Compare commits

..

3 commits

Author SHA1 Message Date
Edward Betts 3493b2b81d Move some more code. 2023-09-25 15:20:01 +01:00
Edward Betts 7760ed0b58 Move code around. 2023-09-25 15:14:05 +01:00
Edward Betts 73e694d373 Show wikidata description on github report 2023-09-25 14:55:08 +01:00
4 changed files with 142 additions and 155 deletions

59
confarchive/query.py Normal file
View file

@ -0,0 +1,59 @@
"""Database queries."""
import typing
from sqlalchemy.orm.query import Query
from sqlalchemy.engine.cursor import CursorResult
from sqlalchemy import func
from . import database, model
def top_speakers() -> Query:
"""Find people who spoke at the most conferences."""
q: Query = (
database.session.query(model.Person, func.count())
.join(model.ConferencePerson)
.filter(model.Person.id != 1046) # FOSDEM Staff
.group_by(model.Person)
.order_by(func.count().desc(), model.Person.name)
)
return q
def top_events() -> Query:
"""Most common titles of events."""
q: Query = (
database.session.query(model.Event.title, func.count())
.group_by(model.Event.title)
.order_by(func.count().desc())
.having(func.count() > 3)
)
return q
def search_for_events(search_for: str) -> Query:
"""Search for events with by title."""
q: Query = model.Event.query.filter(
model.Event.title.ilike(f"%{search_for}%")
).order_by(model.Event.title)
return q
def search_for_people(search_for: str) -> Query:
"""Search for people by name."""
q: Query = model.Person.query.filter(
model.Person.name.ilike(f"%{search_for}%")
).order_by(model.Person.name)
return q
def speaker_counts() -> CursorResult:
"""Speaker/conference frequency distribution."""
sql = """
select num, count(*)
from (select person_id, count(*) as num from conference_person group by person_id) a
group by num
order by num
"""
return typing.cast(CursorResult, database.session.execute(sql))

11
confarchive/utils.py Normal file
View file

@ -0,0 +1,11 @@
"""Utility functions."""
def drop_start(s: str, start: str) -> str:
"""Remove text from the start of a string."""
return s[len(start) :] if s.startswith(start) else s
def plural(num: int, label: str) -> str:
"""Make plural version of label as appropriate."""
return f'{num:,d} {label}{"s" if num != 1 else ""}'

