Add numpy-sphinx docs extensions.

This commit is contained in:
Stefan van der Walt
2009-09-24 12:00:46 +02:00
parent 67e5f0b7b8
commit 92f7b0939b
3 changed files with 908 additions and 0 deletions
+497
View File
@@ -0,0 +1,497 @@
"""Extract reference documentation from the NumPy source tree.
"""
import inspect
import textwrap
import re
import pydoc
from StringIO import StringIO
from warnings import warn
class Reader(object):
"""A line-based string reader.
"""
def __init__(self, data):
"""
Parameters
----------
data : str
String with lines separated by '\n'.
"""
if isinstance(data,list):
self._str = data
else:
self._str = data.split('\n') # store string as list of lines
self.reset()
def __getitem__(self, n):
return self._str[n]
def reset(self):
self._l = 0 # current line nr
def read(self):
if not self.eof():
out = self[self._l]
self._l += 1
return out
else:
return ''
def seek_next_non_empty_line(self):
for l in self[self._l:]:
if l.strip():
break
else:
self._l += 1
def eof(self):
return self._l >= len(self._str)
def read_to_condition(self, condition_func):
start = self._l
for line in self[start:]:
if condition_func(line):
return self[start:self._l]
self._l += 1
if self.eof():
return self[start:self._l+1]
return []
def read_to_next_empty_line(self):
self.seek_next_non_empty_line()
def is_empty(line):
return not line.strip()
return self.read_to_condition(is_empty)
def read_to_next_unindented_line(self):
def is_unindented(line):
return (line.strip() and (len(line.lstrip()) == len(line)))
return self.read_to_condition(is_unindented)
def peek(self,n=0):
if self._l + n < len(self._str):
return self[self._l + n]
else:
return ''
def is_empty(self):
return not ''.join(self._str).strip()
class NumpyDocString(object):
def __init__(self,docstring):
docstring = textwrap.dedent(docstring).split('\n')
self._doc = Reader(docstring)
self._parsed_data = {
'Signature': '',
'Summary': [''],
'Extended Summary': [],
'Parameters': [],
'Returns': [],
'Raises': [],
'Warns': [],
'Other Parameters': [],
'Attributes': [],
'Methods': [],
'See Also': [],
'Notes': [],
'Warnings': [],
'References': '',
'Examples': '',
'index': {}
}
self._parse()
def __getitem__(self,key):
return self._parsed_data[key]
def __setitem__(self,key,val):
if not self._parsed_data.has_key(key):
warn("Unknown section %s" % key)
else:
self._parsed_data[key] = val
def _is_at_section(self):
self._doc.seek_next_non_empty_line()
if self._doc.eof():
return False
l1 = self._doc.peek().strip() # e.g. Parameters
if l1.startswith('.. index::'):
return True
l2 = self._doc.peek(1).strip() # ---------- or ==========
return l2.startswith('-'*len(l1)) or l2.startswith('='*len(l1))
def _strip(self,doc):
i = 0
j = 0
for i,line in enumerate(doc):
if line.strip(): break
for j,line in enumerate(doc[::-1]):
if line.strip(): break
return doc[i:len(doc)-j]
def _read_to_next_section(self):
section = self._doc.read_to_next_empty_line()
while not self._is_at_section() and not self._doc.eof():
if not self._doc.peek(-1).strip(): # previous line was empty
section += ['']
section += self._doc.read_to_next_empty_line()
return section
def _read_sections(self):
while not self._doc.eof():
data = self._read_to_next_section()
name = data[0].strip()
if name.startswith('..'): # index section
yield name, data[1:]
elif len(data) < 2:
yield StopIteration
else:
yield name, self._strip(data[2:])
def _parse_param_list(self,content):
r = Reader(content)
params = []
while not r.eof():
header = r.read().strip()
if ' : ' in header:
