def modelEnumer(): global driver, prevTrail, modelName rootUrl='http://www.asus.com/Networking/AllProducts/' CSSs=driver.find_elements_by_css_selector CSS=driver.find_element_by_css_selector numElm=lambda c:len(CSSs(c)) try: goToUrl(rootUrl) # retryStable(lambda:numElm('#list-table-area a'), 30,2) models = getElems('#list-table-area a') modelNames1 = [getElemText(_) for _ in models] numModels=len(models) ulog('numModels=%s'%numModels) startIdx = getStartIdx() for idx in range(startIdx, numModels): model = models[idx] modelName = retryA(lambda: model.text) ulog('click %s "%s"'%(idx, modelName)) prevTrail+=[idx] retryA(lambda:model.click()) fileEnumer() prevTrail.pop() goToUrl(rootUrl) models = getElems('#list-table-area a') if len(models) != numModels: modelNames2 = [getElemText(_) for _ in models] modelNamesD = set(modelNames1) - set(modelNames2) assert len(models)==numModels except Exception as ex: ipdb.set_trace() traceback.print_exc() driver.save_screenshot(getScriptName()+'_'+getFuncName()+'_excep.png')
def fileEnumer(): global driver,prevTrail,modelName CSS=driver.find_element_by_css_selector CSSs=driver.find_elements_by_css_selector try: try: waitClickable('#lisupport a',15,1.6).click() except TimeoutException: driver.save_screenshot('asus_no_firmware_download.png') ulog('No firmware download for "%s" !'%modelName) return enterFrame('ifame_auto_size') # click 'Driver & Tools' waitClickable('#a_support_tab_Download',40,2).click() # switch to frame enterFrame('ifame_auto_size') # open dropdown list to select "Others" OS waitClickable('#mainzone_Download2_btn_select_os',10,1).click() retryA(lambda:elemWithText('ul.dropdown-menu.os a', "Others").click()) try: # expand firmware dropdown waitClickable('#btn_type_20',20,1).click() except TimeoutException: driver.save_screenshot('asus_no_firmware_download_2.png') ulog('No firmware download for" %s"!'%modelName) return # retryA(lambda:elemWithText('#download a','Firmware').click(), 20,1) waitUntilStable('#div_type_20',3,0.4) tables = [_ for _ in CSSs('#div_type_20 table') if getElemText(_).startswith('Description')] numTables = len(tables) ulog('numTables=%s'%numTables) versions = [getElemText(_) for _ in CSSs('#div_type_20 p')] assert len(versions)==numTables pageUrl=driver.current_url startIdx = getStartIdx() for idx in range(startIdx, numTables): desc = tables[idx].text relDate = guessDate(desc) fileSize = guessFileSize(desc) fwVer = guessVersion(versions[idx]) fileUrl = tables[idx].find_element_by_css_selector('a').get_attribute('href') trailStr=str(prevTrail+[idx]) sql("INSERT OR REPLACE INTO TFiles(" " model, fw_ver, rel_date, file_size, fw_desc, " " page_url, file_url, tree_trail) VALUES" "(:modelName,:fwVer,:relDate, :fileSize, :desc," ":pageUrl, :fileUrl, :trailStr)", glocals()) ulog('UPSERT "%(modelName)s", "%(fwVer)s", "%(relDate)s", ' '%(fileSize)s, "%(fileUrl)s", %(trailStr)s '%glocals()) except Exception as ex: ipdb.set_trace() traceback.print_exc() driver.save_screenshot(getScriptName()+'_'+getFuncName()+'_excep.png')
