pandoc 1.0.0a13

Preamble

This project is not a Python binding for pandoc, the command-line tool. If this is what you need, you can either use pypandoc, pyandoc, etc., or create you own wrapper with subprocess or sh.

Instead, this library provides a Pythonic way to analyze, create and transform documents. It targets the developers which have this kind of need and are more productive in Python than in Haskell (the native language of the pandoc library).

The pandoc Python library only reads/writes documents in the pandoc JSON format. Use the pandoc command-line tool to convert documents in/to markdown, html, latex, etc.

Warning: this library is still in its alpha phase and not yet documented. At this stage, only a few examples are provided to give you a feel of the API.

Common Code

For all examples, we use the following imports

import json
import sys
import pandoc
from pandoc.types import *

and the following depth-first document iterator:

def iter(elt, enter=None, exit=None):
    yield elt
    if enter is not None:
        enter(elt)
    if isinstance(elt, dict):
        elt = elt.items()
    if hasattr(elt, "__iter__"): # exclude strings
        for child in elt:
             for subelt in iter(child, enter, exit):
                 yield subelt
    if exit is not None:
        exit(elt)

Mathematics

Define the file math.py to count the number of math items in documents:

def find_math(doc):
    return [elt for elt in iter(doc) if type(elt) is Math]

if __name__ == "__main__":
    doc = pandoc.read(json.load(sys.stdin))
    print "math:", len(find_math(doc)), "items."

Then, use it on the (markdown) document doc.txt:

$ pandoc -t json doc.txt | python math.py

Implicit Sections

I like to use bold text at the beginning of a paragraph to denote the existence of a low-level section. This pattern can be detected and the sections automatically explicited.

Define a sections.py file ; then, use the hooks defined in the depth-first iterator factory to provide the full path from the root to the element at each step:

def iter_path(elt):
    parents = []
    def enter(elt_):
        parents.append(elt_)
    def exit(elt_):
        parents.pop()
    for elt_ in iter(elt, enter, exit):
        yield parents + [elt_]

Leverage this new iterator to find the parent of an element:

def find_parent(doc, elt):
    for path in iter_path(doc):
        elt_ = path[-1]
        parent = path[-2] if len(path) >= 2 else None
        if elt is elt_:
             return parent

To detect a paragraph that is an implicit section, define:

def match_implicit_section(elt):
    if type(elt) is Para:
        content = elt[0]
        if len(content) >= 1 and type(content[0]) is Strong:
            return True
    return False

The transformation itself:

def explicit_sections(doc, level=6):
    for para in filter(match_implicit_section, iter(doc)):
        blocks = find_parent(doc, para)
        content = para[0].pop(0)[0]
        if len(para[0]) >= 1 and para[0][0] == Space():
            para[0].pop(0)
        index = blocks.index(para)
        header = Header(level, ("", [], []), content)
        blocks.insert(index, header)
    return doc

Finally, provide the command-line API with

if __name__ == "__main__":
    doc = pandoc.read(json.load(sys.stdin))
    doc = explicit_sections(doc)
    print json.dumps(pandoc.write(doc))

and use it like that:

$ pandoc -t json doc.txt | \
> python sections.py | \
> pandoc -f json -o doc2.txt