Compare commits

...

8 commits

Author SHA1 Message Date
25056aaf33 Fix ZeroDivisionError when total mention count is zero
Guard the percentage calculation so it only renders when total > 0.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-11 14:21:15 +01:00
2eef8f480f Fix remaining 'Find Link' references and improve error messages
- Rename index page title and heading to 'Missing Link'
- Give a clear message for 429 rate limit errors
- Include response body snippet in other unexpected API errors
- Log warning when falling back to unauthenticated session

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-11 14:05:35 +01:00
c9b4e2face Add redirect support, live candidate list, per-article save count, and error pages
- Detect redirect targets (e.g. "handling stolen goods" → "Possession of
  stolen goods") and use piped links [[target|title]] in edits; exclude
  articles already linking to the redirect target from candidates
- Remove candidates from the list in real time as they are checked and
  found invalid, with live count update in the summary
- Track and display per-article save count in the stats line
- Rename "Find Link" to "Missing Link" throughout
- Show redirect target in the article heading
- Report save errors to the user via error page instead of crashing
- Filter self-links using case-insensitive first-letter comparison

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-11 13:53:35 +01:00
0239b83555 Track processed candidates in session and use OAuth for API calls
- Record skips, saves, and no-match results in session["skipped"] so
  revisiting an article resumes past already-checked candidates
- Filter self-links (case-insensitive first letter) from hit list
- Use OAuth session for all API reads when logged in for higher rate limits
- Add "for" template to exclusion list to avoid bad edits
- Improve API error handling with HTTP status codes logged to stderr

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-11 12:46:13 +01:00
2c197f5c43 Add live search progress, session counter, and fix URLs
- Search candidates client-side with JS, showing "Checking X..." spinner
  instead of leaving user waiting on a blank page
- Fix broken api_valid_hit endpoint (get_diff returns dict, not tuple)
- Remove server-side get_best_hit; article_page now returns candidate list
  immediately and JS iterates via /api/1/valid_hit
- URL now reflects current article via history.replaceState (?title=X),
  Skip navigates to ?after=X to advance past it
- Track saves in session; show count as green badge in navbar
- Add session counter incremented on each successful save

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-11 12:24:15 +01:00
bc6265d4cd Redesign UI and fix several bugs
- Add navbar with login/logout, search form, and Find Link branding
- Clean up index page: search-only, examples behind ?debug=1
- Improve article page: remove debug clutter, named Wikipedia links, collapsible candidates
- Add SVG favicon (🔗 emoji)
- Fix diff CSS: compact layout, auto table layout to eliminate wide marker column gap
- Catch TokenRequestDenied in OAuth start and show error page
- Store username in session at login; clear bad session on API failure
- Raise NoMatch when diff is empty (edit already applied)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-11 11:30:12 +01:00
7867122326 Fix for match starts inside one link and continues into the next opening link. 2026-05-11 10:51:11 +01:00
fad1ef9e49 Using bootstrap 5 2026-05-11 10:50:38 +01:00
12 changed files with 394 additions and 155 deletions

View file

@ -1,4 +1,5 @@
import re
import sys
import typing
import requests
@ -72,18 +73,33 @@ webpage_error = (
)
def _get_active_session() -> requests.sessions.Session:
"""Return OAuth session if one is available in Flask context, else plain session."""
try:
from flask import g
if hasattr(g, "oauth_session") and g.oauth_session is not None:
return g.oauth_session # type: ignore[return-value]
except RuntimeError:
pass
print("WARNING: using unauthenticated session", file=sys.stderr)
return get_session()
def api_get(params: StrDict) -> StrDict:
"""Make call to Wikipedia API."""
s = get_session()
s = _get_active_session()
r = s.get(get_query_url(), params=params)
try:
ret: StrDict = r.json()
except JSONDecodeError:
print(f"API request failed: HTTP {r.status_code}", file=sys.stderr)
print(f"Response body: {r.text!r}", file=sys.stderr)
if webpage_error in r.text:
raise MediawikiError(webpage_error)
else:
raise MediawikiError("unknown error")
if r.status_code == 429:
raise MediawikiError("Wikipedia rate limit exceeded — wait a moment and try again.")
raise MediawikiError(f"HTTP {r.status_code}: {r.text[:200]!r}")
check_for_error(ret)
return ret
@ -271,7 +287,7 @@ def call_get_diff(title: str, section_num: int, section_text: str) -> str:
"rvdifftotext": section_text.strip(),
}
s = get_session()
s = _get_active_session()
r = s.post(get_query_url(), data=data)
try:
ret = r.json()