arg_name, arg_type = header.split(' : ')[:2]
else:
arg_name, arg_type = header, ''
desc = r.read_to_next_unindented_line()
desc = dedent_lines(desc)
params.append((arg_name,arg_type,desc))
return params
_name_rgx = re.compile(r"^\s*(:(?P<role>\w+):`(?P<name>[a-zA-Z0-9_.-]+)`|"
r" (?P<name2>[a-zA-Z0-9_.-]+))\s*", re.X)
def _parse_see_also(self, content):
"""
func_name : Descriptive text
continued text
another_func_name : Descriptive text
func_name1, func_name2, :meth:`func_name`, func_name3
"""
items = []
def parse_item_name(text):
"""Match ':role:`name`' or 'name'"""
m = self._name_rgx.match(text)
if m:
g = m.groups()
if g[1] is None:
return g[3], None
else:
return g[2], g[1]
raise ValueError("%s is not a item name" % text)
def push_item(name, rest):
if not name:
return
name, role = parse_item_name(name)
items.append((name, list(rest), role))
del rest[:]
current_func = None
rest = []
for line in content:
if not line.strip(): continue
m = self._name_rgx.match(line)
if m and line[m.end():].strip().startswith(':'):
push_item(current_func, rest)
current_func, line = line[:m.end()], line[m.end():]
rest = [line.split(':', 1)[1].strip()]
if not rest[0]:
rest = []
elif not line.startswith(' '):
push_item(current_func, rest)
current_func = None
if ',' in line:
for func in line.split(','):
push_item(func, [])
elif line.strip():
current_func = line
elif current_func is not None:
rest.append(line.strip())
push_item(current_func, rest)
return items
def _parse_index(self, section, content):
"""
.. index: default
:refguide: something, else, and more
"""
def strip_each_in(lst):
return [s.strip() for s in lst]
out = {}
section = section.split('::')
if len(section) > 1:
out['default'] = strip_each_in(section[1].split(','))[0]
for line in content:
line = line.split(':')
if len(line) > 2:
out[line[1]] = strip_each_in(line[2].split(','))
return out
def _parse_summary(self):
"""Grab signature (if given) and summary"""
if self._is_at_section():
return
summary = self._doc.read_to_next_empty_line()
summary_str = " ".join([s.strip() for s in summary]).strip()
if re.compile('^([\w., ]+=)?\s*[\w\.]+\(.*\)$').match(summary_str):
self['Signature'] = summary_str
if not self._is_at_section():
self['Summary'] = self._doc.read_to_next_empty_line()
else:
self['Summary'] = summary
if not self._is_at_section():
self['Extended Summary'] = self._read_to_next_section()
def _parse(self):
self._doc.reset()
self._parse_summary()
for (section,content) in self._read_sections():
if not section.startswith('..'):
section = ' '.join([s.capitalize() for s in section.split(' ')])
if section in ('Parameters', 'Attributes', 'Methods',
'Returns', 'Raises', 'Warns'):
self[section] = self._parse_param_list(content)
elif section.startswith('.. index::'):
self['index'] = self._parse_index(section, content)
elif section == 'See Also':
self['See Also'] = self._parse_see_also(content)
else:
self[section] = content
# string conversion routines
def _str_header(self, name, symbol='-'):
return [name, len(name)*symbol]
def _str_indent(self, doc, indent=4):
out = []
for line in doc:
out += [' '*indent + line]
return out
def _str_signature(self):
if self['Signature']:
return [self['Signature'].replace('*','\*')] + ['']
else:
return ['']
def _str_summary(self):
if self['Summary']:
return self['Summary'] + ['']
else:
return []
def _str_extended_summary(self):
if self['Extended Summary']:
return self['Extended Summary'] + ['']
else:
return []
def _str_param_list(self, name):
out = []
if self[name]:
