg[1]
+        ftype = re.sub(r"\bcv::", "", (farg[0] or ""))
+        stype = re.sub(r"\bcv::", "", (sarg[0] or ""))
+        if ftype != stype:
+            return False, "type of argument #" + str(idx+1) + " mismatch"
+        fname = farg[1] or "arg" + str(idx)
+        sname = sarg[1] or "arg" + str(idx)
+        if fname != sname:
+            return False, "name of argument #" + str(idx+1) + " mismatch"
+        fdef = re.sub(r"\bcv::", "", (farg[2] or ""))
+        sdef = re.sub(r"\bcv::", "", (sarg[2] or ""))
+        if fdef != sdef:
+            return False, "default value of argument #" + str(idx+1) + " mismatch"
+    return True, "match"
+
+def formatSignature(s):
+    _str = ""
+    if s[1]:
+        _str += s[1] + " "
+    else:
+        if not bool(re.match(r"(cv\.)?(?P<cls>\w+)\.(?P=cls)", s[0])):
+            _str += "void "
+    if s[0].startswith("cv."):
+        _str += s[0][3:].replace(".", "::")
+    else:
+        _str += s[0].replace(".", "::")
+    if len(s[3]) == 0:
+        _str += "()"
+    else:
+        _str += "( "
+        for idx, arg in enumerate(s[3]):
+            if idx > 0:
+                _str += ", "
+            _str += re.sub(r"\bcv::", "", arg[0]) + " "
+            if arg[1]:
+                _str += arg[1]
+            else:
+                _str += "arg" + str(idx)
+            if arg[2]:
+                _str += "=" + re.sub(r"\bcv::", "", arg[2])
+        _str += " )"
+    if "/C" in s[2]:
+        _str += " const"
+    return _str
+
+
+def logerror(code, message, doc = None):
+    if code in errors_disabled:
+        return
+    if doc:
+        print doc["file"] + ":" + str(doc["line"]),
+    print "error %03d: %s" % (code, message)
+    #print
+
+def process_module(module, path):
+    hppparser = hp.CppHeaderParser()
+    rstparser = rp.RstParser(hppparser)
+
+    rstparser.parse(module, path)
+    rst = rstparser.definitions
+
+    hdrlist = glob.glob(os.path.join(path, "include", "opencv2", module, "*.h*"))
+    hdrlist.extend(glob.glob(os.path.join(path, "include", "opencv2", module, "detail", "*.h*")))
+
+    decls = []
+    for hname in hdrlist:
+        if not "ts_gtest.h" in hname:
+            decls += hppparser.parse(hname, wmode=False)
+
+    funcs = []
+    # not really needed to hardcode all the namespaces. Normally all they are collected automatically
+    namespaces = ['cv', 'cv.gpu', 'cvflann', 'cvflann.anyimpl', 'cvflann.lsh', 'cv.flann', 'cv.linemod', 'cv.detail', 'cvtest', 'perf', 'cv.videostab']
+    classes = []
+    structs = []
+
+    # collect namespaces and classes/structs
+    for decl in decls:
+        if decl[0].startswith("const"):
+            pass
+        elif decl[0].startswith("class") or decl[0].startswith("struct"):
+            if decl[0][0] == 'c':
+                classes.append(decl)
+            else:
+                structs.append(decl)
+            dotIdx = decl[0].rfind('.')
+            if dotIdx > 0:
+                namespace = decl[0][decl[0].find(' ')+1:dotIdx]
+                if not [c for c in classes if c[0].endswith(namespace)] and not [s for s in structs if s[0].endswith(namespace)]:
+                    if namespace not in namespaces:
+                        namespaces.append(namespace)
+        else:
+            funcs.append(decl)
+
+    clsnamespaces = []
+    # process classes
+    for cl in classes:
+        name = cl[0][cl[0].find(' ')+1:]
+        if name.find('.') < 0 and not name.startswith("Cv"):
+            logerror(ERROR_004_MISSEDNAMESPACE, "class " + name + " from opencv_" + module + " is placed in global namespace but violates C-style naming convention")
+        clsnamespaces.append(name)
+        if do_python_crosscheck and not name.startswith("cv.") and name.startswith("Cv"):
+            clsnamespaces.append("cv." + name[2:])
+        if name.startswith("cv."):
+            name = name[3:]
+        name = name.replace(".", "::")
+        doc = rst.get(name)
+        if not doc:
+            #TODO: class is not documented
+            continue
+        doc[DOCUMENTED_MARKER] = True
+        # verify class marker
+        if not doc.get("isclass"):
+            logerror(ERROR_001_NOTACLASS, "class " + name + " is not marked as \"class\" in documentation", doc)
+        else:
+            # verify base
+            signature = doc.get("class", "")
+            signature = signature.replace(", public ", " ").replace(" public ", " ")
+            signature = signature.replace(", protected ", " ").replace(" protected ", " ")
+            signature = signature.replace(", private ", " ").replace(" private ", " ")
+            signature = ("class " + signature).strip()
+            hdrsignature = (cl[0] + " " +  cl[1]).replace("class cv.", "class ").replace(".", "::").strip()
+            if signature != hdrsignature:
+                logerror(ERROR_003_INCORRECTBASE, "invalid base class documentation\ndocumented: " + signature + "\nactual:     " + hdrsignature, doc)
+
+    # process structs
+    for st in structs:
+        name = st[0][st[0].find(' ')+1:]
+        if name.find('.') < 0 and not name.startswith("Cv"):
+            logerror(ERROR_004_MISSEDNAMESPACE, "struct " + name + " from opencv_" + module + " is placed in global namespace but violates C-style naming convention")
+        clsnamespaces.append(name)
+        if name.startswith("cv."):
+            name = name[3:]
+        name = name.replace(".", "::")
+        doc = rst.get(name)
+        if not doc:
+            #TODO: struct is not documented
+            continue
+        doc[DOCUMENTED_MARKER] = True
+        # verify struct marker
+        if not doc.get("isstruct"):
+            logerror(ERROR_002_NOTASTRUCT, "struct " + name + " is not marked as \"struct\" in documentation", doc)
+        else:
+            # verify base
+            signature = doc.get("class", "")
+            signature = signature.replace(", public ", " ").replace(" public ", " ")
+            signature = signature.replace(", protected ", " ").replace(" protected ", " ")
+            signature = signature.replace(", private ", " ").replace(" private ", " ")
+            signature = ("struct " + signature).strip()
+            hdrsignature = (st[0] + " " +  st[1]).replace("struct cv.", "struct ").replace(".", "::").strip()
+            if signature != hdrsignature:
+                logerror(ERROR_003_INCORRECTBASE, "invalid base struct documentation\ndocumented: " + signature + "\nactual:     " + hdrsignature, doc)
+
+    # process functions and methods
+    flookup = {}
+    for fn in funcs:
+        name = fn[0]
+        parent = None
+        namespace = None
+        for cl in clsnamespaces:
+            if name.startswith(cl + "."):
+                if cl.startswith(parent or ""):
+                    parent = cl
+        if parent:
+            name = name[len(parent) + 1:]
+            for nm in namespaces:
+                if parent.startswith(nm + "."):
+                    if nm.startswith(namespace or ""):
+                        namespace = nm
+            if namespace:
+                parent = parent[len(namespace) + 1:]
+        else:
+            for nm in namespaces:
+                if name.startswith(nm + "."):
+                    if nm.startswith(namespace or ""):
+                        namespace = nm
+            if namespace:
+                name = name[len(namespace) + 1:]
+        #print namespace, parent, name, fn[0]
+        if not namespace and not parent and not name.startswith("cv") and not name.startswith("CV_"):
+            logerror(ERROR_004_MISSEDNAMESPACE, "function " + name + " from opencv_" + module + " is placed in global namespace but violates C-style naming convention")
+        else:
+            fdescr = (namespace, parent, name, fn)
+            flookup_entry = flookup.get(fn[0], [])
+            flookup_entry.append(fdescr)
