Hi Julie,


Not sure if you are still looking for a solution, but I wrote a python script to do the parsing: https://github.com/fishbb/parsedate


You can simply download the script and start using it.


All dates will be formatted as you prefer expect one entry.  "1933, 1937-1938, 1941" will become "1933/1941" instead of "1933/1938". I want to keep the behavior consistent in my script so that every time the script see some ranges of dates in one entry, it will aways pick the widest range possible. You can of course always modify my script and add additional regex for this specific case.


Dan

Dan Lou | Senior Librarian
3700 Middlefield Road | Palo Alto, CA 94301
650.838.2973 [log in to unmask]

________________________________
From: Code for Libraries <[log in to unmask]> on behalf of Sarah Weissman <[log in to unmask]>
Sent: Monday, October 30, 2017 1:45:16 PM
To: [log in to unmask]
Subject: Re: [CODE4LIB] Help with parsing dates?

Hey Julie,

I was able to put together a simple ruby script that takes in a file of dates one per line and uses timetwister to kind of do what you want:

require 'timetwister'
File.open("<path-to-dates-file> ", "r") do |f|
  f.each_line do |line|
    tw = Timetwister.parse(line)
    date_start = tw[0][:date_start]
    date_end = tw[0][:date_end]
    datestr = date_start
    if date_end && (date_start != date_end)
      datestr = datestr + "/" + date_end
    end
    if !datestr
      puts 'ERROR'
    else
      puts datestr
    end
  end
end

If you have ruby installed and can do “gem install TimeTwister” on the command line you can edit the path above to point to your input file and then copy the code into a file called dateparse.rb and you should be able to run it by calling “ruby dateparse.rb”. If you aren’t set up enough with ruby to do this and just want to send me a dump of all your dates off list I can run it for you and send you the output.

When I ran it on your date inputs, I got the following. There are definitely more errors than in your example, but it could be a start.

1947
1947-08
1947-08-03
1947-08-03/1947-08-07
1914-07-24/1915-01-30
1957-05-23/1957-06-20
1947-08
1947-08-03
1947-08-03/1947-08-07
ERROR
1917
ERROR
1947
1971/1972
1980/1989
1947
1939/1940
1944-04/1944-05
1939-11/1940-08
1955-01/1955-06
1939-11-06/1940-08-07
1983-06/1983-12
1988-08-24
ERROR
1986
ERROR
ERROR
1933
1897
ERROR
ERROR
1965/1975
ERROR
ERROR
1932
1960/1969
1930/1959
1954
1970-05-09
ERROR
1973-01-20
ERROR
ERROR
ERROR
ERROR
ERROR
ERROR
ERROR
ERROR
ERROR
1973-05/1973-07

-Sarah

From: Code for Libraries <[log in to unmask]> on behalf of Julie Swierczek <[log in to unmask]>
Reply-To: Code for Libraries <[log in to unmask]>
Date: Monday, October 30, 2017 at 12:04 PM
To: "[log in to unmask]" <[log in to unmask]>
Subject: [CODE4LIB] Help with parsing dates?

Hi all,

I'm looking for some help with parsing dates. I have inherited over 430 HTML finding aids that I need to migrate to ArchivesSpace. I have an excel template that creates EADs from folder lists, and I use that to get the finding aids into ArchivesSpace. (It is a simplified version of the Spreadsheet from Heaven: http://clir.pacscl.org/2012/03/19/excel-to-xml-the-spreadsheet-from-heaven/.<http://clir.pacscl.org/2012/03/19/excel-to-xml-the-spreadsheet-from-heaven/>) Mine has columns for box number, series, folder title, normalized single or begin date, normalized end date, free-form date, and notes. The tricky part is filling in the normalized date columns. They need to be in the format YYYY-MM-DD, YYYY-M, or YYYY. So far, I have found dates in the following formats, and I've indicated after the '|' how I'd like them to be normalized:

1947|1947
August 1947|1947-08
August 3, 1947|1947-08-03
August 3-7, 1947|1947-08-03/1947-08-07
July 24, 1914 - January 30, 1915|1914-07-24/1915-01-30
May 23, 1957-June 20, 1957|1957-05-23/1957-06-20
1947 (August)|1947-08
1947 (August 3)|1947-08-03
1947 (August 3-7)|1947-08-03/1947-08-07
May 14 (?)|ERROR
1917?|1917
May 14, ____|ERROR
ca. 1947|1947
ca. 1971-1972|1971/1972
ca. 1980s|1980/1989
circa 1947|1947
circa 1939-1940|1939/1940
1944 (April - May)|1944-04/1944-05
1939 (November) - 1940 (August)|1939-11/1940-08
1955 (Jan.-June)|1955-01/1955-06
1939 (November 6) - 1940 (August 7)|1939-11-16/1940-08-07
June-December 1983|1983-06/1983-12
August 24 1988; October 31, 1988|1988-08-24/1988-10-31
Winter 1985-1986|1985/1986
1986-|1986
through 1983|ERROR
thru 198|ERROR
1933, 1937-1938, 1941|1933/1938
1897, 1906|1897/1906
pre-1975|ERROR
pre-1975 (May)|ERROR
1965-1975, n.d.|1965/1975
undated|
n.d.|
1932, 1940s-1975, n.d.|1932/1975
1960s|1960/1969
1930s-1950s|1930/1959
1954 and undated|1954
5/9/1970|1970-05-09
Saturday, 9 May 1970|1970-05-09
20 Jan 1973|1973-01-20
1944-1950 [died Aug. 1949]|ERROR
1967-onward|1967
January 27, 1975 [1974?]|ERROR
re: 1906|1906
Easter 1961|1961
May 31, 1964-Fall 1965|1964-05-31/1965
June 2 - ____, 1971|ERROR
n.d.; May 26, 1976|1976-05-26
May 1973 - Jul7 1973|1973-05/1973-07-07
May 1973-July 1973|1973-05/1973-07

I feel like I read to the end of the internet and didn't find any tools that are easy to use.[1][2][3][4] (Where "easy to use" means "does not require a degree in computer science or software engineering or *ahem* several hours of 'spare time' to learn how to use this tool". I know some people have jobs where they get to spend time doing cool stuff like that, but this ain't one of them. This is more of a firefighter job, where you just put out one fire after another (and the arsonists follow you around and set more while you aren't looking).)

I have heard wonderful things about Timetwister (https://github.com/alexduryee/timetwister) but I don't know Ruby, and neither does my technical support team (of 0). I gave myself a day to try to learn just enough Ruby to use this, but, hey, failure is a thing that happens.

I know a little bit of Python, but I Am Not A Programmer™ and I need things spelled out as though it were my first day using Python. (I know Python the same way you would know a foreign language if you memorized one sentence from a tourist phrasebook each week.  You could muddle through reading something if you had all the time in the world and a dictionary, but you wouldn't really *know* the language.)

I am rapidly approaching the point where it would be faster for me to type the dates manually than spend any more time trying to figure this out, but I am concerned about introducing errors into the dates by typing them.

So, while I am off trying to use OpenRefine and Excel, with the world's sloppiest regex and GREL to see if I can somehow transform the majority of these dates - if not all of them - I thought I'd reach out to the group and see if anyone can lend a hand. I will pay you in cookies (or your preferred treat). For some of the dates above, like the ones with underscores, I'd be fine if the process just returned an error, so I could highlight the error cells in Excel and fix those dates by hand.  But I think the vast majority of the dates can be handled programmatically, and I don't feel like reinventing the wheel because I'm sure someone has already figured this out. I just need it to be easier to implement.

Thanks in advance,

Julie Swierczek
Breaker of Dates (and Software)