postparse.py

#!/usr/bin/env python

import re

START = "> "

BLOCKSEP_RE = re.compile("^ - ([a-zA-Z0-9\-_]+) -\s*$")

class Post(object):
    def __init__(self, postid, fields=None):
        self.postid = postid

        if fields is None:
            self.fields = {}
        else:
            self.fields = fields

    def get(self, field):
        if field in self.fields:
            return self.fields[field]
        else:
            return None

    def __getattr__(self, field):
        if field in self.fields:
            return self.fields[field]
        else:
            raise AttributeError


def parsePosts(posts):
    ret = []
    postbuf = []

    for line in posts.split("\n"):
        pass


def parsePost(postid, postlines):
    ret = Post(postid)

    blockvar = None
    blockbuf = []
    blockmatch = None

    for line in postlines:
        line = line.rstrip()

        if not line:
            continue

        blockmatch = BLOCKSEP_RE.match(line)

        # handle the block delimiter, dump to var if necessary
        if blockmatch:
            # start block
            if blockvar is None:
                blockvar = blockmatch.group(1)
                continue

            # end block
            elif blockmatch.group(1) == blockvar:
                ret.fields[blockvar] = "\n".join(blockbuf)
                blockbuf = []
                blockvar = None
                continue

        if blockvar:
            blockbuf.append(line)

        else:
            try:
                key, val = line.split(" = ", 1)
            except ValueError as e:
                raise ValueError("invalid post line: \"{}\"".format(line)) from e

            ret.fields[key] = val

    if blockvar:
        raise SyntaxError("unclosed block: \"{}\"".format(blockvar))


    return ret