A .md previewed as its own source, which is reading the punctuation
rather than the notes. It now opens rendered, with Rendered | Source in
the dialog footer for anyone who wants what the file actually says.
Markdown only; a .txt has nothing to toggle between.
Rendering is `marked`, sanitised by DOMPurify -- the one the app already
carries for mail. Markdown is not a safe subset of anything: raw HTML
passes through it by design, so a <script> in a file somebody uploaded or
shared into the account is a script tag unless something takes it out.
Images become links rather than pictures. An image in a Markdown file is
either a relative path, which has no base to resolve against here, or a
URL somewhere else, which fetches on open and tells that server the file
was read -- the tracking pixel this app blocks in mail. The link keeps
the alt text and the address, so nothing vanishes silently.
Fixes the PDF preview while here, which never worked: securityHeaders
put X-Frame-Options: DENY on every response including the blob route, so
the iframe showed Chrome's "refused to connect" where the file should
have been -- in Files today and in mail attachments long before that.
The middleware now leaves a header the route has set, and a PDF served
inline says SAMEORIGIN. Nothing else on the server is framable.
A gzip response is decompressed before the blob proxy sees the body, but
its content-length still describes the compressed bytes. Copying that
header onto the longer body made the browser stop reading that many bytes
in and call the download complete, so files arrived truncated with nothing
reporting a failure.
It took a hop that compresses to show up, and one that only compresses
above a threshold to look like a race: a Sieve script stayed intact for two
rules and came back cut off mid-rule once the third pushed it past 1 KiB.
Ask upstream for identity, and forward no length at all rather than one
that describes different bytes.
jcoffey-dev