views:

189

answers:

3

I need to re-format a list of UK postcodes and have started with the following to strip whitespace and capitalize:

postcode.upcase.gsub(/\s/,'')

I now need to change the postcode so the new postcode will be in a format that will match the following regexp:

^([A-PR-UWYZ0-9][A-HK-Y0-9][AEHMNPRTVXY0-9]?[ABEHMNPRVWXY0-9]? {1,2}[0-9][ABD-HJLN-UW-Z]{2}|GIR 0AA)$

I would be grateful of any assistance.

A: 

Reformat or pattern match? I suspect the latter, although upcasing it first is a good idea.

Before we proceed though I would point out that you are stripping spaces but your regex contains " {1,2}" which is "one or two space characters". As you have already stripped whitespace you've already caused all to fail the match.

Given a post code as input we can check whether it matches the regex using =~

Here we create some example post codes (taken from the wikipedia page), and test each one against the regex:

post_codes = ["M1 1AA", "M60 1NW", "CR2 6XH", "DN55 1PT", "W1A 1HQ", "EC1A 1BB", "bad one", "cc93h29r2"]
r = /^([A-PR-UWYZ0-9][A-HK-Y0-9][AEHMNPRTVXY0-9]?[ABEHMNPRVWXY0-9]? {1,2}[0-9][ABD-HJLN-UW-Z]{2}|GIR 0AA)$/

post_codes.each do |pc|
  # pc =~ r will return something true if we have a match (specifically the integer of first match position)
  # We use !! to display it as true|false
  puts "#{pc}: #{!!(pc =~ r)}"
end
M1 1AA: true
M60 1NW: true
CR2 6XH: true
DN55 1PT: true
W1A 1HQ: true
EC1A 1BB: true
bad one: false
cc93h29r2: false
The postcodes in the database are valid, they are just not formatted correctly. I need to reformat each postcode and update the database. Typically they have no spaces, with mixed upper and lower case.Taking your example postcode set, I want to reformat dn551Pt to DN55 1PT and update the entry in the database.
noddy
A: 

UK Postcodes aren't consistent, but they are finite - you might be better with a look-up table.

Phil
A: 

If this standards doc is to be believed (and Wikipedia concurs), formatting a valid post code for output is straightforward: the last three characters are the second part, everything before is the first part!

So assuming you have a valid postcode, without any pre-embedded space, you just need

def format_post_code(pc)
  pc.strip.sub(/([A-Z0-9]+)([A-Z0-9]{3})/, '\1 \2')
end

If you want to validate an input post code first, then the regex you gave looks like a good starting point. Perhaps something like this?

NORMAL_POSTCODE_RE = /^([A-PR-UWYZ][A-HK-Y0-9][A-HJKS-UW0-9]?[A-HJKS-UW0-9]?)\s*([0-9][ABD-HJLN-UW-Z]{2})$/i
GIROBANK_POSTCODE_RE = /^GIR\s*0AA$/i
def format_post_code(pc)
  return pc.strip.upcase.sub(NORMAL_POSTCODE_RE, '\1 \2') if pc =~ NORMAL_POSTCODE_RE
  return 'GIR 0AA' if pc =~ GIROBANK_POSTCODE_RE
end

Note that I removed the '0-9' part of the first character, which appears unnecessary according to the sources I quoted. I also changed the alpha sets to match the first-cited document. It's still not perfect: a code of the format 'AAA ANN' validates, for example, and I think a more complex RE is probably required.

I think this might cover it (constructed in stages for easier fixing!)

A1  = "[A-PR-UWYZ]"
A2  = "[A-HK-Y]"
A34 = "[A-HJKS-UW]"        # assume rule for alpha in fourth char is same as for third
A5  = "[ABD-HJLN-UW-Z]"
N   = "[0-9]"
AANN = A1 + A2 + N + N     # the six possible first-part combos
AANA = A1 + A2 + N + A34
ANA  = A1 + N + A34
ANN  = A1 + N + N
AAN  = A1 + A2 + N
AN   = A1 + N
PART_ONE = [AANN, AANA, ANA, ANN, AAN, AN].join('|') 
PART_TWO = N + A5 + A5

NORMAL_POSTCODE_RE = Regexp.new("^(#{PART_ONE})[ ]*(#{PART_TWO})$", Regexp::IGNORECASE)  
Mike Woodhouse
Thanks Mike exactly what I was looking for :)
noddy