fix markdown images

use split-files for embedded images

add path2url

IPython/nbconvert/exporters/exporter.py

@@ -1,519 +1,520 b''
 """This module defines Exporter, a highly configurable converter
 that uses Jinja2 to export notebook files into different formats.
 """
 
 #-----------------------------------------------------------------------------
 # Copyright (c) 2013, the 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
 #-----------------------------------------------------------------------------
 
 from __future__ import print_function, absolute_import
 
 # Stdlib imports
 import io
 import os
 import inspect
 import copy
 import collections
 import datetime
 
 # other libs/dependencies
 from jinja2 import Environment, FileSystemLoader, ChoiceLoader, TemplateNotFound
 
 # IPython imports
 from IPython.config.configurable import LoggingConfigurable
 from IPython.config import Config
 from IPython.nbformat import current as nbformat
 from IPython.utils.traitlets import MetaHasTraits, DottedObjectName, Unicode, List, Dict, Any
 from IPython.utils.importstring import import_item
 from IPython.utils.text import indent
 from IPython.utils import py3compat
 
 from IPython.nbconvert import preprocessors as nbpreprocessors
 from IPython.nbconvert import filters
 
 #-----------------------------------------------------------------------------
 # Globals and constants
 #-----------------------------------------------------------------------------
 
 #Jinja2 extensions to load.
 JINJA_EXTENSIONS = ['jinja2.ext.loopcontrols']
 
 default_filters = {
         'indent': indent,
         'markdown2html': filters.markdown2html,
         'ansi2html': filters.ansi2html,
         'filter_data_type': filters.DataTypeFilter,
         'get_lines': filters.get_lines,
         'highlight2html': filters.highlight2html,
         'highlight2latex': filters.highlight2latex,
         'ipython2python': filters.ipython2python,
         'posix_path': filters.posix_path,
         'markdown2latex': filters.markdown2latex,
         'markdown2rst': filters.markdown2rst,
         'comment_lines': filters.comment_lines,
         'strip_ansi': filters.strip_ansi,
         'strip_dollars': filters.strip_dollars,
         'strip_files_prefix': filters.strip_files_prefix,
         'html2text' : filters.html2text,
         'add_anchor': filters.add_anchor,
         'ansi2latex': filters.ansi2latex,
         'strip_math_space': filters.strip_math_space,
         'wrap_text': filters.wrap_text,
         'escape_latex': filters.escape_latex,
-        'citation2latex': filters.citation2latex
+        'citation2latex': filters.citation2latex,
+        'path2url': filters.path2url,
 }
 
 #-----------------------------------------------------------------------------
 # Class
 #-----------------------------------------------------------------------------
 
 class ResourcesDict(collections.defaultdict):
     def __missing__(self, key):
         return ''
 
 
 class Exporter(LoggingConfigurable):
     """
     Exports notebooks into other file formats.  Uses Jinja 2 templating engine
     to output new formats.  Inherit from this class if you are creating a new
     template type along with new filters/preprocessors.  If the filters/
     preprocessors provided by default suffice, there is no need to inherit from
     this class.  Instead, override the template_file and file_extension
     traits via a config file.
 
     {filters}
     """
     
     # finish the docstring
     __doc__ = __doc__.format(filters = '- '+'\n    - '.join(default_filters.keys()))
 
 
     template_file = Unicode(u'default',
             config=True,
             help="Name of the template file to use")
     def _template_file_changed(self, name, old, new):
         if new=='default':
             self.template_file = self.default_template
         else:
             self.template_file = new
         self.template = None
         self._load_template()
     
     default_template = Unicode(u'')
     template = Any()
     environment = Any()
 
     file_extension = Unicode(
         'txt', config=True, 
         help="Extension of the file that should be written to disk"
         )
 
     template_path = List(['.'], config=True)
     def _template_path_changed(self, name, old, new):
         self._load_template()
 
     default_template_path = Unicode(
         os.path.join("..", "templates"), 
         help="Path where the template files are located.")
 
     template_skeleton_path = Unicode(
         os.path.join("..", "templates", "skeleton"), 
         help="Path where the template skeleton files are located.") 
 
