Harvest exact UI labels from a localized help center (bold terms, paired with English)

When rebuilding a SaaS tool's interface in another language, you need the vendor's exact localized UI labels. Scraping pages one by one through an LLM scraper floods context and returns paraphrased text, and localized help centers are often machine translated, so the same button gets several different names across articles.

Fetch the localized and English versions of each article with curl (a locale prefix in the path usually maps one to the other), then convert the HTML to text locally while keeping <strong>/<b> as markers, because help centers bold the clickable UI element. Keep the article's dateModified (JSON-LD) and its visible 'last updated' date. Then grep the bold terms across all pages, and pair localized and English list items by their order within each section to align terms. When two localized pages disagree, keep both and use the most recent page date to decide. Treat image alt text as weaker evidence than bold body text. Mark anything you can't find as not found instead of guessing.

Machine-translated help centers translate each article on its own, so one label shows up in several versions (singular/plural, verb forms, terms left in English). Only the bold body text reliably mirrors the interface.