xref: /freebsd/contrib/llvm-project/llvm/lib/Support/YAMLParser.cpp (revision 5f757f3ff9144b609b3c433dfd370cc6bdc191ad)
1  //===- YAMLParser.cpp - Simple YAML parser --------------------------------===//
2  //
3  // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4  // See https://llvm.org/LICENSE.txt for license information.
5  // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6  //
7  //===----------------------------------------------------------------------===//
8  //
9  //  This file implements a YAML parser.
10  //
11  //===----------------------------------------------------------------------===//
12  
13  #include "llvm/Support/YAMLParser.h"
14  #include "llvm/ADT/AllocatorList.h"
15  #include "llvm/ADT/ArrayRef.h"
16  #include "llvm/ADT/STLExtras.h"
17  #include "llvm/ADT/SmallString.h"
18  #include "llvm/ADT/SmallVector.h"
19  #include "llvm/ADT/StringExtras.h"
20  #include "llvm/ADT/StringRef.h"
21  #include "llvm/ADT/Twine.h"
22  #include "llvm/Support/Compiler.h"
23  #include "llvm/Support/ErrorHandling.h"
24  #include "llvm/Support/MemoryBuffer.h"
25  #include "llvm/Support/SMLoc.h"
26  #include "llvm/Support/SourceMgr.h"
27  #include "llvm/Support/Unicode.h"
28  #include "llvm/Support/raw_ostream.h"
29  #include <cassert>
30  #include <cstddef>
31  #include <cstdint>
32  #include <map>
33  #include <memory>
34  #include <string>
35  #include <system_error>
36  #include <utility>
37  
38  using namespace llvm;
39  using namespace yaml;
40  
41  enum UnicodeEncodingForm {
42    UEF_UTF32_LE, ///< UTF-32 Little Endian
43    UEF_UTF32_BE, ///< UTF-32 Big Endian
44    UEF_UTF16_LE, ///< UTF-16 Little Endian
45    UEF_UTF16_BE, ///< UTF-16 Big Endian
46    UEF_UTF8,     ///< UTF-8 or ascii.
47    UEF_Unknown   ///< Not a valid Unicode encoding.
48  };
49  
50  /// EncodingInfo - Holds the encoding type and length of the byte order mark if
51  ///                it exists. Length is in {0, 2, 3, 4}.
52  using EncodingInfo = std::pair<UnicodeEncodingForm, unsigned>;
53  
54  /// getUnicodeEncoding - Reads up to the first 4 bytes to determine the Unicode
55  ///                      encoding form of \a Input.
56  ///
57  /// @param Input A string of length 0 or more.
58  /// @returns An EncodingInfo indicating the Unicode encoding form of the input
59  ///          and how long the byte order mark is if one exists.
60  static EncodingInfo getUnicodeEncoding(StringRef Input) {
61    if (Input.empty())
62      return std::make_pair(UEF_Unknown, 0);
63  
64    switch (uint8_t(Input[0])) {
65    case 0x00:
66      if (Input.size() >= 4) {
67        if (  Input[1] == 0
68           && uint8_t(Input[2]) == 0xFE
69           && uint8_t(Input[3]) == 0xFF)
70          return std::make_pair(UEF_UTF32_BE, 4);
71        if (Input[1] == 0 && Input[2] == 0 && Input[3] != 0)
72          return std::make_pair(UEF_UTF32_BE, 0);
73      }
74  
75      if (Input.size() >= 2 && Input[1] != 0)
76        return std::make_pair(UEF_UTF16_BE, 0);
77      return std::make_pair(UEF_Unknown, 0);
78    case 0xFF:
79      if (  Input.size() >= 4
80         && uint8_t(Input[1]) == 0xFE
81         && Input[2] == 0
82         && Input[3] == 0)
83        return std::make_pair(UEF_UTF32_LE, 4);
84  
85      if (Input.size() >= 2 && uint8_t(Input[1]) == 0xFE)
86        return std::make_pair(UEF_UTF16_LE, 2);
87      return std::make_pair(UEF_Unknown, 0);
88    case 0xFE:
89      if (Input.size() >= 2 && uint8_t(Input[1]) == 0xFF)
90        return std::make_pair(UEF_UTF16_BE, 2);
91      return std::make_pair(UEF_Unknown, 0);
92    case 0xEF:
93      if (  Input.size() >= 3
94         && uint8_t(Input[1]) == 0xBB
95         && uint8_t(Input[2]) == 0xBF)
96        return std::make_pair(UEF_UTF8, 3);
97      return std::make_pair(UEF_Unknown, 0);
98    }
99  
100    // It could still be utf-32 or utf-16.