def treeWalker(): global startTrail,prevTrail, driver def getNodes(): return getElems('#psbox3 li a' if prevTrail else '#psbox2 li a') nodes=getNodes() numNodes=len(nodes) if startTrail: startIdx=startTrail.pop(0) else: startIdx=0 ulog('startTrail=%s'%startTrail) ulog('prevTrail=%s'%prevTrail) ulog('startIdx=%d'%startIdx) nodeTxts=[getElemText(_) for _ in getNodes()] ulog('nodes=%s (len=%d)'%([(i,_) for i,_ in enumerate(nodeTxts)],numNodes)) for idx in range(startIdx, numNodes): try: crumbs=waitText('#psa_crumbs').replace('\n',' > ') ulog('crumbs=%s'%crumbs) nodeTxt = getElemText(nodes[idx]) nodeId = nodes[idx].get_attribute('id').strip() ulog('nodeId="%s"'%nodeId) if not nodeId: # isLeaf ulog('Click Leaf "%s"'%nodeTxt) clickElem(nodes[idx]) with suppress(StaleElementReferenceException): nodes[idx].click() prevTrail+=[idx] # if not waitUntil(lambda: not driver.find_elements_by_css_selector('#psbox3')): # nodes[idx].click() selectSoftwareType() prevTrail.pop() assert getCurDepthAtTreeWalker()==len(prevTrail) nodes=getNodes() continue with UntilTextChanged('#psbox3'): ulog('Click branch "%s"'%nodeTxt) clickElem(nodes[idx]) prevTrail+=[idx] treeWalker() prevTrail.pop() nodes=getNodes() except Exception as ex: ipdb.set_trace() print(ex); traceback.print_exc() driver.save_screenshot('cisco_treeWalker.png') # end for crumbs=getElems('#psa_crumbs a') ulog('back to "%s"'%getElemText(crumbs[-1])) ulog('prevTrail=%s'%prevTrail) clickElem(crumbs[-1])
def selectSupport(prev_url): global prevTrail,category,productName,model,driver CSS=driver.find_element_by_css_selector try: waitVisible('.product-name-price') productName=CSS('.product-name-price h2').text.strip() ulog('productName="%s"'%productName) # 'Wireless G Travel Router' model=CSS('.product-name-price p').text.strip() # 'Part # F5D7233' model = model.split('#')[1].strip() ulog('model="%s"'%model) # 'F5D7233' if not productName: ulog('productName is empty, bypass!') driver.get(prev_url) # waitText('.search-results-notification') return try: support = next(_ for _ in getElems('.icon-list-header-container') if getElemText(_).startswith('DOWNLOAD')) except StopIteration: ulog('No download in '+driver.current_url) trailStr=str(prevTrail) sql("INSERT OR REPLACE INTO TFiles(category, product_name, model, tree_trail) VALUES (:category, :model, :productName, :trailStr)", glocals()) ulog('UPSERT "%(category)s", "%(model)s", "%(productName)s" %(prevTrail)s'%glocals()) driver.get(prev_url) # waitText('.search-results-notification') return downloads = support.find_elements_by_css_selector('a') numDownloads = len(downloads) startIdx=getStartIdx() for idx in range(startIdx, numDownloads): txt=downloads[idx].text if model not in txt: ulog('bypass %s,"%s" because it\'s Portal'%(idx,txt)) continue ulog('click %s,"%s"'%(idx,txt)) prevTrail += [idx] enterElem(downloads[idx],selectDownload) prevTrail.pop() support = next(_ for _ in getElems('.icon-list-header-container') if getElemText(_).startswith('DOWNLOAD')) downloads = support.find_elements_by_css_selector('a') driver.get(prev_url) # waitText('.search-results-notification') except Exception as ex: ipdb.set_trace() traceback.print_exc() driver.save_screenshot(getScriptName()+'_'+getFuncName()+'_excep.png')
def getElemTextUntilStabled(elm:WebElement, timeOut:float=60, pollFreq:float=3)->str: import time beginTime = time.time() try: oldText=getElemText(elm, timeOut) except TimeoutException: oldText=None elapsedTime = time.time() - beginTime while elapsedTime < timeOut: beginTime = time.time() time.sleep(pollFreq) newText=getElemText(elm, timeOut) if oldText == newText: return newText oldText = newText elapsedTime += time.time() - beginTime raise TimeoutException('[getElemTextUntilStabled] time out with ' 'timeOut=%f seconds, pollFreq=%f seconds'%(timeOut,pollFreq))
