text recognition

[macroman]

I tried doing text recognition with FireFox and it doesn't work period... image recognition have a better successful rate ratio... how can I just find a certain printed text on a site and then just execute my if then there... it's just not possible in the current version...

[CyberCitizen]

The GetText functions work with many windows apps - it will work with any windows app which uses the Windows TextOut set of functions to render text. But not all apps do so it won't work with all apps. E.g. it doesn't currently read text from Chrome which uses proprietary text rendering methods. I don't know about FireFox but it could be the same. If you use the wizard and point it at the e.g. the desktop, notepad and explorer you'll see that it returns text. So it's a case of suck it and see. Not all apps produce text as far as Windows is concerned. This topic has been dealt with on my blog a number of times:

http://www.mjtnet.com/blog/2009/01/23/s ... scheduler/

[mightycpa]

I had the same problem, but in IE, because of a particular application running some kind of Java or something. I found two solutions for Firefox:

Tesseract OCR. In your script you grab an image of the text, OCR it to a file, and read the file. If the text is too small, do a Ctrl + until it gets big enough. The only proviso is that the text must be black and the background white for the best character recognition.

In Firefox, click Options > Content > Colors, select black and white, and uncheck the box that allows web page to use its colors.

Then you're in business.

[mightycpa]

I have two other suggestions for you.

Ctrl-a Ctrl-c which will copy all visible text in the window to the clipboard. You can parse through that and do what you want.

Or

Ctrl-u which will expose the page source in a separate window. That way, if you need to search both text and tags, that's in there too.