Compare X Trends Across Places in an HTML Dashboard

X Trends Dashboard shows what is trending on X in several places side by side. The script asks for the top 30 trends in each place, marks trends that show up in two or more places, pulls a few of the latest posts for the top shared trends, and writes a static HTML dashboard plus CSV and JSON. Run it again with an earlier snapshot to see what moved.

The problem

A trend list for one city does not say whether that name is local. Comparing places means opening each list, matching names that differ by a hash mark or a space, and then checking again an hour later to see what changed.

What the app does

• Reads places as WOEID and name, a count from 30 to 100, and how many shared trends get sample posts.
• Pulls trends per place. A WOEID mismatch is left out of the shared comparison and is still drawn in the side-by-side table.
• Matches trend names across places, ignoring case, spaces and a leading #.
• Pulls latest posts for the top shared trends. Authors are shown as Account A, Account B and so on.
• Writes an HTML table, CSV, JSON, and a moves file when you pass an earlier snapshot.

How it works

1. Trends. GET /twitter/trends?woeid=<id>&count=30 for each place. Count must be from 30 to 100.
2. Match. Names are normalized so #Topic and Topic count as one.
3. Posts. GET /twitter with sort=Latest for the top shared trends you asked to sample.
4. Compare. With an earlier raw file, each row gets its previous rank or new.
5. Write the dashboard. Sample post text replaces handles with [account] and links with [link].

The code

def run_live(cfg, probes, samples):
    import requests
    key = api_key()
    s = requests.Session(); s.headers["Authorization"] = key
    raw = {"run_at": now(), "cfg": cfg, "trends": [], "x": [], "probes": []}
    for w in cfg["places"]:
        raw["trends"].append(call(s, "/twitter/trends", {"woeid": int(w), "count": cfg["count"]}, f"trends {w}"))
    if samples:
        for name, _ in shared_trends(raw)[: cfg["sample_top_shared"]]:
            raw["x"].append(call(s, "/twitter", {"query": name, "sort": "Latest", "count": cfg["x_count"]}, "x " + name[:24]))
    if probes:
        raw["probes"].append(call(s, "/twitter", {"query": cfg["empty_query"], "sort": "Latest", "count": cfg["x_count"]}, "probe x empty result"))
        raw["probes"].append(call(s, "/twitter/trends", {"woeid": 1, "count": cfg["bad_count"]}, "probe trends count=29"))
        raw["probes"].append(call(s, "/twitter/trends", {"woeid": cfg["fallback_woeid"], "count": cfg["count"]}, "probe trends unknown woeid"))
    return raw, key

The key comes from the environment and is not written to the output. --no-probes skips the check calls. --no-samples skips the post searches.

Endpoints

• GET /twitter/trends
• GET /twitter

Example output

The 2026-10-16 sample used 5 places and returned 150 trend rows. 6 trends were shared by 2 or more places. One trend appeared in all 5. A second snapshot 45 seconds later had 1 new row and 19 rows at a different rank. A full snapshot with 3 sample searches cost $0.027 by response headers.

Run it yourself

The public examples folder linked above includes the script and a README. Install its requirements, set DESEARCH_API_KEY from console.desearch.ai, and use the run command in that README. Open the HTML file in a browser for the side-by-side table.

What it costs

$0.027 for the sample snapshot that included 3 post searches, from the response headers on 2026-10-16. See pricing.

Ideas to extend it

• Run it on a timer and keep the moves CSV.
• Add the top post text under each shared trend in the HTML table.
• Highlight trends that jumped more than a set number of ranks.

{
  "mcpServers": {
    "desearch": {
      "url": "https://mcp.desearch.ai/mcp",
      "headers": {
        "Authorization": "Bearer <DESEARCH_API_KEY>"
      }
    }
  }
}

Use Desearch to fetch the top X trends for each place I list. Mark trends that appear in two or more places, pull a few latest posts for the top shared trends, and describe rank changes if I give you an earlier snapshot. Quote only trends and posts you retrieve. Do not invent trend names.