def harvestPage2(): global modelName modelName=getText('big > strong') print("Page2 modelName=",modelName) global driver numRows = getNumElem('tr#rsq') if numRows==0: return for iRow in range(2, numRows+1): row = waitClickable('tr#rsq:nth-child(%d)'%iRow) rowText = getElemText(row) uprint('Row%d %s'%(iRow, rowText)) if 'firmware' not in rowText.lower(): print(' -- bypass') continue uprint('Click '+rowText) row.click() modelName=getText('big > strong') print('Page3 modelName=%s'%modelName) desc=getText('.prodtd > table:nth-child(4) > tbody:nth-child(1) ' '> tr:nth-child(2) > td:nth-child(2)') uprint("Description="+desc) for fn9 in getElems('.fn9'): fileName = getElemText(fn9) fileExt = path.splitext(fileName)[1].lower() uprint('filaName="%s"'%fileName) if fileExt in ['.doc', '.docx', '.txt','.pdf','.htm','.html','.xls']: uprint(' -- fileName "%s" doesn\'t look like a firmware file'%fileName) global conn csr=conn.cursor() model=modelName csr.execute( "INSERT OR REPLACE INTO dlink(model,file_name,desc)" "VALUES(:model,:fileName,:desc)",locals() ) uprint('INSERT OR REPLACE INTO "%(model)s","%(fileName)s","%(desc)s"'% locals()) # waitDownloading() clickDownloadableElem(fn9) global driver driver.back()
def clickDownloadableElem(elem:WebElement)->str: pollFreq=1 filesOld=os.listdir(dlDir) fileName=getElemText(elem) href = elem.get_attribute('href') assert href is not None uprint('fileName="%s" href="%s"'%(fileName,href)) sql("UPDATE dlink SET href=:href WHERE" " file_name=:fileName",locals()) uprint("UPDATE dlink SET href=%(href)s WHERE" " file_name=%(fileName)s"%locals()) return
def main(): startCatIdx = int(sys.argv[1]) if len(sys.argv)>1 else 0 startFamIdx = int(sys.argv[2]) if len(sys.argv)>2 else 0 startPrdIdx = int(sys.argv[3]) if len(sys.argv)>3 else 0 global driver,conn harvest_utils.driver=getFirefox(dlDir) driver = harvest_utils.driver conn=sqlite3.connect('netgear.sqlite3') csr=conn.cursor() csr.execute("CREATE TABLE IF NOT EXISTS TFiles(" "brand TEXT," "category TEXT," "family TEXT," "product TEXT,"# -- is model "desc TEXT,"# -- is fileName "href TEXT," "file_sha1 TEXT," "PRIMARY KEY (product,desc)" ")"); conn.commit() driver.get('http://downloadcenter.netgear.com/') #click DrillDown waitClickable('#ctl00_ctl00_ctl00_mainContent_localizedContent_bodyCenter_BasicSearchPanel_btnAdvancedSearch').click() # # wait Page2 try: catSel=Select(waitClickable(catSelCss)) numCat=len(catSel.options) for catIdx in range(startCatIdx,numCat): catSel=Select(waitClickable(catSelCss)) print('catIdx=',catIdx) catTxt=catSel.options[catIdx].text uprint('catTxt='+catTxt) catSel.select_by_index(catIdx) waitTextChanged(famSelCss) famSel=Select(waitClickable(famSelCss)) numFam=len(famSel.options) for famIdx in range(startFamIdx,numFam): famSel=Select(waitClickable(famSelCss)) print('famIdx=',famIdx) startFamIdx=0 famTxt =famSel.options[famIdx].text uprint('famTxt='+famTxt) famSel.select_by_index(famIdx) waitTextChanged(prdSelCss) prdSel=Select(waitClickable(prdSelCss)) numPrd=len(prdSel.options) for prdIdx in range(startPrdIdx,numPrd): prdSel=Select(waitClickable(prdSelCss)) startPrdIdx=0 print("catIdx=%d, famIdx=%d, prdIdx=%d"%(catIdx,famIdx,prdIdx)) prdTxt=prdSel.options[prdIdx].text uprint('cat,fam,prd=("%s","%s","%s")'%(catTxt,famTxt,prdTxt)) prdWaiting = waitElem(prdWaitingCss) prdSel.select_by_index(prdIdx) WebDriverWait(driver, 5, poll_frequency=0.5).