View file

@ -78,7 +78,7 @@ re_cite = re.compile(
re.I | re.S,
)
re_cite_template_start = re.compile(r"\{\{(?:cite|citation|short description|gli|defn|annotated link|excerpt|main|see)\b", re.I)
re_cite_template_start = re.compile(r"\{\{(?:cite|citation|short description|gli|defn|annotated link|excerpt|main|see|for)\b", re.I)
re_no_param_template = re.compile(r"\{\{[^|{}]+\}\}")
re_external_link = re.compile(r"\[https?://[^\]]+\]")
# Italic text (work titles in bibliographies). Handles apostrophes in content
@ -252,6 +252,14 @@ def add_link(m: re.Match[str], replacement: str, text: str) -> str:
if matched_text.startswith("[[") and matched_text.endswith("|"):
return m.re.sub(lambda m: f"[[{replacement}|", text, count=1)
split_links = matched_text.find("]] [[")
if split_links > 0 and m.start() >= 2 and text[m.start() - 2 : m.start()] == "[[":
# Match starts inside one link and continues into the next opening link.
# Link only the text from the first link span and leave the second link as-is.
link_dest = replacement.split("|")[0] if "|" in replacement else replacement
visible = matched_text[:split_links]
return text[: m.start() - 2] + f"[[{link_dest}|{visible}]]" + text[m.start() + split_links + 2 :]
inner_bracket = matched_text.find("[[")
if inner_bracket > 0:
prefix = matched_text[:inner_bracket].rstrip()
@ -551,4 +559,6 @@ def get_diff(q: str, title: str, linkto: str | None) -> dict[str, typing.Any]:
)
found["diff"] = call_get_diff(title, found["section_num"], section_text)
if not found["diff"]:
raise NoMatch
return found

View file

@ -1,5 +1,6 @@
"""Wikipedia OAuth."""
import sys
import typing
import urllib
from typing import cast
@ -73,9 +74,8 @@ def api_request(params: typing.Mapping[str, str | int]) -> dict[str, typing.Any]
try:
return cast(dict[str, typing.Any], r.json())
except Exception:
print("text")
print(r.text)
print("---")
print(f"API request failed: HTTP {r.status_code}", file=sys.stderr)
print(f"Response body: {r.text!r}", file=sys.stderr)
raise
@ -99,13 +99,40 @@ def userinfo_call() -> typing.Mapping[str, typing.Any]:
return api_request(params)
def get_oauth_session() -> OAuth1Session | None:
"""Return an OAuth1Session for the current user, or None if not logged in."""
if "owner_key" not in session or "owner_secret" not in session:
return None
app = current_app
client_key = app.config["CLIENT_KEY"]
client_secret = app.config["CLIENT_SECRET"]
oauth = OAuth1Session(
client_key,
client_secret=client_secret,
resource_owner_key=session["owner_key"],
resource_owner_secret=session["owner_secret"],
)
oauth.headers.update({"User-Agent": ua})
oauth.params = typing.cast(
dict[str, str | int],
{"format": "json", "action": "query", "formatversion": 2},
)
return oauth
def get_username() -> None | str:
"""Get the username or None if not logged in."""
if "owner_key" not in session:
return None # not authorized
if "username" not in session:
reply = userinfo_call()
try:
reply = userinfo_call()
except Exception as e:
print(f"get_username failed, clearing session: {e}", file=sys.stderr)
session.pop("owner_key", None)
session.pop("owner_secret", None)
return None
if "query" not in reply:
return None
session["username"] = reply["query"]["userinfo"]["name"]

