Exemplo n.º 1
0
 def run(self):
     self.signalConsole.emit(f"Starting map of {self.url}...")
     self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                            self.configuration, False)
     self.getXPaths(self.url, self.filename)
     self.signalConsole.emit(f"Map complete.")
     self.signalFinished.emit(True)
Exemplo n.º 2
0
 def scan(self, filename):
     self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                            self.configuration, False)
     paths = self.subSystem.loadXPaths(filename)
     #print(paths)
     scans = self.subSystemDatabase.getScanTable()
     #print(scans)
     id = 0
     if len(scans) >= 1:
         for x in scans:
             id += 1
     #table = "site_" + str(id)
     #print(table)
     time = datetime.now()
     dt = time.strftime("%d/%m/%Y %H:%M")
     self.subSystemDatabase.addScan(id, ".com", "Chrome", dt)
     #self.subSystemDatabase.createSiteTable(table)
     self.subSystem.openSite(self.url)
     try:
         count = int(self.subSystemDatabase.getMaxElemID()[0][0]) + 1
     except:
         count = 0
     #print(count)
     for item in paths:
         #print(item)
         ats = self.subSystem.getAllAttributes(self.url, item)
         if ats['type'] != "":
             self.subSystemDatabase.insertSiteTable(
                 id, count, ats['type'], self.url, item, ats['nonText'],
                 ats['decorative'], ats['class'], ats['name'], ats['id'],
                 ats['alt'], ats['aria-label'], ats['src'])
             self.subSystemDatabase.conn.commit()
             count += 1
Exemplo n.º 3
0
    def operationInit(self) -> bool:
        trigger = True
        try:
            #print(self.driver)
            self.subSytemScrap = Scrap(self.url, self.navdepth, self.driver, self.targetDevice, self.configuration, self.screenshots)
            
            for linkData,\
                linkError,\
                imageData,\
                imageError,\
                urlError,\
                invalid,\
                external,\
                duplicate in self.subSytemScrap.scrapSite():

                a = 0
                #print("************************************************************************************" + str(len(imageData)))
                
                for item in imageData:
                    if trigger:
                        self.signalConsole.emit("Saving data into databases...")
                        trigger = False
                    a = item
                    imageDatas = asdict(a)
                    '''if imageDatas['image'] == None:
                        imageDatas['image'] = "Blank"'''
                    self.subSytemDatabase.insertImageTableData(self.imgTable, self.count , imageDatas['url'], imageDatas['id'], imageDatas['name'], imageDatas['src'], imageDatas['text'], imageDatas['altText'], 
                                                    imageDatas['width'], imageDatas['height'], imageDatas['x'], imageDatas['y'], str(imageDatas['displayed']), imageDatas['download'], imageDatas['image'])
                    self.count+=1
             

                self.signalData.emit(
                    {
                        "linkData":len(linkData),
                        "linkError": linkError,
                        "imageData": len(imageData),
                        "imageError": imageError,
                        "urlError": urlError,
                        "invalid": invalid,
                        "external": external,
                        "duplicate": duplicate
                    }
                )
            self.signalConsole.emit("Done Saving...")
        except Exception as e:
            logger.error(f"{e}")
            logger.error(e, exc_info=True)
            return False
        else:
            return True
Exemplo n.º 4
0
 def testing(self, siteID, url):
     #self.subsystem.driver.get("https://www.jackboxgames.com/party-pack-five/")
     self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                            self.configuration, False)
     data = self.subSystemDatabase.getSiteData('elements', siteID)
     #print(data)
     #print(f"Data: {data}")
     back = self.subsystem.nonText(self.url, data)
     #print(f"Back: {back}")
     for item in back:
         #print(item["xPath"])
         self.subSystemDatabase.addNotesToSite('elements', "wcag11",
                                               item["xPath"],
                                               str(item["errors"]), siteID)
Exemplo n.º 5
0
def sanityCheck(args):

    import requests
    from urllib.parse import urlparse

    if not Scrap.checkUrlValid(args.url):
        raise requests.ConnectionError(f"invalid url {args.url}")

    if args.navdepth < -1:
        raise OverflowError("depth range should be [-1, ... ).")