101    if (Input.size() >= 4 && Input[1] == 0 && Input[2] == 0 && Input[3] == 0)
102      return std::make_pair(UEF_UTF32_LE, 0);
103  
104    if (Input.size() >= 2 && Input[1] == 0)
105      return std::make_pair(UEF_UTF16_LE, 0);
106  
107    return std::make_pair(UEF_UTF8, 0);
108  }
109  
110  /// Pin the vtables to this file.
111  void Node::anchor() {}
112  void NullNode::anchor() {}
113  void ScalarNode::anchor() {}
114  void BlockScalarNode::anchor() {}
115  void KeyValueNode::anchor() {}
116  void MappingNode::anchor() {}
117  void SequenceNode::anchor() {}
118  void AliasNode::anchor() {}
119  
120  namespace llvm {
121  namespace yaml {
122  
123  /// Token - A single YAML token.
124  struct Token {
125    enum TokenKind {
126      TK_Error, // Uninitialized token.
127      TK_StreamStart,
128      TK_StreamEnd,
129      TK_VersionDirective,
130      TK_TagDirective,
131      TK_DocumentStart,
132      TK_DocumentEnd,
133      TK_BlockEntry,
134      TK_BlockEnd,
135      TK_BlockSequenceStart,
136      TK_BlockMappingStart,
137      TK_FlowEntry,
138      TK_FlowSequenceStart,
139      TK_FlowSequenceEnd,
140      TK_FlowMappingStart,
141      TK_FlowMappingEnd,
142      TK_Key,
143      TK_Value,
144      TK_Scalar,
145      TK_BlockScalar,
146      TK_Alias,
147      TK_Anchor,
148      TK_Tag
149    } Kind = TK_Error;
150  
151    /// A string of length 0 or more whose begin() points to the logical location
152    /// of the token in the input.
153    StringRef Range;
154  
155    /// The value of a block scalar node.
156    std::string Value;
157  
158    Token() = default;
159  };
160  
161  } // end namespace yaml
162  } // end namespace llvm
163  
164  using TokenQueueT = BumpPtrList<Token>;
165  
166  namespace {
167  
168  /// This struct is used to track simple keys.
169  ///
170  /// Simple keys are handled by creating an entry in SimpleKeys for each Token
171  /// which could legally be the start of a simple key. When peekNext is called,
172  /// if the Token To be returned is referenced by a SimpleKey, we continue
173  /// tokenizing until that potential simple key has either been found to not be
174  /// a simple key (we moved on to the next line or went further than 1024 chars).
175  /// Or when we run into a Value, and then insert a Key token (and possibly
176  /// others) before the SimpleKey's Tok.
177  struct SimpleKey {
178    TokenQueueT::iterator Tok;
179    unsigned Column = 0;
180    unsigned Line = 0;
181    unsigned FlowLevel = 0;
182    bool IsRequired = false;
183  
184    bool operator ==(const SimpleKey &Other) {
185      return Tok == Other.Tok;
186    }
187  };
188  
189  } // end anonymous namespace
190  
191  /// The Unicode scalar value of a UTF-8 minimal well-formed code unit
192  ///        subsequence and the subsequence's length in code units (uint8_t).
193  ///        A length of 0 represents an error.
194  using UTF8Decoded = std::pair<uint32_t, unsigned>;
195  
196  static UTF8Decoded decodeUTF8(StringRef Range) {
197    StringRef::iterator Position= Range.begin();
198    StringRef::iterator End = Range.end();
199    // 1 byte: [0x00, 0x7f]
200    // Bit pattern: 0xxxxxxx
201    if (Position < End && (*Position & 0x80) == 0) {
202      return std::make_pair(*Position, 1);
203    }
204    // 2 bytes: [0x80, 0x7ff]
205    // Bit pattern: 110xxxxx 10xxxxxx
206    if (Position + 1 < End && ((*Position & 0xE0) == 0xC0) &&
207        ((*(Position + 1) & 0xC0) == 0x80)) {
208      uint32_t codepoint = ((*Position & 0x1F) << 6) |
209                            (*(Position + 1) & 0x3F);
210      if (codepoint >= 0x80)
211        return std::make_pair(codepoint, 2);
212    }
213    // 3 bytes: [0x8000, 0xffff]
214    // Bit pattern: 1110xxxx 10xxxxxx 10xxxxxx
215    if (Position + 2 < End && ((*Position & 0xF0) == 0xE0) &&
216        ((*(Position + 1) & 0xC0) == 0x80) &&
217        ((*(Position + 2) & 0xC0) == 0x80)) {
218      uint32_t codepoint = ((*Position & 0x0F) << 12) |
219                           ((*(Position + 1) & 0x3F) << 6) |
220                            (*(Position + 2) & 0x3F);
221      // Codepoints between 0xD800 and 0xDFFF are invalid, as
222      // they are high / low surrogate halves used by UTF-16.
