#!/usr/bin/env python2.5

# Currently using Python 2.5, since PyXML has not been updated
# for Python 2.6

# Currently, the Sword importer, osis2mod, expects the OSIS file to be
# structured like a Bible - i.e. <book>, <chapter>, <verse>
# In general, a commentary might be structured using <div osisRef>
# instead.  For importing, we convert to the format osis2mod expects

import sys
import codecs
from xml.dom import minidom as dom
from swordutils.xml.utils import getFileWriter

def usage():
    print "Usage: genbookOsis2Commentary.py <osisFile>"
    print "Output is OSIS converted for use by osis2mod."

def isRoot(n):
    return (n.nodeType == dom.Document.ELEMENT_NODE
            and n.nodeName == u'osis')

def isVerseDiv(n):
    # we must only match things like: "Ps.1.1" 
    # and not: "Ps.1" 
    return n.nodeName == u'div' and n.attributes.has_key(u'osisID') \
        and len(n.attributes['osisID'].value.split('.')) == 3

class VerseRef(object):
    def __init__(self, ref):
        parts = ref.split('.')
        if len(parts) > 0:
            self.book = parts[0]
        else:
            self.book = None

        if len(parts) > 1:
            self.chapter = parts[1]
        else:
            self.chapter = None

        if len(parts) > 2:
            self.verse = parts[2]
        else:
            self.verse = None

def add_book_node(doc, node, verseRef):
    # Insert <div type="book"> node around verse
    div = doc.createElement(u'div')
    div.attributes[u'type'] = u'book'
    node.parentNode.insertBefore(div, node)
    
    #head = doc.createElement(u'head')
    #head.appendChild(doc.createTextNode(verseRef.book.title()))
    #div.appendChild(head)
    div.appendChild(node)
    return div

def add_chapter_node(doc, node, verseRef):
    # Insert <div type="chapter"> node around verse
    div = doc.createElement(u'div')
    div.attributes[u'type'] = u'chapter'
    node.parentNode.insertBefore(div, node)
    
    #head = doc.createElement(u'head')
    #head.appendChild(doc.createTextNode('Chapter %s' % verseRef.chapter))
    #div.appendChild(head)
    div.appendChild(node)
    return div

def versify_tree(doc, node, curRef=VerseRef(''), curBookNode=None, curChapterNode=None):
    # Traverse tree, modifying divisions to include <div type="book">,
    # <div type="chapter"> and <verse>

    if isVerseDiv(node):        
        node.tagName = u'verse'
        node.attributes.removeNamedItem(u'type')
        vr = VerseRef(node.attributes[u'osisID'].value)
        assert vr.book is not None
        if vr.book != curRef.book:
            curBookNode = add_book_node(doc, node, vr)
            curChapterNode = add_chapter_node(doc, node, vr)
        else:
            if vr.chapter != curRef.chapter:
                curChapterNode = add_chapter_node(doc, node, vr)
            else:
                # move the verse into the current chapter
                # We know that curChapterNode != None here,
                # because VerseRef('') never matches VerseRef(anythingelse)
                curChapterNode.appendChild(node)

        curRef = vr                

    elif node.childNodes.length > 0:
        for n in list(node.childNodes):
            curRef, curBookNode, curChapterNode = versify_tree(doc, n, curRef=curRef, curBookNode=curBookNode, curChapterNode=curChapterNode)

    return (curRef, curBookNode, curChapterNode)


def remove_non_commentary(node):
    """Recursivley removes any body text which is not part of commentary on a verse,
    returns True if the current node contains any commentary"""
    # The preamble before actual commentary is currently included by
    # osis2mod.  This is fairly annoying, and this function will
    # remove such text so that it is not included in the end product

    assert node is not None

    inCommentary = \
        (node.nodeName == u'div' and node.attributes.get('type') is not None and node.attributes['type'].value in ['book','chapter']) or \
        (node.nodeName == u'verse')

    inHeader = \
        (node.nodeName in [u'osis', u'osisText']) or\
        isRoot(node)
    
    if not inHeader and not inCommentary:
        # remove text contents of this item, or entire node
        # if it has no child elements
        for n in list(node.childNodes):
            if n.nodeType == n.TEXT_NODE:
                node.removeChild(n)
        if node.childNodes.length == 0:
            # Nothing more to do now, there is definitely
            # no commentary here.
            return False

    if inCommentary:
        # keep node and everything that is below node
        return True

    childrenHaveCommentary = False
    for n in list(node.childNodes):
        hadCommentary = remove_non_commentary(n)
        if hadCommentary:
            childrenHaveCommentary = True
        else:
            node.removeChild(n)
    return childrenHaveCommentary
            

def main(filename):
    d = dom.parse(filename)
    rootNode = filter(isRoot, d.childNodes)[0]
    versify_tree(d, rootNode)
    remove_non_commentary(rootNode)
    d.writexml(getFileWriter(sys.stdout), encoding="UTF-8")

if __name__ == "__main__":
    if len(sys.argv) != 2:
        usage()
        sys.exit(1)
    main(sys.argv[1])