\ until(lambda x:prdWaiting.is_displayed()==True) WebDriverWait(driver, 60, poll_frequency=0.5).\ until(lambda x:prdWaiting.is_displayed()==False) #waitUntil(lambda:prdWaiting.is_displayed()==True) #waitUntil(lambda:prdWaiting.is_displayed()==False) numResults=waitText(numResultsCss,3) print('numResults=',numResults) if numResults is None: continue numResults=int(re.search(r"\d+", numResults).group(0)) if numResults >10: showMore=waitClickable("#lnkAllDownloadMore",3) showMore.click() try: erItems=getElems('a.register-product.navlistsearch',3) except TimeoutException: erItems=getElems('div#LargeFirmware > ul > li > div > p > a.navlistsearch',3) if len(erItems) != numResults: print('Error, numResults=%d, but len(erItems)=%d' %(numResults,len(erItems))) for erItem in erItems: if not erItem.is_displayed(): continue desc=getElemText(erItem) uprint('desc="%s"'%desc) href=erItem.get_attribute('data-durl') if not href: href=erItem.get_attribute('href') print('href=',href) if not href.startswith('http'): print('Error: href=',href) sql("INSERT OR REPLACE INTO TFiles" "(brand,category,family,product,desc,href)VALUES" "('Netgear',:catTxt,:famTxt,:prdTxt,:desc,:href)", locals()) uprint('INSERT ' '("%(catTxt)s","%(famTxt)s","%(prdTxt)s","%(desc)s","%(href)s")' %locals()) except Exception as ex: import ipdb; ipdb.set_trace() print(ex) import traceback; traceback.print_exc() print('-- terminate firefox') driver.quit()
def main1(catIdx, famIdx, prdIdx, executor): startCatIdx, startFamIdx, startPrdIdx = catIdx, famIdx, prdIdx driver = webdriver.PhantomJS() harvest_utils.driver = driver driver.get('http://downloadcenter.netgear.com/') # click DrillDown waitClickable('#ctl00_ctl00_ctl00_mainContent_localizedContent_bodyCenter' '_BasicSearchPanel_btnAdvancedSearch').click() # # wait Page2 try: catSel = Select(waitClickable(catSelCss)) numCat = len(catSel.options) for catIdx in range(startCatIdx, numCat): catSel = Select(waitClickable(catSelCss)) print('catIdx=', catIdx) startCatIdx = 0 catTxt = catSel.options[catIdx].text uprint('catTxt= ' + catTxt) oldText = getText(famSelCss) catSel.select_by_index(catIdx) waitTextChanged(famSelCss, oldText) famSel = Select(waitClickable(famSelCss)) numFam = len(famSel.options) for famIdx in range(startFamIdx, numFam): famSel = Select(waitClickable(famSelCss)) print('famIdx=', famIdx) startFamIdx = 0 famTxt = famSel.options[famIdx].text uprint('famTxt= ' + famTxt) oldText = getText(prdSelCss) famSel.select_by_index(famIdx) waitTextChanged(prdSelCss, oldText) prdSel = Select(waitClickable(prdSelCss)) numPrd = len(prdSel.options) for prdIdx in range(startPrdIdx, numPrd): prdSel = Select(waitClickable(prdSelCss)) startPrdIdx = 0 print("catIdx=%d, famIdx=%d, prdIdx=%d" % (catIdx, famIdx, prdIdx)) prdTxt = prdSel.options[prdIdx].text uprint('cat,fam,prd=("%s","%s","%s")' % (catTxt, famTxt, prdTxt)) prdWaiting = waitElem(prdWaitingCss) prdSel.select_by_index(prdIdx) try: WebDriverWait(driver, 1, 0.5).