API Reference

The Launch Function

launch(fsroot=None, urlroot='/', multithread=True, logger=None, encoding='utf-8', static=False)

Launch and return a TinCan webapp object. Returns a tuple containing two items: an instance of tincan.TinCan and an error count. This function does not run or serve the webapp; it is the caller’s responsibility to perform one of the latter operations. This function accepts the following keyword arguments:

fsroot

Path to the root directory of this webapp on the filesystem. If no path is specified, the current working directory will be used.

urlroot

This defines a directory prefix that is added to all routes. It is mostly for use by the launch shell command; when running under WSGI this should almost always be left to default to /.

multithread

By default, TinCan assumes that multiple threads per process are being used. Setting this parameter to False will result in code that fails to multithread properly, but which runs more efficiently while not multithreading. It is recommended to set this to False if you are not going to be using multithreading.

logger

If set, the name of the Python logging.Logger object to use for logging. If not set, no messages will be logged.

encoding

Character set used by all text files in the webapp. Note that this is different from the character set that TinCan will use to serve responses; the latter is inevitably UTF-8.

static

Whether or not to create routes to serve static content in the webapp. If you’re using the built-in server, this should probably be True. If you’re running via WSGI, it should probably be False.

The TinCan Object

This is a subclass of bottle.Bottle. It is not intended for end users to create instances of this object themselves; use the launch function above to do that.

This class contains one extra method above and beyond the methods bottle.Bottle objects have:

forward(target)

Perform a programmatic, request-time, server-side redirect to the route specified by target, which may be either relative to the route issuing the redirect or an absolute route. This differs from the #forward header directive in that this method causes a server-side redirect to be set up the time a request is processed, not the when a route is created. That makes this method both more flexible and less efficient than #forward.

One may only forward from a normal page to another normal page; neither the source nor the target of a forward may be an error page. Attempts to create forward loops will also cause an error.

TinCan Configuration Variables

The config dictionary of a TinCan object contains tincan.fsroot, tincan.urlroot, tincan.logger, and tincan.encoding keys (in addition to the standard ones defined by WSGI and Bottle); these contain the values of the corresponding parameters passed to the launch function that created this webapp. In general, any key starting with tincan. is reserved for use by TinCan.

Page, BasePage, and ErrorPage

These classes are typically subclassed in the code-behind for a page. Any page without code-behind will get a bare Page or ErrorPage object associated with it, as appropriate.

class tincan.BasePage

This is the parent class of all code-behind classes. All such classes contain two standard methods:

handle()

The handle method is called by TinCan to handle a request. No arguments are passed to it (other than the implied self argument). By default, this is a no-op method.

export()

This is called by TinCan to export template variables; it must return a dict or dict-like object.

The default implementation exports all non-hidden instance variables; moreover, it does not export attributes that define callable objects (e.g. methods). A “hidden” instance variable means any one whose name does not start with an underscore; the request and response instance variables of class tincan.Page are also considered hidden. Finally, self is exported as the page variable.

Note that the exporting happens after header processing; thus, if there is a conflict between a template variable defined by the #load header directive and this exporting logic, the value exported by this method will always overwrite the earlier one.

If the above exporting behavior is for some reason unsuitable, it is permitted to override this method.

class tincan.Page

The parent class of all normal (non-error) pages.

class tincan.ErrorPage

The parent class of all error pages.

Exceptions

class tincan.TinCanException

The parent class of all exceptions TinCan raises.

class tincan.TemplateHeaderError

Raised upon encountering a syntax error in the template header directives.

class tincan.LoadError

Raised when we run into problems #load’ing something, usually because it doesn’t exist.

class tincan.TinCanError

General-purpose exception raised by TinCan when things go wrong, either when attempting to launch webapps, or attempting to service requests. Often caused by errors in user-created files.

Request and Response Objects

The request and response instance variables of tincan.Page are standard bottle.Request and bottle.Response objects. In the environ attribute of the request object, any key beginning with tincan. is reserved.

Header Directives

A .pspx file is a standard Chameleon template, plus a set of optional header directives that may be present to override the default behavior of the created page or its associated route.

#end

Marks the last line of the headers. This is not currently necessary, as headers are implicitly ended by the first line which does not start with a leading “#”. This directive is mainly here to facilitate future support of alternate templating engines.

#errors

This is an error page which handles the specified error codes; any code-behind associated with this page must be based on the tincan.ErrorPage class.

#forward

Ignore everything else in this template (and any code-behind associated with it), using the specified route to serve it instead. The route specified with #forward may itself contain a #forward, but attempts to create a #forward loop are not allowed and will cause a TinCanError to be raised at initialization time.

#hidden

This is a hidden page; do not create a route for it. The page can only be displayed by a server-side forward.

#load

Load the specified Chameleon template file and make the loaded template available as a template variable. Useful for importing and invoking macros. See Loading Templates below for more information.

#methods

A list of HTTP request methods, separated by whitespace, follows. The route will allow all specified methods. Not specifying this line is equivalent to specifying #methods GET.

#python

What follows is the name of the Python file containing the code-behind for this route; the file name must end in .py.

#template

Ignore the body of this file and instead use the template in the body of the specified file, which must end in .pspx. Any headers in the referred template file are ignored.

Error Pages

Error pages supersede the standard Bottle error handling, and are created by using the #errors page header. The #hidden and #method header directives are ignored in error pages (error pages are effectively hidden anyhow, by virtue of never having normal routes created for them).

The #errors directive takes a list of numeric error codes (values from 400 to 599 are allowed), separated by spaces; the page is created to handle the specified errors. If no error codes are specified, the page will handle all errors. The behavior of specifying multiple error pages for the same error code is undefined; doing so is best avoided.

Templates with No Explicit Code-Behind

Code-behind is optional for both normal and error page templates. If code-behind is not provided, TinCan will use the Page or ErrorPage class as appropriate.

Loading Templates

The #load directive may be used to load additional templates, e.g. ones containing macro definitions. Note that the loaded files are standard Chameleon templates and not TinCan .pspx files (i.e. they cannot contain any header directives); as such, loaded files must have the standard .pt extension for Chameleon template files.

In the normal case, #load foo.pt will load a file relative to the same directory as the page containing the #load directive itself. The loaded template object will be made available as a template variable matching the file name sans extension (e.g. foo). One can change the name of the variable created by prefixing the file specification with a variable name followed by an equals sign, e.g. #load t=foo.pt. If one places the specification inside angle brackets (e.g. #load <t=foo.pt>), files to be loaded are searched for in WEB-INF/tlib instead.

Finally, as is allowed for all arguments to header directives, one may enclose the argument to #load inside single or double quotes and use the normal Python string syntax.

Using Loaded Macros

Once a template has been loaded, it will be available as a sub-attribute of the macros attribute of the associated template object. E.g.:

#load foo.pt
<!DOCTYPE html>
<html>
  <head>
    <title>Macro Example</title>
  </head><body>
    <p metal:use-macro="foo.macros.bar"></p>
  </body>
</html>

Chameleon Templates

TinCan templates are Chameleon templates, as documented here. Note that the templates in .pspx files are processed as strings, so load: expressions will not work in them (hence the #load header directive). Templates loaded via #load are processed as files, so the load: expression will work from within them.

TinCan provides chameleon_template and chameleon_view callables, which follow the standard Bottle conventions for templating. Thus, if it is desired to define routes the “Bottle way”, the following import line will make Chameleon the default templating engine:

from tincan import chameleon_view as view, chameleon_template as template