(The good programmer is lazy; instead of typing something over and over, she writes a script to automate the steps. The wizard programmer writes a set of object classes that model the problem at hand, and tells those objects to behave. The genius programmer writes a new language suitable for the problem domain.)<\/p>\n
I wanted to automate getting info about my pictures<\/a> from the soon-to-be-discontinued Sprint Picture Mail website.<\/p>\n (If you want to read the code, here’s the original un-uglified easier-to-read version<\/a>.) This should pop up a new browser window that either lists your Picture Mail albums, or tells you what to do.<\/p>\n Then for each album,<\/p>\n Once you’re logged in to SPM, a bookmarklet can modify the contents of a web page to do something useful.<\/p>\n Googling found a suitable bookmarklet half-way down a page of bookmarklets from 2003<\/a>:<\/p>\n Copy and paste this babble into your location bar and press enter, and it pops up a new window that lists all the URLs on a page!<\/p>\n It’s hard to read because a bookmarklet is a URL, just like http:\/\/example.com\/path<\/tt> but starting with javascript:<\/tt> , so it has to be a single line. First thing to do is to find a tool to convert between compressed rubbish and a decent multi-line program… UglifyJS<\/a>. Yay, more tools in JavaScript, it’s taking over the world.<\/p>\n Now I’ve got a multi-line JavaScript file. To turn this back into a bookmarklet,<\/p>\n I then paste the contents of the resulting single-line JavaScript into my browser’s location bar; to automate this step, pipe it into a clipboard utility such as xsel -b<\/tt>.<\/p>\n Now, to do the coding…<\/p>\n Once logged in, a useful URL to visit is http:\/\/pictures.sprintpcs.com\/ajax\/view\/getAlbumListResults.do?sortCode=5<\/a><\/p>\n this returns a JSON structure of all my picture albums. To parse it, just<\/p>\n this fails! Sprint took the title of one album, 2512 house (DL’d!), and turned it into<\/p>\n but this is invalid JSON, you don’t<\/em> put a backslash in front of single quote.<\/p>\n OK, so<\/p>\n Now iterate through this, for each album fabricate<\/p>\n The first is easy, since my research found the URL that returns the album detail is just \/getMediaContainerJSON.do?componentType=mediaDetail&sortCode=17&containerID=MMMMMMMM<\/em>)<\/p>\n The second is much harder, as SPM seems to fabricate a hidden form that enumerates the elementIDs and extensions of pictures to download, and then dynamically modifies this to turn it into a POST to a \/downloadzip\/sprintpictures.zip?machineid=xxxx<\/em> URL. This is hard to do with a bookmarklet.<\/p>\n Once you’ve saved an album’s details as Album name<\/em>.json\u00a0 you want to walk the structure looking for interesting info and warning about unexpected values. You could do this in another browser bookmarklet, but I want to change the names and modification times of files based on the info, and you can’t perform those operations in browser JavaScript. But the nice node.js<\/a> engine can. The script is here<\/a>.<\/p>\n Getting album and picture metadata is nice, but the most time-consuming part is selecting and dragging albums and pictures to [Download to my PC].<\/p>\n I started writing a bookmarklet that would make SPM’s download form’s hidden form fields visible and label them, so you could then fill it with your own element IDs and drive your own arbitrary download. But you’d have to interactively step in to provide the machine ID, download URL, etc.<\/p>\n It would be better to do it by driving the whole process from node.js. The script takes a login and password, supplies them in a request to some SPM login URL, then remember cookies sent back by SPM and uses them in subsequent requests. Useful links for this: How to POST in node.js<\/a>, logging in with cookie<\/a>.<\/p>\n But I can’t figure out SPM login, it’s a bear. Their login form’s <noscript> action is to call authenticate.jsp, but instead it does an AJAX in-page call to a different login URL, which returns script to check if cookies are enabled and then redirects. I tried both URLs using curl and node.js and even though I get a pmjsessionid cookie back, my subsequent requests fail. SPM must be looking for user agent or a particular sequence of requests, or something else. Help wanted! These kinds of explorations reveal shortcomings in the source material. In an open source read-write web, you can fix them:<\/p>\n (The good programmer is lazy; instead of typing something over and over, she writes a script to automate the steps. The wizard programmer writes a set of object classes that model the problem at hand, and tells those objects to … Continue reading The script so far<\/h2>\n
\n
javascript:var linkWin=open(\"\",\"albums\",\"width=400,height=600,scrollbars,resizable\"),lw=linkWin.document,isJSON=!0;try{albums=JSON.parse(document.body.innerHTML.replace(\/\\\\'\/,\"'\"))}catch(e){isJSON=!1}if(isJSON&&typeof albums==\"object\"&&Array.isArray(albums)&&albums.length>0){baseURL=document.baseURI.match(\/(.*?\\\/\\\/.*?)\\\/\/)[1],lw.write('<base href=\"'+baseURL+'\" target=\"_blank\">\\n'),lw.write(\"<ol>\\n\");for(i=0;i<albums.length;i++){var album=albums[i];lw.write(\"\u00a0 <li>\"+album.title.link(\"\/ui-refresh\/getMediaContainerJSON.do?componentType=mediaDetail&sortCode=17&containerID=\"+album.containerID)+\" - \"+album.coverTitle+\"<\/li>\\n\\n\")}lw.write(\"<\/ol>\\n\")}else lw.write(\"<p>window contents do not appear to be Sprint Picture Mail JSON info<\/p>\\n\"),lw.write('<p>Try visiting <a href=\"http:\/\/pictures.sprintpcs.com\/ajax\/view\/getAlbumListResults.do?sortCode=5\">http:\/\/pictures.sprintpcs.com\/ajax\/view\/getAlbumListResults.do?sortCode=5<\/a><\/p>\\n');<\/pre>\n\n
Bookmarklet background<\/h2>\n
javascript:WN7z=open('','Z6','width=400,height=200,scrollbars,resizable,menubar');\n DL5e=document.links;with(WN7z.document){write('<base\n target=_blank>');for(lKi=0;lKi<DL5e.length;lKi++){write(DL5e[lKi].toString().link(DL5e[lKi])+'<br><br>')};void(close())}<\/pre>\nnpm install uglify-js\nsed 's\/javascript:\/\/' < pcreview_bookmarklet \\\n\u00a0 | .\/node_modules\/uglify-js\/bin\/uglifyjs -b \\\n\u00a0 > spm.js<\/pre>\n
.\/node_modules\/uglify-js\/bin\/uglifyjs spm.js \\\n \u00a0| sed 's\/^\/javascript:\/'<\/pre>\n
Generating an album list<\/h3>\n
albums=JSON.parse(document.body.innerHTML);<\/pre>\n
\"title\": \"2512 house (DL\\'d!)\"<\/pre>\n
albums=JSON.parse(document.body.innerHTML.replace(\/\\\\'\/, \"'\"));<\/pre>\n
\n
Parsing an album’s media details<\/h2>\n
Attempts\u00a0 to trigger downloads<\/h2>\n
\n<\/em><\/p>\nTo Dos<\/h2>\n
\n