    main(args=args)
Exemplo n.º 6
0
    def _checkValidAddr(self, a0: str) -> bool:

        if not a0:
            self.ui.btn_addrPrefix.setStyleSheet(
                "background-color: white; color: black")
            return False
        else:
            if not Scrap.checkUrlValid(a0):
                logger.error(f"{a0} not valid url")
                self.ui.btn_addrPrefix.setStyleSheet(
                    "background-color: red; color: black")
                return False

            else:
                self.ui.btn_addrPrefix.setStyleSheet(
                    "background-color: green; color: black")
                return True
Exemplo n.º 7
0
class siteMap(QThread):
    signalConsole = pyqtSignal(str)
    signalFinished = pyqtSignal(bool)

    def __init__(self,
                 url: str,
                 filename: str,
                 driver: str,
                 targetDevice: Union[str, None],
                 configuration: Optional[Dict] = None) -> None:

        super().__init__()
        self.targetDevice = targetDevice
        self.url = url
        self.configuration = configuration
        self.driver = driver
        self.filename = filename
        self.subSystemDatabase = WACDatabase("sqlite/database.db")

    def _setupScrap(self) -> bool:

        self.setupDriver()

        if not self.isDriverInitialized():
            print("driver not initialized!")
            return False

        return True

    def __getattribute__(self, name, *args, **kwargs):
        def make_interceptor(callble):
            def func(*args, **kwargs):
                logger.info(f"{name} {args} {kwargs}")
                return callble(*args, **kwargs)

            return func

        returned = object.__getattribute__(self, name)
        if inspect.isfunction(returned) or inspect.ismethod(returned):
            return make_interceptor(returned)
        return returned

    def run(self):
        self.signalConsole.emit(f"Starting map of {self.url}...")
        self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        self.getXPaths(self.url, self.filename)
        self.signalConsole.emit(f"Map complete.")
        self.signalFinished.emit(True)

    def getXPaths(self, url, filename):

        if not self.subSystem._setupScrap():
            logger.critical(f"driver setup failed")

        self.subSystem.driver.get(url)
        #self.subSystem.driver.mana   manage().timeouts().implicitlyWait(10)
        try:
            WebDriverWait(self.subSystem.driver, 10).until(
                EC.presence_of_all_elements_located((By.CSS_SELECTOR, '*')))
        except:
            print("webwait failed")

        #time.sleep(5)
        '''EC.element_to_be_clickable((By.CLASS_NAME, 'playButton'))
        self.subSystem.driver.find_element(By.CLASS_NAME,"ule-onboarding-continue-button").click()
        self.subSystem.driver.find_element(By.CLASS_NAME,"playButton").click()'''
        elements = self.subSystem.driver.find_elements(By.TAG_NAME, 'html')
        tags = []
        self.subSystem.driver.implicitly_wait(0.00001)
        self.subSystem.paths = []
        #self.subSystem.index = 0
        self.subSystem.paths.append('/html')
        self.trigger = True
        for x in elements:
            #print(x.get_attribute("innerHTML"))

            subElems = x.find_elements_by_xpath(".//*")
            self.elemSize = len(subElems)
            print(f'Final Total: {self.xPath(subElems,"/html", 1)}')

        allpaths = ""
        for x in self.subSystem.paths:
            #print(x)
            #x = x.replace('[1]', '')
            #print(x)
            allpaths += f"\n{x}"

        file = open(filename + ".txt", 'w')
        file.write(allpaths)
        file.close()

        return self.subSystem.paths

    def xPath(self, elems, path, size):
        index = 0

        if self.trigger:
            #self.signalConsole.emit(f"Elements to map: {len(elems)}")
            self.trigger = False
        while len(elems) >= index + 1:
            if len(self.subSystem.paths) % (int(
                    self.elemSize / 20)) == 0 and index != 0:
                #print(f"Completed {index} of {len(elems)}" )
                perc = int((len(self.subSystem.paths) / self.elemSize) * 100)
                self.signalConsole.emit(
                    f"Completed {len(self.subSystem.paths)} elements.  {perc}%"
                )

            i = 1
            while True:
                if not (path + '/' + elems[index].tag_name +
                        f'[{i}]') in self.subSystem.paths:
                    new = path + '/' + elems[index].tag_name + f'[{i}]'

