void Node::anchor() {}
void NullNode::anchor() {}
void ScalarNode::anchor() {}
+void BlockScalarNode::anchor() {}
void KeyValueNode::anchor() {}
void MappingNode::anchor() {}
void SequenceNode::anchor() {}
TK_Key,
TK_Value,
TK_Scalar,
+ TK_BlockScalar,
TK_Alias,
TK_Anchor,
TK_Tag
/// of the token in the input.
StringRef Range;
+ /// The value of a block scalar node.
+ std::string Value;
+
Token() : Kind(TK_Error) {}
};
}
/// b-break.
StringRef::iterator skip_b_break(StringRef::iterator Position);
+ /// Skip a single s-space[31] starting at Position.
+ ///
+ /// An s-space is 0x20
+ ///
+ /// @returns The code unit after the s-space, or Position if it's not a
+ /// s-space.
+ StringRef::iterator skip_s_space(StringRef::iterator Position);
+
/// @brief Skip a single s-white[33] starting at Position.
///
/// A s-white is 0x20 | 0x9
StringRef::iterator skip_while( SkipWhileFunc Func
, StringRef::iterator Position);
+ /// Skip minimal well-formed code unit subsequences until Func returns its
+ /// input.
+ void advanceWhile(SkipWhileFunc Func);
+
/// @brief Scan ns-uri-char[39]s starting at Cur.
///
/// This updates Cur and Column while scanning.
/// Pos is whitespace or a new line
bool isBlankOrBreak(StringRef::iterator Position);
+ /// Consume a single b-break[28] if it's present at the current position.
+ ///
+ /// Return false if the code unit at the current position isn't a line break.
+ bool consumeLineBreakIfPresent();
+
/// @brief If IsSimpleKeyAllowed, create and push_back a new SimpleKey.
void saveSimpleKeyCandidate( TokenQueueT::iterator Tok
, unsigned AtColumn
/// @brief Scan a block scalar starting with | or >.
bool scanBlockScalar(bool IsLiteral);
+ /// Scan a chomping indicator in a block scalar header.
+ char scanBlockChompingIndicator();
+
+ /// Scan an indentation indicator in a block scalar header.
+ unsigned scanBlockIndentationIndicator();
+
+ /// Scan a block scalar header.
+ ///
+ /// Return false if an error occurred.
+ bool scanBlockScalarHeader(char &ChompingIndicator, unsigned &IndentIndicator,
+ bool &IsDone);
+
+ /// Look for the indentation level of a block scalar.
+ ///
+ /// Return false if an error occurred.
+ bool findBlockScalarIndent(unsigned &BlockIndent, unsigned BlockExitIndent,
+ unsigned &LineBreaks, bool &IsDone);
+
+ /// Scan the indentation of a text line in a block scalar.
+ ///
+ /// Return false if an error occurred.
+ bool scanBlockScalarIndent(unsigned BlockIndent, unsigned BlockExitIndent,
+ bool &IsDone);
+
/// @brief Scan a tag of the form !stuff.
bool scanTag();
case Token::TK_Scalar:
OS << "Scalar: ";
break;
+ case Token::TK_BlockScalar:
+ OS << "Block Scalar: ";
+ break;
case Token::TK_Alias:
OS << "Alias: ";
break;
return Position;
}
+StringRef::iterator Scanner::skip_s_space(StringRef::iterator Position) {
+ if (Position == End)
+ return Position;
+ if (*Position == ' ')
+ return Position + 1;
+ return Position;
+}
StringRef::iterator Scanner::skip_s_white(StringRef::iterator Position) {
if (Position == End)
return Position;
}
+void Scanner::advanceWhile(SkipWhileFunc Func) {
+ auto Final = skip_while(Func, Current);
+ Column += Final - Current;
+ Current = Final;
+}
+
static bool is_ns_hex_digit(const char C) {
return (C >= '0' && C <= '9')
|| (C >= 'a' && C <= 'z')
return false;
}
+bool Scanner::consumeLineBreakIfPresent() {
+ auto Next = skip_b_break(Current);
+ if (Next == Current)
+ return false;
+ Column = 0;
+ ++Line;
+ Current = Next;
+ return true;
+}
+
void Scanner::saveSimpleKeyCandidate( TokenQueueT::iterator Tok
, unsigned AtColumn
, bool IsRequired) {
return true;
}
-bool Scanner::scanBlockScalar(bool IsLiteral) {
- StringRef::iterator Start = Current;
- skip(1); // Eat | or >
- while(true) {
- StringRef::iterator i = skip_nb_char(Current);
- if (i == Current) {
- if (Column == 0)
- break;
- i = skip_b_break(Current);
- if (i != Current) {
- // We got a line break.