+            flookup[fn[0]] = flookup_entry
+
+    if do_python_crosscheck:
+        for name, doc in rst.iteritems():
+            decls = doc.get("decls")
+            if not decls:
+                continue
+            for signature in decls:
+                if signature[0] == "Python1":
+                    pname = signature[1][:signature[1].find('(')]
+                    try:
+                        fn = getattr(cv2.cv, pname[3:])
+                        docstr = "cv." + fn.__doc__
+                    except AttributeError:
+                        logerror(ERROR_005_MISSINGPYFUNC, "could not load documented function: cv2." + pname, doc)
+                        continue
+                    docstring = docstr
+                    sign = signature[1]
+                    signature.append(DOCUMENTED_MARKER)
+                    # convert old signature to pydoc style
+                    if docstring.endswith("*"):
+                        docstring = docstring[:-1]
+                    s = None
+                    while s != sign:
+                        s = sign
+                        sign = re.sub(r"^(.*\(.*)\(.*?\)(.*\) *->)", "\\1_\\2", sign)
+                    s = None
+                    while s != sign:
+                        s = sign
+                        sign = re.sub(r"\s*,\s*([^,]+)\s*=\s*[^,]+\s*(( \[.*\])?)\)", " [, \\1\\2])", sign)
+                    sign = re.sub(r"\(\s*([^,]+)\s*=\s*[^,]+\s*(( \[.*\])?)\)", "([\\1\\2])", sign)
+
+                    sign = re.sub(r"\)\s*->\s*", ") -> ", sign)
+                    sign = sign.replace("-> convexHull", "-> CvSeq")
+                    sign = sign.replace("-> lines", "-> CvSeq")
+                    sign = sign.replace("-> boundingRects", "-> CvSeq")
+                    sign = sign.replace("-> contours", "-> CvSeq")
+                    sign = sign.replace("-> retval", "-> int")
+                    sign = sign.replace("-> detectedObjects", "-> CvSeqOfCvAvgComp")
+
+                    def retvalRplace(match):
+                        m = match.group(1)
+                        m = m.replace("CvScalar", "scalar")
+                        m = m.replace("CvMemStorage", "memstorage")
+                        m = m.replace("ROIplImage", "image")
+                        m = m.replace("IplImage", "image")
+                        m = m.replace("ROCvMat", "mat")
+                        m = m.replace("CvMat", "mat")
+                        m = m.replace("double", "float")
+                        m = m.replace("CvSubdiv2DPoint", "point")
+                        m = m.replace("CvBox2D", "Box2D")
+                        m = m.replace("IplConvKernel", "kernel")
+                        m = m.replace("CvHistogram", "hist")
+                        m = m.replace("CvSize", "width,height")
+                        m = m.replace("cvmatnd", "matND")
+                        m = m.replace("CvSeqOfCvConvexityDefect", "convexityDefects")
+                        mm = m.split(',')
+                        if len(mm) > 1:
+                            return "(" + ", ".join(mm) + ")"
+                        else:
+                            return m
+
+                    docstring = re.sub(r"(?<=-> )(.*)$", retvalRplace, docstring)
+                    docstring = docstring.replace("( [, ", "([")
+
+                    if sign != docstring:
+                        logerror(ERROR_006_INVALIDPYOLDDOC, "old-style documentation differs from pydoc\npydoc: " + docstring + "\nfixup: " + sign + "\ncvdoc: " + signature[1], doc)
+                elif signature[0] == "Python2":
+                    pname = signature[1][4:signature[1].find('(')]
+                    cvname = "cv." + pname
+                    parent = None
+                    for cl in clsnamespaces:
+                        if cvname.startswith(cl + "."):
+                            if cl.startswith(parent or ""):