222
main.py
View file

@ -3,11 +3,10 @@
import os import os
import flask import flask
import sqlalchemy
from sqlalchemy import func, or_, update from sqlalchemy import func, or_, update
from werkzeug.wrappers import Response from werkzeug.wrappers import Response
from confarchive import database, model, wikidata from confarchive import database, model, wikidata, query, utils
app = flask.Flask(__name__) app = flask.Flask(__name__)
app.debug = True app.debug = True
@ -16,46 +15,6 @@ app.config.from_object("config.default")
database.init_app(app) database.init_app(app)
def top_speakers() -> sqlalchemy.orm.query.Query:
q = (
database.session.query(model.Person, func.count())
.join(model.ConferencePerson)
.filter(model.Person.id != 1046) # FOSDEM Staff
.group_by(model.Person)
.order_by(func.count().desc(), model.Person.name)
.having(func.count() > 4)
)
return q
def top_speakers2() -> sqlalchemy.orm.query.Query:
q = (
database.session.query(model.Person, func.count())
.join(model.ConferencePerson)
.filter(model.Person.name.like("% %"))
.group_by(model.Person)
.order_by(func.count().desc())
.having(func.count() > 2)
)
# .order_by(func.length(model.Person.name).desc())
return q
def top_events() -> sqlalchemy.orm.query.Query:
q = (
database.session.query(model.Event.title, func.count())
.group_by(model.Event.title)
.order_by(func.count().desc())
.having(func.count() > 3)
)
return q
def drop_start(s: str, start: str) -> str:
"""Remove text from the start of a string."""
return s[len(start) :] if s.startswith(start) else s
@app.route("/person/<int:person_id>", methods=["GET", "POST"]) @app.route("/person/<int:person_id>", methods=["GET", "POST"])
def person(person_id: int) -> str | Response: def person(person_id: int) -> str | Response:
item = model.Person.query.get(person_id) item = model.Person.query.get(person_id)
@ -69,7 +28,7 @@ def person(person_id: int) -> str | Response:
if "P18" in wd_item["claims"]: if "P18" in wd_item["claims"]:
claim_p18 = wd_item["claims"]["P18"] claim_p18 = wd_item["claims"]["P18"]
wikidata_photo = [ wikidata_photo = [
drop_start(s["mainsnak"]["datavalue"]["value"], "-") utils.drop_start(s["mainsnak"]["datavalue"]["value"], "-")
for s in claim_p18 for s in claim_p18
] ]
for filename in wikidata_photo: for filename in wikidata_photo:
@ -113,7 +72,7 @@ def person(person_id: int) -> str | Response:
"person.html", "person.html",
item=item, item=item,
Event=model.Event, Event=model.Event,
plural=plural, plural=utils.plural,
wikidata_hits=wikidata_hits, wikidata_hits=wikidata_hits,
is_admin=check_admin_mode, is_admin=check_admin_mode,
) )
@ -162,85 +121,73 @@ def search_people() -> str:
@app.route("/merge", methods=["GET", "POST"]) @app.route("/merge", methods=["GET", "POST"])
def merge() -> str | Response: def merge() -> str | Response:
assert app.config["ADMIN_MODE"] """Merge speakers."""
assert check_admin_mode()
if flask.request.method == "POST": if flask.request.method == "GET":
search_for = flask.request.form["q"]
item_ids_str = flask.request.form.getlist("person_id")
item_ids: list[int] = [int(i) for i in item_ids_str]
merge_to_id = min(item_ids)
other_ids = [i for i in item_ids if i != merge_to_id]
name_from_person_id = flask.request.form["name"]
print(other_ids, "->", merge_to_id)
with database.session.begin():
if merge_to_id != name_from_person_id:
merge_to = model.Person.query.get(merge_to_id)
name_from_person = model.Person.query.get(name_from_person_id)
merge_to.name = name_from_person.name
print("update ConferencePerson")
database.session.execute(
update(model.ConferencePerson)
.where(model.ConferencePerson.person_id.in_(other_ids))
.values(person_id=merge_to_id)
)
print("update EventPerson")
database.session.execute(
update(model.EventPerson)
.where(model.EventPerson.person_id.in_(other_ids))
.values(person_id=merge_to_id)
)
print("delete people")
for person_id in other_ids:
item = model.Person.query.get(person_id)
database.session.delete(item)
endpoint = flask.request.endpoint
assert endpoint
return flask.redirect(flask.url_for(endpoint, q=search_for))
else:
search_for = flask.request.args["q"] search_for = flask.request.args["q"]
assert search_for
search_for = search_for.strip()
q = query.search_for_people(search_for)
return flask.render_template("merge_people.html", q=q, search_for=search_for)
assert search_for assert flask.request.method == "POST"
search_for = search_for.strip() search_for = flask.request.form["q"]
q = model.Person.query.filter(model.Person.name.ilike(f"%{search_for}%")).order_by(
model.Person.name item_ids_str = flask.request.form.getlist("person_id")
) item_ids: list[int] = [int(i) for i in item_ids_str]
return flask.render_template("merge_people.html", q=q, search_for=search_for)
merge_to_id: int = min(item_ids)
other_ids = [i for i in item_ids if i != merge_to_id]
name_from_person_id = int(flask.request.form["name"])
print(other_ids, "->", merge_to_id)
with database.session.begin():
if merge_to_id != name_from_person_id:
merge_to = model.Person.query.get(merge_to_id)
name_from_person = model.Person.query.get(name_from_person_id)
merge_to.name = name_from_person.name
print("update ConferencePerson")
database.session.execute(
update(model.ConferencePerson)
.where(model.ConferencePerson.person_id.in_(other_ids))
.values(person_id=merge_to_id)