                    if elems[index].tag_name == 'svg':
                        clas = elems[index].get_attribute('class')
                        new = f'svg {clas}'  #f'//svg[@class="{clas}"][@viewbox="{viewbox}"]'
                        #print(str(at))

                    if 'path' not in new:
                        self.subSystem.paths.append(new)
                    #print(path + '/' + elems[index].tag_name + f'[{i}]')
                    #print(path + '/' + elems[index].tag_name + f'[{i}]                      Index: {index}')
                    break
                else:
                    i += 1
            try:
                subElems = elems[index].find_elements_by_xpath(".//*")
            except:
                subElems = []
            subSize = self.xPath(subElems,
                                 path + '/' + elems[index].tag_name + f'[{i}]',
                                 size + 1)
            index += subSize + 1

        return index
Exemplo n.º 8
0
    def run(self):
        self.signalConsole.emit(f"Starting scan of {self.url}")
        self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)

        paths = self.subSystem.loadXPaths(self.filename)

        if paths == False:
            self.signalConsole.emit(f"Unable to find path file.")
        else:
            scans = self.subSystemDatabase.getScanTable()

            id = 0
            if len(scans) >= 1:
                for x in scans:
                    #print(f"x: {x}")
                    if x[0] > id:
                        id = x[0]
            id += 1

            time = datetime.now()
            dt = time.strftime("%d/%m/%Y %H:%M")
            self.subSystemDatabase.addScan(id, ".com", "Chrome", dt)
            self.signalConsole.emit(f"New scanID created: {id}")

            self.subSystem.openSite(self.url)
            self.signalConsole.emit(f"Chrome driver started...")
            try:
                count = int(self.subSystemDatabase.getMaxElemID()[0][0]) + 1
            except:
                count = 0
            #print(f'count: {count}')
            total = len(paths)
            completed = 0
            #self.subSystemDatabase.conn.cursor.fast_executemany = True
            elements = []
            self.scanSignal.emit(str(id))
            for item in paths[1:]:
                #print(completed)
                if not ((('header' in item) or
                         ('svg' in item)) and self.headers):
                    #time.sleep(1)
                    #print('inside')
                    ats = self.subSystem.getAllAttributes(self.url, item)
                    #print('mid')
                    if ats['type'] != "":

                        #print(ats)

                        #self.subSystemDatabase.insertSiteTable(id, count, ats['type'], self.url, item, ats['alt'], ats['aria-current'], ats['aria-describedby'], ats['aria-hidden'],ats['aria-label'],ats['aria-required'],
                        #                                        ats['class'], ats['id'], ats['name'], ats['role'], ats['src'], ats['tabindex'], ats['target'],ats['title'], ats['href'], ats['text'], ats['type id'])

                        elements.append(
                            (id, count, ats['type'], self.url, item,
                             ats['alt'], ats['aria-current'],
                             ats['aria-describedby'], ats['aria-hidden'],
                             ats['aria-label'], ats['aria-required'],
                             ats['class'], ats['id'], ats['name'], ats['role'],
                             ats['src'], ats['tabindex'], ats['target'],
                             ats['title'], ats['href'], ats['text'],
                             ats['type id'], ats['scope'], ats['imgText'],
                             ats['imgTable'],
                             ats['imgIcon'], ats['srcset'], ats['lang'],
                             str('html'), ats['fontSize'], ats['data-src']))
                        count += 1
                    #print('next')
                    #print(elements)
                completed += 1
                if completed == 1:
                    self.signalConsole.emit(
                        f"First Element scanned of {total} elements.")
                if completed % (int(len(paths) / 20)) == 0:
                    perc = int((completed / len(paths)) * 100)
                    self.signalConsole.emit(
                        f'Completed scan of {completed}/{total} elements. {perc}%'
                    )
                    #print('before')
                    self.subSystemDatabase.insertLargeSiteTable(elements)
                    self.subSystemDatabase.conn.commit()
                    elements = []
                    #print('after')
                #print('done')

