Re: pdf vs. octet-stream

OK, I know it's been two years, but all that time it's been bugging me that 
Mime_Magic seems to work for most people, but not for me. It really bugs me 
because admin people here use Outlook and pdf attachments are listed as type 
application/octet-stream, and I have to edit them by hand, changing it to 
application/pdf. I was just writing a sed procedure when I couldn't stand it 
any more and just had to figure out why exmh couldn't do the substitution for 
me, like Mime_Magic was intended to do. 

I found the problem, and now I don't understand why Brent and others were not 
affected by it.

To refresh memory, here is the out-of-the-box Mime_Magic in 2.7.2:

1proc Mime_Magic { fileName } {
2    global exmh env
3    Exmh_Debug Mime_Magic $fileName
4
5    foreach m [list $exmh(userLibrary)/.magic \
6	     $exmh(library)/local.magic" "" ] {
7	# Avoid eval in this command to tolerate spaces in file names
8	set cmd {exec file}
9	if {[string length $m]} {
10	    lappend cmd -m $m
11	}
12	lappend cmd $fileName
13	if [catch $cmd result] {
14	    continue
15	}
16
17	set result [string trim [lindex [split $result :] 1]] 
18
19	foreach res { magic_u magic_l magic_ } {
20	    set contentType [option get . $res$result {}]
21	    if [string length $contentType] {
22		return $contentType
23	    }
24	}
25    }
26
27    return
28}

First, note there is an extraneous double quote on line 6. But that's not the 
real problem.

I tested $res$result of line 20 when looking at a pdf file specified as 
octet-stream, and it properly gave:

magic_uPDF document, version 1.4
magic_lPDF document, version 1.4
magic_PDF document, version 1.4

I made sure I had the exact same string in my exmh-defaults (I even copied it 
with my mouse buffer to be sure):

*magic_lPDF document, version 1.4:      application/pdf

and yet, 'option get' turned up no match. GRRR. Thinking maybe it did not like 
periods, I split on the comma inserting 

	set result [string trim [lindex [split $result ,] 0]] 

after line 17 and shortened the magic line to 

*magic_lPDF document:      application/pdf

BINGO! 

I don't know whether matching periods is a general problem; in fact I don't 
know anything about tcl.

Then I tried this on a png file; one for which the file command gives 

PNG image data, 1600 x 1200, 8-bit/color RGB, non-interlaced

This works without modifying Mime_Magic (no periods). But I'm thinking: do we 
really want to add all the possible permutations in the magic variables? That 
seems retarded. So I propose splitting all on commas. For example for PDF's, we 
need only the single line 

*magic_lPDF document:      application/pdf

instead of all the different versions and not to mention having to add new 
versions as they occur.


-- 
rick baartman