out += self._str_header(name)
for param,param_type,desc in self[name]:
out += ['%s : %s' % (param, param_type)]
out += self._str_indent(desc)
out += ['']
return out
def _str_section(self, name):
out = []
if self[name]:
out += self._str_header(name)
out += self[name]
out += ['']
return out
def _str_see_also(self, func_role):
if not self['See Also']: return []
out = []
out += self._str_header("See Also")
last_had_desc = True
for func, desc, role in self['See Also']:
if role:
link = ':%s:`%s`' % (role, func)
elif func_role:
link = ':%s:`%s`' % (func_role, func)
else:
link = "`%s`_" % func
if desc or last_had_desc:
out += ['']
out += [link]
else:
out[-1] += ", %s" % link
if desc:
out += self._str_indent([' '.join(desc)])
last_had_desc = True
else:
last_had_desc = False
out += ['']
return out
def _str_index(self):
idx = self['index']
out = []
out += ['.. index:: %s' % idx.get('default','')]
for section, references in idx.iteritems():
if section == 'default':
continue
out += [' :%s: %s' % (section, ', '.join(references))]
return out
def __str__(self, func_role=''):
out = []
out += self._str_signature()
out += self._str_summary()
out += self._str_extended_summary()
for param_list in ('Parameters','Returns','Raises'):
out += self._str_param_list(param_list)
out += self._str_section('Warnings')
out += self._str_see_also(func_role)
for s in ('Notes','References','Examples'):
out += self._str_section(s)
for param_list in ('Attributes', 'Methods'):
out += self._str_param_list(param_list)
out += self._str_index()
return '\n'.join(out)
def indent(str,indent=4):
indent_str = ' '*indent
if str is None:
return indent_str
lines = str.split('\n')
return '\n'.join(indent_str + l for l in lines)
def dedent_lines(lines):
"""Deindent a list of lines maximally"""
return textwrap.dedent("\n".join(lines)).split("\n")
def header(text, style='-'):
return text + '\n' + style*len(text) + '\n'
class FunctionDoc(NumpyDocString):
def __init__(self, func, role='func', doc=None):
self._f = func
self._role = role # e.g. "func" or "meth"
if doc is None:
doc = inspect.getdoc(func) or ''
try:
NumpyDocString.__init__(self, doc)
except ValueError, e:
print '*'*78
print "ERROR: '%s' while parsing `%s`" % (e, self._f)
print '*'*78
#print "Docstring follows:"
#print doclines
#print '='*78
if not self['Signature']:
func, func_name = self.get_func()
try:
# try to read signature
argspec = inspect.getargspec(func)
argspec = inspect.formatargspec(*argspec)
argspec = argspec.replace('*','\*')
signature = '%s%s' % (func_name, argspec)
except TypeError, e:
signature = '%s()' % func_name
self['Signature'] = signature
def get_func(self):
func_name = getattr(self._f, '__name__', self.__class__.__name__)
if inspect.isclass(self._f):
func = getattr(self._f, '__call__', self._f.__init__)
else:
func = self._f
return func, func_name
def __str__(self):
out = ''
func, func_name = self.get_func()
signature = self['Signature'].replace('*', '\*')
roles = {'func': 'function',
'meth': 'method'}
if self._role:
if not roles.has_key(self._role):
print "Warning: invalid role %s" % self._role
out += '.. %s:: %s\n \n\n' % (roles.get(self._role,''),
func_name)
out += super(FunctionDoc, self).__str__(func_role=self._role)
return out
class ClassDoc(NumpyDocString):
def __init__(self,cls,modulename='',func_doc=FunctionDoc,doc=None):
if not inspect.isclass(cls):
raise ValueError("Initialise using a class. Got %r" % cls)
self._cls = cls
if modulename and not modulename.endswith('.'):
modulename += '.'
