Hey, folks. So as part of the possible move to MozTrap (or, really, it'd be useful for any other TCMS), I've come up with a truly hideous bash script to discover and extract test cases from the wiki. Here it is, in all its glory: #!/bin/bash # Requires https://github.com/dominictarr/JSON.sh available as json.sh # pageid of Category:Test Cases main_cat_id=15824 testcase_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$main_cat_id&cmtype=page&cmlimit=200&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) sub1_cat_ids=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$main_cat_id&cmtype=subcat&cmlimit=200&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) echo "sub1_cat_ids are:" printf -- '%s\n' "${sub1_cat_ids[@]}" for i in "${sub1_cat_ids[@]}" do sub2_cat_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$i&cmtype=subcat&cmlimit=100&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) testcase_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$i&cmtype=page&cmlimit=100&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) done echo "sub2_cat_ids are:" printf -- '%s\n' "${sub2_cat_ids[@]}" for i in "${sub2_cat_ids[@]}" do sub3_cat_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$i&cmtype=subcat&cmlimit=100&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) testcase_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$i&cmtype=page&cmlimit=100&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) done echo "sub3_cat_ids are:" printf -- '%s\n' "${sub3_cat_ids[@]}" for i in "${sub3_cat_ids[@]}" do sub4_cat_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$i&cmtype=subcat&cmlimit=100&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) testcase_ids+=(`curl -s "https://fedoraproject.org/w/api.php?action=query&list=categorymembers&cmpageid=$i&cmtype=page&cmlimit=100&cmprop=ids&redirects&format=json"; | json.sh -b | cut -f2`) done echo "sub4_cat_ids are:" printf -- '%s\n' "${sub4_cat_ids[@]}" echo "testcase_ids are:" printf -- '%s\n' "${testcase_ids[@]}" mkdir -p /home/adamw/local/test_cases for i in "${testcase_ids[@]}" do echo "Page title: `curl -s "https://fedoraproject.org/w/api.php?action=query&pageids=$i&prop=info&format=json"; | json.sh -b | grep title | cut -f2`" > /home/adamw/local/test_cases/$i.mw echo "Test case contents:" >> /home/adamw/local/test_cases/$i.mw curl -s "https://fedoraproject.org/w/index.php?curid=$i&action=raw"; >> /home/adamw/local/test_cases/$i.mw done ...I'm not proud. Yes, this is asking a PHP webapp to produce JSON then parsing it using bash, which is one of the dumbest things you'll see all week, but I suck at writing PHP from scratch. (Yes, yes, I also suck at writing shell scripts, thanks.) You need https://github.com/dominictarr/JSON.sh , and you also probably want to change the hardcoded directory in the second to last line. This whacks the mediawiki API pretty hard, so probably no point running it unless you want to improve it or something; I have it running ATM and dumping all ~900 test cases in the wiki out to my system, once it's done I'll tar them and stick them up somewhere for others to access. Next step would be to write something to convert them to MozTrap's mass import format. It stops at four levels of nesting because that seems to be all we have; sub4_cat_ids comes up empty. mediawiki's API doesn't have any way to say 'give me all the pages in this category and all its subcategories' - no way you can ask it to solve the nesting problem for you - so you pretty much have to do something like this. The 'progress reports' aren't really necessary, they were just to reassure me the thing was working properly as I went along. There are all kinds of potential weaknesses and bugs in the approach, but for the actual set of categories and test cases we have in the Fedora wiki, it seems to work, and that's really all I cared about. It also produces some dupes - there are ~150 test cases in multiple categories...but I figured it's just as easy to wait for the dupes to get downloaded twice as it would be to spend time fixing the script to filter them out before doing the download step. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net http://www.happyassassin.net -- test mailing list test@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
