Explaining New Python Library – Regular Expression in JSON

Hi Guys!

As discussed, here is the continuation of the previous post. We’ll explain the regular expression from the library that I’ve created recently.

First, let me share the calling script for regular expression –

##############################################
#### Written By: SATYAKI DE               ####
#### Written On: 08-Sep-2019              ####
####                                      ####
#### Objective: Main calling scripts.     ####
##############################################

from dnpr.clsDnpr import clsDnpr
import datetime as dt
import json

# Disbling Warning
def warn(*args, **kwargs):
    pass

import warnings
warnings.warn = warn

# Lookup functions from
# Azure cloud SQL DB

def main():
    try:
        # Initializing the class
        t = clsDnpr()
        
        srcJson = [
                    {"FirstName": "Satyaki", "LastName": "De", "Sal": 1000},
                    {"FirstName": "Satyaki", "LastName": "De", "Sal": 1000},
                    {"FirstName": "Archi", "LastName": "Bose", "Sal": 500},
                    {"FirstName": "Archi", "LastName": "Bose", "Sal": 7000},
                    {"FirstName": "Deb", "LastName": "Sen", "Sal": 9500}
                  ]

        print("4. Checking regular expression functionality!")
        print()

        var13 = dt.datetime.now().strftime("%Y-%m-%d %H-%M-%S")
        print("Start Time: ", str(var13))

        print('::Function Regex_Like:: ')
        print()

        tarColumn = 'FirstName'
        print('Target Column for Rexex_Like: ', tarColumn)
        inpPattern = r"\bSa"
        print('Input Pattern: ', str(inpPattern))

        # Invoking the distinct function
        tarJson = t.regex_like(srcJson, tarColumn, inpPattern)

        print('End of Function Regex_Like!')
        print()

        print("*" * 157)
        print("Output Data: ")
        tarJsonFormat = json.dumps(tarJson, indent=1)
        print(str(tarJsonFormat))
        print("*" * 157)

        if not tarJson:
            print()
            print("No relevant output data!")
            print("*" * 157)
        else:
            print()
            print("Relevant output data comes!")
            print("*" * 157)

        var14 = dt.datetime.now().strftime("%Y-%m-%d %H-%M-%S")
        print("End Time: ", str(var14))

        var15 = dt.datetime.now().strftime("%Y-%m-%d %H-%M-%S")
        print("Start Time: ", str(var15))

        print('::Function Regex_Replace:: ')
        print()

        tarColumn = 'FirstName'
        print('Target Column for Rexex_Replace: ', tarColumn)
        inpPattern = r"\bSa"
        print('Input Pattern: ', str(inpPattern))
        replaceString = 'Ka'
        print('Replacing Character: ', replaceString)

        # Invoking the distinct function
        tarJson = t.regex_replace(srcJson, tarColumn, inpPattern, replaceString)

        print('End of Function Rexex_Replace!')
        print()

        print("*" * 157)
        print("Output Data: ")
        tarJsonFormat = json.dumps(tarJson, indent=1)
        print(str(tarJsonFormat))
        print("*" * 157)

        if not tarJson:
            print()
            print("No relevant output data!")
            print("*" * 157)
        else:
            print()
            print("Relevant output data comes!")
            print("*" * 157)

        var16 = dt.datetime.now().strftime("%Y-%m-%d %H-%M-%S")
        print("End Time: ", str(var16))

        var17 = dt.datetime.now().strftime("%Y-%m-%d %H-%M-%S")
        print("Start Time: ", str(var17))

        print('::Function Regex_Substr:: ')
        print()

        tarColumn = 'FirstName'
        print('Target Column for Regex_Substr: ', tarColumn)
        inpPattern = r"\bSa"
        print('Input Pattern: ', str(inpPattern))

        # Invoking the distinct function
        tarJson = t.regex_substr(srcJson, tarColumn, inpPattern)

        print('End of Function Regex_Substr!')
        print()

        print("*" * 157)
        print("Output Data: ")
        tarJsonFormat = json.dumps(tarJson, indent=1)
        print(str(tarJsonFormat))
        print("*" * 157)

        if not tarJson:
            print()
            print("No relevant output data!")
            print("*" * 157)
        else:
            print()
            print("Relevant output data comes!")
            print("*" * 157)

        var18 = dt.datetime.now().strftime("%Y-%m-%d %H-%M-%S")
        print("End Time: ", str(var18))

        print("=" * 157)
        print("End of regular expression function!")
        print("=" * 157)



    except ValueError:
        print("No relevant data to proceed!")

    except Exception as e:
        print("Top level Error: args:{0}, message{1}".format(e.args, e.message))

if __name__ == "__main__":
    main()

As per the library, we’ll discuss the following functionalities –

  1. regex_like
  2. regex_replace
  3. regex_substr

Now, let us check how to call these functions.