View file

@ -5,19 +5,16 @@ span.searchmatch { font-weight: bold; }
table.diff,td.diff-otitle,td.diff-ntitle{background-color:white}
td.diff-otitle,td.diff-ntitle{text-align:center}
td.diff-marker{text-align:right;font-weight:bold;font-size:1.25em}
td.diff-marker{width:1.5em;text-align:center;font-weight:bold;font-size:1.25em;padding:0 0.3em}
td.diff-lineno{font-weight:bold}
td.diff-addedline,td.diff-deletedline,td.diff-context{font-size:88%;vertical-align:top;white-space:-moz-pre-wrap;white-space:pre-wrap}
td.diff-addedline,td.diff-deletedline{border-style:solid;border-width:1px 1px 1px 4px;border-radius:0.33em}
td.diff-addedline{border-color:#a3d3ff}
td.diff-deletedline{border-color:#ffe49c}
td.diff-context{background:#f3f3f3;color:#333333;border-style:solid;border-width:1px 1px 1px 4px;border-color:#e6e6e6;border-radius:0.33em}
td.diff-addedline,td.diff-deletedline{border-left:3px solid}
td.diff-addedline{border-color:#a3d3ff;background:#f0f8ff}
td.diff-deletedline{border-color:#ffe49c;background:#fffaf0}
td.diff-context{color:#555}
.diffchange{font-weight:bold;text-decoration:none}
table.diff{border:none;width:98%;border-spacing:4px; table-layout:fixed}
td.diff-addedline .diffchange,td.diff-deletedline .diffchange{border-radius:0.33em;padding:0.25em 0}
table.diff{border:none;width:100%;border-spacing:0;border-collapse:collapse;table-layout:auto}
td.diff-addedline .diffchange{background:#d8ecff}
td.diff-deletedline .diffchange{background:#feeec8}
table.diff td{padding:0.33em 0.66em}
table.diff col.diff-marker{width:2%}
table.diff col.diff-content{width:48%}
table.diff td div{ word-wrap:break-word; overflow:auto}
table.diff td{padding:0.2em 0.5em}
table.diff td div{word-wrap:break-word;overflow:auto}

3
static/favicon.svg Normal file
View file

@ -0,0 +1,3 @@
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 100 100">
<text y=".9em" font-size="90">🔗</text>
</svg>

After

Width:  |  Height:  |  Size: 114 B

View file

@ -1,10 +1,11 @@
{% extends "base.html" %}
{% block title %}Index{% endblock %}
{% block title %}All done{% endblock %}
{% block content %}
<div class="container">
<h1>All done</h1>
<div><a href="{{ url_for('index') }}">back to index </a></div>
</div>
<div class="container text-center mt-5">
<h1 class="mb-3">All done</h1>
<p class="text-muted mb-4">No more candidates found for this article.</p>
<a href="{{ url_for('index') }}" class="btn btn-primary">Search another article</a>
</div>
{% endblock %}

