v0.2.0

Features

• Redesigned the interface, with a modern app shell
• Split the details panel into tabs, each its own scrollable region, with badged counts
• Aggregated the Links, Images and Hreflang tabs to one row per URL or source
• Added image size and status to the Images tab
• Recorded every hop of a redirect chain
• Recorded the URLs a crawl knows about but never fetched
• Reported how much of a page's link graph only exists after rendering
• Added noindex directives, H2 outline rules, missing body element, canonical and ecommerce rules as overview rules
• Read hreflang from Link headers and sitemaps
• Captured sitemap lastmod and reported unhealthy sitemap entries
• Validated rich result URLs
• Exported the crawl as a PDF report
• Set the report and the SERP preview in Google Sans
• Added bulk CSV export
• Batched live crawl updates during a crawl
• Moved crawl controls into the gear panel, with one fetch per page and paced auxiliary requests

Fixes

• Corrected structured data, hreflang, title, link and content extraction
• Computed crawl depth from a walk of the link graph, and recomputed inlink counts from the full graph after a crawl
• Refreshing the UI from the database when a crawl finishes
• Found the redirects spider hid behind a 200, and treated an unfollowed 3xx as a redirect everywhere
• Decoded HTML entities in hrefs before queueing them, and inflated gzipped sitemaps before parsing
• Modelled canonicalisation as a non indexable state, and counted a missing canonical
• Fixed missing headers, and discarded response headers Chrome reported from another request
• Measured SERP pixel width with real Arial advance widths, titles in Google Sans at 22px and descriptions in Arial at 14px
• Counted characters, not bytes, in the length columns and filters
• Counted overview rules against the population their percentage divides by, image rules in images, and instance derived rules in pages
• Gated the pixel width and self reference rules on content eligibility
• Raised the low content threshold to 200 words
• Read a link's aria-label, or its image alt, as anchor text when it has none
• Exempted subresources, XHR calls and bundled assets from the URL quality rules
• Only reported a missing hreflang return tag when the target was crawled
• Status checked images Chrome could not size, without re-fetching API calls
• Replaced an inline image's payload with a hash and its size
• Removed cookie alerts from the words diff
• Verified updates against checksums and signatures, bundled axe-core, and measured vitals in visible windows
• Fixed table sorting for text, trimmed quotes from HTTP headers, removed content type from redirects and included links from the initial page

Performance

• Wrote each page in one transaction, and loaded a crawl in one query
• Built a crawl's aggregates off the foreground thread, with a loading state
• Built the overview's issue entries once, not once per cell
• Virtualised the link lists in the details panel