Skip to content
Source

The parsers, as plain text.

These are the files that read and remove metadata, exactly as they are compiled into the site. The shipped script carries a source map too, so the developer tools show them under src/lib.

  1. index.tssniffs the format by its first bytes, dispatches, and holds the policy
  2. jpeg.tsJPEG segments: EXIF, XMP, IPTC, comments removed; ICC kept; orientation written back
  3. png.tsPNG chunks: text, eXIf and tIME removed; iCCP kept
  4. ooxml.tsWord, Excel, PowerPoint packages: properties emptied, thumbnail dropped, zip rewritten
  5. pdf.tsPDF: the Info dictionary and uncompressed XMP blanked in place, same length
  6. exif.tsthe TIFF reader behind EXIF, with GPS and the orientation-only writer
  7. xmp.tsXMP packets and Photoshop IRB, IPTC datasets
  8. xml.tsa small XML reader, so the same code runs in the browser, Node and a Worker
  9. zip.tsthe stored zip writer batch uses
  10. bytes.tsbyte helpers: slices, integers, CRC, inflate and deflate
  11. types.tsthe report shape every parser returns

The same files build the command line tool, the API and the extension. No metadata library sits behind any of them.

Watch it run.

Open the network panel, drop a file on the home page, and count the requests.

Clean a file