+                                parent = cl
+                    try:
+                        if parent:
+                            instance, clsname = get_cv2_object(parent)
+                            fn = getattr(instance, cvname[len(parent)+1:])
+                            docstr = fn.__doc__
+                            docprefix = "cv2." + clsname + "."
+                        else:
+                            fn = getattr(cv2, pname)
+                            docstr = fn.__doc__
+                            docprefix = "cv2."
+                    except AttributeError:
+                        if parent:
+                            logerror(ERROR_005_MISSINGPYFUNC, "could not load documented member of " + parent + " class: cv2." + pname, doc)
+                        else:
+                            logerror(ERROR_005_MISSINGPYFUNC, "could not load documented function cv2." + pname, doc)
+                        continue
+                    docstrings = [docprefix + s.replace("([, ", "([") for s in docstr.split("  or  ")]
+                    if not signature[1] in docstrings:
+                        pydocs = "\npydoc: ".join(docstrings)
+                        logerror(ERROR_007_INVALIDPYDOC, "documentation differs from pydoc\npydoc: " + pydocs + "\ncvdoc: " + signature[1], doc)
+                    else:
+                        signature.append(DOCUMENTED_MARKER)
+
+    #build dictionary for functions lookup
+    # verify C/C++ signatures
+    for name, doc in rst.iteritems():
+        decls = doc.get("decls")
+        if not decls:
+            continue
+        for signature in decls:
+            if signature[0] == "C" or signature[0] == "C++":
+                fd = flookup.get(signature[2][0])
+                if not fd:
+                    if signature[2][0].startswith("cv."):
+                        fd = flookup.get(signature[2][0][3:])
+                    if not fd:
+                        continue
+                    else:
+                        signature[2][0] = signature[2][0][3:]
+                if signature[0] == "C":
+                    ffd = [f for f in fd if not f[0] and not f[1]] # filter out C++ stuff
+                    if not ffd:
+                        if fd[0][1]:
+                            logerror(ERROR_008_CFUNCISNOTGLOBAL, "function " + fd[0][2] + " is documented as C function but is actually member of " + fd[0][1] + " class", doc)
+                        elif fd[0][0]:
+                            logerror(ERROR_008_CFUNCISNOTGLOBAL, "function " + fd[0][2] + " is documented as C function but is actually placed in " + fd[0][0] + " namespace", doc)
+                    fd = ffd
+                error = None
+                for f in fd:
+                    match, error = compareSignatures(signature[2], f[3])
+                    if match:
+                        signature.append(DOCUMENTED_MARKER)
+                        break
+                if signature[-1] != DOCUMENTED_MARKER:
+                    candidates = "\n\t".join([formatSignature(f[3]) for f in fd])
+                    logerror(ERROR_009_OVERLOADNOTFOUND, signature[0] + " function " + signature[2][0].replace(".","::") + " is documented but misses in headers (" + error + ").\nDocumented as:\n\t" + signature[1] + "\nCandidates are:\n\t" + candidates, doc)
+    #print hdrlist
+    #for d in decls:
+     #   print d
+    #print rstparser.definitions
+
+if __name__ == "__main__":
+    if len(sys.argv) < 2:
+        print "Usage:\n", os.path.basename(sys.argv[0]), " <module path>"
+        exit(0)
+
+    for module in sys.argv[1:]:
+        selfpath = os.path.dirname(os.path.abspath(sys.argv[0]))
+        module_path = os.path.join(selfpath, "..", "modules", module)
+
+        if not os.path.isdir(module_path):
+            print "Module \"" + module + "\" could not be found."
+            exit(1)
+
+        process_module(module, module_path)

