pandora_cms/ontology/update.py

#!/usr/bin/python3
import json
import os
from collections import defaultdict

base = os.path.abspath(os.path.dirname(__file__))

keywords = json.load(open(os.path.join(base, 'keywords.json')))
ontology = json.load(open(os.path.join(base, 'ontology.json')))

def find_path(parent, root=None, path=None):
    if root is None:
        root = ontology
    if path is None:
        path = []
    if parent == 'location':
        return ['film', 'architecture', 'rhythmanalysis', 'the everyday', 'environment', 'location']
    for key in root:
        if key == parent:
            return path + [key]
        elif root[key]:
            r = find_path(parent, root[key], path + [key])
            if r:
                return r

def get_node(name, children, parent=None):
    children_ = [get_node(child, children[child], name) for child in children]
    children_.sort(key=lambda c: c['name'])
    node = {
        "size": len(children) + 100,
        "name": name,
        "children": children_
    }
    if not node['children']:
        del node['children']
        key = '%s: %s' % (parent, name)
        if key in keywords:
            node['size'] = keywords[key]
    return node


def render_children(root, indent=0):
    txt = ('\t' * indent) + root['name']
    if 'children' in root:
        parts = ''
        for child in root['children']:
            parts += '\n' + render_children(child, indent+1)
        txt += '\n'.join([('\t' * indent) + p for p in parts.split('\n')])
    return '\n'.join([l.rstrip() for l in txt.split('\n')])

if __name__ == '__main__':
    os.chdir(base)

    tree = defaultdict(dict)

    ontology_txt = ''

    for keyword in keywords:
        if ': ' not in keyword:
            parent = 'other'
            child = keyword
        else:
            parent, child = keyword.split(': ', 1)
        path = find_path(parent)
        if path:
            p = tree
            for part in path:
                if part not in p:
                    p[part] = {}
                p = p[part]
            p[child] = {}
        else:
            if parent not in tree['missing']:
                tree['missing'][parent] = {}
            tree['missing'][parent][child] = {}
            #print('missing root - %s: %s' % (parent, child))

    #print(json.dumps(tree, indent=4, sort_keys=True))
    sized_ontology = {
        "size": len(tree),
        "name": "CineUrban",
        "children": []
    }
    for name in tree:
        children = tree[name]
        child = get_node(name, tree[name], name)
        sized_ontology['children'].append(child)


    ontology_txt = render_children(sized_ontology)

    with open('../static/ontology/sized_ontology.json', 'w') as fd:
        json.dump(sized_ontology, fd, indent=4, sort_keys=True)
    with open('../static/ontology/ontology.txt', 'w') as fd:
        fd.write(ontology_txt)