Quoting entire messages (top posting) and HTML attachments are evil

Introduction

In particular a message in top-posting style reproducing my original message is a pain and a waste when archiving ... I have no need to archive again my original message which is already in my folder (and anyhow the message with the appendage is less legible). In the rare case I'd need to distribute the entire correspondence to others, a MIME digest is a much more efficient solution,

Some people use this signature to show how top-posting is silly:

--
Answer: Because it makes conversations flow in a nonsensical order.
Question: Why is top-posting wrong?

Therefore, after some attempts of dealing with this while archiving the mail message to a folder (I use pine), or using procmail, I finally devised this system (based on awk and integrated within my procmail rules.).

• if the message contains quoted text (i.e. if the body or first text/plain attachment contains lines starting with the standard quotation character (>) ...
• ... or if the message contains Content-type references to text/html
  • then I scan it and locate the start and end record of the unwanted parts.
  • an "improper quotation" is a sequence of lines prefixed with the standard quotation character, which are not part of a forwarded message (so far identified by ----- Original Message -----) and are not followed by any non-blank text (excluding a standard signature)
  • an "unwanted attachment" is a part of a MIME multipart message (as defined in RFC 2046) of type text/html. Currently the scan stage parses the start and end of any attachment, while the rejection is deferred to the following stage.
    I added provision to cope with the case an attachment is also a MIME multipart object, and descend recursiverly into it.
• Only if the scan has been tentatively successful, I submit the message to a filter, which passes the message through except for the unwanted parts, which are replaced by a placeholder.
  • IMPROPER QUOTATION (TOP-POSTING) OF n LINES DELETED for a bulky quotation
  • HTML ATTACHMENT DELETED in a one-line text/plain attachment, instead of an HTML attachment

Annoying behaviour by Outlook

Therefore I was compelled (in v1.3) to disable the pre-existing protection for forwarded messages in the particular case of replies generated by Outlook Express.

Download and installation

Please follow these steps for installation :

• Insert in your .procmailrc, in the place most appropriate for it, a line calling my rule, e.g. fi $PMDIR is your procmail rule directory
  INCLUDERC = $PMDIR/rc.quote-html

• Get and install in such directory the following file rc.quote-html
• Optionally customize it according to the comments in the first section of the file, namely :
  • set PMDIR if not already done
  • uncomment the generation of TODAY, an environment variable containing today's date, if you wish to keep dated backups. It is commented in my version since I set it elsewhere (I use it also for spam filtering).
  • set BACKUPFOLDER if you do not like my default. See below for backup operation.

• Get and install in the same procmail directory the analysis awk file cleanbadquote.awk
• Get and install in the same procmail directory the filter awk file cleanbadquote_do.awk

• Test it, sending some "bad manner" mail to yourself. If you do not want to test it on real mail, uncomment in rc.quote-html the line making reference to a Subject: TEST-TEST-TEST and send yourself mail messages with such subject. Comment it again when satisfied !

• If you have implemented dated backups and want to be perfect, arrange to clean up periodically the backup directory getting rid of older folders, inserting in a crontab something like this
  find . -ctime $for -o -size $zero -type f -exec rm -f {} \;
  which deletes the folders older than for days or of zero size (zero is not 0 but has to be found by trial, e.g. if your folder are managed by pine taking into account the "FOLDER INTERNAL DATA" used by pine).

Principles of operation

Update history

Lucio Chiappetti - IASF Milano - INAF