I'm happy to sort this out with a small Python script if you email me: 1. A list of categories (or the parent the others sit under) 2. Explain what you want as metadata (a text file of the image page edit history perhaps?) Cheers, Fae On 15 October 2013 15:32, Jonathan Cardy <jonathan.cardy(a)wikimedia.org.uk> wrote: including a this supply 4LT. over -- faewik(a)gmail.com http://j.mp/faewm Personal and confidential, please do not circulate or re-quote. _______________________________________________ Wikimedia UK mailing list wikimediauk-l(a)wikimedia.org http://mail.wikimedia.org/mailman/listinfo/wikimediauk-l WMUK: http://uk.wikimedia.org

I suggest that anyone with topics they would like to cover in a python/pywikipediabot workshop, consider adding to discussion on the event registration talk page, so that Jonathan can pull ideas and expected outcomes together. He's trying to agree a new date for a workshop and I'm thinking of the value of splitting it into a basics session for, say, 2 hours one evening and a more advanced practical session one afternoon (you can then choose to come to one rather than both). I would be happy for this to be either a weekday or a weekend depending on what most people can make. Go to <https://wiki.wikimedia.org.uk/wiki/Python_and_Wikimedia_bots_workshop_Oct_2013> to add your ideas on dates and content of a workshop(s). I have pasted the code for a recursive dump of Wikipedia Takes Chester that I cobbled together before breakfast below, but it's not all that helpful without getting the basics of python modules, pywikipediabot and the Wikimedia API (that it is built on) in your head first. It is badly written, but works, and I can tweak this to be a general multi cat-dump routine with a couple of minutes work. The idea of having a couple of workshops is to give a group of contributors the basic "bot" writing skills and an effective kit-bag of methods to write anything they can imagine, from clever analytical reports to daily house-keeping bots, even if they use fairly poor code to do so ;-) The main problem we have with pywikipediabot is that documentation is poor (for example, I don't think the function "fileUrl()" is documented anywhere, for several months I was using the API directly to do what this function nicely does, as I didn't know it was available, and yet would probably fail in mysterious ways if used on the wrong class of object, such as a category rather than an image page, something that a manual ought to help the user understand). It would be great if those interested in improving the manuals could play around with the various commands and illustrate with example working code (and highlight common errors!). I would hope that the outcome of the workshop would be to achieve some of this, perhaps even laying down a few short demonstration screen-capture videos of what these tools can do, and how to go about setting yourself up to use them. BTW the "unidecode" bit below was hacked on after the dump fell over trying to write façade in a local file name. It neatly transcribes it into "facade", a clever module for handling non-ascii international characters of all sorts. Fae ---- /* The main part of 'batchCatDump.py', treat as CC-BY-SA. This takes all images recursively under Commons category 'catname' and saves the full size image along with the current text of its associated image page in a local directory. In this case it generated 468 image files and the same number of matching html files, taking just under 2GB on a usb stick. */ catname="Wikipedia Takes Chester" cat = catlib.Category(site, catname) gen = pagegenerators.CategorizedPageGenerator(cat,recurse=True) count=0 savedir="//Volumes/Fae_32GB/Wiki/"+catname+"/" if not os.path.exists(savedir): os.makedirs(savedir) for page in gen: title=page.title() if not title[0:5]=="File:": continue count+=1 utitle=unidecode(title[5:]) saveas=savedir+utitle if os.path.exists(saveas): continue if utitle!=title[5:]: print "Transcribing title as", utitle html = page.get() source=page.fileUrl() urllib.urlretrieve(source, saveas) f=open(saveas+".html","w") f.write(unidecode(html)) f.close() -- faewik(a)gmail.com http://j.mp/faewm _______________________________________________ Wikimedia UK mailing list wikimediauk-l(a)wikimedia.org http://mail.wikimedia.org/mailman/listinfo/wikimediauk-l WMUK: http://uk.wikimedia.org

On 18/10/2013, Michael Maggs &lt;Michael(a)maggs.name&gt; wrote: No need for md5 hash calculations :-) In python the call to fileUrl() actually 'translates' the Commons image page name into the URL for the full size image file to download without having to provide any other information. Behind the scenes this actually uses the API to call a "query" returning "imageinfo". You may find a real example of API calls useful. Starting with the image title "Gay.jpg" [1] we can ask the API to tell us its properties by creating a call to API query [2], within the results (which you could ask for as xml or JSON) this returns us the URL link to the full sized image file which you can download. [3] Rather than full size, you can request a particular size, such as with a width of 100px [4]. In my script to query tineye.com to check for matches to mobile images as possible copyright violations, I only used thumbnails of width 300px, it saved a lot of bandwidth. :-) 1. http://commons.wikimedia.org/wiki/File:Gay.jpg 2. https://commons.wikimedia.org/w/api.php?action=query&titles=File:Gay.jp… 3. https://upload.wikimedia.org/wikipedia/commons/9/90/Gay.jpg 4. https://commons.wikimedia.org/w/api.php?action=query&titles=File:Gay.jp… I would imagine that an introductory workshop should cover the basics of how to use the API and how to find parameters in the manual, this would be particularly useful for folks more comfortable programming in other languages that Python, or using other tools, and would still use all the features of the API to do interesting stuff. Fae -- faewik(a)gmail.com http://j.mp/faewm