1. regex_like:

Following is the base skeleton in order to invoke this function –

regex_like(Input Json, Target Column, Pattern To Match) return Output Json

Here are the key lines in the script –

srcJson = [
            {"FirstName": "Satyaki", "LastName": "De", "Sal": 1000},
            {"FirstName": "Satyaki", "LastName": "De", "Sal": 1000},
            {"FirstName": "Archi", "LastName": "Bose", "Sal": 500},
            {"FirstName": "Archi", "LastName": "Bose", "Sal": 7000},
            {"FirstName": "Deb", "LastName": "Sen", "Sal": 9500}
          ]

# Invoking the distinct function
tarJson = t.regex_like(srcJson, tarColumn, inpPattern)

2. regex_replace:

Following is the base skeleton in order to invoke this function –

regex_replace(Input Json, Target Column, Pattern to Replace) return Output Json

Here are the key lines in the script –

tarColumn = 'FirstName'
print('Target Column for Rexex_Replace: ', tarColumn)
inpPattern = r"\bSa"
print('Input Pattern: ', str(inpPattern))
replaceString = 'Ka'
print('Replacing Character: ', replaceString)

# Invoking the distinct function
tarJson = t.regex_replace(srcJson, tarColumn, inpPattern, replaceString)

As you can see, here ‘Sa’ with ‘Ka’ provided it matches the specific pattern in the JSON.

3. regex_replace:

Following is the base skeleton in order to invoke this function –

regex_substr(Input Json, Target Column, Pattern to substring) return Output Json

Here are the key lines –

tarColumn = 'FirstName'
print('Target Column for Regex_Substr: ', tarColumn)
inpPattern = r"\bSa"
print('Input Pattern: ', str(inpPattern))

# Invoking the distinct function
tarJson = t.regex_substr(srcJson, tarColumn, inpPattern)

In this case, we’ve subtracted a part of the JSON string & return the final result as JSON.

Let us first see the sample input JSON –

SourceJSON_Regex

Let us check how it looks when we run the calling script –

  • regex_like:
Regex_Like

This function will retrieve the elements, which will start with ‘Sa‘. As a result, we’ll see the following two elements in the Payload.

  • regex_replace:
Regex_Replace

In this case, we’re replacing any string which starts with ‘Sa‘ & replaced with the ‘Ka‘.

  • regex_substr:
Regex_Substr

As you can see that the first element FirstName changed the name from “Satyaki” to “tyaki“.

So, finally, we’ve achieved our target.

I’ll post the next exciting concept very soon.

Till then! Happy Avenging! 😀

N.B.: This is demonstrated for RnD/study purposes. All the data posted here are representational data & available over the internet.

Reverse String using Regexp Functions

Hi,

 

Today, we’ll be checking one new area where we can implement regular expression to achieve the same without involving any kind of Macro, Stored-Proc.

 

Many occasion we may have to parse various kind of strings. Assume that, we need to parse the string in reverse order. Until TD 14.0, you don’t have any easy method to implement the same. Off course, this new method also has some limits. It can only able to reverse couple of characters only. I have checked with 9 characters.  It supports that.

 

So, if you have specific string lengths, then you may also try this solution if you are using TD 14.0. That will be handy.

 

Let’s check the Query & Output –

 

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
SELECT SEQ_NO,
                 SRC_STR,
                 regexp_replace(SRC_STR,'([[:alnum:]]{1,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})([[:alnum:]]{0,1})','\9\8\7\6\5\4\3\2\1') AS REV_SRC_STR
FROM WM_CONCAT_TAB
ORDER BY 1;


SEQ_NO	SRC_STR	 REV_SRC_STR
-----   -------  --------------
1	BIRESWAR RAWSERIB
1	TRIDIB	 BIDIRT
1	SUJAY	 YAJUS
1	ANUPAM	 MAPUNA
1	RAM	 MAR
2	PAPU	 UPAP
2	SAYAN	 NAYAS
2	TUNKAI	 IAKNUT
2	BABU	 UBAB
3	ATBIS	 SIBTA
3	SAPMUNDA ADNUMPAS
3	IK	 KI

 

Hope, this will give you another way – to implement the same logic. 🙂

String Manipulation Advanced Using Teradata 14.0 Regular Expression

Today, I’ll show couple of very useful functions or logic implemented in Teradata using It’s Regular Expression.

