Adding PDFFormatter and kernel side handling of PDF display data.

IPython/core/display.py

@@ -1,701 +1,719 b''
 # -*- coding: utf-8 -*-
 """Top-level display functions for displaying object in different formats.
 
 Authors:
 
 * Brian Granger
 """
 
 #-----------------------------------------------------------------------------
 #       Copyright (C) 2013 The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 
 from __future__ import print_function
 
 import os
 import struct
 
 from IPython.utils.py3compat import (string_types, cast_bytes_py2, cast_unicode,
                                      unicode_type)
 
 from .displaypub import publish_display_data
 
 #-----------------------------------------------------------------------------
 # utility functions
 #-----------------------------------------------------------------------------
 
 def _safe_exists(path):
     """Check path, but don't let exceptions raise"""
     try:
         return os.path.exists(path)
     except Exception:
         return False
 
 def _merge(d1, d2):
     """Like update, but merges sub-dicts instead of clobbering at the top level.
 
     Updates d1 in-place
     """
 
     if not isinstance(d2, dict) or not isinstance(d1, dict):
         return d2
     for key, value in d2.items():
         d1[key] = _merge(d1.get(key), value)
     return d1
 
 def _display_mimetype(mimetype, objs, raw=False, metadata=None):
     """internal implementation of all display_foo methods
 
     Parameters
     ----------
     mimetype : str
         The mimetype to be published (e.g. 'image/png')
     objs : tuple of objects
         The Python objects to display, or if raw=True raw text data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     if metadata:
         metadata = {mimetype: metadata}
     if raw:
         # turn list of pngdata into list of { 'image/png': pngdata }
         objs = [ {mimetype: obj} for obj in objs ]
     display(*objs, raw=raw, metadata=metadata, include=[mimetype])
 
 #-----------------------------------------------------------------------------
 # Main functions
 #-----------------------------------------------------------------------------
 
 def display(*objs, **kwargs):
     """Display a Python object in all frontends.
 
     By default all representations will be computed and sent to the frontends.
     Frontends can decide which representation is used and how.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display.
     raw : bool, optional
         Are the objects to be displayed already mimetype-keyed dicts of raw display data,
         or Python objects that need to be formatted before display? [default: False]
     include : list or tuple, optional
         A list of format type strings (MIME types) to include in the
         format data dict. If this is set *only* the format types included
         in this list will be computed.
     exclude : list or tuple, optional
         A list of format type strings (MIME types) to exclude in the format
         data dict. If this is set all format types will be computed,
         except for those included in this argument.
     metadata : dict, optional
         A dictionary of metadata to associate with the output.
         mime-type keys in this dictionary will be associated with the individual
         representation formats, if they exist.
     """
     raw = kwargs.get('raw', False)
     include = kwargs.get('include')
     exclude = kwargs.get('exclude')
     metadata = kwargs.get('metadata')
 
     from IPython.core.interactiveshell import InteractiveShell
 
     if not raw:
         format = InteractiveShell.instance().display_formatter.format
 
     for obj in objs:
 
         # If _ipython_display_ is defined, use that to display this object.
         display_method = getattr(obj, '_ipython_display_', None)
         if display_method is not None:
             try:
                 display_method(**kwargs)
             except NotImplementedError:
                 pass
             else:
                 continue
         if raw:
             publish_display_data('display', obj, metadata)
         else:
             format_dict, md_dict = format(obj, include=include, exclude=exclude)
             if metadata:
                 # kwarg-specified metadata gets precedence
                 _merge(md_dict, metadata)
             publish_display_data('display', format_dict, md_dict)
 
 
 def display_pretty(*objs, **kwargs):
     """Display the pretty (default) representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw text data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('text/plain', objs, **kwargs)
 
 
 def display_html(*objs, **kwargs):
     """Display the HTML representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw HTML data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('text/html', objs, **kwargs)
 
 
 def display_svg(*objs, **kwargs):
     """Display the SVG representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw svg data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('image/svg+xml', objs, **kwargs)
 
 
 def display_png(*objs, **kwargs):
     """Display the PNG representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw png data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('image/png', objs, **kwargs)
 
 
 def display_jpeg(*objs, **kwargs):
     """Display the JPEG representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw JPEG data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('image/jpeg', objs, **kwargs)
 
 
 def display_latex(*objs, **kwargs):
     """Display the LaTeX representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw latex data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('text/latex', objs, **kwargs)
 
 
 def display_json(*objs, **kwargs):
     """Display the JSON representation of an object.
 
     Note that not many frontends support displaying JSON.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw json data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('application/json', objs, **kwargs)
 
 
 def display_javascript(*objs, **kwargs):
     """Display the Javascript representation of an object.
 
     Parameters
     ----------
     objs : tuple of objects
         The Python objects to display, or if raw=True raw javascript data to
         display.
     raw : bool
         Are the data objects raw data or Python objects that need to be
         formatted before display? [default: False]
     metadata : dict (optional)
         Metadata to be associated with the specific mimetype output.
     """
     _display_mimetype('application/javascript', objs, **kwargs)
 
+
+def display_pdf(*objs, **kwargs):
+    """Display the PDF representation of an object.
+
+    Parameters
+    ----------
+    objs : tuple of objects
+        The Python objects to display, or if raw=True raw javascript data to
+        display.
+    raw : bool
+        Are the data objects raw data or Python objects that need to be
+        formatted before display? [default: False]
+    metadata : dict (optional)
+        Metadata to be associated with the specific mimetype output.
+    """
+    _display_mimetype('application/pdf', objs, **kwargs)
+
+
 #-----------------------------------------------------------------------------
 # Smart classes
 #-----------------------------------------------------------------------------
 
 
 class DisplayObject(object):
     """An object that wraps data to be displayed."""
 
     _read_flags = 'r'
 
     def __init__(self, data=None, url=None, filename=None):
         """Create a display object given raw data.
 
         When this object is returned by an expression or passed to the
         display function, it will result in the data being displayed
         in the frontend. The MIME type of the data should match the
         subclasses used, so the Png subclass should be used for 'image/png'
         data. If the data is a URL, the data will first be downloaded
         and then displayed. If
 
         Parameters
         ----------
         data : unicode, str or bytes
             The raw data or a URL or file to load the data from
         url : unicode
             A URL to download the data from.
         filename : unicode
             Path to a local file to load the data from.
         """
         if data is not None and isinstance(data, string_types):
             if data.startswith('http') and url is None:
                 url = data
                 filename = None
                 data = None
             elif _safe_exists(data) and filename is None:
                 url = None
                 filename = data
                 data = None
 
         self.data = data
         self.url = url
         self.filename = None if filename is None else unicode_type(filename)
 
         self.reload()
         self._check_data()
     
     def _check_data(self):
         """Override in subclasses if there's something to check."""
         pass
 
     def reload(self):
         """Reload the raw data from file or URL."""
         if self.filename is not None:
             with open(self.filename, self._read_flags) as f:
                 self.data = f.read()
         elif self.url is not None:
             try:
                 try:
                     from urllib.request import urlopen  # Py3
                 except ImportError:
                     from urllib2 import urlopen
                 response = urlopen(self.url)
                 self.data = response.read()
                 # extract encoding from header, if there is one:
                 encoding = None
                 for sub in response.headers['content-type'].split(';'):
                     sub = sub.strip()
                     if sub.startswith('charset'):
                         encoding = sub.split('=')[-1].strip()
                         break
                 # decode data, if an encoding was specified
                 if encoding:
                     self.data = self.data.decode(encoding, 'replace')
             except:
                 self.data = None
 
 class TextDisplayObject(DisplayObject):
     """Validate that display data is text"""
     def _check_data(self):
         if self.data is not None and not isinstance(self.data, string_types):
             raise TypeError("%s expects text, not %r" % (self.__class__.__name__, self.data))
 
 class Pretty(TextDisplayObject):
 
     def _repr_pretty_(self):
         return self.data
 
 
 class HTML(TextDisplayObject):
 
     def _repr_html_(self):
         return self.data
 
     def __html__(self):
         """
         This method exists to inform other HTML-using modules (e.g. Markupsafe,
         htmltag, etc) that this object is HTML and does not need things like
         special characters (<>&) escaped.
         """
         return self._repr_html_()
 
 
 class Math(TextDisplayObject):
 
     def _repr_latex_(self):
         s = self.data.strip('$')
         return "$$%s$$" % s
 
 
 class Latex(TextDisplayObject):
 
     def _repr_latex_(self):
         return self.data
 
 
 class SVG(DisplayObject):
 
