can i automatically set the timestamp of a file based on part of its filename?

Automatically? AFAIK, no. Standard tools for this task? AFAIK, no. A script can be written to do this, however. If you're familiar with bash, then man touch for details. For more information, here's an semi-related example, though it'd have to be hacked heavily to match your particular requirements and your particular filename format. Here's an example that's a little closer to your requirements.

This works for me on sample files.

#!/bin/bash
#
# process all files with *nnn_YYYY_MM_DD.* format
# Note: Non-current year creation dates do not show time, but Finder
# Get Info will show full creation detail.
for f in "$@"
do
    # strip file extension
    afile=${f%.*}
    #define new separator for filename
    oIFS="$IFS"
    IFS="_"
    # unpack individual filename components in array
    declare -a filedate=($afile)
    IFS="$oIFS"
    unset oIFS
    #printf "%s\n" "${filedate[@]}"
    # Only want the date components
    filedate=("${filedate[@]:1}")
    #printf "%s\n" "${filedate[@]}"
    hhmm=`date "+%H%M"`
    #now concatenate the current time
    filedate=("${filedate[@]}" "${hhmm}")
    # generate date/time string compatible with touch
    tdate=$(printf "%s" "${filedate[@]}")
    #printf "%s\n" $tdate
    # force creation date derived from current filename
    touch -t $tdate $f
    unset filedate
done
exit 0

This script was tested with a file format that you originally offered without spaces in the name (e.g. invoice015_2013_06_25.pdf). The file that you ran the script against is “URSAFF 1er trim 2012_1998_09_19.pdf”, and does not match the contiguous name, filennn_YYYY_MM_DD.extension that the script is designed to correctly function. You probably have files named URSAFF, 1er, trim, and 2012_1998_09_19.pdf, all with the Sep 19 1998 date on them.

If I quoted everything but the file extension on your URSAFF file, the script puts the following information into the array:

URSAFF 1er trim 2012

1998

09

19

After I shift the array elements on line 19 (the [@]:1) syntax, the new array contents are:

1998

09

19

The tdate variable contains: 199809190625.

If I change the touch sequence to: touch -t $tdate "${f}"

The file “URSAFF 1er trim 2012_1998_09_19”.pdf is correctly date stamped as Sep 19 1998.

See if changing the Pass input selection to “as arguments” helps too.

bash uses spaces as delimiters for the "hunks" of a command, and thus accepts spaces within filenames or within strings by quoting the string or the filename, or wants the spaces escaped with a \ character.

For this case, that would be either 'a b c_2010_09_19.pdf' or a\ b\ c_2010_09_19.pdf or (yes) a string with an embedded substring that's quoted as in "a b c_2010_09_19".pdf

FWIW, here are some Bash Quoting Rules, and (more generally) Apple has available an introduction to shell scripting. The latter primer is also as a PDF download, with a zip of the code downloaded separately.

Here is a simplified, but still functional example of my original script. Following MrHoffman's advice on Bash Quoting rules, this script would be run from the Terminal like this:

nameofscript.sh "URSAFF 1er trim 2012_1998_09_19".pdf

nameofscript.sh foo_2009_02_06.py

#!/bin/bash

#

# Process all files with *_YYYY_MM_DD.* format and use the YYYY_MM_DD

# as new creation date. If the input file basename has spaces or other

# characters treated specially by the shell, the basename will require

# single, or double quotes -- either will work with this script.

#

for f in "$@"

do

# return YYYY_MM_DD from filename

justthedate=`echo "${f}" | egrep -o --color=never '([0-9]{4}_[0-9]{2}_[0-9]{2})'`

#define new separator for filename

oIFS="$IFS"

IFS="_"

# split the YYYY_MM_DD string into individual list elements

declare -a filedate=($justthedate)

#restore original separator

IFS="$oIFS"

unset oIFS

#printf "%s\n" "${filedate[@]}"

hhmm=`date "+%H%M"`

#now concatenate the current time into array

filedate+=($hhmm)

# generate date/time string compatible with touch

tdate=$(printf "%s" "${filedate[@]}")

#printf "%s\n" $tdate

# force creation date derived from current filename

touch -t $tdate "${f}"

unset filedate

done

exit 0

Here's a bash script.

Note: works on file in ~/Desktop Adjust to the directory you need

Note: hours and minutes of file set to 00.00 Adjust accordingly

Note: I only tested on 2 files, so test on copies before use

#/bin/bash

for f in ~/Desktop/*
do
   date=${f#*_}
   date=${date%.*}
      if [[ $date = ????_??_?? ]] 
         then
            date=$( echo $date | /usr/bin/sed 's/_//g' )  
            touch -t ${date}0000 $f
      fi
done

Shorter than mine. Cool.

The OP may not have files with multiple ‘_’ in the filename before the date sequence. If they exist, my script will correctly process these filenames without change. (e.g. do_foo_2007_06_01.pdf)

Also, for those with Xcode command line tools installed, there is the /usr/bin/getfileinfo and /usr/bin/setfile combination that can (in combination) set new creation date. No man pages though. Info from an older [^.} superuser post.

VikingOSX wrote:

The OP may not have files with multiple ‘_’ in the filename before the date sequence.

Good point.

This will extract the last 10 chars in the filename (after striping the .ext):

#/bin/bash

for f in ~/Desktop/*
do
   d=${f%.*}
   d=${d:${#d}-10}
   if [[ $d = ????_??_?? ]] ; then
      d=$( echo $d | /usr/bin/sed 's/_//g' )  
      touch -t ${d}0000 $f
   fi
done

😎

A little bit cleaner (no need to call sed, I just remembered that patern substitution can be done in bash)

#/bin/bash

for f in ~/Desktop/*
do
    d=${f%.*}
    d=${d:${#d}-10}
    if [[ $d = ????_??_?? ]] ; then 
        d=${d//_}
        touch -t ${d}0000 $f
    fi
done

Another way to get 10 char from the end of a string is to use an offset that evaluates to a negitive number. So, d=${d:0-10} can be used in place of

d=${d:${#d}-10}, a little bit cleaner.