modules/java/rst_parser.py

@@ -3,6 +3,7 @@ allmodules = ["core", "flann", "imgproc", "ml", "highgui", "video", "features2d"
 verbose = False
 show_warnings = True
 show_errors = True
+show_critical_errors = True
 
 params_blacklist = {
     "fromarray" : ("object", "allowND"), # python only function
@@ -66,7 +67,7 @@ class DeclarationParser(object):
         if line.startswith(".. ocv:jfunction::"):
             return "Java"
         return None
-    
+
     def hasDeclaration(self, line):
         return self.getLang(line) is not None
 
@@ -97,7 +98,7 @@ class ParamParser(object):
             self.comment += "\n" + line.lstrip()
         else:
             self.active = False
-            
+
     def hasDeclaration(self, line):
         return line.lstrip().startswith(":param")
 
@@ -115,7 +116,7 @@ class RstParser(object):
         doclist = glob.glob(os.path.join(module_path,"doc/*.rst"))
         for doc in doclist:
             self.parse_rst_file(module_name, doc)
-            
+
     def parse_section_safe(self, module_name, section_name, file_name, lineno, lines):
         try:
             self.parse_section(module_name, section_name, file_name, lineno, lines)
@@ -184,16 +185,16 @@ class RstParser(object):
                     continue
                 else:
                     skip_code_lines = False
-                    
+
             if ll.startswith(".. code-block::") or ll.startswith(".. image::"):
                 skip_code_lines = True
                 continue
-                
+
             # todo: parse structure members; skip them for now
             if ll.startswith(".. ocv:member::"):
                 skip_code_lines = True
                 continue
-                
+
             #ignore references (todo: collect them)
             if l.startswith(".. ["):
                 continue
@@ -204,7 +205,7 @@ class RstParser(object):
                 # turn on line-skipping mode for code fragments
                 skip_code_lines = True
                 ll = ll[:len(ll)-2]
-                
+
             # continue param parsing (process params after processing .. at the beginning of the line and :: at the end)
             if pdecl.active:
                 pdecl.append(l)
@@ -269,10 +270,10 @@ class RstParser(object):
             if skip_code_lines:
                 func["long"] = func.get("long", "") + "\n"
         # endfor l in lines
-        
+
         if fdecl.balance != 0:
-            if show_errors:
-                print >> sys.stderr, "RST parser error: invalid parentheses balance in \"%s\" File: %s (line %s)" % (section_name, file_name, lineno)
+            if show_critical_errors:
+                print >> sys.stderr, "RST parser error: invalid parentheses balance in \"%s\" File: %s:%s" % (section_name, file_name, lineno)
             return
 
         # save last parameter if needed
@@ -295,7 +296,7 @@ class RstParser(object):
         lineno = 0
         whitespace_warnings = 0
         max_whitespace_warnings = 10
-      
+
         lines = []
         flineno = 0
         fname = ""
@@ -310,7 +311,7 @@ class RstParser(object):
                 if whitespace_warnings <= max_whitespace_warnings and show_warnings:
                     print >> sys.stderr, "RST parser warning: tab symbol instead of space is used at file %s (line %s)" % (doc, lineno)
                 l = l.replace("\t", "    ")
-                
+
             # handle first line
             if prev_line == None:
                 prev_line = l.rstrip()
@@ -325,7 +326,7 @@ class RstParser(object):
                 flineno = lineno-1
                 fname = prev_line.strip()
             elif flineno > 0:
-                lines.append(ll)               
+                lines.append(ll)
             prev_line = ll
         df.close()
 
@@ -346,9 +347,9 @@ class RstParser(object):
         decls =  func.get("decls",[])
         if (decl.lang == "C++" or decl.lang == "C"):
             rst_decl = self.cpp_parser.parse_func_decl_no_wrap(decl.fdecl)
-            decls.append( (decl.lang, decl.fdecl, rst_decl) )
+            decls.append( [decl.lang, decl.fdecl, rst_decl] )
         else:
-            decls.append( (decl.lang, decl.fdecl) )
+            decls.append( [decl.lang, decl.fdecl] )
         func["decls"] = decls
 
     def add_new_pdecl(self, func, decl):
@@ -403,7 +404,7 @@ class RstParser(object):
     def validateParams(self, func):
         documentedParams = func.get("params",{}).keys()
         params = []
-       	
+
         for decl in func.get("decls", []):
             if len(decl) > 2:
                 args = decl[2][3] # decl[2] -> [ funcname, return_ctype, [modifiers], [args] ]
@@ -476,7 +477,7 @@ class RstParser(object):
         return func
 
     def fixOldCFunctionName(self, func):
-        if not "decls" in func: 
+        if not "decls" in func:
             return
         fname = None
         for decl in func["decls"]:
@@ -495,7 +496,7 @@ class RstParser(object):
             elif show_warnings:
                 print >> sys.stderr, "\"%s\" - section name is \"%s\" instead of \"%s\". File: %s (line %s)" % (fname, func["name"], fname[6:], func["file"], func["line"])
                 #self.print_info(func)
-                
+
     def normalizeText(self, s):
         if s is None:
             return s
@@ -503,7 +504,7 @@ class RstParser(object):
         s = re.sub(r"\.\. math::[ \r]*\n+((.|\n)*?)(\n[ \r]*\n|$)", mathReplace2, s)
         s = re.sub(r":math:`([^`]+?)`", mathReplace, s)
         s = re.sub(r" *:sup:", "^", s)
-        
+
         s = s.replace(":ocv:class:", "")
         s = s.replace(":ocv:struct:", "")
         s = s.replace(":ocv:func:", "")
@@ -526,7 +527,7 @@ class RstParser(object):
         s = re.sub(r"`([^`<]+ )<(https?://[^>]+)>`_", "\\1(\\2)", s)
         # remove tailing ::
         s = re.sub(r"::(\n|$)", "\\1", s)
-            
+
         # normalize line endings
         s = re.sub(r"\r\n", "\n", s)
         # remove extra line breaks before/after _ or ,
@@ -554,7 +555,7 @@ class RstParser(object):
         #s = re.sub(r"\.\. \[", "[", s)
         # unescape
         s = re.sub(r"\\(.)", "\\1", s)
-        
+
         # remove whitespace before .
         s = re.sub(r"[ ]+\.", ".", s)
         # remove tailing whitespace
@@ -582,7 +583,7 @@ class RstParser(object):
 