     # wrap data in a property, which extracts the <svg> tag, discarding
     # document headers
     _data = None
 
     @property
     def data(self):
         return self._data
 
     @data.setter
     def data(self, svg):
         if svg is None:
             self._data = None
             return
         # parse into dom object
         from xml.dom import minidom
         svg = cast_bytes_py2(svg)
         x = minidom.parseString(svg)
         # get svg tag (should be 1)
         found_svg = x.getElementsByTagName('svg')
         if found_svg:
             svg = found_svg[0].toxml()
         else:
             # fallback on the input, trust the user
             # but this is probably an error.
             pass
         svg = cast_unicode(svg)
         self._data = svg
 
     def _repr_svg_(self):
         return self.data
 
 
 class JSON(TextDisplayObject):
 
     def _repr_json_(self):
         return self.data
 
 css_t = """$("head").append($("<link/>").attr({
   rel:  "stylesheet",
   type: "text/css",
   href: "%s"
 }));
 """
 
 lib_t1 = """$.getScript("%s", function () {
 """
 lib_t2 = """});
 """
 
 class Javascript(TextDisplayObject):
 
     def __init__(self, data=None, url=None, filename=None, lib=None, css=None):
         """Create a Javascript display object given raw data.
 
         When this object is returned by an expression or passed to the
         display function, it will result in the data being displayed
         in the frontend. If the data is a URL, the data will first be
         downloaded and then displayed.
 
         In the Notebook, the containing element will be available as `element`,
         and jQuery will be available.  The output area starts hidden, so if
         the js appends content to `element` that should be visible, then
         it must call `container.show()` to unhide the area.
 
         Parameters
         ----------
         data : unicode, str or bytes
             The Javascript source code or a URL to download it from.
         url : unicode
             A URL to download the data from.
         filename : unicode
             Path to a local file to load the data from.
         lib : list or str
             A sequence of Javascript library URLs to load asynchronously before
             running the source code. The full URLs of the libraries should
             be given. A single Javascript library URL can also be given as a
             string.
         css: : list or str
             A sequence of css files to load before running the source code.
             The full URLs of the css files should be given. A single css URL
             can also be given as a string.
         """
         if isinstance(lib, string_types):
             lib = [lib]
         elif lib is None:
             lib = []
         if isinstance(css, string_types):
             css = [css]
         elif css is None:
             css = []
         if not isinstance(lib, (list,tuple)):
             raise TypeError('expected sequence, got: %r' % lib)
         if not isinstance(css, (list,tuple)):
             raise TypeError('expected sequence, got: %r' % css)
         self.lib = lib
         self.css = css
         super(Javascript, self).__init__(data=data, url=url, filename=filename)
 
     def _repr_javascript_(self):
         r = ''
         for c in self.css:
             r += css_t % c
         for l in self.lib:
             r += lib_t1 % l
         r += self.data
         r += lib_t2*len(self.lib)
         return r
 
 # constants for identifying png/jpeg data
 _PNG = b'\x89PNG\r\n\x1a\n'
 _JPEG = b'\xff\xd8'
 
 def _pngxy(data):
     """read the (width, height) from a PNG header"""
     ihdr = data.index(b'IHDR')
     # next 8 bytes are width/height
     w4h4 = data[ihdr+4:ihdr+12]
     return struct.unpack('>ii', w4h4)
 
 def _jpegxy(data):
     """read the (width, height) from a JPEG header"""
     # adapted from http://www.64lines.com/jpeg-width-height
     
     idx = 4
     while True:
         block_size = struct.unpack('>H', data[idx:idx+2])[0]
         idx = idx + block_size
         if data[idx:idx+2] == b'\xFF\xC0':
             # found Start of Frame
             iSOF = idx
             break
         else:
             # read another block
             idx += 2
 
     h, w = struct.unpack('>HH', data[iSOF+5:iSOF+9])
     return w, h
 
 class Image(DisplayObject):
 
     _read_flags = 'rb'
     _FMT_JPEG = u'jpeg'
     _FMT_PNG = u'png'
     _ACCEPTABLE_EMBEDDINGS = [_FMT_JPEG, _FMT_PNG]
 
     def __init__(self, data=None, url=None, filename=None, format=u'png', embed=None, width=None, height=None, retina=False):
         """Create a PNG/JPEG image object given raw data.
 
         When this object is returned by an input cell or passed to the
         display function, it will result in the image being displayed
         in the frontend.
 
         Parameters
         ----------
         data : unicode, str or bytes
             The raw image data or a URL or filename to load the data from.
             This always results in embedded image data.
         url : unicode
             A URL to download the data from. If you specify `url=`,
             the image data will not be embedded unless you also specify `embed=True`.
         filename : unicode
             Path to a local file to load the data from.
             Images from a file are always embedded.
         format : unicode
             The format of the image data (png/jpeg/jpg). If a filename or URL is given
             for format will be inferred from the filename extension.
         embed : bool
             Should the image data be embedded using a data URI (True) or be
             loaded using an <img> tag. Set this to True if you want the image
             to be viewable later with no internet connection in the notebook.
 
             Default is `True`, unless the keyword argument `url` is set, then
             default value is `False`.
 
             Note that QtConsole is not able to display images if `embed` is set to `False`
         width : int
             Width to which to constrain the image in html
         height : int
             Height to which to constrain the image in html
         retina : bool
             Automatically set the width and height to half of the measured
             width and height.
             This only works for embedded images because it reads the width/height
             from image data.
             For non-embedded images, you can just set the desired display width
             and height directly.
 
         Examples
         --------
         # embedded image data, works in qtconsole and notebook
         # when passed positionally, the first arg can be any of raw image data,
         # a URL, or a filename from which to load image data.
         # The result is always embedding image data for inline images.
         Image('http://www.google.fr/images/srpr/logo3w.png')
         Image('/path/to/image.jpg')
         Image(b'RAW_PNG_DATA...')
 
         # Specifying Image(url=...) does not embed the image data,
         # it only generates `<img>` tag with a link to the source.
         # This will not work in the qtconsole or offline.
         Image(url='http://www.google.fr/images/srpr/logo3w.png')
 
         """
         if filename is not None:
             ext = self._find_ext(filename)
         elif url is not None:
             ext = self._find_ext(url)
         elif data is None:
             raise ValueError("No image data found. Expecting filename, url, or data.")
         elif isinstance(data, string_types) and (
             data.startswith('http') or _safe_exists(data)
         ):
             ext = self._find_ext(data)
         else:
             ext = None
 
         if ext is not None:
             format = ext.lower()
             if ext == u'jpg' or ext == u'jpeg':
                 format = self._FMT_JPEG
             if ext == u'png':
                 format = self._FMT_PNG
         elif isinstance(data, bytes) and format == 'png':
             # infer image type from image data header,
             # only if format might not have been specified.
             if data[:2] == _JPEG:
                 format = 'jpeg'
 
         self.format = unicode_type(format).lower()
         self.embed = embed if embed is not None else (url is None)
 
         if self.embed and self.format not in self._ACCEPTABLE_EMBEDDINGS:
             raise ValueError("Cannot embed the '%s' image format" % (self.format))
         self.width = width
         self.height = height
         self.retina = retina
         super(Image, self).__init__(data=data, url=url, filename=filename)
         
         if retina:
             self._retina_shape()
     
     def _retina_shape(self):
         """load pixel-doubled width and height from image data"""
         if not self.embed:
             return
         if self.format == 'png':
             w, h = _pngxy(self.data)
         elif self.format == 'jpeg':
             w, h = _jpegxy(self.data)
         else:
             # retina only supports png
             return
         self.width = w // 2
         self.height = h // 2
 
     def reload(self):
         """Reload the raw data from file or URL."""
         if self.embed:
             super(Image,self).reload()
             if self.retina:
                 self._retina_shape()
 
     def _repr_html_(self):
         if not self.embed:
             width = height = ''
             if self.width:
                 width = ' width="%d"' % self.width
             if self.height:
                 height = ' height="%d"' % self.height
             return u'<img src="%s"%s%s/>' % (self.url, width, height)
 
     def _data_and_metadata(self):
         """shortcut for returning metadata with shape information, if defined"""
         md = {}
         if self.width:
             md['width'] = self.width
         if self.height:
             md['height'] = self.height
         if md:
             return self.data, md
         else:
             return self.data
 
     def _repr_png_(self):
         if self.embed and self.format == u'png':
             return self._data_and_metadata()
 
     def _repr_jpeg_(self):
         if self.embed and (self.format == u'jpeg' or self.format == u'jpg'):
             return self._data_and_metadata()
 
     def _find_ext(self, s):
         return unicode_type(s.split('.')[-1].lower())
 
 
 def clear_output(wait=False):
     """Clear the output of the current cell receiving output.
 
     Parameters
     ----------
     wait : bool [default: false]
         Wait to clear the output until new output is available to replace it."""
     from IPython.core.interactiveshell import InteractiveShell
     if InteractiveShell.initialized():
         InteractiveShell.instance().display_pub.clear_output(wait)
     else:
         from IPython.utils import io
         print('\033[2K\r', file=io.stdout, end='')
         io.stdout.flush()
         print('\033[2K\r', file=io.stderr, end='')
         io.stderr.flush()

IPython/core/formatters.py

@@ -1,825 +1,846 b''
 # -*- coding: utf-8 -*-
 """Display formatters.
 
 Inheritance diagram:
 
 .. inheritance-diagram:: IPython.core.formatters
    :parts: 3
 
 Authors:
 
 * Robert Kern
 * Brian Granger
 """
 #-----------------------------------------------------------------------------
 # Copyright (C) 2010-2011, IPython Development Team.
 #
 # Distributed under the terms of the Modified BSD License.
 #
 # The full license is in the file COPYING.txt, distributed with this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 
 # Stdlib imports
 import abc
 import sys
 import warnings
 
 from IPython.external.decorator import decorator
 
 # Our own imports
 from IPython.config.configurable import Configurable
 from IPython.lib import pretty
 from IPython.utils import io
 from IPython.utils.traitlets import (
     Bool, Dict, Integer, Unicode, CUnicode, ObjectName, List,
 )
 from IPython.utils.warn import warn
 from IPython.utils.py3compat import (
     unicode_to_str, with_metaclass, PY3, string_types, unicode_type,
 )
 
 if PY3:
     from io import StringIO
 else:
     from StringIO import StringIO
 
 
 #-----------------------------------------------------------------------------
 # The main DisplayFormatter class
 #-----------------------------------------------------------------------------
 
 class DisplayFormatter(Configurable):
 
     # When set to true only the default plain text formatter will be used.
     plain_text_only = Bool(False, config=True)
     def _plain_text_only_changed(self, name, old, new):
         warnings.warn("""DisplayFormatter.plain_text_only is deprecated.
         
         Use DisplayFormatter.active_types = ['text/plain']
         for the same effect.
         """, DeprecationWarning)
         if new:
             self.active_types = ['text/plain']
         else:
             self.active_types = self.format_types
     
     active_types = List(Unicode, config=True,
         help="""List of currently active mime-types to display.
         You can use this to set a white-list for formats to display.
         
         Most users will not need to change this value.
         """)
     def _active_types_default(self):
         return self.format_types
     
     def _active_types_changed(self, name, old, new):
         for key, formatter in self.formatters.items():
             if key in new:
                 formatter.enabled = True
             else:
                 formatter.enabled = False
     
     # A dict of formatter whose keys are format types (MIME types) and whose
     # values are subclasses of BaseFormatter.
     formatters = Dict()
     def _formatters_default(self):
         """Activate the default formatters."""
         formatter_classes = [
             PlainTextFormatter,
             HTMLFormatter,
             SVGFormatter,
             PNGFormatter,
+            PDFFormatter,
             JPEGFormatter,
             LatexFormatter,
             JSONFormatter,
             JavascriptFormatter
         ]
         d = {}
         for cls in formatter_classes:
             f = cls(parent=self)
             d[f.format_type] = f
         return d
 
     def format(self, obj, include=None, exclude=None):
         """Return a format data dict for an object.
 
         By default all format types will be computed.
 
         The following MIME types are currently implemented:
 
         * text/plain
         * text/html
         * text/latex
         * application/json
         * application/javascript
+        * application/pdf
         * image/png
         * image/jpeg
         * image/svg+xml
 
         Parameters
         ----------
         obj : object
             The Python object whose format data will be computed.
         include : list or tuple, optional
             A list of format type strings (MIME types) to include in the
             format data dict. If this is set *only* the format types included
             in this list will be computed.
         exclude : list or tuple, optional
             A list of format type string (MIME types) to exclude in the format
             data dict. If this is set all format types will be computed,
             except for those included in this argument.
 
         Returns
         -------
         (format_dict, metadata_dict) : tuple of two dicts
         
             format_dict is a dictionary of key/value pairs, one of each format that was
             generated for the object. The keys are the format types, which
             will usually be MIME type strings and the values and JSON'able
             data structure containing the raw data for the representation in
             that format.
             
             metadata_dict is a dictionary of metadata about each mime-type output.
             Its keys will be a strict subset of the keys in format_dict.
         """
         format_dict = {}
         md_dict = {}
 
         for format_type, formatter in self.formatters.items():
             if include and format_type not in include:
                 continue
             if exclude and format_type in exclude:
                 continue
             
             md = None
             try:
                 data = formatter(obj)
             except:
                 # FIXME: log the exception
                 raise
             
             # formatters can return raw data or (data, metadata)
             if isinstance(data, tuple) and len(data) == 2:
                 data, md = data
             
             if data is not None:
                 format_dict[format_type] = data
             if md is not None:
                 md_dict[format_type] = md
             
         return format_dict, md_dict
 
     @property
     def format_types(self):
         """Return the format types (MIME types) of the active formatters."""
         return list(self.formatters.keys())
 
 
 #-----------------------------------------------------------------------------
 # Formatters for specific format types (text, html, svg, etc.)
 #-----------------------------------------------------------------------------
 
 class FormatterWarning(UserWarning):
     """Warning class for errors in formatters"""
 
 @decorator
 def warn_format_error(method, self, *args, **kwargs):
     """decorator for warning on failed format call"""
     try:
         r = method(self, *args, **kwargs)
     except NotImplementedError as e:
         # don't warn on NotImplementedErrors
         return None
     except Exception as e:
         warnings.warn("Exception in %s formatter: %s" % (self.format_type, e),
             FormatterWarning,
         )
         return None
     if r is None or isinstance(r, self._return_type) or \
         (isinstance(r, tuple) and r and isinstance(r[0], self._return_type)):
         return r
     else:
         warnings.warn(
             "%s formatter returned invalid type %s (expected %s) for object: %s" % \
             (self.format_type, type(r), self._return_type, pretty._safe_repr(args[0])),
             FormatterWarning
         )
 
 
 class FormatterABC(with_metaclass(abc.ABCMeta, object)):
     """ Abstract base class for Formatters.
 
     A formatter is a callable class that is responsible for computing the
     raw format data for a particular format type (MIME type). For example,
     an HTML formatter would have a format type of `text/html` and would return
     the HTML representation of the object when called.
     """
 
     # The format type of the data returned, usually a MIME type.
     format_type = 'text/plain'
 
     # Is the formatter enabled...
     enabled = True
     
     @abc.abstractmethod
     @warn_format_error
     def __call__(self, obj):
         """Return a JSON'able representation of the object.
 
         If the object cannot be formatted by this formatter,
         warn and return None.
         """
         return repr(obj)
 
 
 def _mod_name_key(typ):
     """Return a (__module__, __name__) tuple for a type.
     