- Column = 0;
- ++Line;
- Current = i;
- continue;
- } else {
- // There was an error, which should already have been printed out.
+char Scanner::scanBlockChompingIndicator() {
+ char Indicator = ' ';
+ if (Current != End && (*Current == '+' || *Current == '-')) {
+ Indicator = *Current;
+ skip(1);
+ }
+ return Indicator;
+}
+
+/// Get the number of line breaks after chomping.
+///
+/// Return the number of trailing line breaks to emit, depending on
+/// \p ChompingIndicator.
+static unsigned getChompedLineBreaks(char ChompingIndicator,
+ unsigned LineBreaks, StringRef Str) {
+ if (ChompingIndicator == '-') // Strip all line breaks.
+ return 0;
+ if (ChompingIndicator == '+') // Keep all line breaks.
+ return LineBreaks;
+ // Clip trailing lines.
+ return Str.empty() ? 0 : 1;
+}
+
+unsigned Scanner::scanBlockIndentationIndicator() {
+ unsigned Indent = 0;
+ if (Current != End && (*Current >= '1' && *Current <= '9')) {
+ Indent = unsigned(*Current - '0');
+ skip(1);
+ }
+ return Indent;
+}
+
+bool Scanner::scanBlockScalarHeader(char &ChompingIndicator,
+ unsigned &IndentIndicator, bool &IsDone) {
+ auto Start = Current;
+
+ ChompingIndicator = scanBlockChompingIndicator();
+ IndentIndicator = scanBlockIndentationIndicator();
+ // Check for the chomping indicator once again.
+ if (ChompingIndicator == ' ')
+ ChompingIndicator = scanBlockChompingIndicator();
+ Current = skip_while(&Scanner::skip_s_white, Current);
+ skipComment();
+
+ if (Current == End) { // EOF, we have an empty scalar.
+ Token T;
+ T.Kind = Token::TK_BlockScalar;
+ T.Range = StringRef(Start, Current - Start);
+ TokenQueue.push_back(T);
+ IsDone = true;
+ return true;
+ }
+
+ if (!consumeLineBreakIfPresent()) {
+ setError("Expected a line break after block scalar header", Current);
+ return false;
+ }
+ return true;
+}
+
+bool Scanner::findBlockScalarIndent(unsigned &BlockIndent,
+ unsigned BlockExitIndent,
+ unsigned &LineBreaks, bool &IsDone) {
+ unsigned MaxAllSpaceLineCharacters = 0;
+ StringRef::iterator LongestAllSpaceLine;
+
+ while (true) {
+ advanceWhile(&Scanner::skip_s_space);
+ if (skip_nb_char(Current) != Current) {
+ // This line isn't empty, so try and find the indentation.
+ if (Column <= BlockExitIndent) { // End of the block literal.
+ IsDone = true;
+ return true;
+ }
+ // We found the block's indentation.
+ BlockIndent = Column;
+ if (MaxAllSpaceLineCharacters > BlockIndent) {
+ setError(
+ "Leading all-spaces line must be smaller than the block indent",
+ LongestAllSpaceLine);
return false;
}
+ return true;
}
- Current = i;
+ if (skip_b_break(Current) != Current &&
+ Column > MaxAllSpaceLineCharacters) {
+ // Record the longest all-space line in case it's longer than the
+ // discovered block indent.
+ MaxAllSpaceLineCharacters = Column;
+ LongestAllSpaceLine = Current;
+ }
+
+ // Check for EOF.
+ if (Current == End) {
+ IsDone = true;
+ return true;
+ }
+
+ if (!consumeLineBreakIfPresent()) {
+ IsDone = true;
+ return true;
+ }
+ ++LineBreaks;
+ }
+ return true;
+}
+
+bool Scanner::scanBlockScalarIndent(unsigned BlockIndent,
+ unsigned BlockExitIndent, bool &IsDone) {
+ // Skip the indentation.
+ while (Column < BlockIndent) {
+ auto I = skip_s_space(Current);
+ if (I == Current)
+ break;
+ Current = I;
++Column;
}
- if (Start == Current) {
- setError("Got empty block scalar", Start);
+ if (skip_nb_char(Current) == Current)
+ return true;
+
+ if (Column <= BlockExitIndent) { // End of the block literal.
+ IsDone = true;
+ return true;
+ }
+
+ if (Column < BlockIndent) {
+ if (Current != End && *Current == '#') { // Trailing comment.
+ IsDone = true;
+ return true;
+ }
+ setError("A text line is less indented than the block scalar", Current);
return false;
}
+ return true; // A normal text line.
