Module pdoc provides types and functions for accessing the public
documentation of a Python module. This includes modules (and
sub-modules), functions, classes and module, class and instance
variables. Docstrings are taken from modules, functions and classes
using the special __doc__
attribute. Docstrings for variables are
extracted by examining the module's abstract syntax tree.
The public interface of a module is determined through one of two ways.
If __all__
is defined in the module, then all identifiers in that
list will be considered public. No other identifiers will be considered
as public. Conversely, if __all__
is not defined, then pdoc
will
heuristically determine the public interface. There are three rules that
are applied to each identifier in the module:
- If the name starts with an underscore, it is not public.
- If the name is defined in a different module, it is not public.
- If the name refers to an immediate sub-module, then it is public.
Once documentation for a module is created with pdoc.Module
, it can
be output as either HTML or plain text using the covenience functions
pdoc.html
and pdoc.text
, or the corresponding methods
pdoc.Module.html
and pdoc.Module.text
.
Alternatively, you may run an HTTP server with the pdoc
script
included with this module.
pdoc
has been tested on Python 2.6, 2.7 and 3.3. It seems to work on
all three.
pdoc
is on GitHub. Pull
requests and bug reports are welcome.
In your documentation, you may link to other identifiers in your module or submodules. Linking is automatically done for you whenever you surround an identifier with a back quote (grave). The identifier name must be fully qualified. For example, `pdoc.Doc.docstring` is correct while `Doc.docstring` is incorrect.
If the pdoc
script is used to run an HTTP server, then external
linking to other packages installed is possible. No extra work is
necessary; simply use the fully qualified path. For example,
`nflvid.slice` will create a link to the nflvid.slice
function,
which is not a part of pdoc
at all.
Broadly speaking, pdoc
gets everything you see from introspecting
the module. This includes words describing a particular module, class,
function or variable. While pdoc
does some analysis on the source
code of a module, importing the module itself is necessary to use
Python's introspection features.
In Python, objects like modules, functions, classes and methods have a
special attribute named __doc__
which contains that object's
docstring. The docstring comes from a special placement of a string in
your source code. For example, the following code shows how to define a
function with a docstring and access the contents of that docstring:
#!python >>> def test(): ... '''This is a docstring.''' ... pass ... >>> test.__doc__ 'This is a docstring.'
Something similar can be done for classes and modules too. For classes,
the docstring should come on the line immediately following
class ...
. For modules, the docstring should start on the first line
of the file. These docstrings are what you see for each module, class,
function and method listed in the documentation produced by pdoc
.
The above just about covers standard uses of docstrings in Python.
pdoc
extends the above in a few important ways.
Firstly, docstrings can be inherited. Consider the following code sample:
#!python >>> class A (object): ... def test(): ... '''Docstring for A.''' ... >>> class B (A): ... def test(): ... pass ... >>> print(A.test.__doc__) Docstring for A. >>> print(B.test.__doc__) None
In Python, the docstring for B.test
is empty, even though one was
defined in A.test
. If pdoc
generates documentation for the above
code, then it will automatically attach the docstring for A.test
to
B.test
only if B.test
does not have a docstring. In the default
HTML output, an inherited docstring is grey.
Secondly, docstrings can be attached to variables, which includes module (or global) variables, class variables and instance variables. Python by itself does not allow docstrings to be attached to variables. For example:
#!python variable = "SomeValue" '''Docstring for variable.'''
The resulting variable
will have no __doc__
attribute. To
compensate, pdoc
will read the source code when it's available to
infer a connection between a variable and a docstring. The connection is
only made when an assignment statement is followed by a docstring.
Something similar is done for instance variables as well. By convention,
instance variables are initialized in a class's __init__
method.
Therefore, pdoc
adheres to that convention and looks for docstrings
of variables like so:
#!python def __init__(self): self.variable = "SomeValue" '''Docstring for instance variable.'''
Note that pdoc
only considers attributes defined on self
as
instance variables.
Class and instance variables can also have inherited docstrings.
Thirdly and finally, docstrings can be overridden with a special
__pdoc__
dictionary that pdoc
inspects if it exists. The keys of
__pdoc__
should be identifiers within the scope of the module. (In
the case of an instance variable self.variable
for class A
, its
module identifier would be A.variable
.) The values of __pdoc__
should be docstrings.
This particular feature is useful when there's no feasible way of attaching a docstring to something. A good example of this is a namedtuple:
#!python __pdoc__ = {} Table = namedtuple('Table', ['types', 'names', 'rows']) __pdoc__['Table.types'] = 'Types for each column in the table.' __pdoc__['Table.names'] = 'The names of each column in the table.' __pdoc__['Table.rows'] = 'Lists corresponding to each row in the table.'
pdoc
will then show Table
as a class with documentation for the
types
, names
and rows
members.
Note that assignments to __pdoc__
need to placed where they'll be
executed when the module is imported. For example, at the top level of a
module or in the definition of a class.
If __pdoc__[key] = None
, then key
will not be included in the
public interface of the module.
pdoc
is in the public domain via the
UNLICENSE.