     #Jinja block definitions
     jinja_comment_block_start = Unicode("", config=True)
     jinja_comment_block_end = Unicode("", config=True)
     jinja_variable_block_start = Unicode("", config=True)
     jinja_variable_block_end = Unicode("", config=True)
     jinja_logic_block_start = Unicode("", config=True)
     jinja_logic_block_end = Unicode("", config=True)
     
     #Extension that the template files use.    
     template_extension = Unicode(".tpl", config=True)
 
     #Configurability, allows the user to easily add filters and preprocessors.
     preprocessors = List(config=True,
         help="""List of preprocessors, by name or namespace, to enable.""")
 
     filters = Dict(config=True,
         help="""Dictionary of filters, by name and namespace, to add to the Jinja
         environment.""")
 
     default_preprocessors = List([nbpreprocessors.coalesce_streams,
                                  nbpreprocessors.SVG2PDFPreprocessor,
                                  nbpreprocessors.ExtractOutputPreprocessor,
                                  nbpreprocessors.CSSHTMLHeaderPreprocessor,
                                  nbpreprocessors.RevealHelpPreprocessor,
                                  nbpreprocessors.LatexPreprocessor,
                                  nbpreprocessors.SphinxPreprocessor],
         config=True,
         help="""List of preprocessors available by default, by name, namespace, 
         instance, or type.""")
 
 
     def __init__(self, config=None, extra_loaders=None, **kw):
         """
         Public constructor
     
         Parameters
         ----------
         config : config
             User configuration instance.
         extra_loaders : list[of Jinja Loaders]
             ordered list of Jinja loader to find templates. Will be tried in order
             before the default FileSystem ones.
         template : str (optional, kw arg)
             Template to use when exporting.
         """
         if not config:
             config = self.default_config
         
         super(Exporter, self).__init__(config=config, **kw)
 
         #Init
         self._init_template()
         self._init_environment(extra_loaders=extra_loaders)
         self._init_preprocessors()
         self._init_filters()
 
 
     @property
     def default_config(self):
         return Config()
     
     def _config_changed(self, name, old, new):
         """When setting config, make sure to start with our default_config"""
         c = self.default_config
         if new:
             c.merge(new)
         if c != old:
             self.config = c
         super(Exporter, self)._config_changed(name, old, c)
         
 
     def _load_template(self):
         """Load the Jinja template object from the template file
         
         This is a no-op if the template attribute is already defined,
         or the Jinja environment is not setup yet.
         
         This is triggered by various trait changes that would change the template.
         """
         if self.template is not None:
             return
         # called too early, do nothing
         if self.environment is None:
             return
         # Try different template names during conversion.  First try to load the
         # template by name with extension added, then try loading the template
         # as if the name is explicitly specified, then try the name as a 
         # 'flavor', and lastly just try to load the template by module name.
         module_name = self.__module__.rsplit('.', 1)[-1]
         try_names = []
         if self.template_file:
             try_names.extend([
                 self.template_file + self.template_extension,
                 self.template_file,
                 module_name + '_' + self.template_file + self.template_extension,
             ])
         try_names.append(module_name + self.template_extension)
         for try_name in try_names:
             self.log.debug("Attempting to load template %s", try_name)
             try:
                 self.template = self.environment.get_template(try_name)
             except (TemplateNotFound, IOError):
                 pass
             except Exception as e:
                 self.log.warn("Unexpected exception loading template: %s", try_name, exc_info=True)
             else:
                 self.log.info("Loaded template %s", try_name)
                 break
     
     def from_notebook_node(self, nb, resources=None, **kw):
         """
         Convert a notebook from a notebook node instance.
     