     Used as key in Formatter.deferred_printers.
     """
     module = getattr(typ, '__module__', None)
     name = getattr(typ, '__name__', None)
     return (module, name)
 
 
 def _get_type(obj):
     """Return the type of an instance (old and new-style)"""
     return getattr(obj, '__class__', None) or type(obj)
 
 _raise_key_error = object()
 
 
 class BaseFormatter(Configurable):
     """A base formatter class that is configurable.
 
     This formatter should usually be used as the base class of all formatters.
     It is a traited :class:`Configurable` class and includes an extensible
     API for users to determine how their objects are formatted. The following
     logic is used to find a function to format an given object.
 
     1. The object is introspected to see if it has a method with the name
        :attr:`print_method`. If is does, that object is passed to that method
        for formatting.
     2. If no print method is found, three internal dictionaries are consulted
        to find print method: :attr:`singleton_printers`, :attr:`type_printers`
        and :attr:`deferred_printers`.
 
     Users should use these dictionaries to register functions that will be
     used to compute the format data for their objects (if those objects don't
     have the special print methods). The easiest way of using these
     dictionaries is through the :meth:`for_type` and :meth:`for_type_by_name`
     methods.
 
     If no function/callable is found to compute the format data, ``None`` is
     returned and this format type is not used.
     """
 
     format_type = Unicode('text/plain')
     _return_type = string_types
 
     enabled = Bool(True, config=True)
 
     print_method = ObjectName('__repr__')
 
     # The singleton printers.
     # Maps the IDs of the builtin singleton objects to the format functions.
     singleton_printers = Dict(config=True)
 
     # The type-specific printers.
     # Map type objects to the format functions.
     type_printers = Dict(config=True)
 
     # The deferred-import type-specific printers.
     # Map (modulename, classname) pairs to the format functions.
     deferred_printers = Dict(config=True)
     
     @warn_format_error
     def __call__(self, obj):
         """Compute the format for an object."""
         if self.enabled:
             # lookup registered printer
             try:
                 printer = self.lookup(obj)
             except KeyError:
                 pass
             else:
                 return printer(obj)
             # Finally look for special method names
             method = pretty._safe_getattr(obj, self.print_method, None)
             if method is not None:
                 return method()
             return None
         else:
             return None
     
     def __contains__(self, typ):
         """map in to lookup_by_type"""
         try:
             self.lookup_by_type(typ)
         except KeyError:
             return False
         else:
             return True
     
     def lookup(self, obj):
         """Look up the formatter for a given instance.
         
         Parameters
         ----------
         obj  : object instance
 
         Returns
         -------
         f : callable
             The registered formatting callable for the type.
 
         Raises
         ------
         KeyError if the type has not been registered.
         """
         # look for singleton first
         obj_id = id(obj)
         if obj_id in self.singleton_printers:
             return self.singleton_printers[obj_id]
         # then lookup by type
         return self.lookup_by_type(_get_type(obj))
     
     def lookup_by_type(self, typ):
         """Look up the registered formatter for a type.
 
         Parameters
         ----------
         typ  : type or '__module__.__name__' string for a type
 
         Returns
         -------
         f : callable
             The registered formatting callable for the type.
 
         Raises
         ------
         KeyError if the type has not been registered.
         """
         if isinstance(typ, string_types):
             typ_key = tuple(typ.rsplit('.',1))
             if typ_key not in self.deferred_printers:
                 # We may have it cached in the type map. We will have to
                 # iterate over all of the types to check.
                 for cls in self.type_printers:
                     if _mod_name_key(cls) == typ_key:
                         return self.type_printers[cls]
             else:
                 return self.deferred_printers[typ_key]
         else:
             for cls in pretty._get_mro(typ):
                 if cls in self.type_printers or self._in_deferred_types(cls):
                     return self.type_printers[cls]
         
         # If we have reached here, the lookup failed.
         raise KeyError("No registered printer for {0!r}".format(typ))
 
     def for_type(self, typ, func=None):
         """Add a format function for a given type.
         
         Parameters
         -----------
         typ : type or '__module__.__name__' string for a type
             The class of the object that will be formatted using `func`.
         func : callable
             A callable for computing the format data.
             `func` will be called with the object to be formatted,
             and will return the raw data in this formatter's format.
             Subclasses may use a different call signature for the
             `func` argument.
             
             If `func` is None or not specified, there will be no change,
             only returning the current value.
         
         Returns
         -------
         oldfunc : callable
             The currently registered callable.
             If you are registering a new formatter,
             this will be the previous value (to enable restoring later).
         """
         # if string given, interpret as 'pkg.module.class_name'
         if isinstance(typ, string_types):
             type_module, type_name = typ.rsplit('.', 1)
             return self.for_type_by_name(type_module, type_name, func)
         
         try:
             oldfunc = self.lookup_by_type(typ)
         except KeyError:
             oldfunc = None
         
         if func is not None:
             self.type_printers[typ] = func
         
         return oldfunc
 
     def for_type_by_name(self, type_module, type_name, func=None):
         """Add a format function for a type specified by the full dotted
         module and name of the type, rather than the type of the object.
 
         Parameters
         ----------
         type_module : str
             The full dotted name of the module the type is defined in, like
             ``numpy``.
         type_name : str
             The name of the type (the class name), like ``dtype``
         func : callable
             A callable for computing the format data.
             `func` will be called with the object to be formatted,
             and will return the raw data in this formatter's format.
             Subclasses may use a different call signature for the
             `func` argument.
             
             If `func` is None or unspecified, there will be no change,
             only returning the current value.
         
         Returns
         -------
         oldfunc : callable
             The currently registered callable.
             If you are registering a new formatter,
             this will be the previous value (to enable restoring later).
         """
         key = (type_module, type_name)
         
         try:
             oldfunc = self.lookup_by_type("%s.%s" % key)
         except KeyError:
             oldfunc = None
         
         if func is not None:
             self.deferred_printers[key] = func
         return oldfunc
     
     def pop(self, typ, default=_raise_key_error):
         """Pop a formatter for the given type.
 
         Parameters
         ----------
         typ : type or '__module__.__name__' string for a type
         default : object
             value to be returned if no formatter is registered for typ.
 
         Returns
         -------
         obj : object
             The last registered object for the type.
 
         Raises
         ------
         KeyError if the type is not registered and default is not specified.
         """
         
         if isinstance(typ, string_types):
             typ_key = tuple(typ.rsplit('.',1))
             if typ_key not in self.deferred_printers:
                 # We may have it cached in the type map. We will have to
                 # iterate over all of the types to check.
                 for cls in self.type_printers:
                     if _mod_name_key(cls) == typ_key:
                         old = self.type_printers.pop(cls)
                         break
                 else:
                     old = default
             else:
                 old = self.deferred_printers.pop(typ_key)
         else:
             if typ in self.type_printers:
                 old = self.type_printers.pop(typ)
             else:
                 old = self.deferred_printers.pop(_mod_name_key(typ), default)
         if old is _raise_key_error:
             raise KeyError("No registered value for {0!r}".format(typ))
         return old
 
     def _in_deferred_types(self, cls):
         """
         Check if the given class is specified in the deferred type registry.
 
         Successful matches will be moved to the regular type registry for future use.
         """
         mod = getattr(cls, '__module__', None)
         name = getattr(cls, '__name__', None)
         key = (mod, name)
         if key in self.deferred_printers:
             # Move the printer over to the regular registry.
             printer = self.deferred_printers.pop(key)
             self.type_printers[cls] = printer
             return True
         return False
 
 
 class PlainTextFormatter(BaseFormatter):
     """The default pretty-printer.
 