self._mod = modulename
self._name = cls.__name__
self._func_doc = func_doc
if doc is None:
doc = pydoc.getdoc(cls)
NumpyDocString.__init__(self, doc)
if not self['Methods']:
self['Methods'] = [(name, '', '') for name in sorted(self.methods)]
if not self['Attributes']:
self['Attributes'] = [(name, '', '')
for name in sorted(self.properties)]
@property
def methods(self):
return [name for name,func in inspect.getmembers(self._cls)
if not name.startswith('_') and callable(func)]
@property
def properties(self):
return [name for name,func in inspect.getmembers(self._cls)
if not name.startswith('_') and func is None]
+219
View File
@@ -0,0 +1,219 @@
import re, inspect, textwrap, pydoc
import sphinx
from docscrape import NumpyDocString, FunctionDoc, ClassDoc
class SphinxDocString(NumpyDocString):
use_plots = False
# string conversion routines
def _str_header(self, name, symbol='`'):
return ['.. rubric:: ' + name, '']
def _str_field_list(self, name):
return [':' + name + ':']
def _str_indent(self, doc, indent=4):
out = []
for line in doc:
out += [' '*indent + line]
return out
def _str_signature(self):
return ['']
if self['Signature']:
return ['``%s``' % self['Signature']] + ['']
else:
return ['']
def _str_summary(self):
return self['Summary'] + ['']
def _str_extended_summary(self):
return self['Extended Summary'] + ['']
def _str_param_list(self, name):
out = []
if self[name]:
out += self._str_field_list(name)
out += ['']
for param,param_type,desc in self[name]:
out += self._str_indent(['**%s** : %s' % (param.strip(),
param_type)])
out += ['']
out += self._str_indent(desc,8)
out += ['']
return out
@property
def _obj(self):
if hasattr(self, '_cls'):
return self._cls
elif hasattr(self, '_f'):
return self._f
return None
def _str_member_list(self, name):
"""
Generate a member listing, autosummary:: table where possible,
and a table where not.
"""
out = []
if self[name]:
out += ['.. rubric:: %s' % name, '']
prefix = getattr(self, '_name', '')
if prefix:
prefix = '~%s.' % prefix
autosum = []
others = []
for param, param_type, desc in self[name]:
param = param.strip()
if not self._obj or hasattr(self._obj, param):
autosum += [" %s%s" % (prefix, param)]
else:
others.append((param, param_type, desc))
if autosum:
out += ['.. autosummary::', ' :toctree:', '']
out += autosum
if others:
maxlen_0 = max([len(x[0]) for x in others])
maxlen_1 = max([len(x[1]) for x in others])
hdr = "="*maxlen_0 + " " + "="*maxlen_1 + " " + "="*10
fmt = '%%%ds %%%ds ' % (maxlen_0, maxlen_1)
n_indent = maxlen_0 + maxlen_1 + 4
out += [hdr]
for param, param_type, desc in others:
out += [fmt % (param.strip(), param_type)]
out += self._str_indent(desc, n_indent)
out += [hdr]
out += ['']
return out
def _str_section(self, name):
out = []
if self[name]:
out += self._str_header(name)
out += ['']
content = textwrap.dedent("\n".join(self[name])).split("\n")
out += content
out += ['']
return out
def _str_see_also(self, func_role):
out = []
if self['See Also']:
see_also = super(SphinxDocString, self)._str_see_also(func_role)
out = ['.. seealso::', '']
out += self._str_indent(see_also[2:])
return out
def _str_warnings(self):
out = []
if self['Warnings']:
out = ['.. warning::', '']
out += self._str_indent(self['Warnings'])
return out
def _str_index(self):
idx = self['index']
out = []
if len(idx) == 0:
return out
out += ['.. index:: %s' % idx.get('default','')]
for section, references in idx.iteritems():
if section == 'default':
continue
elif section == 'refguide':
out += [' single: %s' % (', '.join(references))]
else:
out += [' %s: %s' % (section, ','.join(references))]
return out
def _str_references(self):
out = []
if self['References']:
out += self._str_header('References')
if isinstance(self['References'], str):
self['References'] = [self['References']]
out.extend(self['References'])