View file

@ -1,48 +1,152 @@
{% extends "base.html" %}
{% block title %}Link '{{ title }}' in '{{ hit_title }}'{% endblock %}
{% block title %}{{ title }}{% endblock %}
{% block style %}
<link href="{{ url_for("static", filename="css/diff.css") }}" rel="stylesheet"/>
{% endblock %}
{% block content %}
<div class="container">
<h1>Link '{{ title }}' in '{{ hit_title }}'</h1>
<form action="{{ url_for("index") }}">
<input name="q">
<input type="submit" value="search">
</form>
<div class="container">
<nav aria-label="breadcrumb" class="mb-3">
<ol class="breadcrumb">
<li class="breadcrumb-item"><a href="{{ url_for('index') }}">Home</a></li>
<li class="breadcrumb-item active">{{ title }}</li>
</ol>
</nav>
<div>Username: {{ g.user }}</div>
<div class="d-flex flex-wrap align-items-baseline gap-3 mb-1">
<h1 class="h4 mb-0">Find links to "{{ title }}"</h1>
<a href="https://en.wikipedia.org/wiki/{{ title }}" target="_blank" class="text-muted small">{{ title }} ↗</a>
{% if redirect_to %}
<span class="text-muted small">→ redirects to <a href="https://en.wikipedia.org/wiki/{{ redirect_to }}" target="_blank">{{ redirect_to }} ↗</a></span>
{% endif %}
</div>
<div><a href="https://en.wikipedia.org/wiki/{{ title }}" target="_blank">view article</a></div>
<div class="d-flex gap-3 mb-3 text-muted small">
<span>{{ total }} mentions total</span>
<span>{{ with_link }} already linked{% if total > 0 %} ({{ "{:.0%}".format(with_link / total) }}){% endif %}</span>
{% if saves_this_session %}
<span class="text-success">{{ saves_this_session }} added this session</span>
{% endif %}
</div>
<div><a href="{{ url_for('index') }}">back to index </a></div>
<div id="search-progress" class="my-4">
<div class="d-flex align-items-center gap-2 text-muted">
<div class="spinner-border spinner-border-sm" role="status">
<span class="visually-hidden">Searching…</span>
</div>
<span id="search-status">Searching…</span>
</div>
</div>
<div>total: {{ total }}</div>
<div>with link: {{ with_link }}</div>
<div>ratio: {{ "{:.1%}".format(with_link / total) }}</div>
{# <div>hit: {{ hit }}</div> #}
<div>replacement: {{ found.replacement }}</div>
<div>section: {{ found.section }}</div>
<table>
{{ diff | safe }}
</table>
<form method="POST">
<input type="hidden" name="hit" value="{{ hit_title }}">
<div class="my-3">
<input type="submit" class="btn btn-primary" value="save"/>
<a href="{{url_for("article_page", url_title=url_title, after=hit_title)}}" class="btn btn-primary">skip</a>
<div id="result" hidden>
<div class="d-flex flex-wrap align-items-baseline gap-2 mb-3">
<span class="text-muted small">Adding link in</span>
<a id="result-hit-link" href="#" target="_blank" class="small"><span id="result-hit-title"></span></a>
</div>
<div class="mb-4">
<table class="diff" id="diff-table"></table>
</div>
<form method="POST" class="mb-4">
<input type="hidden" name="hit" id="hit-input">
<div class="d-flex gap-2">
<button type="submit" class="btn btn-success">Save edit</button>
<a id="skip-link" href="#" class="btn btn-outline-secondary">Skip</a>
</div>
</form>
</div>
<ol>
<div id="all-done" hidden class="text-center mt-4">
<p class="text-muted mb-4">No more candidates found for this article.</p>
<a href="{{ url_for('index') }}" class="btn btn-primary">Search another article</a>
</div>
{% if hits %}
<details id="candidates-section" class="border rounded p-3 mt-2">
<summary class="text-muted small" style="cursor:pointer"><span id="candidates-count">{{ hits | length }}</span> candidates</summary>
<ol class="mt-3 mb-0 small" id="candidates-list">
{% for hit in hits %}
{% set url = url_for("article_page", url_title=url_title, title=hit.title) %}
<li><a href="{{ url }}">{{ hit.title }}</a> &ndash; {{ hit.snippet | safe }}</li>
<li class="mb-1" data-title="{{ hit.title }}">
<a href="{{ url_for("article_page", url_title=url_title, title=hit.title) }}">{{ hit.title }}</a>
</li>
{% endfor %}
</ol>
</div>
</details>
{% endif %}
</div>
{% endblock %}
{% block script %}
<script>
(function () {
const hits = {{ hits | map(attribute='title') | list | tojson }};
const linkTo = {{ title | tojson }};
const redirectTo = {{ redirect_to | tojson }};
const apiUrl = {{ url_for('api_valid_hit') | tojson }};
const pageUrl = new URL(window.location.href);
const elProgress = document.getElementById('search-progress');
const elStatus = document.getElementById('search-status');
const elResult = document.getElementById('result');
const elAllDone = document.getElementById('all-done');
const elList = document.getElementById('candidates-list');
function removeCandidate(title) {
if (!elList) return;
const li = elList.querySelector(`li[data-title="${CSS.escape(title)}"]`);
if (!li) return;
li.remove();
const elCount = document.getElementById('candidates-count');
if (elCount) elCount.textContent = elList.children.length;
}
async function search() {
for (const hitTitle of hits) {
elStatus.textContent = `Checking "${hitTitle}"…`;
let data;
try {
const params = new URLSearchParams({ link_to: linkTo, link_from: hitTitle });
if (redirectTo) params.append('redirect_to', redirectTo);
const resp = await fetch(apiUrl + '?' + params);
if (!resp.ok) continue;
data = await resp.json();
} catch (e) {
continue;
}
if (!data.valid) { removeCandidate(hitTitle); continue; }
elProgress.hidden = true;
document.getElementById('result-hit-title').textContent = hitTitle;
document.getElementById('result-hit-link').href =
'https://en.wikipedia.org/wiki/' + encodeURIComponent(hitTitle.replace(/ /g, '_'));
document.getElementById('diff-table').innerHTML = data.diff;
document.getElementById('hit-input').value = hitTitle;
const skipUrl = new URL(pageUrl);
skipUrl.searchParams.delete('title');
skipUrl.searchParams.set('after', hitTitle);
document.getElementById('skip-link').href = skipUrl.toString();
const currentUrl = new URL(pageUrl);
currentUrl.searchParams.delete('after');
currentUrl.searchParams.set('title', hitTitle);
history.replaceState(null, '', currentUrl.toString());
elResult.hidden = false;
return;
}
elProgress.hidden = true;
const elCandidates = document.getElementById('candidates-section');
if (elCandidates) elCandidates.hidden = true;
elAllDone.hidden = false;
}
search();
}());
</script>
{% endblock %}

