parser: refine IBAN checking

Allow arbitrary spacing, minor optimizations...

Signed-off-by: Simon Rozman <simon@rozman.si>

UnitTests/parser.cpp

@@ -205,11 +205,47 @@ namespace UnitTests
 				Assert::IsTrue(p.match(L"si56 0231 2001 5226 972", 0, SIZE_MAX, match_case_insensitive));
 				Assert::IsTrue(p.is_valid);
 				Assert::IsTrue(p.match(L"SI56 0231 2001 5226 9720", 0, SIZE_MAX));
-				Assert::IsFalse(p.is_valid);
+				Assert::AreEqual(stdex::interval<size_t>(0, 23), p.interval);
+				Assert::IsTrue(p.is_valid);
 				Assert::IsTrue(p.match(L"...SI56 0231 2001 5226 972...", 3, SIZE_MAX));
 				Assert::IsTrue(p.is_valid);
 				Assert::IsTrue(p.match(L"SI56 0231 2001 5226 972", 0, SIZE_MAX)); // no-break space
 				Assert::IsTrue(p.is_valid);
+
+				Assert::IsTrue(p.match(L"BE71 0961 2345 6769", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"BR15 0000 0000 0000 1093 2840 814 P2", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"CR99 0000 0000 0000 8888 88", 0, SIZE_MAX));
+				Assert::IsFalse(p.is_valid);
+				Assert::IsTrue(p.match(L"FR76 3000 6000 0112 3456 7890 189", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"IE12 BOFI 9000 0112 3456 78", 0, SIZE_MAX));
+				Assert::IsFalse(p.is_valid);
+				Assert::IsTrue(p.match(L"DE91 1000 0000 0123 4567 89", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"GR96 0810 0010 0000 0123 4567 890", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"MU43 BOMM 0101 1234 5678 9101 000 MUR", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"PK70 BANK 0000 1234 5678 9000", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"PL10 1050 0099 7603 1234 5678 9123", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"RO09 BCYP 0000 0012 3456 7890", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"LC14 BOSL 1234 5678 9012 3456 7890 1234", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"SA44 2000 0001 2345 6789 1234", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"ES79 2100 0813 6101 2345 6789", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"SE87 3000 0000 0101 2345 6789", 0, SIZE_MAX));
+				Assert::IsFalse(p.is_valid);
+				Assert::IsTrue(p.match(L"CH56 0483 5012 3456 7800 9", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
+				Assert::IsTrue(p.match(L"GB98 MIDL 0700 9312 3456 78", 0, SIZE_MAX));
+				Assert::IsTrue(p.is_valid);
 			}
 
 			{
@@ -322,7 +358,8 @@ namespace UnitTests
 				Assert::IsTrue(p.match("si56 0231 2001 5226 972", 0, SIZE_MAX, match_case_insensitive));
 				Assert::IsTrue(p.is_valid);
 				Assert::IsTrue(p.match("SI56 0231 2001 5226 9720", 0, SIZE_MAX));
-				Assert::IsFalse(p.is_valid);
+				Assert::AreEqual(stdex::interval<size_t>(0, 23), p.interval);
+				Assert::IsTrue(p.is_valid);
 				Assert::IsTrue(p.match("...SI56 0231 2001 5226 972...", 3, SIZE_MAX));
 				Assert::IsTrue(p.is_valid);
 				Assert::IsTrue(p.match("SI56&nbsp;0231&nbsp;2001&nbsp;5226&nbsp;972", 0, SIZE_MAX));

include/stdex/parser.hpp

@@ -4806,7 +4806,8 @@ namespace stdex
 					{ { 'X', 'K' }, {}, 20 }, // Kosovo
 				};
 				const country_t* country_desc = nullptr;
-				size_t n;
+				size_t n, available, next, bban_length;
+				uint32_t nominator;
 
 				this->interval.end = start;
 				for (size_t i = 0; i < 2; ++i, ++this->interval.end) {
@@ -4844,32 +4845,28 @@ namespace stdex
 					(country_desc->check_digits[1] && this->check_digits[1] != country_desc->check_digits[1]))
 					goto error; // unexpected check digits
 
-				for (n = 0; ;) {
+				bban_length = country_desc->length - 4;
-					if (m_space && m_space->match(text, this->interval.end, end, flags))
+				for (n = 0; n < bban_length;) {
-						this->interval.end = m_space->interval.end;
-					for (size_t j = 0; j < 4; ++j) {
 					if (this->interval.end >= end || !text[this->interval.end])
-							goto out;
+						goto error; // bban too short
+					if (m_space && m_space->match(text, this->interval.end, end, flags)) {
+						this->interval.end = m_space->interval.end;
+						continue;
+					}
 					T chr = case_insensitive ? ctype.toupper(text[this->interval.end]) : text[this->interval.end];
 					if (('0' <= chr && chr <= '9') || ('A' <= chr && chr <= 'Z')) {
-							if (n >= _countof(this->bban) - 1)
-								goto error; // bban overflow
 						this->bban[n++] = chr;
 						this->interval.end++;
 					}
 					else
-							goto out;
+						goto error; // invalid bban
-					}
 				}
 			out:
-				if (n < 11)
-					goto error; // bban too short (shorter than Norwegian)
 				this->bban[n] = 0;
 
-				if (n + 4 == country_desc->length) {
 				// Normalize IBAN.
 				T normalized[69];
-					size_t available = 0;
+				available = 0;
 				for (size_t i = 0; ; ++i) {
 					if (!this->bban[i]) {
 						for (i = 0; i < 2; ++i) {
@@ -4908,21 +4905,17 @@ namespace stdex
 				}
 
 				// Calculate modulo 97.
-					size_t next;
+				nominator = stdex::strtou32(normalized, 9, &next, 10);
-					uint32_t nominator = stdex::strtou32(normalized, 9, &next, 10);
 				for (;;) {
 					nominator %= 97;
 					if (!normalized[next]) {
 						this->is_valid = nominator == 1;
 						break;
 					}
-						size_t digit_count = nominator < 10 ? 1 : 2;
+					size_t digit_count = nominator == 0 ? 0 : nominator < 10 ? 1 : 2;
 					for (; digit_count < 9 && normalized[next]; ++next, ++digit_count)
 						nominator = nominator * 10 + (normalized[next] - '0');
 				}
-				}
-				else
-					this->is_valid = false;
 
 				this->interval.start = start;
 				return true;
@@ -5074,7 +5067,7 @@ namespace stdex
 						this->is_valid = nominator == 1;
 						break;
 					}
-					size_t digit_count = nominator < 10 ? 1 : 2;
+					size_t digit_count = nominator == 0 ? 0 : nominator < 10 ? 1 : 2;
 					for (; digit_count < 9 && normalized[next]; ++next, ++digit_count)
 						nominator = nominator * 10 + (normalized[next] - '0');
 				}