* Use input url instead of request url
* Improve cache purging
* Lint
* Fix TS
* Add method to delete from cloudflare
* Clear on all backends
* Fix
* Add global cloudflare tag
* First try
* Lint and format
* Use edge runtime
* Use envs for sentry
* Fix type
* Fix upload of sourcemaps
* Try fixing monitoring calls
* Fix gitbook/monitoring in middleware
* test error in middleware
* Define context on sentry data
* Set SENTRY_ENVIRONMENT to preview or production
* Format
* Pass url and userAgent
* Remove test cases
* Disable debug
* Remove ignoreBuildErrors
* Add tests for getURLLookupAlternatives
* Match correctly revisions and changes URLs
* Use revision and changerequets ID passed from lookup
* Prevent indexing these urls
* Start
* Start rendering toolbar
* Update api client
* Format
* Format
* Mock resolve of snippet
* Add test for revision
* Fix handling urls with direct space ID
* Preserve all search params when navigating in the PDF UI
* Show a trademark on the pdf page
* Generate proper trademark url
* Format
* Cleanup
* Start
* Improve generation
* Improve style
* Lint
* Use tilde instead of dot in the url
* Improve style
* Fix page group
* Format
* Add visual tests for PDF
* Optimize images using Cloudflare Images
* Try skipping middleware
* Make it work with base path
* Fetch image size
* Use a target instead of the cf fetch options
* Fix resizing
* Start optimizing images
* Preload images
* Use it for page cover
* Format and adapt cover
* Fix image resizing serving jpeg
* Lazy load dark mode images
* Timeout at 2s
* Fix iframe CSP
* Implement logic to lazy load images that should be offscreen
* Also lazy load cover images
* Low priority in preloading
* Fix csp
* Set colorSchema metadata
* Try CSP
* Move colorScheme to generateViewport
* Bun install
* Use wildcard for img-src and connect-src
* Allow all unsafe-inline for now for style
* Remove nonce for style
* Remove nonce for style
* Optimize data fetching with parallel requests
* Try to reuse memory cache between middleware and rendering in CF
* Comment
* Fetch document in parallel of collections
* Cleanup
* Log if redis is enabled for caching
* Fix memory cache before redis one
* Optimize alternatives being tested
* Implement a tagging logic for cache value
* Extract cache-tags from HTTP requests
* Auto-tag API cache entry
* Implement route
* Output cache-control and cache-tag
* Make the revalidate route a edge one
* Move back to use the middleware as the main logic
* Fix link
* Improve lookup to support docs.gitbook.com.localhost:3000
* Make it work with dev url in prod for now
* Start lookup with API
* Continue
* Prepare for cloudflare build
* Dummy commit
* Try to abort requests
* Support x-forwarded-host
* Test multiple modes
* Add logs
* Add logs
* Update next
* Default to multi-path
* Switch back to nodejs runtime
* Adapt to staging API
* Make sure to return an absolute url for a redirect
* Update api client and try single request
* Try again with concurrent lookup
Samy Pessé