            ret1, ret2 = self.checkTabbing()
            print('got back')
            if len(ret1) > 0:
                err = "Tab progression respects a logical sequence. Error found in tab numbers: "
                for item in ret1:
                    err = err + f'{item},'
                self.subSystemDatabase.addNotesToSite("elements", "Rule 1",
                                                      '/html', err, id)
            if len(ret2) > 0:
                err = "Each tab should have h2 element even if blank. These did not "
                for item in ret2:
                    err = err + f'{item},'
                self.subSystemDatabase.addNotesToSite("elements", "Rule 2",
                                                      '/html', err, id)
            #self.subSystemDatabase.conn.cursor.fast_executemany = False
            self.signalConsole.emit(f"Scan Complete.")
            self.signalFinished.emit(True)
Exemplo n.º 9
0
 def checkTabbing(self):
     self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                            self.configuration, False)
     return self.subsystem.checkTabbing()
Exemplo n.º 10
0
class wcagScan(QThread):
    signalConsole = pyqtSignal(str)
    scanSignal = pyqtSignal(str)
    signalFinished = pyqtSignal(bool)

    def __init__(self,
                 url: str,
                 filename: str,
                 driver: str,
                 headers: bool,
                 targetDevice: Union[str, None],
                 configuration: Optional[Dict] = None) -> None:

        super().__init__()
        self.targetDevice = targetDevice
        self.headers = headers
        self.url = url
        self.filename = filename
        self.configuration = configuration
        self.driver = driver
        self.subSystemDatabase = WACDatabase("sqlite/database.db")
        self.numxpath = 4
        print(self.targetDevice)

    def _setupScrap(self) -> bool:

        self.setupDriver()

        if not self.isDriverInitialized():
            print("driver not initialized!")
            return False

        return True

    def __getattribute__(self, name, *args, **kwargs):
        def make_interceptor(callble):
            def func(*args, **kwargs):
                logger.info(f"{name} {args} {kwargs}")
                return callble(*args, **kwargs)

            return func

        returned = object.__getattribute__(self, name)
        if inspect.isfunction(returned) or inspect.ismethod(returned):
            return make_interceptor(returned)
        return returned

    def checkTabbing(self):
        self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        return self.subsystem.checkTabbing()

    def run(self):
        self.signalConsole.emit(f"Starting scan of {self.url}")
        self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)

        paths = self.subSystem.loadXPaths(self.filename)

        if paths == False:
            self.signalConsole.emit(f"Unable to find path file.")
        else:
            scans = self.subSystemDatabase.getScanTable()

            id = 0
            if len(scans) >= 1:
                for x in scans:
                    #print(f"x: {x}")
                    if x[0] > id:
                        id = x[0]
            id += 1

            time = datetime.now()
            dt = time.strftime("%d/%m/%Y %H:%M")
            self.subSystemDatabase.addScan(id, ".com", "Chrome", dt)
            self.signalConsole.emit(f"New scanID created: {id}")

            self.subSystem.openSite(self.url)
            self.signalConsole.emit(f"Chrome driver started...")
            try:
                count = int(self.subSystemDatabase.getMaxElemID()[0][0]) + 1
            except:
                count = 0
            #print(f'count: {count}')
            total = len(paths)
            completed = 0
            #self.subSystemDatabase.conn.cursor.fast_executemany = True
            elements = []
            self.scanSignal.emit(str(id))
            for item in paths[1:]:
                #print(completed)
                if not ((('header' in item) or
                         ('svg' in item)) and self.headers):
                    #time.sleep(1)
                    #print('inside')
                    ats = self.subSystem.getAllAttributes(self.url, item)
                    #print('mid')
                    if ats['type'] != "":

                        #print(ats)

                        #self.subSystemDatabase.insertSiteTable(id, count, ats['type'], self.url, item, ats['alt'], ats['aria-current'], ats['aria-describedby'], ats['aria-hidden'],ats['aria-label'],ats['aria-required'],
                        #                                        ats['class'], ats['id'], ats['name'], ats['role'], ats['src'], ats['tabindex'], ats['target'],ats['title'], ats['href'], ats['text'], ats['type id'])

