ToddAndMargo: grep is a decent start, but it actually is much more complicated than that.


To get this data in a format you can read/process, you have to deal with the fact that there is no standard in SMTP for the order of headers, and every client seems to do it differently.  On top of which, some clients might record a "Sent" header instead of a "Date" header, and then you have to deal with control fields, etc.

Anyway, assuming Thunderbird on Linux, you can get to the MBOX files from ~/.thunderbird/<YOUR_PROFILE_ID>/ImapMail/<IMAPSERVERNAME_FOLDER>/<REMOTE_FOLDER>

YOUR_PROFILE_ID would be whatever you see there, it's a random string and if you only have one profile it will end in ".default".
IMAPSERVERNAME_FOLDER would be which email account you are looking for, and if you have multiple it might append a "-2", "-3", etc. "imap.gmail.com" is a good example
REMOTE_FOLDER is the actual folder name of the folder you are trying to scrape.  So "INBOX", "Sent", "Spam", etc.

I wrote this simple combination of grep/awk to convert everything into a CSV that you can import into whatever you want.

If you save this in thunderbird_to_csv.sh, you can execute it like so (the first argument is the Thunderbird MBOX file):

$ ./thunderbird_to_csv.sh .thunderbird/<YOUR_PROFILE_ID>/ImapMail/<IMAPSERVERNAME_FOLDER>/<REMOTE_FOLDER>



#!/bin/bash
grep -E "^((Subject|Date|Sent|From): |From - )" $1 | awk 'BEGIN { print "From,Subject,Date"; } /^From - /{
    subject=""; from=""; date="";
    while(length(from) == 0 || length(date) == 0 || length(subject) == 0){
        getline;
        if(length(from) == 0 && index($0, "From: ") == 1){
            from=gensub("^From: (.*)$", "\\1", $0);
        }
        if(length(subject) == 0 && index($0, "Subject: ") == 1){
            subject=gensub("^Subject: (.*)$", "\\1", $0);
        }
        if(length(date) == 0 && index($0, "Date: ") == 1){
            date=gensub("^Date: (.*)$", "\\1", $0);
        }
        if(length(date) == 0 && index($0, "Sent: ") == 1){
            date=gensub("^Sent: (.*)$", "\\1", $0);
        }
    }
    sub("\"", "\"\"", from);
    sub("\"", "\"\"", subject);
    sub("\"", "\"\"", date);
    from=substr(from, 0, length(from)-1);
    subject=substr(subject, 0, length(subject)-1);
    date=substr(date, 0, length(date)-1);
    print "\"" from "\",\"" subject "\",\"" date "\"";
}'




For those curious what this does, the grep command strips everything down to lines starting with "Subject: ", "From: ", "Date: ", "Sent: ", and "From - ".

I can't recall if "From -" is a part of the MBOX format (I don't remember it being there), but I think it's actually something Thunderbird threw in there.  Glad they did, as it separates each email pretty nicely.

It then loops through every line to see if you can find these headers, and replaces them IF AND ONLY IF THAT HEADER HASN'T BEEN SEEN BEFORE.  So for instance, if you have an email that was originally from Bob, forwarded to you from Alice, if I kept searching through it would say the email was from Bob and not Alice (because Alice is who actually sent that email).

After that, it escapes the double quotes inside to be two double-quotes, the standard for CSV files, and takes off the last character which is an extra newline.

-Brad



On 08/24/2015 06:54 PM, ToddAndMargo wrote:

On 08/24/2015 04:29 PM, Yasha Karant wrote:

My query applies specifically to Mozilla Thunderbird current, but could
have a more general solution.

I need to convert to a plain text file listing (that could be imported
into a word processor, LaTeX or a GUI front end thereto, etc) what
appears in the display of Thunderbird as the columns Subject From and
Date for an internal activity report that I must write.   These columns
appear on the end-user GUI display and allow one to then read specific
messages by "point and click".  As I cannot find a description of the
official Thunderbird nomenclature for the various sections of the GUI
display, I am using the above descriptions.

I could use a screenshot application, select a rectangular region, save
each entity as a PNG image, and then use an OCR application to yield
plain text.  I would prefer that the screenshot application simply
recognizes the text *AS* text, allowing me to copy and paste into a text
editor, etc., all running under X wndows.   Does anyone know of an
application that does this?  A brief perusal on the web as well as a
quick read of the information on the "default" screenshot applications
that come with either MATE or KDE does not seem to reveal a mechanism
for this (but rather the PNG or other image, non-text, route).

The normal mechanism I use -- highlight (select), pointing device button
(to copy), and then point device button (paste) to capture from say a
text HTTP file in a web browser to a word processor application -- does
not seem to work for the above "column" portion of the Thunderbird
display.  This normal mechanism does work if I view source for each
message, displaying the SMTP text source and headers in a box, but is
very time consuming as the information that I need is available in the
"columns" of the basic Thunderbird user interface without having to view
the source.

Any assistance is appreciated.

Yasha Karant

Hi Yasha,

   Something like this?

       grep -i  "subject\|from\|date" Inbox

-T