Question

I need to output the following query as csv.

I can easily write php logic to transpose the rows to columns from my group_concat column

However I am keen to keep as much of the data part in the database and minimize the manipulations on the php side.

I am experimenting with the two columns below the group_concat in the query.

The problem is the abundance value also returns for life_stage column. If there is no way around this other than manipulating the group_concat key values then that’s fine, I just wanted to double check. Thanks in advance

SELECT 
`tr`.`tr_id_pk` as 'RecordKey',
`t`.`tax_name` as `TaxonName`,
`tr`.`tr_date` as 'Date',
`s`.`si_name` as 'SiteName',
`tr`.`tr_grid_reference` as 'GridReference',
`tr`.`tr_is_site_grid` as 'IsSiteGrid',
`r`.`rec_name` as 'Recorder',
`r`.`rec_email` as 'RecorderEmail',
`tr`.`tr_comment` as 'RecordComment',
`tr`.`tr_last_update` as 'LastUpdated',
`tr`.`tr_form_key` as 'FormKey',
`c`.`co_name` as 'County',
`vc`.`vc_name` as 'ViceCounty',
`h`.`hab_name` as 'Habitat',
GROUP_CONCAT(DISTINCT CONCAT_WS('=', `ra`.`ra_name`, `rad`.`rad_value`)) as 'RecordAttributeKeyValuePairs',
`rad`.`rad_value` AS `abundance`,
`rad`.`rad_value` AS `life_stage`
FROM
`taxon_record`as `tr`
    INNER JOIN
`taxon`as `t` ON `tr`.`tax_id_fk` = `t`.`tax_id_pk`
    INNER JOIN
`recorder`as `r` ON `tr`.`rec_id_fk` = `r`.`rec_id_pk`
    INNER JOIN
`site`as `s` ON `tr`.`si_id_fk` = `s`.`si_id_pk`
    LEFT JOIN
`county`as `c` ON `tr`.`co_id_fk` = `c`.`co_id_pk`
    LEFT JOIN
`vice_county`as `vc` ON `tr`.`vc_id_fk` = `vc`.`vc_id_pk`
    LEFT JOIN
`habitat`as `h` ON `tr`.`hab_id_fk` = `h`.`hab_id_pk`
    LEFT JOIN
(`record_attribute_data`as `rad`
INNER JOIN `record_attribute`as `ra` ON (`rad`.`ra_id_fk` = `ra`.`ra_id_pk`)) ON (`tr`.`tr_id_pk` = `rad`.`tr_id_fk`)
WHERE
`r`.`rec_email` = '[e-mail]'
GROUP BY `tr`.`tr_id_pk`; 

Answer 1

What you wish to do is known as “pivoting” your data and is something for which some other RDBMS have native support, but MySQL does not (by design, as the developers feel that such manipulations belong in the presentation layer).

However, as you’ve hinted at, you can construct a rather horrible MySQL query to perform the pivoting operation manually (one needs to join the attributes tables to the query once for each output column):

SELECT tr.tr_id_pk, abundance, life_stage -- etc.
  FROM taxon_record AS tr
  LEFT JOIN (
    SELECT rad.tr_id_fk, rad.rad_value AS abundance
    FROM
      record_attribute_data ra JOIN record_attribute rad
        ON rad.ra_id_fk = ra.ra_id_pk
    WHERE ra.ra_name = 'abundance'
  ) AS tAbundance ON tAbundance.tr_id_fk = tr.tr_id_pk
  LEFT JOIN (
    SELECT rad.tr_id_fk, rad.rad_value AS life_stage
    FROM
      record_attribute_data ra JOIN record_attribute rad
        ON rad.ra_id_fk = ra.ra_id_pk
    WHERE ra.ra_name = 'life_stage'
  ) AS tLife_Stage ON tLife_Stage.tr_id_fk = tr.tr_id_pk
  -- etc.

If you choose to go down this path, you can make your life slightly easier by generating this query using either a looping construct in PHP or a prepared statement in MySQL.