)
print("update EventPerson")
database.session.execute(
update(model.EventPerson)
.where(model.EventPerson.person_id.in_(other_ids))
.values(person_id=merge_to_id)
)
print("delete people")
for person_id in other_ids:
item = model.Person.query.get(person_id)
database.session.delete(item)
endpoint = flask.request.endpoint
assert endpoint
return flask.redirect(flask.url_for(endpoint, q=search_for))
@app.route("/events") @app.route("/events")
def events_page() -> str: def events_page() -> str:
"""Events page."""
search_for = flask.request.args.get("q") search_for = flask.request.args.get("q")
if not search_for: if search_for:
return flask.render_template("top_events.html", top_events=top_events()) q = query.search_for_events(search_for)
return flask.render_template("search_events.html", q=q, search_for=search_for)
q = model.Event.query.filter(model.Event.title.ilike(f"%{search_for}%")).order_by( else:
model.Event.title return flask.render_template("top_events.html", top_events=query.top_events())
)
return flask.render_template("search_events.html", q=q, search_for=search_for)
@app.route("/") @app.route("/")
def index() -> str: def index() -> str:
"""Start page.""" """Start page."""
if False:
q = (
model.Conference.query.order_by(model.Conference.start.desc())
.add_columns(
func.count(model.Event.id), func.count(model.ConferencePerson.person_id)
)
.group_by(model.Conference)
)
q = model.Conference.query.order_by(model.Conference.start.desc()) q = model.Conference.query.order_by(model.Conference.start.desc())
count = { count = {
@ -254,21 +201,6 @@ def index() -> str:
return flask.render_template("index.html", items=q, count=count) return flask.render_template("index.html", items=q, count=count)
def plural(num: int, label: str) -> str:
return f'{num:,d} {label}{"s" if num != 1 else ""}'
def speaker_counts():
sql = """
select num, count(*)
from (select person_id, count(*) as num from conference_person group by person_id) a
group by num
order by num
"""
return database.session.execute(sql)
@app.route("/series") @app.route("/series")
def list_series() -> str: def list_series() -> str:
"""Page showing list of conference series.""" """Page showing list of conference series."""
@ -279,33 +211,15 @@ def list_series() -> str:
@app.route("/speakers") @app.route("/speakers")
def top_speakers_page() -> str: def top_speakers_page() -> str:
top = top_speakers()
"""Top speakers page.""" """Top speakers page."""
photos = [] top = query.top_speakers().having(func.count() > 4)
for person, count in top:
photo = person.photo_filename()
if photo:
photos.append((person, photo))
left_photos = photos[::2]
right_photos = photos[1::2]
photo_person_ids = [person.id for person, photo in photos]
left = photo_person_ids[::2]
right = photo_person_ids[1::2]
return flask.render_template( return flask.render_template(
"top_speakers.html", "top_speakers.html",
top_speakers=top, top_speakers=top,
speaker_counts=speaker_counts(), speaker_counts=query.speaker_counts(),
plural=plural, plural=utils.plural,
person_image_filename=person_image_filename, person_image_filename=person_image_filename,
# photo_person_ids=photo_person_ids,
left=left,
right=right,
left_photos=left_photos,
right_photos=right_photos,
) )
@ -335,11 +249,13 @@ def add_venue(city_id: int) -> str | Response:
@app.route("/wikidata") @app.route("/wikidata")
def link_to_wikidata() -> str: def link_to_wikidata() -> str:
items = [] items = []
for person, num in top_speakers2(): top = (
if person.wikidata_qid: query.top_speakers()
continue .filter(model.Person.name.like("% %"), model.Person.wikidata_qid.is_(None))
q = person.name + " haswbstatement:P31=Q5" .having(func.count() > 2)
search_hits = wikidata.search(q) )
for person, num in top:
search_hits = wikidata.search(person.name + " haswbstatement:P31=Q5")
if not search_hits: if not search_hits:
continue continue
@ -452,11 +368,11 @@ def github_wikidata() -> str:
"""Look for speakers on Wikidata based on the GitHub property.""" """Look for speakers on Wikidata based on the GitHub property."""
items = [] items = []
for line in open("found_wikidata_github"): for line in open("found_wikidata_github"):
person_id, person_name, qid, wd_name, github, photo = eval(line) person_id, person_name, qid, wd_name, github, desc = eval(line)
person = model.Person.query.get(person_id) person = model.Person.query.get(person_id)
if person.wikidata_qid: if person.wikidata_qid:
continue continue
items.append((person, qid, wd_name, photo)) items.append((person, qid, wd_name, desc))
items.sort(key=lambda i: len(i[0].name)) items.sort(key=lambda i: len(i[0].name))

View file

@ -6,12 +6,13 @@
<h1>Conference archive</h1> <h1>Conference archive</h1>
<p>{{ items | count }} matches found</p> <p>{{ items | count }} matches found</p>
{% for person, qid, wd_name, photo in items %} {% for person, qid, wd_name, desc in items %}
<div> <div>
<a href="{{ url_for("person", person_id=person.id) }}">{{ person.name }}</a> <a href="{{ url_for("person", person_id=person.id) }}">{{ person.name }}</a>
## ##
<a href="https://www.wikidata.org/wiki/{{ qid }}">{{ wd_name }} ({{ qid }})</a> <a href="https://www.wikidata.org/wiki/{{ qid }}">{{ wd_name }} ({{ qid }})</a>
{% if photo %}📷{% endif %} ##
{{ desc }}
</div> </div>
{% endfor %} {% endfor %}