                        elements.append(
                            (id, count, ats['type'], self.url, item,
                             ats['alt'], ats['aria-current'],
                             ats['aria-describedby'], ats['aria-hidden'],
                             ats['aria-label'], ats['aria-required'],
                             ats['class'], ats['id'], ats['name'], ats['role'],
                             ats['src'], ats['tabindex'], ats['target'],
                             ats['title'], ats['href'], ats['text'],
                             ats['type id'], ats['scope'], ats['imgText'],
                             ats['imgTable'],
                             ats['imgIcon'], ats['srcset'], ats['lang'],
                             str('html'), ats['fontSize'], ats['data-src']))
                        count += 1
                    #print('next')
                    #print(elements)
                completed += 1
                if completed == 1:
                    self.signalConsole.emit(
                        f"First Element scanned of {total} elements.")
                if completed % (int(len(paths) / 20)) == 0:
                    perc = int((completed / len(paths)) * 100)
                    self.signalConsole.emit(
                        f'Completed scan of {completed}/{total} elements. {perc}%'
                    )
                    #print('before')
                    self.subSystemDatabase.insertLargeSiteTable(elements)
                    self.subSystemDatabase.conn.commit()
                    elements = []
                    #print('after')
                #print('done')

            ret1, ret2 = self.checkTabbing()
            print('got back')
            if len(ret1) > 0:
                err = "Tab progression respects a logical sequence. Error found in tab numbers: "
                for item in ret1:
                    err = err + f'{item},'
                self.subSystemDatabase.addNotesToSite("elements", "Rule 1",
                                                      '/html', err, id)
            if len(ret2) > 0:
                err = "Each tab should have h2 element even if blank. These did not "
                for item in ret2:
                    err = err + f'{item},'
                self.subSystemDatabase.addNotesToSite("elements", "Rule 2",
                                                      '/html', err, id)
            #self.subSystemDatabase.conn.cursor.fast_executemany = False
            self.signalConsole.emit(f"Scan Complete.")
            self.signalFinished.emit(True)

    def mapSite(self, filename):
        self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        self.subSystem.getXPaths(self.url, filename)

    def openElem(self, xPath):
        self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        self.signalConsole.emit("Opening Element")
        self.subsystem.openSite(self.url)
        self.subsystem.openWebElement(self.url, str(xPath))

    def output(self, scanID, filename):
        errors = []
        print(scanID)
        #self.subSystemDatabase2 = WACDatabase("sqlite/database.db")
        data = self.subSystemDatabase.getSiteErrors('elements', scanID)
        #print(data[0:200])
        for item in data:
            #print(item[self.numxpath])
            add = False
            ers = []

            ers.append(item[3])
            ers.append(item[4])
            ers.append(item[2])

            for i in range(60):
                #print(f'item: {item[i + 31]} i = {i}')
                if item[i + 30] != None:

                    if len(item[i + 30]) > 2:
                        ers.append(f"Rule {i}, {item[i + 30]}")
                        print(item[i + 30])
                        add = True
            if add:
                print(f"{item[0:4]} +    {ers}")
                errors.append(ers)

        print(f"errors: {errors[0:10]}")
        self.ws = Xlsx(filename + '_errors.xlsx', "Sheet1")
        for x in errors:
            self.ws.addRows(x)

        self.ws.set_col_widths_errors(100)
        self.ws.saveWorkbook()
Exemplo n.º 11
0
 def openElem(self, xPath):
     self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                            self.configuration, False)
     self.signalConsole.emit("Opening Element")
     self.subsystem.openSite(self.url)
     self.subsystem.openWebElement(self.url, str(xPath))
Exemplo n.º 12
0
 def mapSite(self, filename):
     self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                            self.configuration, False)
     self.subSystem.getXPaths(self.url, filename)
Exemplo n.º 13
0
class OperationScrapSite(QThread):

    signalConsole = pyqtSignal(str)
    signalData = pyqtSignal(dict)
    signalTeardown = pyqtSignal(bool)

    def __init__(
        self,
        db: str,
        url: str,
        navdepth: int,
        driver: str,
        targetDevice: str,
        screenshots: bool,
        configuration: Optional[Dict] = None,
    ) -> None:

        super().__init__()

        
        self.count = 0
        self.url = url
        self.navdepth = navdepth
        self.driver = driver
        self.targetDevice = targetDevice
        self.configuration = configuration
        self.db = db
        self.screenshots = screenshots
        

        #
        self.emid: Optional[int] = None
        self.ssid: Optional[int] = None
        self.urlTable: str = ""
        self.imgTable: str = ""
        self.btnTable: str = ""