out += ['']
# Latex collects all references to a separate bibliography,
# so we need to insert links to it
if sphinx.__version__ >= "0.6":
out += ['.. only:: latex','']
else:
out += ['.. latexonly::','']
items = []
for line in self['References']:
m = re.match(r'.. \[([a-z0-9._-]+)\]', line, re.I)
if m:
items.append(m.group(1))
out += [' ' + ", ".join(["[%s]_" % item for item in items]), '']
return out
def _str_examples(self):
examples_str = "\n".join(self['Examples'])
if (self.use_plots and 'import matplotlib' in examples_str
and 'plot::' not in examples_str):
out = []
out += self._str_header('Examples')
out += ['.. plot::', '']
out += self._str_indent(self['Examples'])
out += ['']
return out
else:
return self._str_section('Examples')
def __str__(self, indent=0, func_role="obj"):
out = []
out += self._str_signature()
out += self._str_index() + ['']
out += self._str_summary()
out += self._str_extended_summary()
for param_list in ('Parameters', 'Returns', 'Raises'):
out += self._str_param_list(param_list)
out += self._str_warnings()
out += self._str_see_also(func_role)
out += self._str_section('Notes')
out += self._str_references()
out += self._str_examples()
for param_list in ('Attributes', 'Methods'):
out += self._str_member_list(param_list)
out = self._str_indent(out,indent)
return '\n'.join(out)
class SphinxFunctionDoc(SphinxDocString, FunctionDoc):
pass
class SphinxClassDoc(SphinxDocString, ClassDoc):
pass
class SphinxObjDoc(SphinxDocString):
def __init__(self, obj, doc):
self._f = obj
SphinxDocString.__init__(self, doc)
def get_doc_object(obj, what=None, doc=None):
if what is None:
if inspect.isclass(obj):
what = 'class'
elif inspect.ismodule(obj):
what = 'module'
elif callable(obj):
what = 'function'
else:
what = 'object'
if what == 'class':
return SphinxClassDoc(obj, '', func_doc=SphinxFunctionDoc, doc=doc)
elif what in ('function', 'method'):
return SphinxFunctionDoc(obj, '', doc=doc)
else:
if doc is None:
doc = pydoc.getdoc(obj)
return SphinxObjDoc(obj, doc)
+192
View File
@@ -0,0 +1,192 @@
"""
========
numpydoc
========
Sphinx extension that handles docstrings in the Numpy standard format. [1]
It will:
- Convert Parameters etc. sections to field lists.
- Convert See Also section to a See also entry.
- Renumber references.
- Extract the signature from the docstring, if it can't be determined otherwise.
.. [1] http://projects.scipy.org/numpy/wiki/CodingStyleGuidelines#docstring-standard
"""
import os, re, pydoc
from docscrape_sphinx import get_doc_object, SphinxDocString
from sphinx.util.compat import Directive
import inspect
def mangle_docstrings(app, what, name, obj, options, lines,
reference_offset=[0]):
if what == 'module':
# Strip top title
title_re = re.compile(ur'^\s*[#*=]{4,}\n[a-z0-9 -]+\n[#*=]{4,}\s*',
re.I|re.S)
lines[:] = title_re.sub(u'', u"\n".join(lines)).split(u"\n")
else:
doc = get_doc_object(obj, what, u"\n".join(lines))
doc.use_plots = app.config.numpydoc_use_plots
lines[:] = unicode(doc).split(u"\n")
if app.config.numpydoc_edit_link and hasattr(obj, '__name__') and \
obj.__name__:
if hasattr(obj, '__module__'):
v = dict(full_name=u"%s.%s" % (obj.__module__, obj.__name__))
else:
v = dict(full_name=obj.__name__)
lines += [u'', u'.. htmlonly::', '']
lines += [u' %s' % x for x in
(app.config.numpydoc_edit_link % v).split("\n")]
# replace reference numbers so that there are no duplicates
references = []
for line in lines:
line = line.strip()
m = re.match(ur'^.. \[([a-z0-9_.-])\]', line, re.I)
if m:
references.append(m.group(1))
# start renaming from the longest string, to avoid overwriting parts
references.sort(key=lambda x: -len(x))
if references:
for i, line in enumerate(lines):
for r in references:
if re.match(ur'^\d+$', r):
new_r = u"R%d" % (reference_offset[0] + int(r))
else:
new_r = u"%s%d" % (r, reference_offset[0])
