Extracting Data from a TestStand HTML report

[ToddK]

Does anyone know of a utility that allows a TestStand 3.0 formatted HTML report to be parsed to obtain the measurements and limits?  I have a utility program that was written by the engineers at Digalog Systems Inc. which worked with HTML reports from TS 1.0 and 2.0 and allowed the results to be extracted into an Excel file.  The new release of TestStand 3.0 has changed the HTML formatting of the standard report slightly, making the previous utility useless.  I could change the TS "reportgen_html.seq" to match the formatting of previous versions but I have report files that have already been generated and need to extract the data.

[Santiago D]

Unfortunately, I haven't heard of any tool to obtain the measurement and limits from an HTML report created by TestStand.  Does the application from Digalog Systems provide any configuration files?  I wonder if there is a way to tweak the way it parses through the HTML to deal with the differences between the old and new HTML reports.

Of course the other solution would be to write a parser yourself, but I'm not sure if that would be time efficient or would be justified by the number of files you need to parse.

Santiago D

National Instruments

[ToddK]

I was able to obtain the source code from the original utility program and modify the parsing routine to work with the new HTML files. The problem was with the formatting of the Step Name. The parser was looking for the step name to be bolded using the <B></B> syntax which is how the default reports from TS were created in Version 1 & 2.  To use the original program with TS 3.0, I found a way to modify the HTML report to look like previous versions by modifing TestStand's, reportgen_html.seq.  I modified the "Put OneResultInReport" callback to change the default TS3.0 report formatting to look more like previous versions.  I changed the function step "Add Step Name" which creates the HTML line that builds the step name in the report to include the bolding key tags and changed the TS report options to not use the DLL version of modelsupport2.dll to create the reports. With this utility, I can quickly take large reports and convert them into Excel where the results can be evaluated.

[Dennis_Knutson]

Saving test data as HTML is actually a very ineffecient method. They are big text files that contain as much formating information as useable data. You'd be better off using XML with TestStand 3.1 and much, much better off using a database to store the results. I don't know how many testers you're running or the complexity of your tests, but the 64K limit to the number of rows in Excel might also prove to be limiting at some point. I know that in my case, that number of rows is less than a day's worth of data.

[Gemini]

Someone mentioned using the XML output format from TestStand in place of HTML.  My company, VI Technology, develops Arendar Test Data Management software.  Arendar comes with a TestStand XML file parser so that you can merge/compare/analyze data from multiple stations using an SQL Server or Oracle database. There is close to a zero integration effort since the parser not only imports data but also builds the database schema. If the XML file format changes, the importer is intelligent and modifies the schema as required (only additions, no deletions). The schema can be locked at any time.

Once the data is captured, querying/reporting/analyzing is easy and does not require SQL knowledge.

Message Edited by Support on 05-26-2010 01:45 PM

[jacky Wang]

Hi
I'm also trying to have a tool to extract Teststand html report to excel format. At the same time, I have the questions below:
1, As far as I know, Teststand provides only two ways of repor format: web page(html) and txt. I can not understand why
Somebody here has the expression of "You'd be better off using XML with TestStand 3.1 and much, much better off using a database to store the results.” Of course, I admit HTML is a very inefficient method and I have worked out a module embedded into process model to directly created Excel format test report output.
2, I can not find proper way to parse html report, for example, how to get low limit, measured value and high limit. I'm using LabVIEW8.2 and CVI8. This question has puzzled me a long time and appreciate somebody could get me out of the trouble.


Thanks!
Jacky

[Dennis_Knutson]

TestStand also provides the ability to create a report in xml. What version are you using? I'm using 2.0 and generating xml with a sequence I wrote myself. If you use xml, then it's a matter of changing the schema to view it in Excel.

And since you agree that html is ineffecient, why do not understand my previous statement? To get meaningful yield and other process data, you would have to process many different files. My latest TestStand project is a single tester that tests 6 uut's simultaneously. That's about 250 test results in a single shift. By the time a couple of the html files could be read and processed, I've run a single query on the database to provide all of the information anyone could possibly need. I'll take a couple second query over minutes or hours any time. If it's cost that is a concern, try MySQL. TestStand has a schema for it and MySQL is open source so it doesn't cost anything.

[jacky Wang]

Hi Dennis Knutson
Thanks for your reply.
I checked my Teststand again, and found: Teststand3.1 only provides html and txt report output; and Teststand3.5 provides XML,ATML, HTML and txt report ouput. It seems that Teststand3.1 do have no XML output.
As to the effiency for different Teststand report formats,I can not agree with you and think database is still not a good solution to rapidly save test result. I once tried on Microsoft SQL 2000 and found it still could be time-consuming, especially for large test items, besides, it has to depend on the speed of internet. From some kind of views, it could result into low speed or jam of network when there's more than 1 test stations.
In practice, I'm used to insert a DLL in process model and have excel format test result output, while disabling the functions of built-in report and database, which results into easier and more convenient analysis on Six sigma for test.
Sometimes, what I concerns most is cycle time, so I don't like the html format, even it's easy to understand. I always think binary format is faster than txt format. So I do think binary format report could be suppored in future, that's also the reason why now I have the plan to develop TDMS(Diadem test data format) test report output for Teststand.
I do hope NI here could provide an analysis on diferent teststand report formats, from views of effiency,view and so on.
Attached is what I'm thinking: about how to manage test data.

Thanks!
Jacky

[Dennis_Knutson]

If I go to Configure>Report Options and then to the Report Format options, I see XML there in TestStand 3.1. As I said, I'm using it with TestStand 2.0. NI even published an add-on sequence before 3.1 was released. TestStand 3.5 adds ATML.

I guess we'll agree to disagree. I have over 50 testers (local and remote) all logging to the same SQL Server database and don't find it any more time consuming than the creation of the report. Combining the reports from all of those testers would be a nightmare. I don't have any netwrok problems or 'data jams'. I guess it comes down to correct database design and network implmentation.

[Joji]

Hi Todd,

Would you be willing to share your modified parser with me? We have are trying to analyze test results that we have the HTML reports from TestStand. It is TestStand 3.5, if it doesn't work right of the box, I will be willing to modify it to make it work, just didn't want to recreate the wheel. Is the parser done in LabVIEW?