conference-archive/load_conference.py

#!/usr/bin/python3

import datetime

import lxml.etree

from confarchive import database, model

DB_URL = "postgresql:///confarchive"

database.init_db(DB_URL)

Element = lxml.etree._Element

meals = {"lunch", "dinner", "breakfast"}
non_talk_titles = {"afternoon break", "cheese and wine party", "debcamp", "job fair"}


def not_a_talk(title: str) -> bool:
    """Event with this title is not a talk."""
    return is_meal(title) or title.lower() in non_talk_titles


def is_meal(title: str) -> bool:
    """Event title represents a meal."""
    return title.lower() in meals


def read_field(root: Element, field: str) -> str | None:
    """Get conference field."""
    value = root.findtext(".//" + field)
    if value is None:
        return None
    assert isinstance(value, str)
    return value


def read_date_field(root: Element, field: str) -> datetime.date | None:
    """Read date from a field."""
    value = read_field(root, field)
    return parse_isodate(value) if value is not None else None


def read_required_field(root: Element, field: str) -> str:
    """Read a required field."""
    value = read_field(root, field)
    assert value
    return value


def parse_isodate(iso_date: str) -> datetime.date:
    """Read a date in ISO format."""
    return datetime.datetime.fromisoformat(iso_date).date()


def conference_obj(root: Element) -> model.Conference:
    """Build conference object."""
    e = root.find(".//conference")
    assert e is not None

    return model.Conference(
        title=read_required_field(e, "title"),
        start=read_date_field(e, "start"),
        end=read_date_field(e, "end"),
        timezone=read_field(e, "time_zone_name"),
    )


def build_event_object(
    e: Element, person_lookup: dict[int, model.Person]
) -> model.Event | None:
    """Build an event object."""
    title = read_required_field(e, "title")
    if not_a_talk(title):
        return None

    room = read_field(e, "room")
    slug = read_field(e, "slug")
    description = read_field(e, "description")
    event_type = read_field(e, "type")
    url = read_field(e, "url")

    persons_element = e.find(".//persons")
    if persons_element is None:
        return None
    people = []
    for p in persons_element:
        id_str = p.get("id")
        assert id_str is not None
        people.append(person_lookup[int(id_str)])

    print(title, people)

    return model.Event(
        title=title,
        room=room,
        slug=slug,
        description=description,
        event_type=event_type,
        url=url,
        people=people,
    )


def get_all_people(root: Element) -> list[tuple[int, str]]:
    people: dict[int, str] = {}
    for person in root.findall(".//person"):
        assert person.text
        person_id_str = person.get("id")
        assert person_id_str
        person_id = int(person_id_str)
        existing = people.get(person_id)
        if existing:
            assert person.text == existing
            continue
        people[person_id] = person.text

    return sorted(people.items())


def load(filename: str) -> None:
    """Load conference schedule."""
    root = lxml.etree.parse(filename).getroot()
    conf = conference_obj(root)
    database.session.add(conf)

    event_count = 0
    people = get_all_people(root)
    person_lookup = {}
    for person_id, name in people:
        person = model.Person.query.filter_by(name=name).first()
        if not person:
            person = model.Person(name=name)
            database.session.add(person)
        person_lookup[person_id] = person

    for day in root.findall(".//day"):
        for room in root.findall(".//room"):
            for event_element in root.findall(".//event"):
                event = build_event_object(event_element, person_lookup)
                if not event:
                    continue
                event.conference = conf
                print()
                database.session.add(event)
                event_count += 1
                if event_count > 10:
                    return None


load("/home/edward/src/2022/conference-gender-mix/schedules/debconf22")

database.session.commit()