+}
+
+bool Scanner::scanBlockScalar(bool IsLiteral) {
+ // Eat '|' or '>'
+ assert(*Current == '|' || *Current == '>');
+ skip(1);
+
+ char ChompingIndicator;
+ unsigned BlockIndent;
+ bool IsDone = false;
+ if (!scanBlockScalarHeader(ChompingIndicator, BlockIndent, IsDone))
+ return false;
+ if (IsDone)
+ return true;
+
+ auto Start = Current;
+ unsigned BlockExitIndent = Indent < 0 ? 0 : (unsigned)Indent;
+ unsigned LineBreaks = 0;
+ if (BlockIndent == 0) {
+ if (!findBlockScalarIndent(BlockIndent, BlockExitIndent, LineBreaks,
+ IsDone))
+ return false;
+ }
+
+ // Scan the block's scalars body.
+ SmallString<256> Str;
+ while (!IsDone) {
+ if (!scanBlockScalarIndent(BlockIndent, BlockExitIndent, IsDone))
+ return false;
+ if (IsDone)
+ break;
+
+ // Parse the current line.
+ auto LineStart = Current;
+ advanceWhile(&Scanner::skip_nb_char);
+ if (LineStart != Current) {
+ Str.append(LineBreaks, '\n');
+ Str.append(StringRef(LineStart, Current - LineStart));
+ LineBreaks = 0;
+ }
+
+ // Check for EOF.
+ if (Current == End)
+ break;
+
+ if (!consumeLineBreakIfPresent())
+ break;
+ ++LineBreaks;
+ }
+
+ if (Current == End && !LineBreaks)
+ // Ensure that there is at least one line break before the end of file.
+ LineBreaks = 1;
+ Str.append(getChompedLineBreaks(ChompingIndicator, LineBreaks, Str), '\n');
+
+ // New lines may start a simple key.
+ if (!FlowLevel)
+ IsSimpleKeyAllowed = true;
Token T;
- T.Kind = Token::TK_Scalar;
+ T.Kind = Token::TK_BlockScalar;
T.Range = StringRef(Start, Current - Start);
+ T.Value = Str.str().str();
TokenQueue.push_back(T);
return true;
}
case NK_Null:
return "tag:yaml.org,2002:null";
case NK_Scalar:
+ case NK_BlockScalar:
// TODO: Tag resolution.
return "tag:yaml.org,2002:str";
case NK_Mapping:
, AnchorInfo.Range.substr(1)
, TagInfo.Range
, T.Range);
+ case Token::TK_BlockScalar:
+ getNext();
+ return new (NodeAllocator)
+ BlockScalarNode(stream.CurrentDoc, AnchorInfo.Range.substr(1),
+ TagInfo.Range, T.Value, T.Range);
case Token::TK_Key:
// Don't eat the TK_Key, KeyValueNode expects it.
return new (NodeAllocator)
ExpectParseSuccess("Array of arrays", "[[]]");
}
+TEST(YAMLParser, ParsesBlockLiteralScalars) {
+ ExpectParseSuccess("Block literal scalar", "test: |\n Hello\n World\n");
+ ExpectParseSuccess("Block literal scalar EOF", "test: |\n Hello\n World");
+ ExpectParseSuccess("Empty block literal scalar header EOF", "test: | ");
+ ExpectParseSuccess("Empty block literal scalar", "test: |\ntest2: 20");
+ ExpectParseSuccess("Empty block literal scalar 2", "- | \n \n\n \n- 42");
+ ExpectParseSuccess("Block literal scalar in sequence",
+ "- |\n Testing\n Out\n\n- 22");
+ ExpectParseSuccess("Block literal scalar in document",
+ "--- |\n Document\n...");
+ ExpectParseSuccess("Empty non indented lines still count",
+ "- |\n First line\n \n\n Another line\n\n- 2");
+ ExpectParseSuccess("Comment in block literal scalar header",
+ "test: | # Comment \n No Comment\ntest 2: | # Void");
+ ExpectParseSuccess("Chomping indicators in block literal scalar header",
+ "test: |- \n Hello\n\ntest 2: |+ \n\n World\n\n\n");
+ ExpectParseSuccess("Indent indicators in block literal scalar header",
+ "test: |1 \n \n Hello \n World\n");
+ ExpectParseSuccess("Chomping and indent indicators in block literals",
+ "test: |-1\n Hello\ntest 2: |9+\n World");
+ ExpectParseSuccess("Trailing comments in block literals",
+ "test: |\n Content\n # Trailing\n #Comment\ntest 2: 3");
+ ExpectParseError("Invalid block scalar header", "test: | failure");
+ ExpectParseError("Invalid line indentation", "test: |\n First line\n Error");
+ ExpectParseError("Long leading space line", "test: |\n \n Test\n");
+}
+
TEST(YAMLParser, HandlesEndOfFileGracefully) {
ExpectParseError("In string starting with EOF", "[\"");
ExpectParseError("In string hitting EOF", "[\" ");