MarkdownProperty for Riakkit

markdownproperty.py

from riakkit import DictProperty, BaseProperty
import markdown
from lxml.html.clean import Cleaner

_cleaner = Cleaner(add_nofollow=True, style=True)
_markdown = markdown.Markdown(safe_mode="escape")

class MarkdownProperty(BaseProperty):
  class MD(DictProperty.DotDict):
    def get(self, markdown):
      return self.markdown if markdown else self.html

  @staticmethod
  def mdconverter(text):
    """Converts some text to a DotDict object with markdown and html as
    attribute. Note that the argument text is unstripped. All security
    have to go through this converter.

    Args:
      text: The input text probably directly submitted by the client.

    Returns:
      A DocDict object with an html field.
    """
    if isinstance(text, DictProperty.DotDict):
      return text
    elif isinstance(text, dict):
      return MarkdownProperty.MD(text)

    md = text
    # TODO: https://github.com/waylan/Python-Markdown/issues/101#issuecomment-5882555
    html = _markdown.convert(md)
    if len(html) > 0:
      html = _cleaner.clean_html(html)

    # TODO: Process Youtube Link

    return MarkdownProperty.MD({"markdown" : md, "html" : html})

  def __init__(self, mdconverter=None, **kwargs):
    BaseProperty.__init__(self, **kwargs)
    if mdconverter is not None:
      self.mdconverter = mdconverter

  def standardize(self, value):
    value = BaseProperty.standardize(self, value)
    return self.mdconverter(value)

  def convertFromDb(self, value):
    value = BaseProperty.convertFromDb(self, value)
    if value is None:
      return MarkdownProperty.MD({"markdown" : "", "html" : ""})
    return MarkdownProperty.MD(value)

  def defaultValue(self):
    return MarkdownProperty.MD({"markdown" : "", "html" : ""})