         Parameters
         ----------
         nb : Notebook node
         resources : dict (**kw) 
             of additional resources that can be accessed read/write by 
             preprocessors and filters.
         """
         nb_copy = copy.deepcopy(nb)
         resources = self._init_resources(resources)
 
         # Preprocess
         nb_copy, resources = self._preprocess(nb_copy, resources)
 
         self._load_template()
 
         if self.template is not None:
             output = self.template.render(nb=nb_copy, resources=resources)
         else:
             raise IOError('template file "%s" could not be found' % self.template_file)
         return output, resources
 
 
     def from_filename(self, filename, resources=None, **kw):
         """
         Convert a notebook from a notebook file.
     
         Parameters
         ----------
         filename : str
             Full filename of the notebook file to open and convert.
         """
 
         #Pull the metadata from the filesystem.
         if resources is None:
             resources = ResourcesDict()
         if not 'metadata' in resources or resources['metadata'] == '':
             resources['metadata'] = ResourcesDict()
         basename = os.path.basename(filename)
         notebook_name = basename[:basename.rfind('.')]
         resources['metadata']['name'] = notebook_name
 
         modified_date = datetime.datetime.fromtimestamp(os.path.getmtime(filename))
         resources['metadata']['modified_date'] = modified_date.strftime("%B %d, %Y")
         
         with io.open(filename) as f:
             return self.from_notebook_node(nbformat.read(f, 'json'), resources=resources,**kw)
 
 
     def from_file(self, file_stream, resources=None, **kw):
         """
         Convert a notebook from a notebook file.
     
         Parameters
         ----------
         file_stream : file-like object
             Notebook file-like object to convert.
         """
         return self.from_notebook_node(nbformat.read(file_stream, 'json'), resources=resources, **kw)
 
 
     def register_preprocessor(self, preprocessor, enabled=False):
         """
         Register a preprocessor.
         Preprocessors are classes that act upon the notebook before it is
         passed into the Jinja templating engine.  Preprocessors are also
         capable of passing additional information to the Jinja
         templating engine.
     
         Parameters
         ----------
         preprocessor : preprocessor
         """
         if preprocessor is None:
             raise TypeError('preprocessor')
         isclass = isinstance(preprocessor, type)
         constructed = not isclass
 
         #Handle preprocessor's registration based on it's type
         if constructed and isinstance(preprocessor, py3compat.string_types):
             #Preprocessor is a string, import the namespace and recursively call
             #this register_preprocessor method
             preprocessor_cls = import_item(preprocessor)
             return self.register_preprocessor(preprocessor_cls, enabled)
         
         if constructed and hasattr(preprocessor, '__call__'):
             #Preprocessor is a function, no need to construct it.
             #Register and return the preprocessor.
             if enabled:
                 preprocessor.enabled = True
             self._preprocessors.append(preprocessor)
             return preprocessor
 
         elif isclass and isinstance(preprocessor, MetaHasTraits):
             #Preprocessor is configurable.  Make sure to pass in new default for 
             #the enabled flag if one was specified.
             self.register_preprocessor(preprocessor(parent=self), enabled)
 
         elif isclass:
             #Preprocessor is not configurable, construct it
             self.register_preprocessor(preprocessor(), enabled)
 
         else:
             #Preprocessor is an instance of something without a __call__ 
             #attribute.  
             raise TypeError('preprocessor')
 
 
     def register_filter(self, name, jinja_filter):
         """
         Register a filter.
         A filter is a function that accepts and acts on one string.  
         The filters are accesible within the Jinja templating engine.
     
         Parameters
         ----------
         name : str
             name to give the filter in the Jinja engine
         filter : filter
         """
         if jinja_filter is None:
             raise TypeError('filter')
         isclass = isinstance(jinja_filter, type)
         constructed = not isclass
 
