From d5848df776da903afaa688a252a7035838249922 Mon Sep 17 00:00:00 2001
From: Olivier Dony <odo@openerp.com>
Date: Fri, 14 Feb 2014 12:48:45 +0100
Subject: [PATCH] Add support for 2013 extension of Irish PPS Numbers

References:
- https://www.welfare.ie/en/Pages/PPSN.aspx
- http://www.citizensinformation.ie/en/social_welfare/irish_social_welfare_system/personal_public_service_number.html
---
 stdnum/ie/pps.py          | 37 ++++++++++++++++++++++++++++++-------
 stdnum/ie/vat.py          | 23 +++++++++++++++--------
 tests/test_eu_vat.doctest |  2 ++
 3 files changed, 47 insertions(+), 15 deletions(-)

diff --git a/stdnum/ie/pps.py b/stdnum/ie/pps.py
index 0447bf3f..a98f94db 100644
--- a/stdnum/ie/pps.py
+++ b/stdnum/ie/pps.py
@@ -19,11 +19,28 @@
 
 """PPS No (Personal Public Service Number, Irish personal number).
 
-The Personal Public Service number consists of 8 digits. The first seven
-are numeric and the last is the check character. The number is sometimes
-be followed by an extra letter that can be a 'W', 'T' or an 'X' and is
-ignored for the check algorithm.
+The Personal Public Service number consists of 7 digits, and one or
+two letters. The first letter is a check character.
+When present (which should be the case for new numbers as of 2013),
+the second letter can be 'A' (for individuals) or 'H' (for
+non-individuals, such as limited companies, trusts, partnerships
+and unincorporated bodies). Pre-2013 values may have 'W', 'T',
+or 'X' as the second letter ; it is ignored by the check.
 
+>>> validate('6433435F') # pre-2013
+'6433435F'
+>>> validate('6433435FT') # pre-2013 with special final 'T'
+'6433435FT'
+>>> validate('6433435FW') # pre-2013 format for married women
+'6433435FW'
+>>> validate('6433435OA') # 2013 format (personal)
+'6433435OA'
+>>> validate('6433435IH') # 2013 format (non-personal)
+'6433435IH'
+>>> validate('6433435VH') # 2013 format (incorrect check)
+Traceback (most recent call last):
+    ...
+InvalidChecksum: ...
 >>> validate('6433435F')
 '6433435F'
 >>> validate('6433435E')  # incorrect check digit
@@ -39,7 +56,7 @@
 from stdnum.util import clean
 
 
-pps_re = re.compile('^\d{7}[A-W][WTX]?$')
+pps_re = re.compile('^\d{7}[A-W][AHWTX]?$')
 """Regular expression used to check syntax of PPS numbers."""
 
 
@@ -55,8 +72,14 @@ def validate(number):
     number = compact(number)
     if not pps_re.match(number):
         raise InvalidFormat()
-    if number[7] != vat.calc_check_digit(number[:7]):
-        raise InvalidChecksum()
+    if len(number) == 9 and number[8] in 'AH':
+        # new 2013 format
+        if number[7] != vat.calc_check_digit(number[:7] + number[8:]):
+            raise InvalidChecksum()
+    else:
+        # old format, last letter ignored
+        if number[7] != vat.calc_check_digit(number[:7]):
+            raise InvalidChecksum()
     return number
 
 
diff --git a/stdnum/ie/vat.py b/stdnum/ie/vat.py
index 1988093d..39b5207c 100644
--- a/stdnum/ie/vat.py
+++ b/stdnum/ie/vat.py
@@ -22,8 +22,10 @@
 The Irish VAT number consists of 8 digits. The last digit is a check
 letter, the second digit may be a number, a letter, "+" or "*".
 
->>> validate('IE 6433435F')
+>>> validate('IE 6433435F') # pre-2013 format
 '6433435F'
+>>> validate('IE 6433435OA') # 2013 format
+'6433435OA'
 >>> validate('6433435E')  # incorrect check digit
 Traceback (most recent call last):
     ...
@@ -53,9 +55,14 @@ def calc_check_digit(number):
     """Calculate the check digit. The number passed should not have the
     check digit included."""
     alphabet = 'WABCDEFGHIJKLMNOPQRSTUV'
-    number = (7 - len(number)) * '0' + number
-    return alphabet[sum((8 - i) * int(n) for i, n in enumerate(number)) % 23]
-
+    number = number.zfill(7)
+    if len(number) < 8 or number[7] in ' W':
+        # for pre-2013 compatibility, W or space are not significant
+        extra_number = 0
+    else:
+        extra_number = 9 * (ord(number[7]) - 64)
+    return alphabet[(sum((8 - i) * int(n) for i, n in enumerate(number[:7])) +
+        + extra_number) % 23]
 
 def validate(number):
     """Checks to see if the number provided is a valid VAT number. This checks
@@ -63,19 +70,19 @@ def validate(number):
     number = compact(number)
     if not number[:1].isdigit() or not number[2:7].isdigit():
         raise InvalidFormat()
-    if len(number) != 8:
+    if len(number) not in (8, 9):
         raise InvalidLength()
     if number[:7].isdigit():
         # new system
-        if number[-1] != calc_check_digit(number[:-1]):
+        if number[7] != calc_check_digit(number[:7] + number[8:]):
             raise InvalidChecksum()
     elif number[1] in 'ABCDEFGHIJKLMNOPQRSTUVWXYZ+*':
         # old system
-        if number[-1] != calc_check_digit(number[2:-1] + number[0]):
+        if number[7] != calc_check_digit(number[2:7] + number[0]):
             raise InvalidChecksum()
     else:
         raise InvalidFormat()
-    return number
+    return compact(number)
 
 
 def is_valid(number):
diff --git a/tests/test_eu_vat.doctest b/tests/test_eu_vat.doctest
index b1c67a06..098dfd1d 100644
--- a/tests/test_eu_vat.doctest
+++ b/tests/test_eu_vat.doctest
@@ -387,6 +387,8 @@ These have been found online and should all be valid numbers.
 ... IE6599001W
 ... IE8D79739I
 ... IE9Y71814N
+... IE6433435OA
+... IE6433435IH
 ...
 ... IT - 01404480202
 ... IT 0 0 6 1 8 2 8 0 4 9 9