View file

@ -2,21 +2,38 @@
<html lang="en">
<head>
<meta charset="utf-8">
<link href="{{ url_for("static", filename="bootstrap/css/bootstrap.min.css") }}" rel="stylesheet">
<link href="{{ url_for("static", filename="bootstrap5/css/bootstrap.min.css") }}" rel="stylesheet">
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>
{% block title %}{% endblock %}
</title>
<link rel="icon" type="image/svg+xml" href="{{ url_for('static', filename='favicon.svg') }}">
<title>{% block title %}{% endblock %} Missing Link</title>
{% block style %}{% endblock %}
</head>
<body>
<nav class="navbar navbar-expand-md navbar-dark bg-dark mb-4">
<div class="container">
<a class="navbar-brand fw-semibold" href="{{ url_for('index') }}">Missing Link</a>
<div class="d-flex align-items-center gap-2">
<form class="d-flex" action="{{ url_for('index') }}">
<input class="form-control form-control-sm me-2" name="q" placeholder="Article title…" style="width:240px">
<button class="btn btn-outline-light btn-sm" type="submit">Go</button>
</form>
{% if g.user %}
<span class="text-light small opacity-75">{{ g.user }}</span>
{% if session.get("saves") %}
<span class="badge bg-success">{{ session["saves"] }} saved</span>
{% endif %}
<a class="btn btn-outline-light btn-sm" href="{{ url_for('oauth_disconnect') }}">Log out</a>
{% else %}
<a class="btn btn-outline-light btn-sm" href="{{ url_for('start_oauth') }}">Log in with Wikipedia</a>
{% endif %}
</div>
</div>
</nav>
{% block content %}{% endblock %}
<script src="{{ url_for("static", filename="bootstrap/js/bootstrap.bundle.min.js")}}></script>
<script src="{{ url_for("static", filename="bootstrap5/js/bootstrap.bundle.min.js") }}"></script>
{% block script %}{% endblock %}
</body>
</html>

17
templates/error.html Normal file
View file

@ -0,0 +1,17 @@
{% extends "base.html" %}
{% block title %}Error{% endblock %}
{% block content %}
<div class="container mt-5">
<div class="row justify-content-center">
<div class="col-md-7">
<div class="alert alert-danger">
<h4 class="alert-heading">Something went wrong</h4>
<p class="mb-0"><code>{{ message }}</code></p>
</div>
<a href="{{ url_for('index') }}" class="btn btn-outline-secondary btn-sm">Back to home</a>
</div>
</div>
</div>
{% endblock %}

