pyspark - How to TRUNCATE and / or use wildcards with Databrick

Question

I'm trying to write a script in databricks that will select a file based on certain characters in the name of the file or just on the datestamp in the file.

For example, the following file looks as follows:

LCMS_MRD_Delta_LoyaltyAccount_1992_2018-12-22 06-07-31

I have created the following code in Databricks:

import datetime
now1 = datetime.datetime.now()
now = now1.strftime("%Y-%m-%d")

Using the above code I tried to select the file using following:

LCMS_MRD_Delta_LoyaltyAccount_1992_%s.csv'% now

However, if you look closely you will notice that there is a space between the datestamp and the timestamp, i.e between 22 and 06

LCMS_MRD_Delta_LoyaltyAccount_1992_2018-12-22 06-07-31

It is because if this space that is preventing my code above from working.

I don't think Databricks supports wildcards so the following won't work:

LCMS_MRD_Delta_LoyaltyAccount_1992_%s.csv'% now

Someone once suggested TRUNCATING the timestamp.

Can someone let me know if:

A.TRUNCATING will solve this problem B.Is there a way to my code LCMS_MRD_Delta_LoyaltyAccount_1992_%s.csv'% now

To select the whole file? Bearing in mind I definitely need to select based on current date.. I just want to be able to use my code to select on the file.

Answer 1

You can read filenames with dbutils and can check if a pattern matches in an if-statement: if now in filname. So instead of reading files with a specific pattern directly, you get a list of files and then copy the concrete files matching your required pattern.

The following code works in a databricks python notebook:

1. Writing three files to the filesystem:

data = """
{"a":1, "b":2, "c":3}
{"a":{, b:3} 
{"a":5, "b":6, "c":7}

"""

dbutils.fs.put("/mnt/adls2/demo/files/file1-2018-12-22 06-07-31.json", data, True)
dbutils.fs.put("/mnt/adls2/demo/files/file2-2018-02-03 06-07-31.json", data, True)
dbutils.fs.put("/mnt/adls2/demo/files/file3-2019-01-03 06-07-31.json", data, True)

2. Reading the filnames as a list:

files = dbutils.fs.ls("/mnt/adls2/demo/files/")

3. Getting the actual date:

import datetime

now = datetime.datetime.now().strftime("%Y-%m-%d")
print(now)

Output: 2019-01-03

4. Copy actual files:

for i in range (0, len(files)):
  file = files[i].name
  if now in file:  
    dbutils.fs.cp(files[i].path,'/mnt/adls2/demo/target/' + file)
    print ('copied     ' + file)
  else:
    print ('not copied ' + file)

Output:

not copied file1-2018-12-22 06-07-31.json

not copied file2-2018-02-03 06-07-31.json

copied file3-2019-01-03 06-07-31.json