     This uses :mod:`IPython.lib.pretty` to compute the format data of
     the object. If the object cannot be pretty printed, :func:`repr` is used.
     See the documentation of :mod:`IPython.lib.pretty` for details on
     how to write pretty printers.  Here is a simple example::
 
         def dtype_pprinter(obj, p, cycle):
             if cycle:
                 return p.text('dtype(...)')
             if hasattr(obj, 'fields'):
                 if obj.fields is None:
                     p.text(repr(obj))
                 else:
                     p.begin_group(7, 'dtype([')
                     for i, field in enumerate(obj.descr):
                         if i > 0:
                             p.text(',')
                             p.breakable()
                         p.pretty(field)
                     p.end_group(7, '])')
     """
 
     # The format type of data returned.
     format_type = Unicode('text/plain')
 
     # This subclass ignores this attribute as it always need to return
     # something.
     enabled = Bool(True, config=False)
 
     # Look for a _repr_pretty_ methods to use for pretty printing.
     print_method = ObjectName('_repr_pretty_')
 
     # Whether to pretty-print or not.
     pprint = Bool(True, config=True)
 
     # Whether to be verbose or not.
     verbose = Bool(False, config=True)
 
     # The maximum width.
     max_width = Integer(79, config=True)
 
     # The newline character.
     newline = Unicode('\n', config=True)
 
     # format-string for pprinting floats
     float_format = Unicode('%r')
     # setter for float precision, either int or direct format-string
     float_precision = CUnicode('', config=True)
 
     def _float_precision_changed(self, name, old, new):
         """float_precision changed, set float_format accordingly.
 
         float_precision can be set by int or str.
         This will set float_format, after interpreting input.
         If numpy has been imported, numpy print precision will also be set.
 
         integer `n` sets format to '%.nf', otherwise, format set directly.
 
         An empty string returns to defaults (repr for float, 8 for numpy).
 
         This parameter can be set via the '%precision' magic.
         """
 
         if '%' in new:
             # got explicit format string
             fmt = new
             try:
                 fmt%3.14159
             except Exception:
                 raise ValueError("Precision must be int or format string, not %r"%new)
         elif new:
             # otherwise, should be an int
             try:
                 i = int(new)
                 assert i >= 0
             except ValueError:
                 raise ValueError("Precision must be int or format string, not %r"%new)
             except AssertionError:
                 raise ValueError("int precision must be non-negative, not %r"%i)
 
             fmt = '%%.%if'%i
             if 'numpy' in sys.modules:
                 # set numpy precision if it has been imported
                 import numpy
                 numpy.set_printoptions(precision=i)
         else:
             # default back to repr
             fmt = '%r'
             if 'numpy' in sys.modules:
                 import numpy
                 # numpy default is 8
                 numpy.set_printoptions(precision=8)
         self.float_format = fmt
 
     # Use the default pretty printers from IPython.lib.pretty.
     def _singleton_printers_default(self):
         return pretty._singleton_pprinters.copy()
 
     def _type_printers_default(self):
         d = pretty._type_pprinters.copy()
         d[float] = lambda obj,p,cycle: p.text(self.float_format%obj)
         return d
 
     def _deferred_printers_default(self):
         return pretty._deferred_type_pprinters.copy()
 
     #### FormatterABC interface ####
 
     @warn_format_error
     def __call__(self, obj):
         """Compute the pretty representation of the object."""
         if not self.pprint:
             return pretty._safe_repr(obj)
         else:
             # This uses use StringIO, as cStringIO doesn't handle unicode.
             stream = StringIO()
             # self.newline.encode() is a quick fix for issue gh-597. We need to
             # ensure that stream does not get a mix of unicode and bytestrings,
             # or it will cause trouble.
             printer = pretty.RepresentationPrinter(stream, self.verbose,
                 self.max_width, unicode_to_str(self.newline),
                 singleton_pprinters=self.singleton_printers,
                 type_pprinters=self.type_printers,
                 deferred_pprinters=self.deferred_printers)
             printer.pretty(obj)
             printer.flush()
             return stream.getvalue()
 
 
 class HTMLFormatter(BaseFormatter):
     """An HTML formatter.
 
     To define the callables that compute the HTML representation of your
     objects, define a :meth:`_repr_html_` method or use the :meth:`for_type`
     or :meth:`for_type_by_name` methods to register functions that handle
     this.
 
     The return value of this formatter should be a valid HTML snippet that
     could be injected into an existing DOM. It should *not* include the
     ```<html>`` or ```<body>`` tags.
     """
     format_type = Unicode('text/html')
 
     print_method = ObjectName('_repr_html_')
 
 
 class SVGFormatter(BaseFormatter):
     """An SVG formatter.
 
     To define the callables that compute the SVG representation of your
     objects, define a :meth:`_repr_svg_` method or use the :meth:`for_type`
     or :meth:`for_type_by_name` methods to register functions that handle
     this.
 
     The return value of this formatter should be valid SVG enclosed in
     ```<svg>``` tags, that could be injected into an existing DOM. It should
     *not* include the ```<html>`` or ```<body>`` tags.
     """
     format_type = Unicode('image/svg+xml')
 
     print_method = ObjectName('_repr_svg_')
 
 
 class PNGFormatter(BaseFormatter):
     """A PNG formatter.
 
     To define the callables that compute the PNG representation of your
     objects, define a :meth:`_repr_png_` method or use the :meth:`for_type`
     or :meth:`for_type_by_name` methods to register functions that handle
     this.
 
     The return value of this formatter should be raw PNG data, *not*
     base64 encoded.
     """
     format_type = Unicode('image/png')
 
     print_method = ObjectName('_repr_png_')
     
     _return_type = (bytes, unicode_type)
 
 
 class JPEGFormatter(BaseFormatter):
     """A JPEG formatter.
 
     To define the callables that compute the JPEG representation of your
     objects, define a :meth:`_repr_jpeg_` method or use the :meth:`for_type`
     or :meth:`for_type_by_name` methods to register functions that handle
     this.
 
     The return value of this formatter should be raw JPEG data, *not*
     base64 encoded.
     """
     format_type = Unicode('image/jpeg')
 
     print_method = ObjectName('_repr_jpeg_')
 
     _return_type = (bytes, unicode_type)
 
 
 class LatexFormatter(BaseFormatter):
     """A LaTeX formatter.
 
     To define the callables that compute the LaTeX representation of your
     objects, define a :meth:`_repr_latex_` method or use the :meth:`for_type`
     or :meth:`for_type_by_name` methods to register functions that handle
     this.
 
     The return value of this formatter should be a valid LaTeX equation,
     enclosed in either ```$```, ```$$``` or another LaTeX equation
     environment.
     """
     format_type = Unicode('text/latex')
 
     print_method = ObjectName('_repr_latex_')
 
 
 class JSONFormatter(BaseFormatter):
     """A JSON string formatter.
 
     To define the callables that compute the JSON string representation of
     your objects, define a :meth:`_repr_json_` method or use the :meth:`for_type`
     or :meth:`for_type_by_name` methods to register functions that handle
     this.
 
     The return value of this formatter should be a valid JSON string.
     """
     format_type = Unicode('application/json')
 
     print_method = ObjectName('_repr_json_')
 
 
 class JavascriptFormatter(BaseFormatter):
     """A Javascript formatter.
 
     To define the callables that compute the Javascript representation of
     your objects, define a :meth:`_repr_javascript_` method or use the
     :meth:`for_type` or :meth:`for_type_by_name` methods to register functions
     that handle this.
 
     The return value of this formatter should be valid Javascript code and
     should *not* be enclosed in ```<script>``` tags.
     """
     format_type = Unicode('application/javascript')
 
     print_method = ObjectName('_repr_javascript_')
 
+
+class PDFFormatter(BaseFormatter):
+    """A PDF formatter.
+
+    To defined the callables that compute to PDF representation of your
+    objects, define a :meth:`_repr_pdf_` method or use the :meth:`for_type`
+    or :meth:`for_type_by_name` methods to register functions that handle
+    this.
+
+    The return value of this formatter should be raw PDF data, *not*
+    base64 encoded.
+    """
+    format_type = Unicode('application/pdf')
+
+    print_method = ObjectName('_repr_pdf_')
+
+
 FormatterABC.register(BaseFormatter)
 FormatterABC.register(PlainTextFormatter)
 FormatterABC.register(HTMLFormatter)
 FormatterABC.register(SVGFormatter)
 FormatterABC.register(PNGFormatter)
+FormatterABC.register(PDFFormatter)
 FormatterABC.register(JPEGFormatter)
 FormatterABC.register(LatexFormatter)
 FormatterABC.register(JSONFormatter)
 FormatterABC.register(JavascriptFormatter)
 
 
 def format_display_data(obj, include=None, exclude=None):
     """Return a format data dict for an object.
 