         s = s.strip()
         return s
-        
+
     def printSummary(self):
         print
         print "RST Parser Summary:"
@@ -623,7 +624,7 @@ def matrixReplace(match):
     m = match.group(2)
     m = re.sub(r" *& *", "   ", m)
     return m
-        
+
 def mathReplace(match):
     m = match.group(1)
 
@@ -645,7 +646,7 @@ def mathReplace(match):
     m = re.sub(r"\\begin{(?P<gtype>array|bmatrix)}(?:{[\|lcr\. ]+})? *(.*?)\\end{(?P=gtype)}", matrixReplace, m)
     m = re.sub(r"\\hdotsfor{(\d+)}", hdotsforReplace, m)
     m = re.sub(r"\\vecthreethree{(.*?)}{(.*?)}{(.*?)}{(.*?)}{(.*?)}{(.*?)}{(.*?)}{(.*?)}{(.*?)}", "<BR>|\\1 \\2 \\3|<BR>|\\4 \\5 \\6|<BR>|\\7 \\8 \\9|<BR>", m)
-    
+
     m = re.sub(r"\\left[ ]*\\lfloor[ ]*", "[", m)
     m = re.sub(r"[ ]*\\right[ ]*\\rfloor", "]", m)
     m = re.sub(r"\\left[ ]*\([ ]*", "(", m)
@@ -696,7 +697,7 @@ if __name__ == "__main__":
     if len(sys.argv) < 2:
         print "Usage:\n", os.path.basename(sys.argv[0]), " <module path>"
         exit(0)
-        
+
     if len(sys.argv) >= 3:
         if sys.argv[2].lower() == "verbose":
             verbose = True
@@ -714,7 +715,7 @@ if __name__ == "__main__":
         exit(1)
 
     parser = RstParser(hdr_parser.CppHeaderParser())
-    
+
     if module == "all":
         for m in allmodules:
             parser.parse(m, os.path.join(rst_parser_dir, "../" + m))

modules/python/src2/hdr_parser.py

@@ -242,7 +242,7 @@ class CppHeaderParser(object):
         bases = ll[2:]
         return classname, bases, modlist
 
-    def parse_func_decl_no_wrap(self, decl_str):
+    def parse_func_decl_no_wrap(self, decl_str, static_method = False):
         fdecl = decl_str.replace("CV_OUT", "").replace("CV_IN_OUT", "")
         fdecl = fdecl.strip().replace("\t", " ")
         while "  " in fdecl:
@@ -273,9 +273,16 @@ class CppHeaderParser(object):
 
         fname = "cv." + fname.replace("::", ".")
         decl = [fname, rettype, [], []]
+
+        # inline constructor implementation
+        implmatch = re.match(r"(\(.*?\))\s*:\s*(\w+\(.*?\),?\s*)+", fdecl[apos:])
+        if bool(implmatch):
+            fdecl = fdecl[:apos] + implmatch.group(1)
+
         args0str = fdecl[apos+1:fdecl.rfind(")")].strip()
 
-        if args0str != "":
+        if args0str != "" and args0str != "void":
+            args0str = re.sub(r"\([^)]*\)", lambda m: m.group(0).replace(',', "@comma@"), args0str)
             args0 = args0str.split(",")
 