View file

@ -1,25 +1,44 @@
{% extends "base.html" %}
{% block title %}Index{% endblock %}
{% block title %}Missing Link{% endblock %}
{% block content %}
<div class="container">
<h1>Index</h1>
<form>
<input name="q">
<input type="submit" value="search">
</form>
<div>Username: {{ g.user }}</div>
<table class="table w-auto">
{% for item in examples %}
<tr>
<td><a href="{{ article_url(item.title) }}">{{ item.title }}</a></td>
<td>{{ item.total }}</td>
<td>{{ "{:.1%}".format(item.with_links / item.total) }}</td>
</tr>
{% endfor %}
</table>
<div class="container">
<div class="row justify-content-center mt-5">
<div class="col-md-6 text-center">
<h1 class="mb-2">Missing Link</h1>
<p class="text-muted mb-4">Find unlinked mentions of a Wikipedia article and add the links.</p>
<form class="d-flex gap-2 justify-content-center" action="{{ url_for('index') }}">
<input class="form-control" name="q" placeholder="Article title…" style="max-width:360px" autofocus>
<button class="btn btn-primary" type="submit">Search</button>
</form>
</div>
</div>
{% if debug %}
<div class="row mt-5">
<div class="col">
<h2 class="h6 text-muted text-uppercase mb-3">Examples</h2>
<table class="table table-sm table-hover w-auto">
<thead class="table-light">
<tr>
<th>Article</th>
<th class="text-end">Total</th>
<th class="text-end">% linked</th>
</tr>
</thead>
<tbody>
{% for item in examples %}
<tr>
<td><a href="{{ article_url(item.title) }}">{{ item.title }}</a></td>
<td class="text-end text-muted">{{ item.total }}</td>
<td class="text-end text-muted">{{ "{:.0%}".format(item.with_links / item.total) }}</td>
</tr>
{% endfor %}
</tbody>
</table>
</div>
</div>
{% endif %}
</div>
{% endblock %}

View file

@ -1,10 +1,11 @@
{% extends "base.html" %}
{% block title %}Index{% endblock %}
{% block title %}Edit saved{% endblock %}
{% block content %}
<div class="container">
<h1>Save done</h1>
<div>Save is complete.</div>
</div>
<div class="container text-center mt-5">
<h1 class="mb-3">Edit saved</h1>
<p class="text-muted mb-4">Your edit has been saved to Wikipedia.</p>
<a href="{{ url_for('index') }}" class="btn btn-primary">Search another article</a>
</div>
{% endblock %}