There is two very popular demand comes from most of the Developer across different databases regarding the following two cases –

1. How to Split Comma Separated Values in each rows 

2. How to bind separate values in 1 row (Just opposite of Step 1)

2nd Options are very demanding as Cross platform database professional specially Oracle Developers looking for these kind of implementation as Oracle has directly built-in functions to do the same. Those functions are Listagg, wm_concat, group_concat.

Let’s check the solution –

Case 1,

Let’s create the table & prepare some data –

 

1
2
3
4
5
6
7
CREATE MULTISET TABLE ETL_DATA.PARSE_STR
  (
     SEQ_NO       INTEGER,
     SRC_STR     VARCHAR(70)
  );
 
CREATE TABLE completed. 0 rows processed. Elapsed Time =  00:00:01.864

 

Let’s insert some data –

 

1
2
3
INSERT INTO ETL_DATA.PARSE_STR VALUES(1,'RAM,TRIDIB,ANUPAM,BIRESWAR,SUJAY')
;INSERT INTO ETL_DATA.PARSE_STR VALUES(2,'TUNKAI,SAYAN,BABU,PAPU')
;INSERT INTO ETL_DATA.PARSE_STR VALUES(3,'IK,ATBIS,SAPMUNDA');

 

Let’s check the value –

 

1
2
3
4
5
SEQ_NO          SRC_STR
------  ----------------------------------
    1   RAM,TRIDIB,ANUPAM,BIRESWAR,SUJAY
    2   TUNKAI,SAYAN,BABU,PAPU
    3   IK,ATBIS,SAPMUNDA

 

Fine, Now our objective will be split these comma separated values in each lines.

 

1
2
3
4
5
6
SELECT b.SEQ_NO,
       regexp_substr(b.SRC_STR,'[^,]+',1,day_of_calendar) AS SRC_STR
FROM sys_calendar.calendar ,
     PARSE_STR b
WHERE day_of_calendar BETWEEN 1 AND  (LENGTH(b.SRC_STR) - LENGTH(regexp_replace(b.SRC_STR,'[^A-Z]+','',1,0,'i'))+1 )
ORDER BY 1,2;

 

And, let’s check the output –

 

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
SEQ_NO  SRC_STR
-----   ----------------------
1       ANUPAM
1       BIRESWAR
1       RAM
1       SUJAY
1       TRIDIB
2       BABU
2       PAPU
2       SAYAN
2       TUNKAI
3       ATBIS
3       IK
3       SAPMUNDA

 

Gr8! I guess, result is coming as per my expectation. 🙂

 

Case 2(Subsitute Of Listagg, wm_concat, group_concat in Oracle),

This we’ve to do it in Two small Steps for better understanding & performance.

First, let us create another table –

 

1
2
3
4
5
6
7
CREATE MULTISET TABLE ETL_DATA.WM_CONCAT_TAB
   (
      SEQ_NO   INTEGER,
      SRC_STR VARCHAR(20)
   );
    
CREATE TABLE completed. 0 rows processed. Elapsed Time =  00:00:01.230

 

Good. Now we’ll populate some data into this table. We’ll populate data from Step 1 as this will provide the exact data that we’re expecting as input test data for Case 2.

Let’s insert those data –

 

1
2
3
4
5
6
INSERT INTO ETL_DATA.WM_CONCAT_TAB
SELECT b.SEQ_NO,
       regexp_substr(b.SRC_STR,'[^,]+',1,day_of_calendar) AS SRC_STR
FROM sys_calendar.calendar ,
     PARSE_STR b
WHERE day_of_calendar BETWEEN 1 AND  (LENGTH(b.SRC_STR) - LENGTH(regexp_replace(b.SRC_STR,'[^A-Z]+','',1,0,'i'))+1 );

 

Let’s check the data –

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
SEQ_NO  SRC_STR
------  --------------------
1       ANUPAM
1       BIRESWAR
1       RAM
1       SUJAY
1       TRIDIB
2       BABU
2       PAPU
2       SAYAN
2       TUNKAI
3       ATBIS
3       IK
3       SAPMUNDA

 

As you know in TD we’ve significant restcriction regarding Hirarchical Queries & Recursive Queries. So, In this step we’ll build one relationship like employee & manager in popular employee table. So, if we have that kind of relation then we can easily establish & fit that in TD model.

