[repo-monitor] Medium: Interest categories systematically mislabeled — off-by-one in tab sequence

[Liohtml]

Summary

In extract_interests, the JS scraper runs before the correct tab is activated, causing every interest record to have the category of the previous iteration's tab.

Location

• File: src/scrapers/person.rs
• Line(s): 434–498

Severity

Medium

Details

The loop iterates ["companies", "groups", "schools", ...]. In each iteration it: (1) runs JS scraper on the currently-active tab, (2) clicks the corresponding tab for the next iteration. Result: items from the "companies" tab are labeled "groups", "groups" items are labeled "schools", etc. The first iteration's category is undefined.

Suggested Fix

Reverse the order — click the tab first, sleep, then scrape:

for category in categories {
    // 1. click tab for this category
    // 2. sleep for page to load
    // 3. run JS scraper and label results with this category
}

---

Automated finding by repo-monitor

[Liohtml]

Analysis

Confirmed. src/scrapers/person.rs:434-498 — the loop order is:

1. Run JS scraper (on currently-active tab → wrong category)
2. Click tab for current iteration's category
3. Sleep

So the first iteration scrapes whatever tab is active by default, but labels it as the first category ("companies"). Every subsequent iteration is off by one — you get companies' data labeled as "groups", groups labeled as "schools", etc.

This is a real data quality bug. Anyone using Person.interests will have structurally wrong data — the names and logos are correct but the category metadata is inverted.

Fix — reverse the order: Click first, sleep, then scrape.

for (i, category) in categories.iter().enumerate() {
    // 1. Click the tab for THIS category
    if let Ok(tab_el) = page.find_element(&format!("[data-view-name='{}']", category)).await {
        let _ = tab_el.click().await;
        sleep(Duration::from_millis(1000)).await; // wait for content to update
    }

    // 2. NOW scrape (content is for this category)
    let items_js = format!("(function() {{ {} }})()", INTERESTS_SCRAPER_JS);
    let result = self.base.page.evaluate(&items_js).await?;
    // ... parse and attach to category
}

Also worth noting: The "companies" tab is usually active by default on LinkedIn interests pages, so the last category in your list will actually get scraped correctly (because the click lands on it and you scrape it on the next iteration). Check what the last category in categories is — if it's "companies", the off-by-one partially cancels itself for that case.

Effort: 15 min for the fix.