if you want this to work anywhere in the UK, there are several formats

the following regex :-

[A-Z]{1,2}[A-Z]?\d+\s\d

should match the 6 possible patterns

• EC1A 1BB
• W1A 0AX
• M1 1AE
• B33 8TH
• CR2 6XH
• DN55 1PT

The following works, assuming your field is called pcd.

You can use the regexp_substr() function in QGIS expressions, but this won't work unless

• there is an outer group (i.e. brackets around the whole expression)
• backslashes are doubled (to escape them)

You can use this expression to extract the bit you want

regexp_substr("pcd",'([A-Z]{1,2}\\d+[A-Z]?\\s\\d)') 

I found a few which didn't match, I'm not sure where I got the data from (either OS or ONS; the latter includes lots of expired postcodes which may explain some discrepancies)

A simpler way (and probably faster too, with large datasets) is to simply find the space and take everything up to and including the character after it

 left("pcd" ,strpos("pcd",' ')+1)

if you want this to work anywhere in the UK, there are several formats

the following regex :-

[A-Z]{1,2}\d+[A-Z]?\s\d

should match the 6 possible patterns

• EC1A 1BB
• W1A 0AX
• M1 1AE
• B33 8TH
• CR2 6XH
• DN55 1PT

The following works, assuming your field is called pcd.

You can use the regexp_substr() function in QGIS expressions, but this won't work unless

• there is an outer group (i.e. brackets around the whole expression)
• backslashes are doubled (to escape them)

You can use this expression to extract the bit you want

regexp_substr("pcd",'([A-Z]{1,2}\\d+[A-Z]?\\s\\d)') 

I found a few which didn't match, I'm not sure where I got the data from (either OS or ONS; the latter includes lots of expired postcodes which may explain some discrepancies)

A simpler way (and probably faster too, with large datasets) is to simply find the space and take everything up to and including the character after it

 left("pcd" ,strpos("pcd",' ')+1)

if you want this to work anywhere in the UK, there are several formats

the following regex :-

[A-Z]{1,2}[A-Z]?\d+\s\d

should match the 6 possible patterns

• EC1A 1BB
• W1A 0AX
• M1 1AE
• B33 8TH
• CR2 6XH
• DN55 1PT

The following works, assuming your field is called pcd.

Note that the regexp_substr won't work unless

• there is an outer group (i.e. brackets around the whole expression)
• backslashes are doubled (to escape them)

You can use this expression to extract the bit you want

regexp_substr("pcd",'([A-Z]{1,2}\\d+[A-Z]?\\s\\d)') 

I found a few which didn't match, I'm not sure where I got the data from (either OS or ONS; the latter includes lots of expired postcodes which may explain some discrepancies)

A simpler way (and probably faster too, with large datasets) is to simply find the space and take everything up to and including the character after it

 left("pcd" ,strpos("pcd",' ')+1)

if you want this to work anywhere in the UK, there are several formats

the following regex :-

[A-Z]{1,2}[A-Z]?\d+\s\d

should match the 6 possible patterns

• EC1A 1BB
• W1A 0AX
• M1 1AE
• B33 8TH
• CR2 6XH
• DN55 1PT

The following works, assuming your field is called pcd.

You can use the regexp_substr() function in QGIS expressions, but this won't work unless

• there is an outer group (i.e. brackets around the whole expression)
• backslashes are doubled (to escape them)

You can use this expression to extract the bit you want

regexp_substr("pcd",'([A-Z]{1,2}\\d+[A-Z]?\\s\\d)') 

I found a few which didn't match, I'm not sure where I got the data from (either OS or ONS; the latter includes lots of expired postcodes which may explain some discrepancies)

A simpler way (and probably faster too, with large datasets) is to simply find the space and take everything up to and including the character after it

 left("pcd" ,strpos("pcd",' ')+1)