String matching in URL using Hive / Spark SQL - regex

I have two tables, one containing list of URL and other having a list of words. My requirement is to filter out the URLs containing the words.
For eg:
URL
https://www.techhive.com/article/3409153/65-inch-oled-4k-tv-from-lg-at-a-1300-dollar-discount.html
https://www.techradar.com/in/news/lg-c9-oled-65-inch-4ktv-price-drop
https://www.t3.com/news/cheap-oled-tv-deals-currys-august
https://indianexpress.com/article/technology/gadgets/lg-bets-big-on-oled-tvs-in-india-to-roll-out-rollable-tv-by-year-end-5823635/
https://www.sony.co.in/electronics/televisions/a1-series
https://www.amazon.in/Sony-138-8-inches-Bravia-KD-55A8F/dp/B07BWKVBYW
https://www.91mobiles.com/list-of-tvs/sony-oled-tv
Words
Sony
Samsung
Deal
Bravia
Now I want to filter any URL that has any of the words. Normally i would do a
Select url from url_table where url not like '%Sony%' or url not like '%Samsung%' or url not like '%Deal%' or not like '%Bravia%';
But that's a cumbersome and not scalable way to do it. What is the best way to achieve this? How do I use a not like function to the words table?

Using regex:
where url not rlike '(?i)Sony|Samsung|Deal|Bravia'
(?i) means case insesitive.
And now let's build the same regexp from the table with words.
You can aggregate list of words from the table and pass it to the rlike. See this example:
with
initial_data as (--replace with your table
select stack(7,
'https://www.techhive.com/article/3409153/65-inch-oled-4k-tv-from-lg-at-a-1300-dollar-discount.html',
'https://www.techradar.com/in/news/lg-c9-oled-65-inch-4ktv-price-drop',
'https://www.t3.com/news/cheap-oled-tv-deals-currys-august',
'https://indianexpress.com/article/technology/gadgets/lg-bets-big-on-oled-tvs-in-india-to-roll-out-rollable-tv-by-year-end-5823635/',
'https://www.sony.co.in/electronics/televisions/a1-series',
'https://www.amazon.in/Sony-138-8-inches-Bravia-KD-55A8F/dp/B07BWKVBYW',
'https://www.91mobiles.com/list-of-tvs/sony-oled-tv'
) as url ) ,
words as (-- replace with your words table
select stack (4, 'Sony','Samsung','Deal','Bravia') as word
),
sub as (--aggregate list of words for rlike
select concat('''','(?i)',concat_ws('|',collect_set(word)),'''') words_regex from words
)
select s.url
from initial_data s cross join sub --cross join with words_regex
where url not rlike sub.words_regex --rlike works fine
Result:
OK
url
https://www.techhive.com/article/3409153/65-inch-oled-4k-tv-from-lg-at-a-1300-dollar-discount.html
https://www.techradar.com/in/news/lg-c9-oled-65-inch-4ktv-price-drop
https://indianexpress.com/article/technology/gadgets/lg-bets-big-on-oled-tvs-in-india-to-roll-out-rollable-tv-by-year-end-5823635/
Time taken: 10.145 seconds, Fetched: 3 row(s)
Also you can calculate sub subquery separately and pass it's result as a variable instead of cross join in my example. Hope you got the idea.

Related

MYSQL get substring

I'm trying to get substring dynamically and group by it. So if my uri column contains records like: /uri1/uri2 and /somelongword/someotherlongword I would like to get everything up to second delimiter, namely up to second / and count it. I'm using this query but obviously it is cutting string statically (6 letters after the first one).
SELECT substr(uri, 1, 6) as URI,
COUNT(*) as COUNTER
FROM staging
GROUP BY substr(uri, 1, 6)
ORDER BY COUNTER DESC
How can I achieve that?
You can use combination of SUBSTRING() and POSITION()
schema:
CREATE TABLE Table1
(`uri` varchar(10))
;
INSERT INTO Table1
(`uri`)
VALUES
('some/text'),
('some/text1'),
('some/text2'),
('aa/bb'),
('aa/cc'),
('bb/cc')
;
query
SELECT
SUBSTRING(uri,1,POSITION('/' IN uri)-1),
COUNT(*)
FROM Table1
GROUP BY SUBSTRING(uri,1,POSITION('/' IN uri)-1);
http://sqlfiddle.com/#!9/293dd3/3/0
edit: here I found amazon athena documentation: https://docs.aws.amazon.com/athena/latest/ug/presto-functions.html and here is the string function documentation: https://prestodb.io/docs/0.217/functions/string.html
my answer above still stands, but you might need to change SUBSTRING to SUBSTR
edit 2: it seems there's a special function to achieve this in amazon athena called SPLIT_PART()
query:
SELECT SPLIT_PART(uri, '/', 1), COUNT(*) FROM tbl GROUP BY SPLIT_PART(uri, '/', 1)
from docs:
split_part(string, delimiter, index) → varchar
Splits string on delimiter and returns the field index. Field indexes start with 1. If the index is larger than than the number of fields, then null is returned.