         #Handle filter's registration based on it's type
         if constructed and isinstance(jinja_filter, py3compat.string_types):
             #filter is a string, import the namespace and recursively call
             #this register_filter method
             filter_cls = import_item(jinja_filter)
             return self.register_filter(name, filter_cls)
         
         if constructed and hasattr(jinja_filter, '__call__'):
             #filter is a function, no need to construct it.
             self.environment.filters[name] = jinja_filter
             return jinja_filter
 
         elif isclass and isinstance(jinja_filter, MetaHasTraits):
             #filter is configurable.  Make sure to pass in new default for 
             #the enabled flag if one was specified.
             filter_instance = jinja_filter(parent=self)
             self.register_filter(name, filter_instance )
 
         elif isclass:
             #filter is not configurable, construct it
             filter_instance = jinja_filter()
             self.register_filter(name, filter_instance)
 
         else:
             #filter is an instance of something without a __call__ 
             #attribute.  
             raise TypeError('filter')
 
         
     def _init_template(self):
         """
         Make sure a template name is specified.  If one isn't specified, try to
         build one from the information we know.
         """
         self._template_file_changed('template_file', self.template_file, self.template_file)
         
 
     def _init_environment(self, extra_loaders=None):
         """
         Create the Jinja templating environment.
         """
         here = os.path.dirname(os.path.realpath(__file__))
         loaders = []
         if extra_loaders:
             loaders.extend(extra_loaders)
 
         paths = self.template_path
         paths.extend([os.path.join(here, self.default_template_path),
                       os.path.join(here, self.template_skeleton_path)])
         loaders.append(FileSystemLoader(paths))
 
         self.environment = Environment(
             loader= ChoiceLoader(loaders),
             extensions=JINJA_EXTENSIONS
             )
         
         #Set special Jinja2 syntax that will not conflict with latex.
         if self.jinja_logic_block_start:
             self.environment.block_start_string = self.jinja_logic_block_start
         if self.jinja_logic_block_end:
             self.environment.block_end_string = self.jinja_logic_block_end
         if self.jinja_variable_block_start:
             self.environment.variable_start_string = self.jinja_variable_block_start
         if self.jinja_variable_block_end:
             self.environment.variable_end_string = self.jinja_variable_block_end
         if self.jinja_comment_block_start:
             self.environment.comment_start_string = self.jinja_comment_block_start
         if self.jinja_comment_block_end:
             self.environment.comment_end_string = self.jinja_comment_block_end
 
     
     def _init_preprocessors(self):
         """
         Register all of the preprocessors needed for this exporter, disabled
         unless specified explicitly.
         """
         self._preprocessors = []
 
         #Load default preprocessors (not necessarly enabled by default).
         if self.default_preprocessors:
             for preprocessor in self.default_preprocessors:
                 self.register_preprocessor(preprocessor)
 
         #Load user preprocessors.  Enable by default.
         if self.preprocessors:
             for preprocessor in self.preprocessors:
                 self.register_preprocessor(preprocessor, enabled=True)
  
 
     def _init_filters(self):
         """
         Register all of the filters required for the exporter.
         """
         
         #Add default filters to the Jinja2 environment
         for key, value in default_filters.items():
             self.register_filter(key, value)
 
         #Load user filters.  Overwrite existing filters if need be.
         if self.filters:
             for key, user_filter in self.filters.items():
                 self.register_filter(key, user_filter)
 
 
     def _init_resources(self, resources):
 
         #Make sure the resources dict is of ResourcesDict type.
         if resources is None:
             resources = ResourcesDict()
         if not isinstance(resources, ResourcesDict):
             new_resources = ResourcesDict()
             new_resources.update(resources)
             resources = new_resources 
 
         #Make sure the metadata extension exists in resources
         if 'metadata' in resources:
             if not isinstance(resources['metadata'], ResourcesDict):
                 resources['metadata'] = ResourcesDict(resources['metadata'])
         else:
             resources['metadata'] = ResourcesDict()
             if not resources['metadata']['name']: 
                 resources['metadata']['name'] = 'Notebook'
 
         #Set the output extension
         resources['output_extension'] = self.file_extension
         return resources
         
 
     def _preprocess(self, nb, resources):
         """
         Preprocess the notebook before passing it into the Jinja engine.
         To preprocess the notebook is to apply all of the 
     
         Parameters
         ----------
         nb : notebook node
             notebook that is being exported.
         resources : a dict of additional resources that
             can be accessed read/write by preprocessors
             and filters.
         """
         