\ until(lambda x:prdWaiting.is_displayed() is True) except TimeoutException: pass try: WebDriverWait(driver, 5, 0.5).\ until(lambda x:prdWaiting.is_displayed() is False) except TimeoutException as ex: pass numResults = waitText(numResultsCss, 3, 0.5) if numResults is None: continue numResults = int(re.search(r"\d+", numResults).group(0)) print('numResults=', numResults) if numResults > 10: waitClickable("#lnkAllDownloadMore", 3).click() try: erItems = getElems('a.register-product.navlistsearch', 3, 0.5) except TimeoutException: erItems = getElems( 'div#LargeFirmware > ul > li > div > p > a.navlistsearch', 3) if len(erItems) != numResults: print('Error, numResults=%d, but len(erItems)=%d' % (numResults, len(erItems))) for itemIdx, erItem in enumerate(erItems): if not erItem.is_displayed(): print('itemIdx=%d is not displayed()' % itemIdx) continue desc = getElemText(erItem) uprint('desc="%s"' % desc) if 'firmware' not in desc.lower(): continue fw_url = erItem.get_attribute('data-durl') if not fw_url: fw_url = erItem.get_attribute('fw_url') print('fw_url=', fw_url) if not fw_url: continue if not fw_url.startswith('http'): print('Error: fw_url=', fw_url) continue executor.submit(download_file, prdTxt, desc, fw_url) # download_file(prdTxt, desc, fw_url) catIdx, famIdx, prdIdx = None, None, None return catIdx, famIdx, prdIdx except BaseException as ex: traceback.print_exc() dumpSnapshot('netgear_crawler.py.png') finally: driver.quit() return catIdx, famIdx, prdIdx
def main(): startModelIdx = int(sys.argv[1]) if len(sys.argv)>1 else 0 startRevisionIdx = int(sys.argv[2]) if len(sys.argv)>2 else 0 brand='Linksys' global driver,conn harvest_utils.driver=getFirefox() driver = harvest_utils.driver conn=sqlite3.connect('Linksys.sqlite3') csr=conn.cursor() csr.execute( "CREATE TABLE IF NOT EXISTS TFiles(" "brand TEXT," "model TEXT," "revision TEXT," # hardware version "fw_date DATE," "fw_ver TEXT," "file_title TEXT," "file_size INTEGER," "href TEXT," "file_sha1 TEXT," "PRIMARY KEY (brand,model,revision,file_title)" ");") conn.commit() driver.get('http://www.linksys.com/us/support/sitemap/') try: numModels = getNumElem('.item ul li a') print('numModels=',numModels) for modelIdx in range(startModelIdx, numModels): startModelIdx=0 modelElm = getElems('.item ul li a')[modelIdx] modelText = getElemText(modelElm, 5) print('modelIdx=',modelIdx) uprint('modelText="%s"'%modelText) # guess Possible Model model = guessModel(modelText) print('model=',model) rows = csr.execute( "SELECT model from TFiles WHERE model=:model",locals() ).fetchall() if rows: print('model "%s" already in TFiles, bypass!!'%model) continue modelElm.click() # click 'Download Software' try: waitClickable('a[title="Download Software"]', 40).click() except TimeoutException: print('No "Download Software" link found, bypass!!') csr.execute( "INSERT INTO TFiles(brand,model,revision)VALUES" "(:brand,:model,'')", locals()) conn.commit() print('INSERT model="%s"'%model) driver.back() continue # enumerate all accordians accordians = getElems('.article-accordian', 10) numAccordians=len(accordians) print('numAccordians=',numAccordians) print('driver.current_url=', driver.current_url) for revisionIdx in range(startRevisionIdx, numAccordians): startRevisionIdx=0 accordians = getElems('.article-accordian') # expand