diagnostic: don't squiggle for unicode lines
If the line contains non-ascii characters, then we cannot assume that a single utf8 code unit represents a single glyph, so don't attempt to draw squiggles.
Bug: tint:1437
Change-Id: Ibd911200b03297f3e1b059ad7cc53fcba59e7714
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/80846
Reviewed-by: David Neto <dneto@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
diff --git a/src/diagnostic/formatter.cc b/src/diagnostic/formatter.cc
index 0e9f217..1998523 100644
--- a/src/diagnostic/formatter.cc
+++ b/src/diagnostic/formatter.cc
@@ -201,19 +201,31 @@
auto& line = src.file->content.lines[line_num - 1];
auto line_len = line.size();
+ bool is_ascii = true;
for (auto c : line) {
if (c == '\t') {
state.repeat(' ', style_.tab_width);
} else {
state << c;
}
+ if (c & 0x80) {
+ is_ascii = false;
+ }
}
state.newline();
+
+ // If the line contains non-ascii characters, then we cannot assume that
+ // a single utf8 code unit represents a single glyph, so don't attempt to
+ // draw squiggles.
+ if (!is_ascii) {
+ continue;
+ }
+
state.set_style({Color::kCyan, false});
// Count the number of glyphs in the line span.
- // start and end use 1-based indexing .
+ // start and end use 1-based indexing.
auto num_glyphs = [&](size_t start, size_t end) {
size_t count = 0;
start = (start > 0) ? (start - 1) : 0;
diff --git a/src/diagnostic/formatter_test.cc b/src/diagnostic/formatter_test.cc
index 966b888..a97f57e 100644
--- a/src/diagnostic/formatter_test.cc
+++ b/src/diagnostic/formatter_test.cc
@@ -37,43 +37,81 @@
return d;
}
-constexpr const char* content = // Note: words are tab-delimited
+constexpr const char* ascii_content = // Note: words are tab-delimited
R"(the cat says meow
the dog says woof
the snake says quack
the snail says ???
)";
+constexpr const char* utf8_content = // Note: words are tab-delimited
+ "the \xf0\x9f\x90\xb1 says meow\n" // NOLINT: tabs
+ "the \xf0\x9f\x90\x95 says woof\n" // NOLINT: tabs
+ "the \xf0\x9f\x90\x8d says quack\n" // NOLINT: tabs
+ "the \xf0\x9f\x90\x8c says ???\n"; // NOLINT: tabs
+
class DiagFormatterTest : public testing::Test {
public:
- Source::File file{"file.name", content};
- Diagnostic diag_note =
+ Source::File ascii_file{"file.name", ascii_content};
+ Source::File utf8_file{"file.name", utf8_content};
+ Diagnostic ascii_diag_note =
Diag(Severity::Note,
- Source{Source::Range{Source::Location{1, 14}}, &file},
+ Source{Source::Range{Source::Location{1, 14}}, &ascii_file},
"purr",
System::Test);
- Diagnostic diag_warn = Diag(Severity::Warning,
- Source{Source::Range{{2, 14}, {2, 18}}, &file},
- "grrr",
- System::Test);
- Diagnostic diag_err = Diag(Severity::Error,
- Source{Source::Range{{3, 16}, {3, 21}}, &file},
- "hiss",
- System::Test,
- "abc123");
- Diagnostic diag_ice = Diag(Severity::InternalCompilerError,
- Source{Source::Range{{4, 16}, {4, 19}}, &file},
- "unreachable",
- System::Test);
- Diagnostic diag_fatal = Diag(Severity::Fatal,
- Source{Source::Range{{4, 16}, {4, 19}}, &file},
- "nothing",
- System::Test);
+ Diagnostic ascii_diag_warn =
+ Diag(Severity::Warning,
+ Source{Source::Range{{2, 14}, {2, 18}}, &ascii_file},
+ "grrr",
+ System::Test);
+ Diagnostic ascii_diag_err =
+ Diag(Severity::Error,
+ Source{Source::Range{{3, 16}, {3, 21}}, &ascii_file},
+ "hiss",
+ System::Test,
+ "abc123");
+ Diagnostic ascii_diag_ice =
+ Diag(Severity::InternalCompilerError,
+ Source{Source::Range{{4, 16}, {4, 19}}, &ascii_file},
+ "unreachable",
+ System::Test);
+ Diagnostic ascii_diag_fatal =
+ Diag(Severity::Fatal,
+ Source{Source::Range{{4, 16}, {4, 19}}, &ascii_file},
+ "nothing",
+ System::Test);
+
+ Diagnostic utf8_diag_note =
+ Diag(Severity::Note,
+ Source{Source::Range{Source::Location{1, 15}}, &utf8_file},
+ "purr",
+ System::Test);
+ Diagnostic utf8_diag_warn =
+ Diag(Severity::Warning,
+ Source{Source::Range{{2, 15}, {2, 19}}, &utf8_file},
+ "grrr",
+ System::Test);
+ Diagnostic utf8_diag_err =
+ Diag(Severity::Error,
+ Source{Source::Range{{3, 15}, {3, 20}}, &utf8_file},
+ "hiss",
+ System::Test,
+ "abc123");
+ Diagnostic utf8_diag_ice =
+ Diag(Severity::InternalCompilerError,
+ Source{Source::Range{{4, 15}, {4, 18}}, &utf8_file},
+ "unreachable",
+ System::Test);
+ Diagnostic utf8_diag_fatal =
+ Diag(Severity::Fatal,