         # Do a copy.deepcopy first, 
         # we are never safe enough with what the preprocessors could do.
         nbc =  copy.deepcopy(nb)
         resc = copy.deepcopy(resources)
 
         #Run each preprocessor on the notebook.  Carry the output along
         #to each preprocessor
         for preprocessor in self._preprocessors:
             nbc, resc = preprocessor(nbc, resc)
         return nbc, resc

IPython/nbconvert/exporters/markdown.py

@@ -1,31 +1,38 b''
 """
 Exporter that will export your ipynb to Markdown.
 """
 #-----------------------------------------------------------------------------
 # Copyright (c) 2013, the 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
 #-----------------------------------------------------------------------------
 
+from IPython.config import Config
 from IPython.utils.traitlets import Unicode
 
 from .exporter import Exporter
 
 #-----------------------------------------------------------------------------
 # Classes
 #-----------------------------------------------------------------------------
 
 class MarkdownExporter(Exporter):
     """
     Exports to a markdown document (.md)
     """
     
     file_extension = Unicode(
         'md', config=True, 
         help="Extension of the file that should be written to disk")
+
+    @property
+    def default_config(self):
+        c = Config({'ExtractOutputPreprocessor':{'enabled':True}})
+        c.merge(super(MarkdownExporter,self).default_config)
+        return c

IPython/nbconvert/filters/strings.py

@@ -1,183 +1,190 b''
 # coding: utf-8
 """String filters.
 
 Contains a collection of useful string manipulation filters for use in Jinja
 templates.
 """
 #-----------------------------------------------------------------------------
 # Copyright (c) 2013, the 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
 #-----------------------------------------------------------------------------
 
 import os
 import re
 import textwrap
+from urllib2 import quote
 from xml.etree import ElementTree
 
 from IPython.core.interactiveshell import InteractiveShell
 from IPython.utils import py3compat
 
 #-----------------------------------------------------------------------------
 # Functions
 #-----------------------------------------------------------------------------
 
 __all__ = [
     'wrap_text',
     'html2text',
     'add_anchor',
     'strip_dollars',
     'strip_files_prefix',
     'comment_lines',
     'get_lines',
     'ipython2python',
     'posix_path',
+    'path2url',
 ]
 
 
 def wrap_text(text, width=100):
     """ 
     Intelligently wrap text.
     Wrap text without breaking words if possible.
     
     Parameters
     ----------
     text : str
         Text to wrap.
     width : int, optional
         Number of characters to wrap to, default 100.
     """
 
     split_text = text.split('\n')
     wrp = map(lambda x:textwrap.wrap(x,width), split_text)
     wrpd = map('\n'.join, wrp)
     return '\n'.join(wrpd)
 
 
 def html2text(element):
     """extract inner text from html
     
     Analog of jQuery's $(element).text()
     """
     if isinstance(element, py3compat.string_types):
         element = ElementTree.fromstring(element)
     
     text = element.text or ""
     for child in element:
         text += html2text(child)
     text += (element.tail or "")
     return text
 
 
 def add_anchor(html):
     """Add an anchor-link to an html header tag
     
     For use in heading cells
     """
     h = ElementTree.fromstring(py3compat.cast_bytes_py2(html, encoding='utf-8'))
     link = html2text(h).replace(' ', '-')
     h.set('id', link)
     a = ElementTree.Element("a", {"class" : "anchor-link", "href" : "#" + link})
     a.text = u'¶'
     h.append(a)
 
     # Known issue of Python3.x, ElementTree.tostring() returns a byte string
     # instead of a text string.  See issue http://bugs.python.org/issue10942
     # Workaround is to make sure the bytes are casted to a string.
     return py3compat.decode(ElementTree.tostring(h), 'utf-8')
 
 
 def strip_dollars(text):
     """
     Remove all dollar symbols from text
     
     Parameters
     ----------
     text : str
         Text to remove dollars from
     """
 
     return text.strip('$')
 
 
 files_url_pattern = re.compile(r'(src|href)\=([\'"]?)files/')
 
 def strip_files_prefix(text):
     """
     Fix all fake URLs that start with `files/`,
     stripping out the `files/` prefix.
     
