A couple of days ago in the "web3py important!" thread I posted some thoughts on URL route matching syntax for the @expose(....) method, but my post is yet to surface. Posting a more-considered version here for consideration and feedback.
I've been looking into some other frameworks, and Alloy stands out to me (http://alloyframework.org/manual/url-router/), it's the main inspiration for the syntax below. (Also looked at Flask, Silex, Codeignighter, Django and considered trade offs each one brings) In a URL route lets say that '/' delimits segments, and anything between matching '<' and '>' pairs is a segment pattern. Each segment may contain either static text or a segment pattern. Segment patterns cannot span '/' separators. 1. Suggest use symbols to define string and integer types in segment patterns. ":" = "str", "#" = int (are other types necessary?). So ':' is no longer a delimiter. @expose('/index/<:company>/<#ssd>') def index(company, ssd): '''company is a string, ssd is an int''' 2. Up to one question mark may be included immediately before any '/' to indicate the remaining segment patterns are all optional (and consequently map to non-required attributes which may have defaults). By specifying a single point in the route beyond which matching segments are optional, it's harder to write ambiguous route definitions. @expose('/index/<:company>/<#ssd>?/<#favourite>') def index(company, sad, favourite=None): @expose('index/<:company>/<#ssd>?/<#favourite>/<#last_visited_article>') def index(company, ssd, favourite=None, last_visited_article=1): 3. Perhaps use "*" as wildcard to match text (including '/') but there may be good security reasons not to allow? The constraint could be that no segment patterns may follow. Also only one per route. Always a string. @expose('index/<:company>/<#ssd>?/<*rest_of_url>') def index(company, ssd, part_of_url='not provided'): @expose('index/<:company>/<#ssd>/<*part_of_url>/some-static-text') def index(company, ssd, part_of_url): 4. Regex. Must match whole segments - i.e. shouldn't be able to split a section into multiple parameters with a regex. Keep type - so regex match is still passed through as string or int. (E.g. the 4-digit year is passed through as 'int' here.). It's not really a validator but could be abused that way. @expose('index/<:company|[A-Za-z ]{3,20}>/<#year|[0-9]{4}>') def index(company, year): Cheers Rob On Tuesday, November 27, 2012 1:39:06 AM UTC-3, User wrote: > > I noticed a thread over in web2py-developers web3py - > important!<https://groups.google.com/forum/?fromgroups=#!topic/web2py-developers/RCeiRd3Rzs0> > which > was exciting to read. I've flirted with web2py and there's a lot that I > like about it. For some reason I find web2py exciting whereas django > doesn't provide that. I've used Yii on the php side which is great > framework as far as php goes and asp.net mvc which is great as well. I'd > love to work with python but the main thing making me hesitate with web2py > is critical mass. > > It seems like it wouldn't be hard for web2py to really dominate the python > web framework space if some of the core criticisms were addressed. I'm not > fully up to speed on what they are but I usually hear about unit testing > and global variables. It feels like there is a roadblock preventing the > project from skyrocketing. Python needs a rails. I understand that the > design decisions are by choice with pros and cons. > > My questions are: > 1. Will web3py likely address these often repeated core criticisms? (I saw > point 5 from the thread linked to above: "5) No more global environment. > Apps will do "from web3py import *" (see below)") > 2. The developer thread is over in the developers section. Will you have > a more open forum for users (as opposed to developers) to have input on > web3py? > > > --