View file

@ -4,11 +4,13 @@ import html
import itertools
import json
import re
import sys
import typing
import flask
import werkzeug
from requests_oauthlib import OAuth1Session
from requests_oauthlib.oauth1_session import TokenRequestDenied
from werkzeug.wrappers.response import Response
from add_links import api, core, mediawiki_api, mediawiki_oauth
@ -94,14 +96,20 @@ def search_count(q: str) -> int:
return get_hit_count(article_title_to_search_query(q)) - 1
def search_count_with_link(q: str) -> int:
def search_count_with_link(q: str, redirect_to: str | None = None) -> int:
"""Articles in Wikipedia that include this search term and a link."""
return get_hit_count(article_title_to_search_query(q) + f' linksto:"{q}"')
count = get_hit_count(article_title_to_search_query(q) + f' linksto:"{q}"')
if redirect_to:
count += get_hit_count(article_title_to_search_query(q) + f' linksto:"{redirect_to}"')
return count
def search_no_link(q: str) -> tuple[int, list[Hit]]:
def search_no_link(q: str, redirect_to: str | None = None) -> tuple[int, list[Hit]]:
"""Search for mentions of article title with no link included."""
query = run_search(article_title_to_search_query(q) + f' -linksto:"{q}"', "max")
exclude = f' -linksto:"{q}"'
if redirect_to:
exclude += f' -linksto:"{redirect_to}"'
query = run_search(article_title_to_search_query(q) + exclude, "max")
return (query["searchinfo"]["totalhits"], query["search"])
@ -109,6 +117,7 @@ def search_no_link(q: str) -> tuple[int, list[Hit]]:
def global_user() -> None:
"""Make username available everywhere."""
flask.g.user = mediawiki_oauth.get_username()
flask.g.oauth_session = mediawiki_oauth.get_oauth_session()
@app.route("/")
@ -118,17 +127,20 @@ def index() -> str | Response:
url = flask.url_for("oauth_callback", **flask.request.args) # type: ignore
return flask.redirect(url)
examples = load_examples()
examples.sort(
key=lambda i: float(i["with_links"]) / float(i["total"]), reverse=True
)
if q := flask.request.args.get("q"):
if q_trimmed := q.strip():
return flask.redirect(article_url(q_trimmed))
debug = flask.request.args.get("debug")
examples: list[dict[str, str | int]] = []
if debug:
examples = load_examples()
examples.sort(
key=lambda i: float(i["with_links"]) / float(i["total"]), reverse=True
)
return flask.render_template(
"index.html", examples=examples, article_url=article_url
"index.html", examples=examples, article_url=article_url, debug=debug
)
@ -187,7 +199,12 @@ def start_oauth() -> Response:
oauth = OAuth1Session(client_key, client_secret=client_secret, callback_uri="oob")
oauth.headers.update({"User-Agent": api.ua})
fetch_response = oauth.fetch_request_token(request_token_url)
try:
fetch_response = oauth.fetch_request_token(request_token_url)
except TokenRequestDenied as e:
return flask.make_response(
flask.render_template("error.html", message=str(e)), 502
)
flask.session["owner_key"] = fetch_response.get("oauth_token")
flask.session["owner_secret"] = fetch_response.get("oauth_token_secret")
@ -229,7 +246,8 @@ def oauth_callback() -> werkzeug.wrappers.response.Response:
flask.session["owner_key"] = oauth_tokens.get("oauth_token")
flask.session["owner_secret"] = oauth_tokens.get("oauth_token_secret")
print("login successful")
username = mediawiki_oauth.get_username()
print(f"login successful: {username}", file=sys.stderr)
next_page = flask.session.get("after_login")
return flask.redirect(next_page if next_page else flask.url_for("index"))
@ -281,31 +299,16 @@ def match_type(q: str, snippet: str) -> str | None:
return match
class NoGoodHit(Exception):
"""No good hit."""
def get_best_hit(title: str, hits: list[Hit]) -> tuple[Hit, dict[str, typing.Any]]:
"""Find the best hit within the search results."""
for hit in hits:
if hit["title"].lower() == title.lower():
continue
# if match_type(title, hit["snippet"]) != "exact":
# continue
try:
print(f'get diff: {hit["title"]}, {title}')
found = get_diff(title, hit["title"], None)
except NoMatch:
print("no match")
continue
except api.MediawikiError as e:
print(f"MediawikiError for {hit['title']!r}: {e}")
continue
return (hit, found)
raise NoGoodHit
def _record_skip(from_title: str, hit_title: str) -> None:
"""Record that a candidate was skipped or saved for this article."""
skipped: dict[str, list[str]] = flask.session.get("skipped", {})
article_skipped = skipped.get(from_title, [])
if hit_title not in article_skipped:
skipped[from_title] = article_skipped + [hit_title]
flask.session["skipped"] = skipped
flask.session.modified = True
def handle_post(url_title: str) -> Response:
