I am trying to capture Text from UI. However in inspect element tool the text is missing. Text is coming from some other div. When I indicate elements from UIPath it says validated but inside the element the text is not there. It's somewhere in other div. I tried editing CSS shown below, which is the exact location of the text. However the data is blank.
Midified CSS:
<webctrl css-selector='div[class*=is-cherry]h1v' idx='2' isleaf='1' parentid='root' tag='DIV' /
The page is using React. So there is no proper way to simply read the text. Instead do it with OCR.
Use a Selector
The best way for you would be using a selector. Here it's a little bit tricky to find the proper selector as the React Framework tries to hide several elements by itself. But when you found the pattern you are good to go.
Find your value under the level root as you can see in the image:
So now you simply use the Get Text activity:
Make sure that you edit the selector in the following schema:
<webctrl tag='H1' parentclass='level-item is-blue fadeInUp' />
This is the selector for your blue value. Now if you want the red one take this one:
<webctrl tag='H1' parentclass='level-item is-cherry fadeInUp' />
I believe you got the point now how that page is working and how to selection of the different colours works.
OCR technology (make sure you are using the Profile Scan)
I would not recommend you to use OCR as you never know if the elements is switching its visual area. If so your process would fail.
Related
I'm developing logs viewer web program with Vue.js
I receive log data with ajax and display it with Slickgrid.
What i need to do is highlighting keyword after searching.
I found some examples highlighting whole cells or row but couldn't find highlighting specific keyword in cell.
ex)When i search a word 'cat', slickgrid shows cells which include 'cat'.
And i need to highlight the word 'cat' in the cell.
Anyone knows how to do this? or any examples??
Thank you.
You'll need to write a custom formatter. See here for an example page. Make sure you're using the 6pac repo - it's up to date, the MLeibman repo is unmaintained now.
Re highlighting a word, you'll need to return HTML from the formatter, and just have a special span to hilight the word, eg:
we will build a <span class="hilight">wall<span/>
It's a tricky business finding a full word, that is making sure it's not part of another word, if that's what you want eg.
did you buy the <span class="hilight">wall<span/>paper yet?
That's a whole 'nother Google search in itself.
I am trying to create a UIPath project to automate filling out a simple form in Microsoft Dynamics Web Client 2016. I have been running into problems with strange behavior due to what I believe to be unreliable selectors. The Dynamics web client is written with Angular and seems to be a single page application. None of the inputs have any type of ID or specific data-attribute that UIPath can use as a selector as far as I can tell. It looks like the program is using the DOM tree to count the number of nested divs in order to assign each input with a "parentId" attribute. This seems to cause unreliable behavior when there are popup windows and I am worried that going forward with an automation with this particular web application will not be scalable.
Example of a selector:
<html title="Microsoft Dynamics GP" />
<webctrl parentid='a000000000000000056800083584c00030100000000f1550000000000'
tag='INPUT' />
Where are this selector and parentId attribute coming from and will it be reliable and scalable through updates and form changes?
If the element itself can't be found reliably, just work around the issue by using an anchor on an element that can. For example, the input elements on http://www.rpachallenge.com/ can never be found reliably as their id changes every few minutes. So, if you need to type text into the field for "Company Name", we need to work with anchors - in our case there will always be a label with a distinctive text.
You will need to use an Anchor Base along with a Find Element activity (plus the desired action). Here's the selector for the anchor - using the element's descriptive text:
<webctrl aaname='Company Name' tag='LABEL' />
Then, the Type Into activity just refers to an Input element:
<webctrl tag='INPUT' />
The only thing to keep in mind are multiple languages in which case you will need to either consider them in the selector, or select the element by id, if applicable.
I'm newest in web.
I want customize default scroll-bar in ag-grid. I try use ngx-scrollbar for this. In sources ngx-scrollbar I found that ngx-scrollbar using ng-content for wrapping elements(link to github source file). If wrap ag-grid-angular element then scrolling even doesn't shows because content doesn't overflow ag-grid-angular content because oveflow happen in div with class .ag-body-viewport where using stock srolls. In order to achieve needed effect I wish pass DOM element with class .ag-body-viewport to ng-content of ngx-scrollbar. Is it possible?
More info here github issue but I don't use Nice Scroll or Perfect Scrollbar
I want use ngx-scrollbar because it has capability of customization.
UPDATE
I can pass DOM element to ng-content using next sintax(agGridViewport is native element):
<ng-scrollbar>
{{ agGridViewport }}
<ng-scrollbar>
but it pass element like a copy of real DOM object but I want pass this like a refence.
Now I got UI like a stack style:
[rendered real ag-grid-angular]
[rendered ng-scrollbar with his content]
But it isn't that I need. Also I got bad rendering with artifacts and even some components doesn't appear. Now I want to try use Renderer2 for putting ng-scrollbar element between two DOM elements in tree(.ag-body-viewport-wrapper and .ag-body-viewport). But I think it's very bad idea because Renderer2 doesn't have methods for putting elements between DOM elements in tree and my code will be very unlify and complicated :( It's must be like injection.
