A clone of WWW::Mechanize for Erlang, with perhaps a bit of Test::WWW::Mechanize and Webtest for good measure. Named according to the seeming defacto convention of inserting "erl" in silly places in the name. This is primarily a convenience wrapper around httpc.
erl -pa ebin -boot start_sasl -s inets -s xmerl -s mechanizerl
1> Mech = mechanizerl:new().
{mechanizerl,<0.64.0>}
2> Mech:get("http://google.com/").
ok
3> Mech:submit_form("search", [{q, "mechanizerl"}]).
ok
4> Mech:body().
<<"<html>...">>- The interface isn't as Erlangy as it could be (it uses the lowly tuple calls), as I tried to follow the WWW::Mechanize API closely to facilitate translating old tests I had written in perl into Erlang. Part of the efficacy of Mechanize's interface is its statefulness.
- This uses xmerl to parse the HTML it receives. This works for me, since I tend to write HTML in an XHTML-ish fashion, but be warned that it will choke on lots of other HTML. I considered using one of the HTML parsers from YAWS or Mochiweb, but it seemed unnecessary for my purposes.
- Strings are used throughout rather than binaries, for convenience rather than efficiency.
- I've only implemented so far the subset of WWW::Mechanize's functionality that I actually use. For example, cookies are always enabled. Patches Thoughtfully Considered.
- Lists are used in numerous places where more sophisticated data structures would be more efficient. Making the simplest thing that would work trumped efficiency in this first version.
- Enforce constraints on fields (select options, maxlength, et cetera);
- More extensive testing: replay-based testing of redirects, et cetera;
- File upload;
- Examples and documentation.
Julian Squires <[email protected]>
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent