How to honor Content-Disposition filenames in wget

Some download endpoints keep one stable URL and send the real export name in the HTTP Content-Disposition header. Using wget with --content-disposition saves the file with that header-provided name instead of falling back to the URL path or query string.

GNU wget 1.25.0 still documents --content-disposition as experimental. When the server sends a filename= value in the header, wget uses it for the local file name. This is separate from --trust-server-names, which follows redirect URLs rather than header-provided filenames.

The remote server chooses the suggested name, so this option is safest on trusted download endpoints. The GNU manual also warns that the feature can trigger extra HEAD round-trips in some cases, and -O still overrides the header completely when one fixed local path matters more than the advertised filename.

1. Download the file with --content-disposition and --server-response so the header-selected name is visible in the same run.

   $ wget --content-disposition --server-response \
     'https://downloads.example.test/exports/monthly-ledger?account_id=ACCT-48271&period=2026-03&format=csv'
   ##### snipped #####
   HTTP request sent, awaiting response...
     HTTP/1.1 200 OK
     Content-Disposition: attachment; filename="acct-48271-ledger-2026-03.csv"
   Length: 43224 [text/csv]
   Saving to: 'acct-48271-ledger-2026-03.csv'
   
   ##### snipped #####
   
   'acct-48271-ledger-2026-03.csv' saved [43224/43224]

   The Content-Disposition header and the Saving to: line should agree. If the site renames files only through redirects, use --trust-server-names instead.

2. Confirm the chosen local name before handing the file to another script.

   $ ls -lh acct-48271-ledger-2026-03.csv
   -rw-r--r--  1 user user 42K Mar 29 09:02 acct-48271-ledger-2026-03.csv

   This is the quickest way to catch collisions when one URL produces dated or account-specific exports.

3. Add content_disposition = on to ~/.wgetrc when most downloads from that account should keep the server-provided name.

   ~/.wgetrc
   content_disposition = on

   GNU wget 1.25.0 still documents this behavior as experimental and warns it can trigger extra HEAD requests for some servers.

   Related: How to configure default options in ~/.wgetrc

4. Override the header-selected name with -O whenever one fixed local path matters more than the server suggestion.

   $ wget --content-disposition \
     --output-document=acct-48271-latest.csv \
     'https://downloads.example.test/exports/monthly-ledger?account_id=ACCT-48271&period=2026-03&format=csv'
   HTTP request sent, awaiting response... 200 OK
   Length: 43224 [text/csv]
   Saving to: 'acct-48271-latest.csv'
   
   ##### snipped #####
   
   'acct-48271-latest.csv' saved [43224/43224]

   -O wins over --content-disposition every time, so use it when a later job expects one stable local filename.