Hi,
I bypassed this problem by restarting a new application with the "Internet explorer" web browser.
I don't know if you are currently evaluating the product but Blueprism is a really bad solution for webscraping.
I've tried to just make a search and extract the result data, i took 3 days to manage it with really a lot of trick shots (C# serialization, dynamic values, tricking the stages behavior...). (That took me one afternoon with puppeteer).
Blueprism is dealing badly with Chromium but could be ok with Internet explorer (that is ending his support end of january).
This is some points that I figured out during my test and why this solution can't be used efficiently for a Webscraping system:
- No untricky way to check if an element exist on page (you need to use the wait stage, set the timeout to 0 and use the timeout)
- Blueprism doesn't wait the complete loading of the page to perform action (it made it really unstable)
- The default matching system attribute is too restrictive (XPATH (what if one div is added ?) + link match (that make all fail if you pass params in get))
- 37 sec vs 1.9 sec to perform the same task as puppeteer (outch)
- By default Blueprism match on "Equals" for class attribute and not making "Include" (that is the base of the web)
- The identify system on Chrome made me match on parent div really often (then the "click" action didn't work)
- No way to extract a simple data (example: "You have 52$", you nee to write a regex to get the 52 value and then use a code stage)
- If you want to pass from a Chrome project to a Internet explorer project you will need to identify again all elements (they are not managed on same way)
- When like me you need to dynamically iterate on multiple level (for each table, for each row, for each cell) and that each elements have conditions of matching the dynamic path start be fast a mess and a complicate way to perform it
- The "getText" function of Chrome project doesn't work (return nothing) when the field is in "display: none" (no sens the plugin should have access to the code)
- Blueprism with Internet Explorer look like not really cleaning data from previous navigation i go really strange cases that never happen with puppeteer (like random redirect)
- Blueprism don't have "Dictionary" object, then you need to write code to manage this kind of data (for more advance stuff i needed to serialize object in String vars)
- The standalone is including .dll lib for dealing with the C# runtime that they are old of 2 years ago (and doesn't include Linq)
Some minor's problems:
- When the debug mode is passed in "Fast" mode, the steps are jumping more fast and then the UI freeze by moment and we can't see correctly what is happening anymore
- When you select a huge quantity of stage that go even out of window and that you move them a little you will go far down for miles
I suppose that Blueprism have a lot of good points around the software automation but definitely not for a simple web scraping.
------------------------------
Lucas S
------------------------------