     By default all format types will be computed.
 
     The following MIME types are currently implemented:
 
     * text/plain
     * text/html
     * text/latex
     * application/json
     * application/javascript
+    * application/pdf
     * image/png
     * image/jpeg
     * image/svg+xml
 
     Parameters
     ----------
     obj : object
         The Python object whose format data will be computed.
 
     Returns
     -------
     format_dict : dict
         A dictionary of key/value pairs, one or each format that was
         generated for the object. The keys are the format types, which
         will usually be MIME type strings and the values and JSON'able
         data structure containing the raw data for the representation in
         that format.
     include : list or tuple, optional
         A list of format type strings (MIME types) to include in the
         format data dict. If this is set *only* the format types included
         in this list will be computed.
     exclude : list or tuple, optional
         A list of format type string (MIME types) to exclue in the format
         data dict. If this is set all format types will be computed,
         except for those included in this argument.
     """
     from IPython.core.interactiveshell import InteractiveShell
 
     InteractiveShell.instance().display_formatter.format(
         obj,
         include,
         exclude
     )
 

IPython/core/tests/test_formatters.py

@@ -1,282 +1,291 b''
 """Tests for the Formatters."""
 
 from math import pi
 
 try:
     import numpy
 except:
     numpy = None
 import nose.tools as nt
 
-from IPython.core.formatters import PlainTextFormatter, HTMLFormatter, _mod_name_key
+from IPython.core.formatters import (
+    PlainTextFormatter, HTMLFormatter, PDFFormatter, _mod_name_key
+)
 from IPython.utils.io import capture_output
 
 class A(object):
     def __repr__(self):
         return 'A()'
 
 class B(A):
     def __repr__(self):
         return 'B()'
 
 class C:
     pass
 
 class BadPretty(object):
     _repr_pretty_ = None
 
 class GoodPretty(object):
     def _repr_pretty_(self, pp, cycle):
         pp.text('foo')
 
     def __repr__(self):
         return 'GoodPretty()'
 
 def foo_printer(obj, pp, cycle):
     pp.text('foo')
 
 def test_pretty():
     f = PlainTextFormatter()
     f.for_type(A, foo_printer)
     nt.assert_equal(f(A()), 'foo')
     nt.assert_equal(f(B()), 'foo')
     nt.assert_equal(f(GoodPretty()), 'foo')
     # Just don't raise an exception for the following:
     f(BadPretty())
 
     f.pprint = False
     nt.assert_equal(f(A()), 'A()')
     nt.assert_equal(f(B()), 'B()')
     nt.assert_equal(f(GoodPretty()), 'GoodPretty()')
 
 
 def test_deferred():
     f = PlainTextFormatter()
 
 def test_precision():
     """test various values for float_precision."""
     f = PlainTextFormatter()
     nt.assert_equal(f(pi), repr(pi))
     f.float_precision = 0
     if numpy:
         po = numpy.get_printoptions()
         nt.assert_equal(po['precision'], 0)
     nt.assert_equal(f(pi), '3')
     f.float_precision = 2
     if numpy:
         po = numpy.get_printoptions()
         nt.assert_equal(po['precision'], 2)
     nt.assert_equal(f(pi), '3.14')
     f.float_precision = '%g'
     if numpy:
         po = numpy.get_printoptions()
         nt.assert_equal(po['precision'], 2)
     nt.assert_equal(f(pi), '3.14159')
     f.float_precision = '%e'
     nt.assert_equal(f(pi), '3.141593e+00')
     f.float_precision = ''
     if numpy:
         po = numpy.get_printoptions()
         nt.assert_equal(po['precision'], 8)
     nt.assert_equal(f(pi), repr(pi))
 
 def test_bad_precision():
     """test various invalid values for float_precision."""
     f = PlainTextFormatter()
     def set_fp(p):
         f.float_precision=p
     nt.assert_raises(ValueError, set_fp, '%')
     nt.assert_raises(ValueError, set_fp, '%.3f%i')
     nt.assert_raises(ValueError, set_fp, 'foo')
     nt.assert_raises(ValueError, set_fp, -1)
 
 def test_for_type():
     f = PlainTextFormatter()
     
     # initial return, None
     nt.assert_is(f.for_type(C, foo_printer), None)
     # no func queries
     nt.assert_is(f.for_type(C), foo_printer)
     # shouldn't change anything
     nt.assert_is(f.for_type(C), foo_printer)
     # None should do the same
     nt.assert_is(f.for_type(C, None), foo_printer)
     nt.assert_is(f.for_type(C, None), foo_printer)
 
 def test_for_type_string():
     f = PlainTextFormatter()
     
     mod = C.__module__
     
     type_str = '%s.%s' % (C.__module__, 'C')
     
     # initial return, None
     nt.assert_is(f.for_type(type_str, foo_printer), None)
     # no func queries
     nt.assert_is(f.for_type(type_str), foo_printer)
     nt.assert_in(_mod_name_key(C), f.deferred_printers)
     nt.assert_is(f.for_type(C), foo_printer)
     nt.assert_not_in(_mod_name_key(C), f.deferred_printers)
     nt.assert_in(C, f.type_printers)
 
 def test_for_type_by_name():
     f = PlainTextFormatter()
     
     mod = C.__module__
     
     # initial return, None
     nt.assert_is(f.for_type_by_name(mod, 'C', foo_printer), None)
     # no func queries
     nt.assert_is(f.for_type_by_name(mod, 'C'), foo_printer)
     # shouldn't change anything
     nt.assert_is(f.for_type_by_name(mod, 'C'), foo_printer)
     # None should do the same
     nt.assert_is(f.for_type_by_name(mod, 'C', None), foo_printer)
     nt.assert_is(f.for_type_by_name(mod, 'C', None), foo_printer)
 
 def test_lookup():
     f = PlainTextFormatter()
     
     f.for_type(C, foo_printer)
     nt.assert_is(f.lookup(C()), foo_printer)
     with nt.assert_raises(KeyError):
         f.lookup(A())
 
 def test_lookup_string():
     f = PlainTextFormatter()
     type_str = '%s.%s' % (C.__module__, 'C')
     
     f.for_type(type_str, foo_printer)
     nt.assert_is(f.lookup(C()), foo_printer)
     # should move from deferred to imported dict
     nt.assert_not_in(_mod_name_key(C), f.deferred_printers)
     nt.assert_in(C, f.type_printers)
 
 def test_lookup_by_type():
     f = PlainTextFormatter()
     f.for_type(C, foo_printer)
     nt.assert_is(f.lookup_by_type(C), foo_printer)
     type_str = '%s.%s' % (C.__module__, 'C')
     with nt.assert_raises(KeyError):
         f.lookup_by_type(A)
 
 def test_lookup_by_type_string():
     f = PlainTextFormatter()
     type_str = '%s.%s' % (C.__module__, 'C')
     f.for_type(type_str, foo_printer)
     
