Converting inconsistent strings to date - postgresql

Question

I'm working with a data set that has a date column in the form of string. I thought it would a simple by doing something like this:
to_date(date_of_birth,'YYYY-01-01')
date_of_birth is in the format DD/MM/YYYY, and is of type 'text'
HOWEVER, I stumbled upon some crazy cases where you have information like
//1980
Another case was:
0/0/1980
When I run my solution, I receive the following error:
ERROR: invalid value "//19" for "YYYY"
DETAIL: Value must be an integer.
My goal actually is simply to collect the year, since that's at least consistent. How do you handle such cases with Postgres?
EDIT:
Switched it to the following:
to_date(date_of_birth,'01/01/YYYY')
My query is this:
SELECT to_date(date_of_birth,'01/01/YYYY') AS year, COUNT(*) AS yearTotal FROM student WHERE date_of_birth LIKE '%/%/1980' GROUP BY year;
The result turns out like this:
year | yeartotal
---------------+-----------
0030-01-01 | 3
0001-01-01 BC | 1

Answer 1

The problem through manipulating the date information (which was of type string) into 'YYYY' in order to access the proper information through date_part.
Using PostgreSQL's "right" function, I was able to collect the year from the string written in the format "DD/MM/YYYY", where right('10/12/2013',4) will return '2013'.
My query looked like this:
SELECT date_part('year', to_date(right(date_of_birth,4),'YYYY')) AS year, COUNT(*) AS total FROM student GROUP BY year
Another way I could've solved the problem was through regex by ensuring that I only work with 'valid' date statements. Something along the line of (using Python regex as an example):
regexp_matches(date_of_birth, '.\/.\/\d{4}')

Answer 2

try TO_CHAR instead of TO_DATE, it worked for me
SELECT replace(date_of_birth,'//','') AS year, COUNT(*) AS yearTotal FROM student GROUP BY year;