Forum Discussion

ory_z's avatar
ory_z
Contributor
13 years ago

LogParams changed behaviour on TC 8.5

Hi there,



I've noticed a change in the behaviour of the LogParams object in TC 8.5.

We use the OnLogError event to write the details of the error to an external text log. This is useful for when TC hangs or crashes (and the in built logs are lost) to help determine cause and at what stage of the test execution the failure happened.



My function looks something like this:



function GeneralEvents_OnLogError(Sender, LogParams)

{ 

  var errorMessage 

  //Log error text & additional text to external log

  errorMessage = LogParams.MessageText + "(" + LogParams.AdditionalText + ")"

  WriteToExternalLog( "!!! ERROR - " + errorMessage )

}



This worked fine for TC8, 8.1 and 8.2. However in TC8.5 the LogParams.AdditionalText return a very padded string, for example:



<html><style>body {font-family : Courier New, Arial; font-size : 10pt; font-weight : normal; padding : 14px 8px 8px 14px; margin : 0 0 0 0;} p {margin : 4px 0 4px 0; padding : 0 0 0 0;} h4 {font-size: 10pt; font-weight : bold; margin : 12px 0 0 16px; padding : 0 0 0 0;} table {margin-left : 16px; border-top: 1px solid grey;} td {font-size : 10pt; font-weight : normal; padding-left : 8px;} </style><body><p>An error occurred while calling the "ClickButton" method or property of the "<a href="lnamemapping://{09E818B6-F3E0-4BED-B332-DB57AC0208CC}.{49DF2E34-97FE-4320-8A5E-6B759ED384C2}">PlayBtn</a>" object.<br/>The object or one of its parent objects does not exist.</p><h4>Tested Object</h4><table cellpadding="0" cellspacing="0" noborder><tbody><tr><td><b>Alias:</b></td><td>Aliases.RIOS_SSC32.InviteUnit.PlayBtn</td><tr><td><b>Mapping item:</b></td><td>NameMapping.Sys.RIOS_SSC32.InviteUnit.PlayBtn</td></tr></tbody></table><h4>Missing Object</h4><table><tbody><tr><td><b>Alias:</b></td><td>Aliases.RIOS_SSC32</td><tr><td><b>Mapping item:</b></td><td>NameMapping.Sys.RIOS_SSC32</td></tr></tbody></table><p style="margin-top: 12px;"><a href="aqa-help://2201">Possible causes of the error</a></p></body></html>



I've attached a screenshot of an inspection of the LogParams object.



Thanks,

Ory

18 Replies

Sort By
  • AlexeyKolosov's avatar
    AlexeyKolosov
    Staff
    13 years ago

    Hi Ory,


    In TestComplete 8.50, the Additional Info panel contains HTML text for error messages. This is the correct behavior of the tool. You can write helper code to parse it or use as it is when logging the information to a file.

  • ory_z's avatar
    ory_z
    Contributor
    13 years ago
    Hi,



    Can you supply me a code example of how to parse the message in JScript?



    Thanks
  • AlexeyKolosov's avatar
    AlexeyKolosov
    Staff
    13 years ago

    Hi Ory,


    Often, HTML is parsed by using regular expressions. For instance:




       fileText = LogParams.AdditionalText;

       var re = /<p>.*<\/p>/;

       Log.Message(fileText.match(re));


    However, after more thorough thinking, I would suggest that you just use an HTML file to log errors in the OnLogError event handler. Below, is a sample script that creates a file with a two-column table. The first column contains error messages, the second column contains corresponding additional information text. The sample uses an existing HTML file as a template. The template file contains one empty table (<table border="1" style="width: 100%;"></table>).




    function makeRow(errText, additionalInfo)

    {

       var aRow = new String("<tr><td>errData</td><td>aiData</td></tr>\r\n");

       aRow = aRow.replace("errData", errText);

       aRow = aRow.replace("aiData", additionalInfo);

       return aRow; 

    }


    function GeneralEvents_OnLogError(Sender, LogParams)

    {

        var ForReading = 1, ForWriting = 2;

        var fso = new ActiveXObject("Scripting.FileSystemObject");

        filename = "d:\\Results\\HTMLPage1.htm";

        var f = fso.OpenTextFile(filename, ForReading);

        var fileText = new String();

        fileText = f.ReadAll();

        f.Close();

        var pos = fileText.lastIndexOf("</table>");

        var newFileText = fileText.slice(0, pos-1);

        newFileText += makeRow(LogParams.Str, LogParams.StrEx);

        newFileText += fileText.slice(pos);

        f = fso.OpenTextFile(filename, ForWriting, true);

        f.Write(newFileText);

        f.Close(); 

    }


    function Test()

    {

     fso = new ActiveXObject("Scripting.FileSystemObject");

     fso.DeleteFile("d:\\Results\\HTMLPage1.htm");

     fso.CopyFile ("d:\\Results\\SampleFiles\\HTMLPage1.htm", "d:\\Results\\HTMLPage1.htm")


      var notepad;

      notepad = Aliases.notepad;

      //notepad.wndNotepad.MainMenu.Click("Help|About Notepad");

      notepad.dlgAboutNotepad.btnOK.ClickButton();

    }

  • ory_z's avatar
    ory_z
    Contributor
    13 years ago
    Thanks for the response.



