feat: switch to JSON API for fetching contest data

Updated the contest data fetching logic to use the JSON API instead of scraping the HTML directly. This change improves the reliability and performance of data retrieval by utilizing a structured API endpoint.

- Replaced HTML parsing with JSON decoding
- Simplified the contest details extraction
- Updated template to display new data structure

Addresses issues with parsing inconsistencies and changes in HTML structure.
This commit is contained in:
Kumi 2024-10-02 13:48:25 +02:00
parent 380041ec1b
commit c633a3ec47
Signed by: kumi
GPG key ID: ECBCC9082395383F
2 changed files with 61 additions and 95 deletions

View file

@ -49,7 +49,7 @@ def init_contest_routes(app):
"total_pages": total_pages,
"has_prev": page > 1,
"has_next": page < total_pages,
"limit": limit
"limit": limit,
}
return render_template(
@ -125,68 +125,33 @@ def init_contest_routes(app):
@app.route("/contest/")
def route_contests():
try:
data = urlopen("https://www.instructables.com/contest/")
# Fetch current contests from the JSON API
response = urlopen(
"https://www.instructables.com/json-api/getCurrentContests?limit=50&offset=0"
)
data = json.loads(response.read().decode())
except HTTPError as e:
abort(e.code)
except Exception as e:
abort(500) # Handle other exceptions such as JSON decode errors
soup = BeautifulSoup(data.read().decode(), "html.parser")
contest_count = "0"
contests = []
for contest in soup.select("div#cur-contests div.row-fluid div.contest-banner"):
link = contest.select("div.contest-banner-inner a")[0].get("href")
img = proxy(contest.select("div.contest-banner-inner a img")[0].get("src"))
alt = contest.select("div.contest-banner-inner a img")[0].get("alt")
deadline = contest.select("span.contest-meta-deadline")[0].get(
"data-deadline"
)
prizes = contest.select("span.contest-meta-count")[0].text
entries = contest.select("span.contest-meta-count")[1].text
contests.append(
{
"link": link,
"img": img,
"alt": alt,
"deadline": deadline,
"prizes": prizes,
"entries": entries,
}
)
closed = []
for display in soup.select("div.contest-winner-display"):
link = display.select("div.contest-banner-inner a")[0].get("href")
img = proxy(display.select("div.contest-banner-inner a img")[0].get("src"))
alt = display.select("div.contest-banner-inner a img")[0].get("alt")
featured_items = []
for featured_item in display.select("ul.featured-items li"):
item_link = featured_item.select("div.ible-thumb a")[0].get("href")
item_img = proxy(
featured_item.select("div.ible-thumb a img")[0].get("src")
)
item_title = featured_item.select("a.title")[0].text
item_author = featured_item.select("a.author")[0].text
item_author_link = featured_item.select("a.author")[0].get("href")
featured_items.append(
{
"link": item_link,
"img": item_img,
"title": item_title,
"author": item_author,
"author_link": item_author_link,
}
)
closed.append(
{"link": link, "img": img, "alt": alt, "featured_items": featured_items}
)
contests = data.get("contests", [])
contest_list = []
for contest in contests:
contest_details = {
"link": f"https://www.instructables.com/{contest['urlString']}",
"img": contest["bannerUrlMedium"],
"alt": contest["title"],
"title": contest["title"],
"deadline": contest["deadline"],
"prizes": contest["prizeCount"],
"entries": contest["numEntries"],
}
contest_list.append(contest_details)
return render_template(
"contests.html",
title="Contests",
contest_count=contest_count,
contests=contests,
closed=closed,
contest_count=len(contest_list),
contests=contest_list,
)

View file

@ -1,40 +1,41 @@
{% extends "base.html" %}
{% block content %}
<center>
<h1>{{ title }}</h1>
{{ contest_count|safe }}
<br>
<div class="contest-list">
{% for contest in contests %}
<div class="contest-list-item">
<a href="{{ contest.link }}">
<img src="{{ contest.img }}" alt="{{ contest.alt }}" style="max-width:500px;">
</a>
<p>Closes {{ contest.deadline }}</p>
<p class="ible-small">{{ contest.prizes }} Prizes, {{ contest.entries }} Entries</p>
</div>
{% endfor %}
</div>
<div class="closed-contests" id="contest-winners">
<h2>Winner's Circle</h2>
{% for closed in closed %}
<div class="closed-contest">
<a href="{{ closed.link }}"><img class="closed-contest-contest" src="{{ closed.img }}"
alt="{{ closed.alt }}"></a>
{% for featured_items in closed.featured_items %}
<div class="closed-contest-winner">
<a href="{{ featured_items.link }}">
<img class="closed-contest-winner-img" src="{{ featured_items.img }}"
alt="{{ featured_items.title}}">
<br>
<b>{{ featured_items.title }}</b>
</a>
<p>by <a href="{{ featured_items.author_link }}">{{ featured_items.author }}</a></p>
</div>
{% endfor %}
</div>
{% endfor %}
</div>
</center>
<center>
<h1>{{ title }}</h1>
<p>Total Running Contests: {{ contest_count }}</p>
<br>
<div class="contest-list">
{% for contest in contests %}
<div class="contest-list-item">
<a href="{{ contest.link }}">
<img src="{{ contest.img }}" alt="{{ contest.alt }}" style="max-width:500px;">
</a>
<h2>{{ contest.title }}</h2>
<p>Closes: {{ contest.deadline }}</p>
<p class="ible-small">{{ contest.prizes }} Prizes, {{ contest.entries }} Entries</p>
</div>
{% endfor %}
</div>
<div class="closed-contests" id="contest-winners">
<h2>Winner's Circle</h2>
{% for closed in closed %}
<div class="closed-contest">
<a href="{{ closed.link }}"><img class="closed-contest-contest" src="{{ closed.img }}"
alt="{{ closed.alt }}"></a>
{% for featured_items in closed.featured_items %}
<div class="closed-contest-winner">
<a href="{{ featured_items.link }}">
<img class="closed-contest-winner-img" src="{{ featured_items.img }}"
alt="{{ featured_items.title}}">
<br>
<b>{{ featured_items.title }}</b>
</a>
<p>by <a href="{{ featured_items.author_link }}">{{ featured_items.author }}</a></p>
</div>
{% endfor %}
</div>
{% endfor %}
</div>
</center>
{% endblock %}