accordian (one-based) accordian = accordians[revisionIdx] revisionTxt = getElemText(accordian) print('revisionIdx=',revisionIdx) uprint('revisionTxt="%s"'%revisionTxt) revision = guessRevision(revisionTxt) print('revision=',revision) divId = accordian.get_attribute('data-collapse-target') # expand accordian 'revision'='Hardware Version' driver.execute_script( "document.querySelectorAll('.article-accordian')[%d].click()" %(revisionIdx)) divElm = waitVisible('#'+divId) divTxt = getElemTextUntilStabled(divElm,10,2.5) assert divTxt uprint('divTxt="%s"'%divTxt) numDowns = getCount(divTxt, 'Download') if numDowns ==0: csr.execute( "INSERT INTO TFiles(brand,model,revision)VALUES" "(:brand,:model,:revision)",locals()) conn.commit() print('INSERT "%(model)s","%(revision)s"'%locals()) continue downElms =iter(divElm.find_elements_by_css_selector('a')) lastSpanEnd=0 for downIdx in range(numDowns): spanBegin = getNthIndex(divTxt, downIdx, 'Download') spanEnd = divTxt.find('\n', spanBegin+len('Download')) if spanEnd==-1: spanEnd=len(divTxt) foreword='\n'.join(reversed(divTxt[lastSpanEnd:spanEnd].splitlines())) fwDate=guessDate(foreword) fileSize = guessFileSize(foreword) fwVer = guessVersion(foreword) if fwVer: fileTitle = guessFileTitle(foreword, fwVer) else: fileTitle = guessFileTitle2(foreword) while True: downElm = next(downElms) if downElm.text.strip().startswith('Download'): break href=downElm.get_attribute('href') lastSpanEnd=spanEnd csr.execute( "INSERT OR REPLACE INTO TFiles(brand,model,revision," "fw_date, fw_ver, file_title, file_size, " "href) VALUES (:brand,:model,:revision," ":fwDate, :fwVer, :fileTitle," ":fileSize, :href)", locals()) conn.commit() uprint("INSERT '%(model)s', '%(revision)s', '%(fwDate)s'" ", '%(fwVer)s', '%(fileTitle)s', '%(fileSize)d'" ", '%(href)s'" %locals()) driver.back() driver.back() except http.client.IncompleteRead as ex: print(ex) import traceback; traceback.print_exc() print('-- Selenium exhausted') driver.quit() except Exception as ex: import ipdb; ipdb.set_trace() print(ex) print('driver.current_url=',driver.current_url) import traceback; traceback.print_exc() print('-- terminate firefox') driver.quit()
def selectSoftwareType(): """ This page would be jumped to versionWalker() or either jumped back to treeWalker forward: may auto jump backward: not auto jump """ global startTrail,prevTrail,driver try: waitText('.csProductSelectorBreadcrumb', 5, 1) waitUntilStable('.csProductSelectorBreadcrumb', 1, 0.3) depth = getDepth() jumpedLevels =depth - len(prevTrail) ulog('jumpedLevels=%d'%jumpedLevels) assert jumpedLevels>=0 ulog('depth=%d, but prevTrail=%s'%(depth, prevTrail)) startIdxFromStartTrail=False def getStartIdx()->int: if startTrail: nonlocal startIdxFromStartTrail startIdxFromStartTrail=True return startTrail.pop(0) else: return 0 if jumpedLevels>0: while depth>len(prevTrail): startIdx=getStartIdx() prevTrail+=[startIdx] else: startIdx=getStartIdx() assert depth==len(prevTrail) ulog('startTrail=%s'%startTrail) ulog('prevTrail=%s'%prevTrail) ulog('startIdx=%d'%startIdx) ulog('url=%s'%driver.current_url) crumbs = waitText('.csProductSelectorBreadcrumb') uprint('crumbs=%s'%(crumbs.replace('\n',' > '))) if not hasElem('table#imageTableContainer', 1.5,0.4): if jumpedLevels>0: startIdx=getStartIdx() if depth > len(prevTrail): prevTrail+=[startIdx] sdpBannerTitle=waitText('td.SDPBannerTitle').strip() ulog('SDBBannerTitle="%s"'%sdpBannerTitle) assert sdpBannerTitle.lower().startswith('select ') waitUntil(lambda: getNumElem('div.csWrapper li a') > 0) swtypes = getElems('div.csWrapper li a') ulog('%s'%[(i,getElemText(_)) for i,_ in enumerate(swtypes)]) numSwTypes=len(swtypes) assert numSwTypes > 0 for idx in range(startIdx, numSwTypes): ulog('goto Trail=%s'%(prevTrail+[idx])) swtypes = getElems('div.csWrapper li a') ulog('Click "%s"'% getElemText(swtypes[idx])) clickElem(swtypes[idx]) prevTrail+=[idx] selectSoftwareType() prevTrail.pop() # Select a Product -> Select a Software type -> Select a Platform # https://software.cisco.com/download/type.html?mdfid=277873153&flowid=170&softwareid=283724313 # Downloads Home >Products >Cisco Interfaces and Modules >WAN Interface Cards >1700/2600/3600/3700 Series 2-Port Analog Modem WAN Interface Card >Analog Firmware Loader >Windows 2000-v6780 # not auto back to treeWalker # go back manually crumbs = getElems('.csProductSelectorBreadcrumb a') ulog('manually backto "%s"'%getElemText(crumbs[-1])) ulog('prevTail=%s'%prevTrail) clickElem(crumbs[-1]) # do I need to pop prevTrail? # prevTrail.pop() else: ulog('auto forward to versionWalker') if startIdxFromStartTrail: startTrail.insert(0, startIdx) for i in range(jumpedLevels): if not startTrail: break startTrail.pop(0) versionWalker() for i in range(jumpedLevels): crumbs = getElems('.csProductSelectorBreadcrumb a') ulog('manually backto "%s"'%getElemText(crumbs[-1])) ulog('prevTail=%s'%prevTrail) clickElem(crumbs[-1]) prevTrail.pop() except Exception as ex: ipdb.set_trace() print(ex); traceback.print_exc() driver.save_screenshot('cisco_selectSoftwareType.png')
def versionWalker(): global startTrail,prevTrail,driver try: waitClickable('.treeLinks > a:nth-child(1)') ulog('current_url=%s'%driver.current_url) crumbs=waitText('.csProductSelectorBreadcrumb').replace('\n', ' > ') ulog('crumbs=%s'%crumbs) # click Expand All numNodes = len(driver.find_elements_by_css_selector('.tree a')) ulog('number of versions=%d'%numNodes) try: with UntilTextChanged('.tree'): clickElem(waitClickable('.treeLinks > a:nth-child(1)')) except TimeoutException: pass treeText=waitText('.tree') ulog('treeText="%s"'%treeText) if startTrail: startIdx=startTrail.pop(0) else: startIdx=1 ulog('startTrail=%s'%startTrail) ulog('prevTrail=%s'%prevTrail) ulog('startIdx=%d'%startIdx) assert startIdx >= 1 try: prevFwVer=waitText('.tree a.nodeSel', 5) except TimeoutException: uprint("css='.tree a.nodeSel' not found") prevFwVer=None ulog('prevFwVer="%s"'%prevFwVer) for idx in range(startIdx, numNodes): nodes = driver.find_elements_by_css_selector('.tree a') isLeaf = (nodes[idx-1].text != '') ulog('goto Trail=%s'%(prevTrail+[idx])) if isLeaf: if not nodes[idx].text.strip(): continue fwVer=nodes[idx].text.strip() nodeClass=nodes[idx].get_attribute('class') ulog('fwVer="%s", nodeClass="%s"'%(fwVer,nodeClass)) if 'nodeSel' not in nodeClass: noWait= (prevFwVer==fwVer) if prevFwVer else False try: with UntilTextChanged('table#imageTableContainer',10,1,noWait): ulog('Click "%s"'%fwVer) clickElem(nodes[idx]) except TimeoutException: with UntilTextChanged('table#imageTableContainer',10,1,noWait): ulog('Click "%s" twice'%fwVer) clickElem(nodes[idx]) prevTrail+=[idx] tableRowWalker(fwVer) prevTrail.pop() prevFwVer=fwVer # go back page crumbs=getElems('.csProductSelectorBreadcrumb a') ulog('backto "%s"'%getElemText(crumbs[-1])) ulog('prevTail=%s'%prevTrail) if prevTrail==[2, 1, 0, 1, 2, 0]: ipdb.set_trace() clickElem(crumbs[-1]) except Exception as ex: ipdb.set_trace() print(ex); traceback.print_exc() driver.save_screenshot('cisco_versionWalker.png')
