Question

I have just set about the task of stripping out HTML entities from our database, as we do a lot of crawling and some of the crawlers didn’t do this at input time 🙁

So I started writing a bunch of queries that look like;

UPDATE nodes SET name=regexp_replace(name, 'à', 'à', 'g') WHERE name LIKE '%#xe0%';
UPDATE nodes SET name=regexp_replace(name, 'á', 'á', 'g') WHERE name LIKE '%#xe1%';
UPDATE nodes SET name=regexp_replace(name, 'â', 'â', 'g') WHERE name LIKE '%#xe2%';

Which is clearly a pretty naive approach. I’ve been trying to figure out if there is something clever I can do with the decode function; maybe grabbing the html entity by regex like /&#x(..);/, then passing just the %1 part to the ascii decoder, and reconstructing the string…or something…

Shall I just press on with the queries? There will probably only be 40 or so of them.

Answer 1

Write a function using pl/perlu and use this module https://metacpan.org/pod/HTML::Entities

Of course you need to have perl installed and pl/perl available.

1)
First of all create the procedural language pl/perlu:

CREATE EXTENSION plperlu;

2) Then create a function like this:

CREATE FUNCTION decode_html_entities(text) RETURNS TEXT AS $$
    use HTML::Entities;
    return decode_entities($_[0]);
$$ LANGUAGE plperlu;

3) Then you can use it like this:

select decode_html_entities('aaabbb&.... asasdasdasd …');
   decode_html_entities    
---------------------------
 aaabbb&.... asasdasdasd …
(1 row)