             args = []
@@ -293,9 +300,19 @@ class CppHeaderParser(object):
                 defval = ""
                 if dfpos >= 0:
                     defval = arg[dfpos+1:].strip()
+                else:
+                    dfpos = arg.find("CV_DEFAULT")
+                    if dfpos >= 0:
+                        defval, pos3 = self.get_macro_arg(arg, dfpos)
+                    else:
+                        dfpos = arg.find("CV_WRAP_DEFAULT")
+                        if dfpos >= 0:
+                            defval, pos3 = self.get_macro_arg(arg, dfpos)
+                if dfpos >= 0:
+                    defval = defval.replace("@comma@", ",")
                     arg = arg[:dfpos].strip()
                 pos = len(arg)-1
-                while pos >= 0 and (arg[pos] == "_" or arg[pos].isalpha() or arg[pos].isdigit()):
+                while pos >= 0 and (arg[pos] in "_[]" or arg[pos].isalpha() or arg[pos].isdigit()):
                     pos -= 1
                 if pos >= 0:
                     aname = arg[pos+1:].strip()
@@ -308,6 +325,10 @@ class CppHeaderParser(object):
                     aname = "param"
                 decl[3].append([atype, aname, defval, []])
 
+        if static_method:
+            decl[2].append("/S")
+        if decl_str.endswith("const"):
+            decl[2].append("/C")
         return decl
 
     def parse_func_decl(self, decl_str):
@@ -328,7 +349,7 @@ class CppHeaderParser(object):
                 return []
 
         # ignore old API in the documentation check (for now)
-        if "CVAPI(" in decl_str:
+        if "CVAPI(" in decl_str and self.wrap_mode:
             return []
 
         top = self.block_stack[-1]
@@ -378,6 +399,10 @@ class CppHeaderParser(object):
                 sys.exit(-1)
             decl_start = decl_str[:args_begin].strip()
 
+        # constructor/destructor case
+        if bool(re.match(r'(\w+::)*(?P<x>\w+)::~?(?P=x)', decl_start)):
+            decl_start = "void " + decl_start
+
         rettype, funcname, modlist, argno = self.parse_arg(decl_start, -1)
 
         if argno >= 0:
@@ -385,8 +410,16 @@ class CppHeaderParser(object):
             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 bool(re.match('\w+\s+\(\*\w+\)\s*\(.*\)', decl_str)):
+                    return [] # function typedef
+                elif bool(re.match('[A-Z_]+', decl_start)):
+                    return [] # it seems to be a macro instantiation
+                elif "__declspec" == decl_start:
+                    return []
+                else:
+                    #print rettype, funcname, modlist, argno
+                    print "Error at %d in %s. the function/method name is missing: '%s'" % (self.lineno, self.hname, decl_start)
+                    sys.exit(-1)
 
         if self.wrap_mode and (("::" in funcname) or funcname.startswith("~")):
             # if there is :: in function name (and this is in the header file),
@@ -399,7 +432,7 @@ class CppHeaderParser(object):
         funcname = self.get_dotted_name(funcname)
 
         if not self.wrap_mode:
-            decl = self.parse_func_decl_no_wrap(decl_str)
+            decl = self.parse_func_decl_no_wrap(decl_str, static_method)
             decl[0] = funcname
             return decl
 
@@ -515,7 +548,7 @@ class CppHeaderParser(object):
                 sys.exit(-1)
             if block_name:
                 n += block_name + "."
-        return n + name
+        return n + name.replace("::", ".")
 
     def parse_stmt(self, stmt, end_token):
         """
@@ -559,7 +592,7 @@ class CppHeaderParser(object):
                 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) or (not self.wrap_mode and ("CV_EXPORTS" in stmt)):
+                if ("CV_EXPORTS_W" in stmt) or ("CV_EXPORTS_AS" in stmt) or (not self.wrap_mode):# and ("CV_EXPORTS" in stmt)):
                     decl = [stmt_type + " " + self.get_dotted_name(classname), "", modlist, []]
                     if bases:
                         decl[1] = ": " + " ".join(bases)
@@ -570,6 +603,8 @@ class CppHeaderParser(object):
 
             if stmt.startswith("namespace"):
                 stmt_list = stmt.split()
+                if len(stmt_list) < 2:
+                    stmt_list.append("<unnamed>")
                 return stmt_list[0], stmt_list[1], True, None
             if stmt.startswith("extern") and "\"C\"" in stmt:
                 return "namespace", "", True, None
@@ -633,6 +668,7 @@ class CppHeaderParser(object):
         The main method. Parses the input file.
         Returns the list of declarations (that can be print using print_decls)
         """
+        self.hname = hname
         decls = []
         f = open(hname, "rt")
         linelist = list(f.readlines())