     Parameters
     ----------
     text : str
         Text in which to replace 'src="files/real...' with 'src="real...'
     """
     return files_url_pattern.sub(r"\1=\2", text)
 
 
 def comment_lines(text, prefix='# '):
     """
     Build a Python comment line from input text.
     
     Parameters
     ----------
     text : str
         Text to comment out.
     prefix : str
         Character to append to the start of each line.
     """
     
     #Replace line breaks with line breaks and comment symbols.
     #Also add a comment symbol at the beginning to comment out
     #the first line.
     return prefix + ('\n'+prefix).join(text.split('\n')) 
 
 
 def get_lines(text, start=None,end=None):
     """
     Split the input text into separate lines and then return the 
     lines that the caller is interested in.
     
     Parameters
     ----------
     text : str
         Text to parse lines from.
     start : int, optional
         First line to grab from.
     end : int, optional
         Last line to grab from.
     """
     
     # Split the input into lines.
     lines = text.split("\n")
     
     # Return the right lines.
     return "\n".join(lines[start:end]) #re-join
 
 def ipython2python(code):
     """Transform IPython syntax to pure Python syntax
 
     Parameters
     ----------
 
     code : str
         IPython code, to be transformed to pure Python
     """
     shell = InteractiveShell.instance()
     return shell.input_transformer_manager.transform_cell(code)
 
 def posix_path(path):
     """Turn a path into posix-style path/to/etc
     
     Mainly for use in latex on Windows,
     where native Windows paths are not allowed.
     """
     if os.path.sep != '/':
         return path.replace(os.path.sep, '/')
     return path
+
+def path2url(path):
+    """Turn a file path into a URL"""
+    parts = path.split(os.path.sep)
+    return '/'.join(quote(part) for part in parts)

IPython/nbconvert/templates/markdown.tpl

@@ -1,72 +1,72 b''
 {% extends 'display_priority.tpl' %}
 
 
 {% block in_prompt %}
 {% endblock in_prompt %}
 
 {% block output_prompt %}
 {%- endblock output_prompt %}
 
 {% block input %}
 {{ cell.input | indent(4)}}
 {% endblock input %}
 
 {% block pyerr %}
 {{ super() }}
 {% endblock pyerr %}
 
 {% block traceback_line %}
 {{ line | indent | strip_ansi }}
 {% endblock traceback_line %}
 
 {% block pyout %}
 
 {% block data_priority scoped %}
 {{ super() }}
 {% endblock %}
 {% endblock pyout %}
 
 {% block stream %}
 {{ output.text | indent }}
 {% endblock stream %}
 
 {% block data_svg %}
-[!image]({{ output.svg_filename }})
+![svg]({{ output.svg_filename | path2url }})
 {% endblock data_svg %}
 
 {% block data_png %}
-[!image]({{ output.png_filename }})
+![png]({{ output.png_filename | path2url }})
 {% endblock data_png %}
 
 {% block data_jpg %}
-[!image]({{ output.jpg_filename }})
+![jpeg]({{ output.jpeg_filename | path2url }})
 {% endblock data_jpg %}
 
 {% block data_latex %}
 {{ output.latex }}
 {% endblock data_latex %}
 
 {% block data_html scoped %}
 {{ output.html }}
 {% endblock data_html %}
 
 {% block data_text scoped %}
 {{ output.text | indent }}
 {% endblock data_text %}
 
 {% block markdowncell scoped %}
 {{ cell.source | wrap_text(80) }}
 {% endblock markdowncell %}
 
 
 {% block headingcell scoped %}
 {{ '#' * cell.level }} {{ cell.source | replace('\n', ' ') }}
 {% endblock headingcell %}
 
 {% block rawcell scoped %}
 {{ cell.source }}
 {% endblock rawcell %}
 
 {% block unknowncell scoped %}
 unknown type  {{ cell.type }}
 {% endblock unknowncell %}
  No newline at end of file