lines[i] = lines[i].replace(u'[%s]_' % r,
u'[%s]_' % new_r)
lines[i] = lines[i].replace(u'.. [%s]' % r,
u'.. [%s]' % new_r)
reference_offset[0] += len(references)
def mangle_signature(app, what, name, obj, options, sig, retann):
# Do not try to inspect classes that don't define `__init__`
if (inspect.isclass(obj) and
'initializes x; see ' in pydoc.getdoc(obj.__init__)):
return '', ''
if not (callable(obj) or hasattr(obj, '__argspec_is_invalid_')): return
if not hasattr(obj, '__doc__'): return
doc = SphinxDocString(pydoc.getdoc(obj))
if doc['Signature']:
sig = re.sub(u"^[^(]*", u"", doc['Signature'])
return sig, u''
def initialize(app):
try:
app.connect('autodoc-process-signature', mangle_signature)
except:
monkeypatch_sphinx_ext_autodoc()
def setup(app, get_doc_object_=get_doc_object):
global get_doc_object
get_doc_object = get_doc_object_
app.connect('autodoc-process-docstring', mangle_docstrings)
app.connect('builder-inited', initialize)
app.add_config_value('numpydoc_edit_link', None, True)
app.add_config_value('numpydoc_use_plots', None, False)
# Extra mangling directives
name_type = {
'cfunction': 'function',
'cmember': 'attribute',
'cmacro': 'function',
'ctype': 'class',
'cvar': 'object',
'class': 'class',
'function': 'function',
'attribute': 'attribute',
'method': 'function',
'staticmethod': 'function',
'classmethod': 'function',
}
for name, objtype in name_type.items():
app.add_directive('np-' + name, wrap_mangling_directive(name, objtype))
#------------------------------------------------------------------------------
# Input-mangling directives
#------------------------------------------------------------------------------
from docutils.statemachine import ViewList
def get_directive(name):
from docutils.parsers.rst import directives
try:
return directives.directive(name, None, None)[0]
except AttributeError:
pass
try:
# docutils 0.4
return directives._directives[name]
except (AttributeError, KeyError):
raise RuntimeError("No directive named '%s' found" % name)
def wrap_mangling_directive(base_directive_name, objtype):
base_directive = get_directive(base_directive_name)
if inspect.isfunction(base_directive):
base_func = base_directive
class base_directive(Directive):
required_arguments = base_func.arguments[0]
optional_arguments = base_func.arguments[1]
final_argument_whitespace = base_func.arguments[2]
option_spec = base_func.options
has_content = base_func.content
def run(self):
return base_func(self.name, self.arguments, self.options,
self.content, self.lineno,
self.content_offset, self.block_text,
self.state, self.state_machine)
class directive(base_directive):
def run(self):
env = self.state.document.settings.env
name = None
if self.arguments:
m = re.match(r'^(.*\s+)?(.*?)(\(.*)?', self.arguments[0])
name = m.group(2).strip()
if not name:
name = self.arguments[0]
lines = list(self.content)
mangle_docstrings(env.app, objtype, name, None, None, lines)
self.content = ViewList(lines, self.content.parent)
return base_directive.run(self)
return directive
#------------------------------------------------------------------------------
# Monkeypatch sphinx.ext.autodoc to accept argspecless autodocs (Sphinx < 0.5)
#------------------------------------------------------------------------------
def monkeypatch_sphinx_ext_autodoc():
global _original_format_signature
import sphinx.ext.autodoc
if sphinx.ext.autodoc.format_signature is our_format_signature:
return
print "[numpydoc] Monkeypatching sphinx.ext.autodoc ..."
_original_format_signature = sphinx.ext.autodoc.format_signature
sphinx.ext.autodoc.format_signature = our_format_signature
def our_format_signature(what, obj):
r = mangle_signature(None, what, None, obj, None, None, None)
if r is not None:
return r[0]
else:
return _original_format_signature(what, obj)