     # verify insertion
     nt.assert_in(_mod_name_key(C), f.deferred_printers)
     nt.assert_not_in(C, f.type_printers)
     
     nt.assert_is(f.lookup_by_type(type_str), foo_printer)
     # lookup by string doesn't cause import
     nt.assert_in(_mod_name_key(C), f.deferred_printers)
     nt.assert_not_in(C, f.type_printers)
     
     nt.assert_is(f.lookup_by_type(C), foo_printer)
     # should move from deferred to imported dict
     nt.assert_not_in(_mod_name_key(C), f.deferred_printers)
     nt.assert_in(C, f.type_printers)
 
 def test_in_formatter():
     f = PlainTextFormatter()
     f.for_type(C, foo_printer)
     type_str = '%s.%s' % (C.__module__, 'C')
     nt.assert_in(C, f)
     nt.assert_in(type_str, f)
 
 def test_string_in_formatter():
     f = PlainTextFormatter()
     type_str = '%s.%s' % (C.__module__, 'C')
     f.for_type(type_str, foo_printer)
     nt.assert_in(type_str, f)
     nt.assert_in(C, f)
 
 def test_pop():
     f = PlainTextFormatter()
     f.for_type(C, foo_printer)
     nt.assert_is(f.lookup_by_type(C), foo_printer)
     nt.assert_is(f.pop(C, None), foo_printer)
     f.for_type(C, foo_printer)
     nt.assert_is(f.pop(C), foo_printer)
     with nt.assert_raises(KeyError):
         f.lookup_by_type(C)
     with nt.assert_raises(KeyError):
         f.pop(C)
     with nt.assert_raises(KeyError):
         f.pop(A)
     nt.assert_is(f.pop(A, None), None)
 
 def test_pop_string():
     f = PlainTextFormatter()
     type_str = '%s.%s' % (C.__module__, 'C')
     
     with nt.assert_raises(KeyError):
         f.pop(type_str)
     
     f.for_type(type_str, foo_printer)
     f.pop(type_str)
     with nt.assert_raises(KeyError):
         f.lookup_by_type(C)
     with nt.assert_raises(KeyError):
         f.pop(type_str)
 
     f.for_type(C, foo_printer)
     nt.assert_is(f.pop(type_str, None), foo_printer)
     with nt.assert_raises(KeyError):
         f.lookup_by_type(C)
     with nt.assert_raises(KeyError):
         f.pop(type_str)
     nt.assert_is(f.pop(type_str, None), None)
     
 
 def test_warn_error_method():
     f = HTMLFormatter()
     class BadHTML(object):
         def _repr_html_(self):
             return 1/0
     bad = BadHTML()
     with capture_output() as captured:
         result = f(bad)
     nt.assert_is(result, None)
     nt.assert_in("FormatterWarning", captured.stderr)
     nt.assert_in("text/html", captured.stderr)
     nt.assert_in("zero", captured.stderr)
 
 def test_nowarn_notimplemented():
     f = HTMLFormatter()
     class HTMLNotImplemented(object):
         def _repr_html_(self):
             raise NotImplementedError
             return 1/0
     h = HTMLNotImplemented()
     with capture_output() as captured:
         result = f(h)
     nt.assert_is(result, None)
     nt.assert_not_in("FormatterWarning", captured.stderr)
 
 def test_warn_error_for_type():
     f = HTMLFormatter()
     f.for_type(int, lambda i: name_error)
     with capture_output() as captured:
         result = f(5)
     nt.assert_is(result, None)
     nt.assert_in("FormatterWarning", captured.stderr)
     nt.assert_in("text/html", captured.stderr)
     nt.assert_in("name_error", captured.stderr)
 
 def test_warn_error_pretty_method():
     f = PlainTextFormatter()
     class BadPretty(object):
         def _repr_pretty_(self):
             return "hello"
     bad = BadPretty()
     with capture_output() as captured:
         result = f(bad)
     nt.assert_is(result, None)
     nt.assert_in("FormatterWarning", captured.stderr)
     nt.assert_in("text/plain", captured.stderr)
     nt.assert_in("argument", captured.stderr)
 
+class MakePDF(object):
+    def _repr_pdf_(self):
+        return 'PDF'
 
+def test_pdf_formatter():
+    pdf = MakePDF()
+    f = PDFFormatter()
+    nt.assert_equal(f(pdf), 'PDF')

IPython/utils/jsonutil.py

@@ -1,247 +1,256 b''
 """Utilities to manipulate JSON objects.
 """
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010-2011  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING.txt, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 # stdlib
 import math
 import re
 import types
 from datetime import datetime
 
 try:
     # base64.encodestring is deprecated in Python 3.x
     from base64 import encodebytes
 except ImportError:
     # Python 2.x
     from base64 import encodestring as encodebytes
 
 from IPython.utils import py3compat
 from IPython.utils.py3compat import string_types, unicode_type, iteritems
 from IPython.utils.encoding import DEFAULT_ENCODING
 next_attr_name = '__next__' if py3compat.PY3 else 'next'
 
 #-----------------------------------------------------------------------------
 # Globals and constants
 #-----------------------------------------------------------------------------
 
 # timestamp formats
 ISO8601 = "%Y-%m-%dT%H:%M:%S.%f"
 ISO8601_PAT=re.compile(r"^(\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{1,6})Z?([\+\-]\d{2}:?\d{2})?$")
 
 #-----------------------------------------------------------------------------
 # Classes and functions
 #-----------------------------------------------------------------------------
 
 def rekey(dikt):
     """Rekey a dict that has been forced to use str keys where there should be
     ints by json."""
     for k in dikt:
         if isinstance(k, string_types):
             ik=fk=None
             try:
                 ik = int(k)
             except ValueError:
                 try:
                     fk = float(k)
                 except ValueError:
                     continue
             if ik is not None:
                 nk = ik
             else:
                 nk = fk
             if nk in dikt:
                 raise KeyError("already have key %r"%nk)
             dikt[nk] = dikt.pop(k)
     return dikt
 
 def parse_date(s):
     """parse an ISO8601 date string
     
     If it is None or not a valid ISO8601 timestamp,
     it will be returned unmodified.
     Otherwise, it will return a datetime object.
     """
     if s is None:
         return s
     m = ISO8601_PAT.match(s)
     if m:
         # FIXME: add actual timezone support
         # this just drops the timezone info
         notz = m.groups()[0]
         return datetime.strptime(notz, ISO8601)
     return s
 
 def extract_dates(obj):
     """extract ISO8601 dates from unpacked JSON"""
     if isinstance(obj, dict):
         new_obj = {} # don't clobber
         for k,v in iteritems(obj):
             new_obj[k] = extract_dates(v)
         obj = new_obj
     elif isinstance(obj, (list, tuple)):
         obj = [ extract_dates(o) for o in obj ]
     elif isinstance(obj, string_types):
         obj = parse_date(obj)
     return obj
 
 def squash_dates(obj):
     """squash datetime objects into ISO8601 strings"""
     if isinstance(obj, dict):
         obj = dict(obj) # don't clobber
         for k,v in iteritems(obj):
             obj[k] = squash_dates(v)
     elif isinstance(obj, (list, tuple)):
         obj = [ squash_dates(o) for o in obj ]
     elif isinstance(obj, datetime):
         obj = obj.isoformat()
     return obj
 
 def date_default(obj):
     """default function for packing datetime objects in JSON."""
     if isinstance(obj, datetime):
         return obj.isoformat()
     else:
         raise TypeError("%r is not JSON serializable"%obj)
 
 
 # constants for identifying png/jpeg data
 PNG = b'\x89PNG\r\n\x1a\n'
 # front of PNG base64-encoded
 PNG64 = b'iVBORw0KG'
 JPEG = b'\xff\xd8'
 # front of JPEG base64-encoded
 JPEG64 = b'/9'
+# front of PDF base64-encoded
+PDF64 = b'JVBER'
 
 def encode_images(format_dict):
     """b64-encodes images in a displaypub format dict
 
     Perhaps this should be handled in json_clean itself?
 
     Parameters
     ----------
 
     format_dict : dict
         A dictionary of display data keyed by mime-type
 
     Returns
     -------
 
     format_dict : dict
         A copy of the same dictionary,
-        but binary image data ('image/png' or 'image/jpeg')
+        but binary image data ('image/png', 'image/jpeg' or 'application/pdf')
         is base64-encoded.
 
     """
     encoded = format_dict.copy()
 
     pngdata = format_dict.get('image/png')
     if isinstance(pngdata, bytes):
         # make sure we don't double-encode
         if not pngdata.startswith(PNG64):
             pngdata = encodebytes(pngdata)
         encoded['image/png'] = pngdata.decode('ascii')
 
     jpegdata = format_dict.get('image/jpeg')
     if isinstance(jpegdata, bytes):
         # make sure we don't double-encode
         if not jpegdata.startswith(JPEG64):
             jpegdata = encodebytes(jpegdata)
         encoded['image/jpeg'] = jpegdata.decode('ascii')
 
+    pdfdata = format_dict.get('application/pdf')
+    if isinstance(pdfdata, bytes):
+        # make sure we don't double-encode
+        if not pdfdata.startswith(PDF64):
+            pdfdata = encodebytes(pdfdata)
+        encoded['application/pdf'] = pdfdata.decode('ascii')
+
     return encoded
 
 
 def json_clean(obj):
     """Clean an object to ensure it's safe to encode in JSON.
 
