authorgravatar for dbandstra@protonmail.comdbandstra <dbandstra@protonmail.com> 2020-08-19 13:10:05-07:00
committergravatar for andrew@ziglang.orgAndrew Kelley <andrew@ziglang.org> 2020-08-19 16:30:42-04:00
loga237283d90ea6799331a7268f474511f352a23d7
tree770d5650e4359e331f4a9c39137e79d821340297
parent083c0f1cebc763e4e43529b50f6df9839c32c1c7

fixes and improvements for parseCharLiteral

1 files changed, 25 insertions(+), 17 deletions(-)

lib/std/zig.zig+25-17
......@@ -84,7 +84,7 @@ pub fn binNameAlloc(
8484/// Slice must be valid utf8 starting and ending with "'" and exactly one codepoint in between.
8585pub fn parseCharLiteral(
8686 slice: []const u8,
87 bad_index: *usize, // populated if error.InvalidCharacter is returned)
87 bad_index: *usize, // populated if error.InvalidCharacter is returned
8888) error{InvalidCharacter}!u32 {
8989 std.debug.assert(slice.len >= 3 and slice[0] == '\'' and slice[slice.len - 1] == '\'');
9090
......@@ -101,24 +101,23 @@ pub fn parseCharLiteral(
101101 bad_index.* = slice.len - 2;
102102 return error.InvalidCharacter;
103103 }
104
105104 var value: u32 = 0;
106105 for (slice[3..5]) |c, i| {
107 switch (slice[3]) {
106 switch (c) {
108107 '0'...'9' => {
109108 value *= 16;
110109 value += c - '0';
111110 },
112111 'a'...'f' => {
113112 value *= 16;
114 value += c - 'a';
113 value += c - 'a' + 10;
115114 },
116115 'A'...'F' => {
117116 value *= 16;
118 value += c - 'a';
117 value += c - 'A' + 10;
119118 },
120119 else => {
121 bad_index.* = i;
120 bad_index.* = 3 + i;
122121 return error.InvalidCharacter;
123122 },
124123 }
......@@ -126,16 +125,12 @@ pub fn parseCharLiteral(
126125 return value;
127126 },
128127 'u' => {
129 if (slice.len < 6 or slice[3] != '{') {
128 if (slice.len < "'\\u{0}'".len or slice[3] != '{' or slice[slice.len - 2] != '}') {
130129 bad_index.* = 2;
131130 return error.InvalidCharacter;
132131 }
133132 var value: u32 = 0;
134 for (slice[4..]) |c, i| {
135 if (value > 0x10ffff) {
136 bad_index.* = i;
137 return error.InvalidCharacter;
138 }
133 for (slice[4 .. slice.len - 2]) |c, i| {
139134 switch (c) {
140135 '0'...'9' => {
141136 value *= 16;
......@@ -143,25 +138,28 @@ pub fn parseCharLiteral(
143138 },
144139 'a'...'f' => {
145140 value *= 16;
146 value += c - 'a';
141 value += c - 'a' + 10;
147142 },
148143 'A'...'F' => {
149144 value *= 16;
150 value += c - 'A';
145 value += c - 'A' + 10;
151146 },
152 '}' => break,
153147 else => {
154 bad_index.* = i;
148 bad_index.* = 4 + i;
155149 return error.InvalidCharacter;
156150 },
157151 }
152 if (value > 0x10ffff) {
153 bad_index.* = 4 + i;
154 return error.InvalidCharacter;
155 }
158156 }
159157 return value;
160158 },
161159 else => {
162160 bad_index.* = 2;
163161 return error.InvalidCharacter;
164 }
162 },
165163 }
166164 }
167165 return std.unicode.utf8Decode(slice[1 .. slice.len - 1]) catch unreachable;
......@@ -172,13 +170,23 @@ test "parseCharLiteral" {
172170 std.testing.expectEqual(try parseCharLiteral("'a'", &bad_index), 'a');
173171 std.testing.expectEqual(try parseCharLiteral("'ä'", &bad_index), 'ä');
174172 std.testing.expectEqual(try parseCharLiteral("'\\x00'", &bad_index), 0);
173 std.testing.expectEqual(try parseCharLiteral("'\\x4f'", &bad_index), 0x4f);
174 std.testing.expectEqual(try parseCharLiteral("'\\x4F'", &bad_index), 0x4f);
175175 std.testing.expectEqual(try parseCharLiteral("'ぁ'", &bad_index), 0x3041);
176 std.testing.expectEqual(try parseCharLiteral("'\\u{0}'", &bad_index), 0);
176177 std.testing.expectEqual(try parseCharLiteral("'\\u{3041}'", &bad_index), 0x3041);
178 std.testing.expectEqual(try parseCharLiteral("'\\u{7f}'", &bad_index), 0x7f);
179 std.testing.expectEqual(try parseCharLiteral("'\\u{7FFF}'", &bad_index), 0x7FFF);
177180
178181 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\x0'", &bad_index));
182 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\x000'", &bad_index));
179183 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\y'", &bad_index));
180184 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\u'", &bad_index));
185 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\uFFFF'", &bad_index));
186 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\u{}'", &bad_index));
181187 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\u{FFFFFF}'", &bad_index));
188 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\u{FFFF'", &bad_index));
189 std.testing.expectError(error.InvalidCharacter, parseCharLiteral("'\\u{FFFF}x'", &bad_index));
182190}
183191
184192test "" {