+ Source{Source::Range{{4, 15}, {4, 18}}, &utf8_file},
+ "nothing",
+ System::Test);
};
TEST_F(DiagFormatterTest, Simple) {
Formatter fmt{{false, false, false, false}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(1:14: purr
2:14: grrr
3:16 abc123: hiss)";
@@ -82,7 +120,7 @@
TEST_F(DiagFormatterTest, SimpleNewlineAtEnd) {
Formatter fmt{{false, false, false, true}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(1:14: purr
2:14: grrr
3:16 abc123: hiss
@@ -100,7 +138,7 @@
TEST_F(DiagFormatterTest, WithFile) {
Formatter fmt{{true, false, false, false}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(file.name:1:14: purr
file.name:2:14: grrr
file.name:3:16 abc123: hiss)";
@@ -109,7 +147,7 @@
TEST_F(DiagFormatterTest, WithSeverity) {
Formatter fmt{{false, true, false, false}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(1:14 note: purr
2:14 warning: grrr
3:16 error abc123: hiss)";
@@ -118,7 +156,7 @@
TEST_F(DiagFormatterTest, WithLine) {
Formatter fmt{{false, false, true, false}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(1:14: purr
the cat says meow
^
@@ -134,9 +172,24 @@
ASSERT_EQ(expect, got);
}
+TEST_F(DiagFormatterTest, UnicodeWithLine) {
+ Formatter fmt{{false, false, true, false}};
+ auto got = fmt.format(List{utf8_diag_note, utf8_diag_warn, utf8_diag_err});
+ auto* expect =
+ "1:15: purr\n"
+ "the \xf0\x9f\x90\xb1 says meow\n"
+ "\n"
+ "2:15: grrr\n"
+ "the \xf0\x9f\x90\x95 says woof\n"
+ "\n"
+ "3:15 abc123: hiss\n"
+ "the \xf0\x9f\x90\x8d says quack\n";
+ ASSERT_EQ(expect, got);
+}
+
TEST_F(DiagFormatterTest, BasicWithFileSeverityLine) {
Formatter fmt{{true, true, true, false}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(file.name:1:14 note: purr
the cat says meow
^
@@ -153,9 +206,9 @@
}
TEST_F(DiagFormatterTest, BasicWithMultiLine) {
- auto multiline =
- Diag(Severity::Warning, Source{Source::Range{{2, 9}, {4, 15}}, &file},
- "multiline", System::Test);
+ auto multiline = Diag(Severity::Warning,
+ Source{Source::Range{{2, 9}, {4, 15}}, &ascii_file},
+ "multiline", System::Test);
Formatter fmt{{false, false, true, false}};
auto got = fmt.format(List{multiline});
auto* expect = R"(2:9: multiline
@@ -169,9 +222,23 @@
ASSERT_EQ(expect, got);
}
+TEST_F(DiagFormatterTest, UnicodeWithMultiLine) {
+ auto multiline = Diag(Severity::Warning,
+ Source{Source::Range{{2, 9}, {4, 15}}, &utf8_file},
+ "multiline", System::Test);
+ Formatter fmt{{false, false, true, false}};
+ auto got = fmt.format(List{multiline});
+ auto* expect =
+ "2:9: multiline\n"
+ "the \xf0\x9f\x90\x95 says woof\n"
+ "the \xf0\x9f\x90\x8d says quack\n"
+ "the \xf0\x9f\x90\x8c says ???\n";
+ ASSERT_EQ(expect, got);
+}
+
TEST_F(DiagFormatterTest, BasicWithFileSeverityLineTab4) {
Formatter fmt{{true, true, true, false, 4u}};
- auto got = fmt.format(List{diag_note, diag_warn, diag_err});
+ auto got = fmt.format(List{ascii_diag_note, ascii_diag_warn, ascii_diag_err});
auto* expect = R"(file.name:1:14 note: purr
the cat says meow
^
@@ -188,9 +255,9 @@
}
TEST_F(DiagFormatterTest, BasicWithMultiLineTab4) {
- auto multiline =
- Diag(Severity::Warning, Source{Source::Range{{2, 9}, {4, 15}}, &file},
- "multiline", System::Test);
+ auto multiline = Diag(Severity::Warning,
+ Source{Source::Range{{2, 9}, {4, 15}}, &ascii_file},
+ "multiline", System::Test);
Formatter fmt{{false, false, true, false, 4u}};
auto got = fmt.format(List{multiline});
auto* expect = R"(2:9: multiline
@@ -206,7 +273,7 @@
TEST_F(DiagFormatterTest, ICE) {
Formatter fmt{{}};
- auto got = fmt.format(List{diag_ice});
+ auto got = fmt.format(List{ascii_diag_ice});
auto* expect = R"(file.name:4:16 internal compiler error: unreachable
the snail says ???
^^^
@@ -217,7 +284,7 @@
TEST_F(DiagFormatterTest, Fatal) {
Formatter fmt{{}};
- auto got = fmt.format(List{diag_fatal});
+ auto got = fmt.format(List{ascii_diag_fatal});
auto* expect = R"(file.name:4:16 fatal: nothing
the snail says ???
^^^
@@ -229,7 +296,8 @@
TEST_F(DiagFormatterTest, RangeOOB) {
Formatter fmt{{true, true, true, true}};
diag::List list;
- list.add_error(System::Test, "oob", Source{{{10, 20}, {30, 20}}, &file});
+ list.add_error(System::Test, "oob",
+ Source{{{10, 20}, {30, 20}}, &ascii_file});
auto got = fmt.format(list);
auto* expect = R"(file.name:10:20 error: oob