    Didn't quite have time to look at it until today...



    We need the result in Text format (and not HTML) due to the fact that the whole idea behind the secondary logging functionality is that it is text based and can be tailed easily.



    Adding ontop of that is the complexity of the functionality: When I use the Log.Error operation, the AdditionalInfo are passed as text. However when TestComplete logs an error (such as object not found), it passes it as HTML.



    In any case I have changed my code as follows:




      var HTML_Text = LogParams.AdditionalText;

      var errorMessage = LogParams.MessageText

      

      if ( HTML_Text.indexOf( "<html>" ) != -1 )

      { 

        var re = /<p>.*<\/p>/;

        errorMessage += "(" + HTML_Text.match(re) + ")";     

      }

      else

        errorMessage += "(" + HTML_Text + ")";       

        

      VIX.Logs.writeToExternalLog( "!!! ERROR - " + errorMessage );




    But the result of the operation you suggested is still not that "clean", here is what HTML_Text.match(re) evaluates as:

    The object does not exist. See Additional Information for details.(<p>An error occurred while calling the "Click" method or property of the "<a href="lnamemapping://{225B2166-C880-4BD1-BCCF-7A46171069A3}.{73626755-8CE9-4667-90D2-14EC71C4BB1B}">EnableDisableBtn</a>" object.<br/>The object or one of its parent objects does not exist.</p><h4>Tested Object</h4><table cellpadding="0" cellspacing="0" noborder><tbody><tr><td><b>Alias:</b></td><td>Aliases.RIOS_SSC32.Play.EnableDisableBtn</td><tr><td><b>Mapping item:</b></td><td>NameMapping.Sys.RIOS_SSC32.Play.EnableDisableBtn</td></tr></tbody></table><h4>Missing Object</h4><table><tbody><tr><td><b>Alias:</b></td><td>Aliases.RIOS_SSC32</td><tr><td><b>Mapping item:</b></td><td>NameMapping.Sys.RIOS_SSC32</td></tr></tbody></table><p style="margin-top: 12px;"><a href="aqa-help://2201">Possible causes of the error</a></p>)



    Unfortunately I'm not proficient enough with Regular Expression in order to try and fix it.



    Other suggestions?
  • rminnich's avatar
    rminnich
    Contributor
    13 years ago

    You could parse it using something besides a regex.  You can manually step through the string extracting or removing extraneous information.  For example, you know that the <html> and </html> tabs can be discarded.  Everything between <style> and </style> can be discarded.  The <br/> tag can be replaced with a linefeed/newline.

    It gets slightly trickier for <a> tags in that you only care about the element value and not the attributes.


    I could also see treating this html as XML and using a SAX parser to grab the text needed.  That would be a more complex solution.

  • ory_z's avatar
    ory_z
    Contributor
    13 years ago
    Thanks for the suggestion.



     I'm looking for the simplest possible solution, let's see what AQA Support reckons.
  • AlexKaras's avatar
    AlexKaras
    Champion Level 3
    13 years ago
    Hi Ory,



    Just a quick idea: have you tried to set LogAttributes.ExtendedMessageAsPlainText to false (see relevant help topic for code sample) ?
  • AlexeyKolosov's avatar
    AlexeyKolosov
    Staff
    13 years ago
    Hi Ory,



    I have discussed the matter with the developers, and they say that there are no plans to modify the LogParams.StrEx value to make it support both the HTML and plain text formats (like the Log.Error operation can do). This would affect the performance of the text execution.



    In most cases, if you need to parse the HTML text, it should be sufficient to create a script that will just remove all tags from the text and replace the </br> tag with a line feed.
     

    Unfortunately, this approach will not work for errors with complex data in the Additional Information panel (for instance, for checkpoints that post tables).
  • ory_z's avatar
    ory_z
    Contributor
    13 years ago
    Hi Allen,



    Thanks for your answer.



    Does that mean that what Alexei suggested would not work?



    Can you supply me with example code in JScript that would strip the HTML formatting from a string?



    Thanks
  • AlexeyKolosov's avatar
    AlexeyKolosov
    Staff
    13 years ago

    Hi Ory,





    Valid HTML stripping is quite a complex task, but you can try using something like this - should be enough for the majority of use cases:



    function stripHtml(html)

    {

      var stripAllTags = new RegExp("(<style[^>]*?>(.*)(</style[^>]*?>|$))|(<script[^>]*?>(.*)(</script[^>]*?>|$))|(<[^<]*?>)", "g");

      var text = html.replace(stripAllTags, " ");

      

      // Expand common entities

      var entityMapping = {};       

      entityMapping["&quot;"] = "\"";

      entityMapping["&apos;"] = "'";

      entityMapping["&lt;"] = "<";

      entityMapping["&gt;"] = ">";

      entityMapping["&nbsp;"] = " ";

      entityMapping["&amp;"] = "&";  

      

      for (var key in entityMapping)

      {

        text = text.replace(new RegExp(key, "g"), entityMapping[key]);

      }

      

      // Remove duplicated spaces

      text = text.replace(new RegExp("(\\s){2,}", "g"), " ");

      

      return text;

    }









    Does that mean that what Alexei suggested would not work?



    This will make TestComplete display the message as formatted HTML, but will not strip HTML tags from the message.