        #
        self.countUrlTotal:int = 0
        self.countUrlDomain:int = 0
        self.countUrlExternal:int = 0
        self.countUrlInvalid:int = 0
        self.countUrlErrors:int = 0
        self.countImagesTotal:int = 0
        self.countImagesScrapped:int = 0
        self.countImagesDownloaded:int = 0
        self.countImagesScreenshot:int = 0
        self.countImagesErrors:int = 0

        
        self.subSytemDatabase = WACDatabase(db)
        



    def __getattribute__(self, name, *args, **kwargs):

        def make_interceptor(callble):
            def func(*args, **kwargs):
                logger.info(f"{name} {args} {kwargs}")
                return callble(*args, **kwargs)
            return func

        returned = object.__getattribute__(self, name)

        if inspect.isfunction(returned) or inspect.ismethod(returned):

            return make_interceptor(returned)

        return returned

    '''def updateData(self,data:dict):
        self.signalData.emit(data)'''

    def run(self):

        ret = self.operationWarmup()

        if ret:
            try:
                self.subSytemDatabase.conn.commit()
            except Exception as e:
                logger.error(f"{e}")
                logger.error(e, exc_info=True)
                return
            else:
                self.signalConsole.emit(f"warmup complete, scrapping now...")

        else:
            self.signalConsole.emit(f"error: warmup failed, not scrapping")
            return

        ret = self.operationInit()

        if ret:
            self.signalConsole.emit(f"scrap completed, performing post process functions")
        else:
            self.signalConsole.emit(f"error: scrap process failed!")
            return

        self.subSytemScrap.driver.close()


    def operationWarmup(self) -> bool:

        # hash key to match previous configuration entry
        hKey = createHashMD5(
            [self.url, self.driver, self.targetDevice, self.navdepth]
        )

        # get id from entry
        self.emid = self.subSytemDatabase.selectEMIDInEntry(self.url)
        flagScrapSetting = True
        print("EMID = " + str(self.emid))

        # if id in entry present, search for entries in scrap configs
        if self.emid:
            self.signalConsole.emit(f"previous entry found for {self.url}")
            ret = self.subSytemDatabase.selectAllScrapSetting(EMID=self.emid)

            # if configurations present , check for hashkey
            if ret:
                self.signalConsole.emit(f"{len(ret)} scrap entries found for {self.url}")

                for ret_ in ret:
                    r_ = nt_ScrapSetting._make(ret_)

                    if r_.HASHKEY == hKey:
                        self.signalConsole.emit(f"scrap found for current config HASHKEY:{hKey} on date:{r_.date}")

                        # if a hashkey present for same date skip
                        if r_.date == str(datetime.datetime.now().date()):
                            self.signalConsole.emit(f"choose different scrap configuration (for same date)")
                            return False
        

        # if entry not present , create new
        elif not self.emid:
            try:
                self.emid = self.subSytemDatabase.getMaxId("Entry") + 1
            except:
                self.emid = 1
            self.signalConsole.emit(f"creating new entry for {self.url}")
            flag = self.subSytemDatabase.insertInEntry(self.url, status="processing", desc="None")

            if not flag:
                self.signalConsole.emit(f"error: could not create new entry for {self.url}")
                return False

        # get id of new entry
         
        # if configuration not present create new
        if flagScrapSetting:
            self.signalConsole.emit(f"creating new scrap configuration " + str(self.emid))
            flag = self.subSytemDatabase.insertInScrapSetting(
                self.emid, hKey, self.driver, self.targetDevice, self.navdepth  #type:ignore
            )

            if not flag:
                self.signalConsole.emit(f"error: could not create new scrap configuration")
                return False

            else:
                self.signalConsole.emit(f"proceeding to create tables")
        else:
            print("not flagged")
        # get ssid for new configuration entry
        self.ssid = self.subSytemDatabase.getMaxId("ScrapSetting")

        # create table for elements
        self.signalConsole.emit(f"creating element tables entries")
        etrid = self.subSytemDatabase.getMaxId("ElementTableRef")

        if isinstance(etrid, int):
            if etrid == 0: etrid = 1
            else: etrid+=1

        else:
            self.signalConsole.emit(f"error: fetching element table id")
            return False