Let’s create this intermediate table. In this case we’ll go for mapping between current rows with next rows. This is also very useful process. In Oracle, they have LEAD or LAG functions to achieve the same. But, here we’ve to work a little bit more to achive the same.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
CREATE MULTISET VOLATILE TABLE VT_SRC_ARRNG
AS
     (
            SELECT SEQ_NO,
                   SRC_STR,
                   MAX(SRC_STR) OVER(
                                        PARTITION BY SEQ_NO
                                        ORDER BY SEQ_NO, SRC_STR
                                        ROWS BETWEEN 1 FOLLOWING AND 1 FOLLOWING 
                                    ) AS PREV_SRC_STR,
                   COUNT(*)  OVER(
                                    PARTITION BY SEQ_NO
                                 ) AS MAX_RECUR_CNT
            FROM WM_CONCAT_TAB
      )
WITH DATA
ON COMMIT
PRESERVE ROWS;
 
CREATE TABLE completed. 0 rows processed. Elapsed Time =  00:00:01.102

 

Let’s look the output –

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
SELECT *
FROM VT_SRC_ARRNG
ORDER BY 1,2;
 
 
 
 
SEQ_NO  SRC_STR  PREV_SRC_STR    MAX_RECUR_CNT
-----   -------  --------------- ---------------------
1       ANUPAM      BIRESWAR     5
1       BIRESWAR    RAM          5
1       RAM         SUJAY        5
1       SUJAY       TRIDIB       5
1       TRIDIB      ?            5
2       BABU        PAPU         4
2       PAPU        SAYAN        4
2       SAYAN       TUNKAI       4
2       TUNKAI      ?            4
3       ATBIS       IK           3
3       IK          SAPMUNDA     3
3       SAPMUNDA    ?            3

 

Fine. From the above VT we can see every Source String has one Previous Source String. Also, we’ve noted down that in each window of SEQ_NO how many levels are there by MAX_RECUR_CNT. We’ll use this column later.

Let’s move to the 2nd & final part –

Let’s aggregate the values based on SEQ_NO & club them with comma –

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
WITH RECURSIVE WM_CONCAT(SEQ_NO, SRC_STR, PREV_SRC_STR, MAX_RECUR_CNT, LVL,  COMMA_SEP_STR)
AS
     (
        SELECT SEQ_NO,
               SRC_STR,
               PREV_SRC_STR,
               MAX_RECUR_CNT,
               1 AS LVL,
               CAST( '' AS VARCHAR(100)) AS COMMA_SEP_STR
       FROM VT_SRC_ARRNG
       WHERE  PREV_SRC_STR IS NULL
       UNION ALL
       SELECT  b.SEQ_NO,
               b.SRC_STR,
               b.PREV_SRC_STR,
               b.MAX_RECUR_CNT,
               c.LVL+1 AS LVL,
               c.COMMA_SEP_STR||b.SRC_STR||',' AS COMMA_SEP_STR
       FROM VT_SRC_ARRNG b,
               WM_CONCAT c
       WHERE c.SRC_STR =  b.PREV_SRC_STR
     )
SELECT k.SEQ_NO,
       k.AGGR_STR
FROM (               
    SELECT SEQ_NO,
           SRC_STR,
           LVL,
           MAX_RECUR_CNT,
           MIN(CASE
                 WHEN LVL = 1 THEN
                    SRC_STR
               ELSE
                  'ZZZZZ'
               END   ) OVER(
                                 PARTITION BY SEQ_NO
                                 ORDER BY LVL ASC
                           ) ROOT_SRC_STR,
           COMMA_SEP_STR||ROOT_SRC_STR AS AGGR_STR
    FROM WM_CONCAT
    )  k
WHERE k.LVL = k.MAX_RECUR_CNT
ORDER BY 1,2;

 

Let’s check the output –

1
2
3
4
5
SEQ_NO  AGGR_STR
------- ---------------------------
1       SUJAY,RAM,BIRESWAR,ANUPAM,TRIDIB
2       SAYAN,PAPU,BABU,TUNKAI
3       IK,ATBIS,SAPMUNDA

 

I guess, We’ve done it. 😀

So, You can achieve the same without writing any UDF.