def tableRowWalker(fwVer:str): global startTrail,prevTrail,driver try: try: cells = getElems('table#imageTableContainer tr td') except TimeoutException: ulog("Timeout at getElems('table#imageTableContainer tr td'); bypass!") return assert len(cells)%4==0 numFiles=int(len(cells)/4) cellTxt = [getElemText(_) for _ in cells] try: carts = getElems('table#imageTableContainer tr td input', 10) cartsTxt = [getElemAttr(_,'title').lower().strip() for _ in carts] # https://software.cisco.com/download/release.html?mdfid=282822110&flowid=266&softwareid=280805680&release=15.0.2-SE8&relind=AVAILABLE&rellifecycle=MD&reltype=latest # fwVer='15.0.2-SE1(ED)' has len(cells)==6 and numFiles==4 cartsTxt = flatCarts(cartsTxt) assert len(cartsTxt)==numFiles needContracts=[ int('cart' in _) for _ in cartsTxt] # needContract=int(int(len(cells2)/numFiles)==2) except TimeoutException: ulog('needContract= "Deferral"') needContracts= [-1]*numFiles # Deferral ulog('needContracts=%s'%needContracts) quats = combinePerFour(cellTxt) assert len(quats)==len(needContracts) quats = list(zip(quats, needContracts)) fileDescs=[] for quat,needContract in quats: fileTitle=quat[0].split('\n')[0].strip() fileName=quat[0].split('\n')[1].strip() relDate=guessDate(quat[1]) fileSize=guessFileSize(quat[2]) fileDescs.append((fileTitle,fileName,relDate,fileSize, needContract)) model = waitText('td.SDPBannerTitle') ulog('model="%s"'%model) pageUrl=driver.current_url if startTrail: startIdx=startTrail.pop(0) else: startIdx=0 ulog('startIdx=%d'%startIdx) try: imageRhSide=driver.find_element_by_css_selector('div#imageRhSide') driver.execute_script( "arguments[0].scroll(0,%d);"%(62*startIdx), imageRhSide) except: pass spans=getElems('#imageTableContainer tr td span.overlay_img') assert len(spans)==numFiles for idx in range(startIdx, numFiles): ulog('getSha512 Trail=%s'%(prevTrail+[idx])) infos = getSha512(spans[idx]) if infos: infos.update(dict(needContract=needContracts[idx], model=model, pageUrl=pageUrl,treeTrail=str(prevTrail+[idx]))) else: fileTitle,fileName,relDate,fileSize,needContract=fileDescs[idx] infos=dict(fileTitle=fileTitle,fileName=fileName,relDate=relDate, fileSize=fileSize,needContract=needContract, pageUrl=pageUrl,treeTrail=str(prevTrail+[idx]), md5=None,sha512=None,model=model,fwVer=fwVer) sql("INSERT OR REPLACE INTO TFiles(model," " fw_date,fw_ver,file_title,file_name,file_size," " need_contract, page_url, tree_trail, md5, sha512) " "VALUES (:model," " :relDate,:fwVer,:fileTitle,:fileName,:fileSize," " :needContract, :pageUrl, :treeTrail, :md5, :sha512)", infos) ulog('UPSERT "%(model)s", "%(relDate)s", "%(fwVer)s", ' '"%(fileTitle)s", "%(fileName)s", %(needContract)d,' ' "%(treeTrail)s", %(pageUrl)s, "%(md5)s", "%(sha512)s"' %infos) except Exception as ex: ipdb.set_trace() print(ex); traceback.print_exc() driver.save_screenshot('cisco_tableRowWalker.png')