        # table names
        self.imgTable = prefix_EleImgTable+str(etrid)
        self.urlTable = prefix_EleUrlTable+str(etrid)
        self.btnTable = prefix_EleBtnTable+str(etrid)

        flag = self.subSytemDatabase.insertInElementTableRef(
            self.ssid,  #type:ignore
            self.urlTable, self.imgTable, self.btnTable
        )

        if not flag:
            self.signalConsole.emit(f"error: could not create table entries")
            return False

        else:
            self.signalConsole.emit(f"creating new element tables")
            
            if not self.subSytemDatabase.creatTableImg(self.imgTable):
                self.signalConsole.emit(f"error: cant creat new element table {self.imgTable}")
                return False

            if not self.subSytemDatabase.creatTableUrl(self.urlTable):
                self.signalConsole.emit(f"error: cant creat new element table {self.urlTable}")
                return False

        return True


    def operationInit(self) -> bool:
        trigger = True
        try:
            #print(self.driver)
            self.subSytemScrap = Scrap(self.url, self.navdepth, self.driver, self.targetDevice, self.configuration, self.screenshots)
            
            for linkData,\
                linkError,\
                imageData,\
                imageError,\
                urlError,\
                invalid,\
                external,\
                duplicate in self.subSytemScrap.scrapSite():

                a = 0
                #print("************************************************************************************" + str(len(imageData)))
                
                for item in imageData:
                    if trigger:
                        self.signalConsole.emit("Saving data into databases...")
                        trigger = False
                    a = item
                    imageDatas = asdict(a)
                    '''if imageDatas['image'] == None:
                        imageDatas['image'] = "Blank"'''
                    self.subSytemDatabase.insertImageTableData(self.imgTable, self.count , imageDatas['url'], imageDatas['id'], imageDatas['name'], imageDatas['src'], imageDatas['text'], imageDatas['altText'], 
                                                    imageDatas['width'], imageDatas['height'], imageDatas['x'], imageDatas['y'], str(imageDatas['displayed']), imageDatas['download'], imageDatas['image'])
                    self.count+=1
             

                self.signalData.emit(
                    {
                        "linkData":len(linkData),
                        "linkError": linkError,
                        "imageData": len(imageData),
                        "imageError": imageError,
                        "urlError": urlError,
                        "invalid": invalid,
                        "external": external,
                        "duplicate": duplicate
                    }
                )
            self.signalConsole.emit("Done Saving...")
        except Exception as e:
            logger.error(f"{e}")
            logger.error(e, exc_info=True)
            return False
        else:
            return True
Exemplo n.º 14
0
class wcag11(QThread):
    signalConsole = pyqtSignal(str)

    def __init__(self,
                 url: str,
                 driver: str,
                 targetDevice: Union[str, None],
                 configuration: Optional[Dict] = None) -> None:

        super().__init__()
        self.targetDevice = targetDevice
        self.url = url
        self.configuration = configuration
        self.driver = driver
        self.subSystemDatabase = WACDatabase("sqlite/database.db")

    def _setupScrap(self) -> bool:

        self.setupDriver()

        if not self.isDriverInitialized():
            print("driver not initialized!")
            return False

        return True

    #def run

    def __getattribute__(self, name, *args, **kwargs):
        def make_interceptor(callble):
            def func(*args, **kwargs):
                logger.info(f"{name} {args} {kwargs}")
                return callble(*args, **kwargs)

            return func

        returned = object.__getattribute__(self, name)
        if inspect.isfunction(returned) or inspect.ismethod(returned):
            return make_interceptor(returned)
        return returned

    def start(self):
        self.subSystemDatabase.createSiteTable("Test")

    def testing(self, siteID, url):
        #self.subsystem.driver.get("https://www.jackboxgames.com/party-pack-five/")
        self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        data = self.subSystemDatabase.getSiteData('elements', siteID)
        #print(data)
        #print(f"Data: {data}")
        back = self.subsystem.nonText(self.url, data)
        #print(f"Back: {back}")
        for item in back:
            #print(item["xPath"])
            self.subSystemDatabase.addNotesToSite('elements', "wcag11",
                                                  item["xPath"],
                                                  str(item["errors"]), siteID)

