Added tests for unicode character classes; more tests for hex characters

regex/re_test.go

@@ -430,6 +430,7 @@ var reTests = []struct {
 	{`^(.+)?B`, []ReFlag{RE_CASE_INSENSITIVE}, `ab`, []Group{{0, 2}}},
 	{`\0009`, []ReFlag{RE_CASE_INSENSITIVE}, "\x009", []Group{{0, 2}}},
 	{`\0141`, []ReFlag{RE_CASE_INSENSITIVE}, "A", []Group{{0, 1}}},
+	{`\0141\0141`, []ReFlag{RE_CASE_INSENSITIVE}, "AA", []Group{{0, 2}}},
 
 	{`a[-]?c`, []ReFlag{RE_CASE_INSENSITIVE}, `AC`, []Group{{0, 2}}},
 
@@ -460,8 +461,10 @@ var reTests = []struct {
 	{`[\D5]+`, nil, `1234abc5678`, []Group{{4, 8}}},
 	{`[\da-fA-F]+`, nil, `123abc`, []Group{{0, 6}}},
 	{`\xff`, nil, "\u00ff", []Group{{0, 1}}},
+	{`\xff+`, nil, "\u00ff\u00ff", []Group{{0, 2}}},
 	{`\xFF`, nil, "\u00ff", []Group{{0, 1}}},
 	{`\x00ff`, nil, "\u00ff", []Group{}},
+	{`\x{0000ff}+`, nil, "\u00ff\u00ff", []Group{{0, 2}}},
 	{`\x{0000ff}`, nil, "\u00ff", []Group{{0, 1}}},
 	{`\x{0000FF}`, nil, "\u00ff", []Group{{0, 1}}},
 	{"\t\n\v\r\f\a", nil, "\t\n\v\r\f\a", []Group{{0, 6}}},
@@ -516,6 +519,14 @@ var reTests = []struct {
 	{`<389-400`, nil, `-`, nil},
 	{`<389-400>`, nil, `391`, []Group{{0, 3}}},
 	{`\b<1-10000>\b`, nil, `America declared independence in 1776.`, []Group{{33, 37}}},
+
+	{`\p{Tamil}+`, nil, `உயிரெழுத்து`, []Group{{0, 11}}}, // Each letter and matra is counted as a separate rune, so 'u', 'ya', 'e (matra), 'ra', 'e (matra)', 'zha', (oo (matra), 'tha', 'ith', 'tha', 'oo (matra)'.
+	{`\P{Tamil}+`, nil, `vowel=உயிரெழுத்து`, []Group{{0, 6}}},
+	{`\P`, nil, `உயிரெழுத்து`, nil},
+	{`\PM\pM*`, nil, `உயிரெழுத்து`, []Group{{0, 1}, {1, 3}, {3, 5}, {5, 7}, {7, 9}, {9, 11}}},
+	{`\pN+`, nil, `123abc456def`, []Group{{0, 3}, {6, 9}}},
+	{`\PN+`, nil, `123abc456def`, []Group{{3, 6}, {9, 12}}},
+	{`[\p{Greek}\p{Cyrillic}]`, nil, `ΣωШД`, []Group{{0, 1}, {1, 2}, {2, 3}, {3, 4}}},
 }
 
 var groupTests = []struct {