How to replace all occurrences of target in matched group?

Question

I want to replace all the quoted strings in a blankets into double quoted strings in the PostgreSQL.

This is my current scripts and sample input.

select regexp_replace(
    'input string',
    '\[(("([^"]+)"|''([^'''']+)'')(, )?)+]',
    '["\3\4"]',
    'g'
)

["The D'Ascoyne Family: The Duke", 'The Banker', 'The Parson', 'The General', 'The Admiral', 'Young Ascoyne', 'Young Henry', 'Lady Agatha']
['Fabian', '...']
['Peter Griffin', 'Stewie Griffin', 'Brian Griffin', 'Carter Pewterschmidt']

However, it replaces only last quoted string and removes everything else.

["Lady Agatha"]
["..."]
["Carter Pewterschmidt"]

The desired outputs are like the following

["The D'Ascoyne Family: The Duke", "The Banker", "The Parson", "The General", "The Admiral", "Young Ascoyne", "Young Henry", "Lady Agatha"]
["Fabian", "..."]
["Peter Griffin", "Stewie Griffin", "Brian Griffin", "Carter Pewterschmidt"]

https://regex101.com/r/dvAIBt/1

How can I overcome this issue?

Answer 1

In case your strings do not contain """ substrings, you can use

REPLACE(REGEXP_REPLACE(col, '("[^"\\]*(?:\\.[^"\\]*)*")|''([^''\\]*(?:\\.[^''\\]*)*)''', '\1"\2"'), '"""', '"')

The ("[^"\\]*(?:\\.[^"\\]*)*")|'([^'\\]*(?:\\.[^'\\]*)*)' regex matches

• ( - start of Group 1:
• " - a " char
• [^"\\]* - zero or more chars other than " and \
• (?:\\.[^"\\]*)* - zero or more sequences of a \ followed with any single char and then zero or more chars other than " and \
• " - a " char
• ) - end of Group 1
• | - or
• ' - a ' char
• ([^'\\]*(?:\\.[^'\\]*)*) - Group 2: zero or more chars other than ' and \ and then zero or more sequences of a \ followed with any single char and then zero or more chars other than ' and \
• ' - a ' char.

Since the "..." string literals do not need changing, they will be replaced with themselves + "" , so they will have """ on the right, and thus, all you need is to add a REPLACE that will replace all """ s with " s.