    def scan(self, filename):
        self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        paths = self.subSystem.loadXPaths(filename)
        #print(paths)
        scans = self.subSystemDatabase.getScanTable()
        #print(scans)
        id = 0
        if len(scans) >= 1:
            for x in scans:
                id += 1
        #table = "site_" + str(id)
        #print(table)
        time = datetime.now()
        dt = time.strftime("%d/%m/%Y %H:%M")
        self.subSystemDatabase.addScan(id, ".com", "Chrome", dt)
        #self.subSystemDatabase.createSiteTable(table)
        self.subSystem.openSite(self.url)
        try:
            count = int(self.subSystemDatabase.getMaxElemID()[0][0]) + 1
        except:
            count = 0
        #print(count)
        for item in paths:
            #print(item)
            ats = self.subSystem.getAllAttributes(self.url, item)
            if ats['type'] != "":
                self.subSystemDatabase.insertSiteTable(
                    id, count, ats['type'], self.url, item, ats['nonText'],
                    ats['decorative'], ats['class'], ats['name'], ats['id'],
                    ats['alt'], ats['aria-label'], ats['src'])
                self.subSystemDatabase.conn.commit()
                count += 1

    def nonText(self, url, data):
        altNum = 10
        nameNum = 8
        idNum = 9
        ariaNum = 11
        classNum = 7

        nonTextElems = ['img', 'button', 'input', 'video']

        elements = []
        #print(data)
        for item in data:

            #print(item)

            #if nonText
            if item[5] == "True":
                xPath = item[4]  #.replace('[1]', '')
                #elem = self.driver.find_element_by_xpath(xPath)
                #elements.append(self.driver.find_element_by_xpath(xPath))
                elements.append({
                    'type': item[2],
                    'attributes': item,
                    'element': None,
                    'errors': [],
                    'xPath': xPath
                })
                #print(elements)
            #except:
            #print(f'Failed Path: {item}')

        #print(elements)
        for item in elements:
            #print(item)

            # Get name and alt text
            try:
                alt = item['attributes'][altNum]
                #print(f"alt: {alt}")
            except:
                item['errors'].append(
                    "WCAG 1.1.1 - All non-text content should have an alt text."
                )
                alt = "Broke"
            #print(f"alt: {alt}")
            try:
                name = item['attributes'][nameNum]
                #print(f"name: {name}")
            except:
                item['errors'].append(
                    "Non-text content should have a name. Controls Must have a name."
                )
                name = ""
            #item['errors'].append(f"Id: {item['attributes'][idNum]}")
            try:
                if item['type'] == "input" or item['type'] == "button":
                    aria = item['attributes'][ariaNum]
                    if not aria == "":
                        alt = aria
                    else:
                        item['errors'].append(
                            "WCAG 1.1.1 - Control did not have an aria label to replace alt."
                        )
                elif "captcha" in item['attributes'][classNum]:
                    item['errors'].append(
                        "WCAG 1.1.1 - Exception: Captcha not tested for alt text."
                    )
            except:

                pass

            #Is displayed
            #Error not displayed - never displayed?

            #item['errors'].append(f"alt = {alt}")

            item['attributes'] = (f"Name: {name}   Alt: {alt}.  ")
            #print(item['attributes'])

            # Test for name and alt text
            if not len(str(alt)) >= 1:
                item['errors'].append(
                    "WCAG 1.1.1 - All non-text content should have an alt text."
                )

            if not len(str(name)) >= 1:
                item['errors'].append(
                    "Non-text content should have a name. Controls Must have a name."
                )

            # compare textdistance
            temp = textdistance.jaro_winkler(
                str(name).lower(),
                str(alt).lower())
            if temp <= .2:
                item['errors'].append(
                    f"WCAG 1.1.1 - Name of element should match alt text or aria-label."
                )

        #print(elements)
        return elements
        '''self.ws = Xlsx('WWC.xlsx', "Sheet1")
        for x in elems:
            self.ws.addRows([x['type'], x['attributes'], x['errors']])
    
        self.ws.set_col_widths(80)
        self.ws.saveWorkbook()
        self.driver.close()'''

    def mapSite(self, filename):
        self.subSystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        self.subSystem.getXPaths(self.url, filename)

    def openElem(self, xPath):
        self.subsystem = Scrap(self.url, 0, self.driver, self.targetDevice,
                               self.configuration, False)
        self.signalConsole.emit("Opening Element")
        self.subsystem.openWebElement(self.url, str(xPath))