How to use the pandocfilters.elt function in pandocfilters

# - lift first H1 element (if it starts the document) to the title
# - lift all other headers by one level
# - upcase all L2 headers
#
# This script requires the 'pandocfilters' package (pip install pandocfilters)

import json
import sys
import re
from collections import OrderedDict

import pandocfilters
from pandocfilters import walk, stringify, Header, Str, Strong, Span, attributes


MetaString = pandocfilters.elt('MetaString', 1)


def liftTitle(doc):
    "Lift the title from the document."
    meta = doc[0]
    content = doc[1]
    heading = None

    if content[0]['t'] == 'Header':
        if content[0]['c'][0] == 1:
            heading = content[0]

    if heading is None:
        print >> sys.stderr, 'warning: first block not a heading'
        sys.exit(1)

metadata = document['meta']
    elif document[0]:                # old API
        metadata = document[0]['unMeta']

    if 'draft' in metadata:
        DRAFT = metadata['draft']['c']
    else:
        DRAFT = False

    newDocument = document
    newDocument = walk(newDocument, handle_comments, format, metadata)

    # Need to ensure the LaTeX/beamer template knows if `mdframed` package is
    # required (when `` has been used).
    if (format == 'latex' or format == 'beamer') and USED_BOX:
        MetaList = elt('MetaList', 1)
        MetaInlines = elt('MetaInlines', 1)
        rawinlines = [MetaInlines([RawInline('tex',
                                             '\\RequirePackage{mdframed}')])]
        if 'header-includes' in metadata:
            headerIncludes = metadata['header-includes']
            if headerIncludes['t'] == 'MetaList':
                rawinlines += headerIncludes['c']
            else:  # headerIncludes['t'] == 'MetaInlines'
                rawinlines += [headerIncludes]
        metadata['header-includes'] = MetaList(rawinlines)
        newDocument['meta'] = metadata

    json.dump(newDocument, sys.stdout)

def isheader(key, value):
    return (key == 'Header')


math_label = r'\\label{(.*?)}'

def islabeledmath(key, value):
    return (key == 'Math' and re.search(math_label, value[1]))


def isattr(string):
    return string.startswith('{') and string.endswith('}')


# define a new Figure type - an image with attributes
Figure = pf.elt('Figure', 3)  # caption, target, attrs


def isfigure(key, value):
    return (key == 'Para' and len(value) == 2 and value[0]['t'] == 'Image')


def isattrfigure(key, value):
    return (key == 'Para'
            and value[0]['t'] == 'Image'
            and isattr(pf.stringify(value[1:])))


def isdivfigure(key, value):
    """Matches images contained in a Div with 'figure' as a class."""
    return (key == 'Div' and 'figure' in value[0][1])

def RawBlock(format, string):
    """Overwrite pandocfilters RawBlock so that html5
    and html raw output both use the html writer.
    """
    if format == 'html5':
        format = 'html'
    return pf.RawBlock(format, string)


# define a new type for internal references [pre, label, post]
InternalRef = pf.elt('InternalRef', 3)
# and multiple references [pre, [label], post]
# https://tex.stackexchange.com/questions/15728/multiple-references-with-autoref
# https://github.com/mathjax/MathJax/issues/71
# http://docs.mathjax.org/en/latest/tex.html#automatic-equation-numbering
MultiInternalRef = pf.elt('MultiInternalRef', 3)
# define a new Figure type - an image with attributes
Figure = pf.elt('Figure', 3)  # caption, target, attrs


def create_pandoc_multilink(strings, refs):
    inlines = [[pf.Str(str(s))] for s in strings]
    targets = [(r, "") for r in refs]
    links = [pf.Link(inline, target)
             for inline, target in zip(inlines, targets)]

    return join_items(links)


def join_items(items, method='append'):
    """Join the list of items together in the format

if format == 'html5':
        format = 'html'
    return pf.RawInline(format, string)


def RawBlock(format, string):
    """Overwrite pandocfilters RawBlock so that html5
    and html raw output both use the html writer.
    """
    if format == 'html5':
        format = 'html'
    return pf.RawBlock(format, string)


# define a new type for internal references [pre, label, post]
InternalRef = pf.elt('InternalRef', 3)
# and multiple references [pre, [label], post]
# https://tex.stackexchange.com/questions/15728/multiple-references-with-autoref
# https://github.com/mathjax/MathJax/issues/71
# http://docs.mathjax.org/en/latest/tex.html#automatic-equation-numbering
MultiInternalRef = pf.elt('MultiInternalRef', 3)
# define a new Figure type - an image with attributes
Figure = pf.elt('Figure', 3)  # caption, target, attrs


def create_pandoc_multilink(strings, refs):
    inlines = [[pf.Str(str(s))] for s in strings]
    targets = [(r, "") for r in refs]
    links = [pf.Link(inline, target)
             for inline, target in zip(inlines, targets)]

    return join_items(links)

and html raw output both use the html writer.
    """
    if format == 'html5':
        format = 'html'
    return pf.RawBlock(format, string)


# define a new type for internal references [pre, label, post]
InternalRef = pf.elt('InternalRef', 3)
# and multiple references [pre, [label], post]
# https://tex.stackexchange.com/questions/15728/multiple-references-with-autoref
# https://github.com/mathjax/MathJax/issues/71
# http://docs.mathjax.org/en/latest/tex.html#automatic-equation-numbering
MultiInternalRef = pf.elt('MultiInternalRef', 3)
# define a new Figure type - an image with attributes
Figure = pf.elt('Figure', 3)  # caption, target, attrs


def create_pandoc_multilink(strings, refs):
    inlines = [[pf.Str(str(s))] for s in strings]
    targets = [(r, "") for r in refs]
    links = [pf.Link(inline, target)
             for inline, target in zip(inlines, targets)]

    return join_items(links)


def join_items(items, method='append'):
    """Join the list of items together in the format

    'item[0]' if len(items) == 1
    'item[0] and item[1]' if len(items) == 2

elif document[0]:                # old API
        metadata = document[0]['unMeta']

    if 'draft' in metadata:
        DRAFT = metadata['draft']['c']
    else:
        DRAFT = False

    newDocument = document
    newDocument = walk(newDocument, handle_comments, format, metadata)

    # Need to ensure the LaTeX/beamer template knows if `mdframed` package is
    # required (when `` has been used).
    if (format == 'latex' or format == 'beamer') and USED_BOX:
        MetaList = elt('MetaList', 1)
        MetaInlines = elt('MetaInlines', 1)
        rawinlines = [MetaInlines([RawInline('tex',
                                             '\\RequirePackage{mdframed}')])]
        if 'header-includes' in metadata:
            headerIncludes = metadata['header-includes']
            if headerIncludes['t'] == 'MetaList':
                rawinlines += headerIncludes['c']
            else:  # headerIncludes['t'] == 'MetaInlines'
                rawinlines += [headerIncludes]
        metadata['header-includes'] = MetaList(rawinlines)
        newDocument['meta'] = metadata

    json.dump(newDocument, sys.stdout)