No, I can not do injection in DOM with Angular tools. It's bad idea do it and it is not possible now. AgGrid doesn't support customization with tools like ngx-scrollbar which using ng-content for wrapping components and other elements. May be better way using other tools which working by another way or using webkit customization which supports not all web browsers.
UPDATE
Also I try to use smooth-scrollbar. I managed to get better result. I bind him to element .ag-body-viewport. It works fine but it scrolling only content of grid without header be careful. smooth-scroll bar doesn't have options for horizontal and vertical scrollbar as a different units. I know how issue can be solve. Vertical scrollbar must be bind to .ag-body-viewport and horizaontal scrollbar must be bind to .ag-root. If your can find scrollbar which let you do it that you can solve this problem. May be I write special component for Angular in near future and then add link to this answer. If you do it faster you can add yourself link or you can add link to already existing packages.
I have spied an input text box using the Application Modeller of Blue Prism and was able to successfully highlight the text box using the below XPath:
/HTML/BODY(1)/DIV(4)/main(1)/DIV(1)/DIV(1)/DIV(1)/DIV(2)/DIV(1)/DIV(1)/DIV(2)/IFRAME(1)/HTML/BODY(1)/DIV(2)/FORM(1)/DIV(3)/TABLE(2)/TBODY(1)/TR(1)/TD(1)/DIV(1)/DIV(1)/DIV(1)/DIV(2)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/SPAN(1)/DIV(1)/DIV(2)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/DIV(1)/TABLE(1)/TBODY(1)/TR(1)/TD(1)/INPUT(1)
I wanted to use a more robust XPath and to achieve that I was trying to use the below XPath:
//*[#id="CT"]/div/div/div/div[1]/div[1]/table/tbody[1]/tr/td/input[1]
The above XPath was identifying the element correctly in Chrome but was getting the below error message when trying the same in Blue Prism:
Error - Highlighting results - Object reference not set to an instance of an object.
Let me know if I am doing anything incorrectly.
Sorry for replying to a pretty old one! The workaround we've devised for this scenario (where making the path dynamic requires too long of a loop / search) is to use Jquery snippets. If the page is using jquery it is trivial to execute these queries very quickly using the blue prism capability of executing javascript functions.
And we put in an enhancement request, because it'd be supremely useful functionality.
Update: As a user points out below, the vanilla js querySelector method is probably safer and more future proof than using jquery if it is possible to be used.
Blue Prism does not fully support the XPath spec; alas the construct you're attempting to use here won't work.
Alternatively, you can set the Path attribute of an application modeler entry to be Dynamic, which allows you to insert dynamic parameters from the process/object level to pinpoint elements you'd like to interact with.
Unfortunately Blue Prism doesn't actually use "real" XPaths, but only an extremely limited subset: Absolute paths without wildcards. (Note: It is technically possible to match the XPath to a string with wildcards, but this seemingly causes BP to check every single element in the document, and is so slow it is almost never the right solution.)
For cases where an element can't be robustly identified via the BP application modeler (maybe because it requires complex or dynamic selectors), my workaround is to inject a JS snippet. JS can select elements much more reliably, and it can then generate the BluePrism path for that element.
Returning data from JS to BluePrism is not trivial, but one of the nicer solutions is to have JS create a <script id="_output"> element, put JSON inside it, then have BluePrism read the contents of this element.
I am writing a CKEDITOR plugin that needs to wrap certain pieces of text in a tag. From a webservice, I have an array of items that need to be wrapped. The array is just the plain text strings. Such as:
"[best buy", "horrible migraine", "eat cake"]
I need to find the instances of this text in the editor and wrap them in a span tag.
This is further complicated because the text may be marked up. So the HTML for "best buy" might be
"<strong>best</strong> buy"
but the text returned from the web service is stripped of any markup.
I started trying to use a CKEDITOR.htmlParser() object, and that seems like it is moderately successful. I am able to catch the parser.onText event and check if the text contains anything in my array.
But then I cannot modify that text. Modifications are not persisted back to the source html. So I think using the htmlParser() is a dead-end.
What is the best way to accomplish this task?
Oh, and as a bonus, I also do not want to lose my user's current cursor position when the changes are displayed.
Here is what I wound up doing and it seems to be working so far.
I created a text filter rule that searches through my array of items for any item that is contained (or partially contained) in the text. If so, it wraps the element in my span.
A drawback here is that I wind up with two spans for items with markup. But in my usecase, this is tolerable.
Then I set the results using:
editor.document.getBody().setHtml(results);
Because of this, I also have to strip this markup back out when this text gets read. I do this using an elements filter on editor.dataProcessor.htmlFilter.
This seems to be working well for my (so far limited) test cases.