@ -316,8 +319,16 @@ def handle_post(url_title: str) -> Response:
do_save(from_title, hit_title)
except mediawiki_oauth.LoginNeeded:
return flask.redirect(flask.url_for("start_oauth"))
except mediawiki_api.APIError as e:
return flask.make_response(f"Save failed: {e}", 502)
except (mediawiki_api.APIError, api.MediawikiError) as e:
return flask.make_response(
flask.render_template("error.html", message=f"Save failed: {e}"), 502
)
flask.session["saves"] = flask.session.get("saves", 0) + 1
saves_by_title: dict[str, int] = flask.session.get("saves_by_title", {})
saves_by_title[from_title] = saves_by_title.get(from_title, 0) + 1
flask.session["saves_by_title"] = saves_by_title
flask.session.modified = True
_record_skip(from_title, hit_title)
return flask.redirect(
flask.url_for("article_page", url_title=url_title, after=hit_title)
)
@ -330,48 +341,55 @@ def article_page(url_title: str) -> str | Response:
return handle_post(url_title)
from_title = url_title.replace("_", " ").strip()
article_title = flask.request.args.get("title")
total = search_count(from_title)
with_link = search_count_with_link(from_title)
try:
redirect_to = api.get_wiki_info(from_title)
except (api.MissingPage, api.MultipleRedirects, api.MediawikiError):
redirect_to = None
no_link_count, hits = search_no_link(from_title)
try:
total = search_count(from_title)
with_link = search_count_with_link(from_title, redirect_to)
_no_link_count, hits = search_no_link(from_title, redirect_to)
except api.MediawikiError as e:
return flask.make_response(
flask.render_template("error.html", message=str(e)), 502
)
by_title = {hit["title"]: hit for hit in hits}
# Filter out candidates already processed this session
session_skipped: set[str] = set(
flask.session.get("skipped", {}).get(from_title, [])
)
found = None
if article_title in by_title:
hit = by_title[article_title]
try:
found = get_diff(from_title, hit["title"], None)
except NoMatch:
pass
# If a specific candidate was requested, move it to the front
title_param = flask.request.args.get("title")
if title_param:
hits = [h for h in hits if h["title"] == title_param] + \
[h for h in hits if h["title"] != title_param]
if not found:
after = flask.request.args.get("after")
if after:
print(after)
hits_iter = itertools.dropwhile(lambda hit: hit["title"] != after, hits)
skip = next(hits_iter, None)
if skip:
hits = list(hits_iter)
# Record and apply explicit skip-past
after = flask.request.args.get("after")
if after:
_record_skip(from_title, after)
session_skipped.add(after)
try:
hit, found = get_best_hit(from_title, hits)
except NoGoodHit:
return flask.render_template("all_done.html")
hits = [h for h in hits if h["title"] not in session_skipped
and h["title"] != from_title and h["title"] != case_flip_first(from_title)]
if not hits:
return flask.render_template("all_done.html")
saves_this_session = flask.session.get("saves_by_title", {}).get(from_title, 0)
return flask.render_template(
"article.html",
title=from_title,
redirect_to=redirect_to,
total=total,
with_link=with_link,
hit_title=hit["title"],
hits=hits,
replacement=found["replacement"],
diff=found["diff"],
found=found,
url_title=url_title,
saves_this_session=saves_this_session,
)
@ -379,7 +397,12 @@ def do_save(title: str, hit_title: str) -> str:
"""Update page on Wikipedia."""
token = mediawiki_oauth.get_token()
found = get_match(title, hit_title, None)
try:
redirect_to = api.get_wiki_info(title)
except (api.MissingPage, api.MultipleRedirects, api.MediawikiError):
redirect_to = None
found = get_match(title, hit_title, redirect_to)
summary = (
f"link [[{found['replacement']}]] using [[:en:User:Edward/Find link|Find link]]"
@ -417,16 +440,20 @@ def api_hits() -> werkzeug.wrappers.response.Response:
@app.route("/api/1/valid_hit")
def api_valid_hit() -> werkzeug.wrappers.response.Response:
"""Return candidates for the given article title."""
link_from = flask.request.args["link_from"]
"""Check if a candidate article has a valid unlinked mention."""
link_to = flask.request.args["link_to"]
link_from = flask.request.args["link_from"]
redirect_to = flask.request.args.get("redirect_to") or None
try:
diff, replacement = get_diff(link_to, link_from, None)
found = get_diff(link_to, link_from, redirect_to)
except NoMatch:
_record_skip(link_to, link_from)
return flask.jsonify(valid=False)
except api.MediawikiError as e:
return flask.jsonify(valid=False, error=str(e))
return flask.jsonify(valid=True, diff=diff, replacement=replacement)
return flask.jsonify(valid=True, diff=found["diff"], replacement=found["replacement"])
@app.route("/favicon.ico")