Filter Out Rows With 3 Or More Sub-Folders

I want to filter out entries of URLs that are the third level or higher. An example is https://example.com/folder1/folder2/folder3 would be filtered out but https://example.com/folder1/folder2 would not be filtered. I have attempted to create a filter using regex:
=filter(A2:A10000, regexmatch(A2:A10000 "https://[^/]+/[^/]+/[^/]+/.*"))
There's no output that indicates I'm close as no rows get filtered.
I believe you should simply alter your regex like this:
=filter(A2:A,not(regexmatch(A2:A,"https?:\/(\/.+?){4}.*")))
This takes into consideration possible trailing slashes to your URLs, see https://regex101.com/r/BEJYTF/1.
=ARRAYFORMULA(SUBSTITUTE(REGEXREPLACE(TRIM(TRANSPOSE(QUERY(TRANSPOSE(QUERY(
IF(IFERROR(SPLIT(A2:A, "/"))<>"", IFERROR(SPLIT(A2:A, "/"))&"/", ),
"where Col5 is not null", 0)), , 999^99))), "\s|/$", ), ":/", "://"))

How to add a string on a specific string by using regex_replace method in Oracle

I am trying to add a string '_$' to a index name and a table name as follows. I need to use a method 'regexp_replace' in SELECT statement.
select regexp_replace(input_string......)
# Input
CREATE UNIQUE INDEX "SCOTT"."PK_EMP" ON "SCOTT"."EMP" ("EMP_NO")
# Desired Output
CREATE UNIQUE INDEX "SCOTT"."PK_EMP_$" ON "SCOTT"."EMP_$" ("EMP_NO")
Can you help me to build a regular expression for that?
Fairly brute solution would be using the following pattern:
(.*)(" ON ".*)(" \(.*)
with the following replace string:
\1_$\2_$\3
The pattern works by splitting the input in the places where you need to insert the _$ token, and then joining it back placing the tokens in the places we split the input:
CREATE UNIQUE INDEX "SCOTT"."PK_EMP|" ON "SCOTT"."EMP|" ("EMP_NO")
Full SELECT query would look like that:
SELECT REGEXP_REPLACE(
'CREATE UNIQUE INDEX "SCOTT"."PK_EMP" ON "SCOTT"."EMP" ("EMP_NO")',
'(.*)(" ON ".*)(" \(.*)',
'\1_$\2_$\3'
) RX
FROM dual;

REGEX help needed in Oracle

How to get all the table names from the below Sql? My sql returns only the last table name.
with t as
(select 'select col1,
(select max(col3) from dd3) max_timestamp
from dd1,
dd2
where dd1.col1 = dd2.col1
and dd1.col1 in(select col1 from dd4)' sql_text from dual)
select regexp_substr(regexp_substr(upper(sql_text), '\sFROM\s*(\w|\.|_)*'), '(\w|_|\.)+', 1,2)
from t
Thanks,
DD.
This is a more of a regex question than an Oracle question.
If you can run the sql through REPLACE(REPLACE(sql,CHR(13),' '),CHR(10),NULL) to replace all newlines with a space, so that the query fits on a single line, here is regex that will return all the tables in group 1 (for the ones after FROM) and group 3 for subsequent items in a list:
/FROM ([A-Z0-9$#_]+)(,[\s]*([A-Z0-9$#_]+))*/gi
Having multiple groups is not ideal, so I would look at the full match instead, see https://regex101.com/r/OZUalH/1/ for an example (see full match on the right, where every match has from followed by one or more tables).
But let me warn you this is not going to be robust, as these valid FROM clause expressions are not handled:
"my_table"
MY_TABLE AS A
MY_TABLE AS "a"
etc...
If it were me, I would write a function to run the query through explain plan (execute immediate 'explain plan for ...') and extract the tables from the plan tables (or possibly using SYS.DBMS_XPLAN)

Matching number sequences in SQLite with random character separators

I have an sqlite database which has number sequences with random separators. For example
_id data
0 123-45/678>90
1 11*11-22-333
2 4-4-5-67891
I want to be able to query the database "intelligently" with and without the separators. For example, both these queries returning _id=0
SELECT _id FROM myTable WHERE data LIKE '%123-45%'
SELECT _id FROM myTable WHERE data LIKE '%12345%'
The 1st query works as is, but the 2nd query is the problem. Because the separators appear randomly in the database there are too many combinations to loop through in the search term.
I could create two columns, one with separators and one without, running each query against each column, but the database is huge so I want to avoid this if possible.
Is there some way to structure the 2nd query to achieve this as is ? Something like a regex on each row during the query ? Pseudo code
SELECT _id
FROM myTable
WHERE REPLACEALL(data,'(?<=\\d)[-/>*](?=\\d)','') LIKE '%12345%'
Ok this is far from being nice, but you could straightforwardly nest the REPLACE function. Example:
SELECT _id FROM myTable
WHERE REPLACE(..... REPLACE(REPLACE(data,'-',''),'_',''), .... '<all other separators>','') = '12345'
When using this in practice (--not that I would recommend it, but at least its simple), you surely might wrap it inside a function.
EDIT: for a small doc on the REPLACE function, see here, for example.
If I get it right, is this what you want?
SELECT _id
FROM myTable
WHERE Replace(Replace(Replace(data, '?', ''), '/', ''), '-', '') LIKE '%12345%'