     Atomic, immutable objects are returned unmodified.  Sets and tuples are
     converted to lists, lists are copied and dicts are also copied.
 
     Note: dicts whose keys could cause collisions upon encoding (such as a dict
     with both the number 1 and the string '1' as keys) will cause a ValueError
     to be raised.
 
     Parameters
     ----------
     obj : any python object
 
     Returns
     -------
     out : object
 
       A version of the input which will not cause an encoding error when
       encoded as JSON.  Note that this function does not *encode* its inputs,
       it simply sanitizes it so that there will be no encoding errors later.
 
     Examples
     --------
     >>> json_clean(4)
     4
     >>> json_clean(list(range(10)))
     [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
     >>> sorted(json_clean(dict(x=1, y=2)).items())
     [('x', 1), ('y', 2)]
     >>> sorted(json_clean(dict(x=1, y=2, z=[1,2,3])).items())
     [('x', 1), ('y', 2), ('z', [1, 2, 3])]
     >>> json_clean(True)
     True
     """
     # types that are 'atomic' and ok in json as-is.
     atomic_ok = (unicode_type, type(None))
 
     # containers that we need to convert into lists
     container_to_list = (tuple, set, types.GeneratorType)
 
     if isinstance(obj, float):
         # cast out-of-range floats to their reprs
         if math.isnan(obj) or math.isinf(obj):
             return repr(obj)
         return float(obj)
     
     if isinstance(obj, int):
         # cast int to int, in case subclasses override __str__ (e.g. boost enum, #4598)
         if isinstance(obj, bool):
             # bools are ints, but we don't want to cast them to 0,1
             return obj
         return int(obj)
 
     if isinstance(obj, atomic_ok):
         return obj
 
     if isinstance(obj, bytes):
         return obj.decode(DEFAULT_ENCODING, 'replace')
 
     if isinstance(obj, container_to_list) or (
         hasattr(obj, '__iter__') and hasattr(obj, next_attr_name)):
         obj = list(obj)
 
     if isinstance(obj, list):
         return [json_clean(x) for x in obj]
 
     if isinstance(obj, dict):
         # First, validate that the dict won't lose data in conversion due to
         # key collisions after stringification.  This can happen with keys like
         # True and 'true' or 1 and '1', which collide in JSON.
         nkeys = len(obj)
         nkeys_collapsed = len(set(map(str, obj)))
         if nkeys != nkeys_collapsed:
             raise ValueError('dict can not be safely converted to JSON: '
                              'key collision would lead to dropped values')
         # If all OK, proceed by making the new dict that will be json-safe
         out = {}
         for k,v in iteritems(obj):
             out[str(k)] = json_clean(v)
         return out
 
     # If we get here, we don't know how to handle the object, so we just get
     # its repr and return that.  This will catch lambdas, open sockets, class
     # objects, and any other complicated contraption that json can't encode
     return repr(obj)

IPython/utils/tests/test_jsonutil.py

@@ -1,147 +1,149 b''
 """Test suite for our JSON utilities.
 """
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010-2011  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING.txt, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 # stdlib
 import datetime
 import json
 from base64 import decodestring
 
 # third party
 import nose.tools as nt
 
 # our own
 from IPython.utils import jsonutil, tz
 from ..jsonutil import json_clean, encode_images
 from ..py3compat import unicode_to_str, str_to_bytes, iteritems
 
 #-----------------------------------------------------------------------------
 # Test functions
 #-----------------------------------------------------------------------------
 class Int(int):
     def __str__(self):
         return 'Int(%i)' % self
 
 def test():
     # list of input/expected output.  Use None for the expected output if it
     # can be the same as the input.
     pairs = [(1, None), # start with scalars
              (1.0, None),
              ('a', None),
              (True, None),
              (False, None),
              (None, None),
              # complex numbers for now just go to strings, as otherwise they
              # are unserializable
              (1j, '1j'),
              # Containers
              ([1, 2], None),
              ((1, 2), [1, 2]),
              (set([1, 2]), [1, 2]),
              (dict(x=1), None),
              ({'x': 1, 'y':[1,2,3], '1':'int'}, None),
              # More exotic objects
              ((x for x in range(3)), [0, 1, 2]),
              (iter([1, 2]), [1, 2]),
              (Int(5), 5),
              ]
     
     for val, jval in pairs:
         if jval is None:
             jval = val
         out = json_clean(val)
         # validate our cleanup
         nt.assert_equal(out, jval)
         # and ensure that what we return, indeed encodes cleanly
         json.loads(json.dumps(out))
 
 
 
 def test_encode_images():
     # invalid data, but the header and footer are from real files
     pngdata = b'\x89PNG\r\n\x1a\nblahblahnotactuallyvalidIEND\xaeB`\x82'
     jpegdata = b'\xff\xd8\xff\xe0\x00\x10JFIFblahblahjpeg(\xa0\x0f\xff\xd9'
+    pdfdata = b'%PDF-1.\ntrailer<</Root<</Pages<</Kids[<</MediaBox[0 0 3 3]>>]>>>>>>'
     
     fmt = {
         'image/png'  : pngdata,
         'image/jpeg' : jpegdata,
+        'application/pdf' : pdfdata
     }
     encoded = encode_images(fmt)
     for key, value in iteritems(fmt):
         # encoded has unicode, want bytes
         decoded = decodestring(encoded[key].encode('ascii'))
         nt.assert_equal(decoded, value)
     encoded2 = encode_images(encoded)
     nt.assert_equal(encoded, encoded2)
     
     b64_str = {}
     for key, encoded in iteritems(encoded):
         b64_str[key] = unicode_to_str(encoded)
     encoded3 = encode_images(b64_str)
     nt.assert_equal(encoded3, b64_str)
     for key, value in iteritems(fmt):
         # encoded3 has str, want bytes
         decoded = decodestring(str_to_bytes(encoded3[key]))
         nt.assert_equal(decoded, value)
 
 def test_lambda():
     jc = json_clean(lambda : 1)
     assert isinstance(jc, str)
     assert '<lambda>' in jc
     json.dumps(jc)
 
 def test_extract_dates():
     timestamps = [
         '2013-07-03T16:34:52.249482',
         '2013-07-03T16:34:52.249482Z',
         '2013-07-03T16:34:52.249482Z-0800',
         '2013-07-03T16:34:52.249482Z+0800',
         '2013-07-03T16:34:52.249482Z+08:00',
         '2013-07-03T16:34:52.249482Z-08:00',
         '2013-07-03T16:34:52.249482-0800',
         '2013-07-03T16:34:52.249482+0800',
         '2013-07-03T16:34:52.249482+08:00',
         '2013-07-03T16:34:52.249482-08:00',
     ]
     extracted = jsonutil.extract_dates(timestamps)
     ref = extracted[0]
     for dt in extracted:
         nt.assert_true(isinstance(dt, datetime.datetime))
         nt.assert_equal(dt, ref)
 
 def test_parse_ms_precision():
     base = '2013-07-03T16:34:52.'
     digits = '1234567890'
     
     for i in range(len(digits)):
         ts = base + digits[:i]
         parsed = jsonutil.parse_date(ts)
         if i >= 1 and i <= 6:
             assert isinstance(parsed, datetime.datetime)
         else:
             assert isinstance(parsed, str)
 
 def test_date_default():
     data = dict(today=datetime.datetime.now(), utcnow=tz.utcnow())
     jsondata = json.dumps(data, default=jsonutil.date_default)
     nt.assert_in("+00", jsondata)
     nt.assert_equal(jsondata.count("+00"), 1)
     extracted = jsonutil.extract_dates(json.loads(jsondata))
     for dt in extracted.values():
         nt.assert_true(isinstance(dt, datetime.datetime))
 
 def test_exception():
     bad_dicts = [{1:'number', '1':'string'},
                  {True:'bool', 'True':'string'},
                  ]
     for d in bad_dicts:
         nt.assert_raises(ValueError, json_clean, d)