223      if (codepoint >= 0x800 &&
224          (codepoint < 0xD800 || codepoint > 0xDFFF))
225        return std::make_pair(codepoint, 3);
226    }
227    // 4 bytes: [0x10000, 0x10FFFF]
228    // Bit pattern: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
229    if (Position + 3 < End && ((*Position & 0xF8) == 0xF0) &&
230        ((*(Position + 1) & 0xC0) == 0x80) &&
231        ((*(Position + 2) & 0xC0) == 0x80) &&
232        ((*(Position + 3) & 0xC0) == 0x80)) {
233      uint32_t codepoint = ((*Position & 0x07) << 18) |
234                           ((*(Position + 1) & 0x3F) << 12) |
235                           ((*(Position + 2) & 0x3F) << 6) |
236                            (*(Position + 3) & 0x3F);
237      if (codepoint >= 0x10000 && codepoint <= 0x10FFFF)
238        return std::make_pair(codepoint, 4);
239    }
240    return std::make_pair(0, 0);
241  }
242  
243  namespace llvm {
244  namespace yaml {
245  
246  /// Scans YAML tokens from a MemoryBuffer.
247  class Scanner {
248  public:
249    Scanner(StringRef Input, SourceMgr &SM, bool ShowColors = true,
250            std::error_code *EC = nullptr);
251    Scanner(MemoryBufferRef Buffer, SourceMgr &SM_, bool ShowColors = true,
252            std::error_code *EC = nullptr);
253  
254    /// Parse the next token and return it without popping it.
255    Token &peekNext();
256  
257    /// Parse the next token and pop it from the queue.
258    Token getNext();
259  
260    void printError(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Message,
261                    ArrayRef<SMRange> Ranges = std::nullopt) {
262      SM.PrintMessage(Loc, Kind, Message, Ranges, /* FixIts= */ std::nullopt,
263                      ShowColors);
264    }
265  
266    void setError(const Twine &Message, StringRef::iterator Position) {
267      if (Position >= End)
268        Position = End - 1;
269  
270      // propagate the error if possible
271      if (EC)
272        *EC = make_error_code(std::errc::invalid_argument);
273  
274      // Don't print out more errors after the first one we encounter. The rest
275      // are just the result of the first, and have no meaning.
276      if (!Failed)
277        printError(SMLoc::getFromPointer(Position), SourceMgr::DK_Error, Message);
278      Failed = true;
279    }
280  
281    /// Returns true if an error occurred while parsing.
282    bool failed() {
283      return Failed;
284    }
285  
286  private:
287    void init(MemoryBufferRef Buffer);
288  
289    StringRef currentInput() {
290      return StringRef(Current, End - Current);
291    }
292  
293    /// Decode a UTF-8 minimal well-formed code unit subsequence starting
294    ///        at \a Position.
295    ///
296    /// If the UTF-8 code units starting at Position do not form a well-formed
297    /// code unit subsequence, then the Unicode scalar value is 0, and the length
298    /// is 0.
299    UTF8Decoded decodeUTF8(StringRef::iterator Position) {
300      return ::decodeUTF8(StringRef(Position, End - Position));
301    }
302  
303    // The following functions are based on the gramar rules in the YAML spec. The
304    // style of the function names it meant to closely match how they are written
305    // in the spec. The number within the [] is the number of the grammar rule in
306    // the spec.
307    //
308    // See 4.2 [Production Naming Conventions] for the meaning of the prefixes.
309    //
310    // c-
311    //   A production starting and ending with a special character.
312    // b-
313    //   A production matching a single line break.
314    // nb-
315    //   A production starting and ending with a non-break character.
316    // s-
317    //   A production starting and ending with a white space character.
318    // ns-
319    //   A production starting and ending with a non-space character.
320    // l-
321    //   A production matching complete line(s).
322  
323    /// Skip a single nb-char[27] starting at Position.
324    ///
325    /// A nb-char is 0x9 | [0x20-0x7E] | 0x85 | [0xA0-0xD7FF] | [0xE000-0xFEFE]
326    ///                  | [0xFF00-0xFFFD] | [0x10000-0x10FFFF]
327    ///
328    /// @returns The code unit after the nb-char, or Position if it's not an
329    ///          nb-char.
330    StringRef::iterator skip_nb_char(StringRef::iterator Position);
331  
332    /// Skip a single b-break[28] starting at Position.
333    ///
334    /// A b-break is 0xD 0xA | 0xD | 0xA
335    ///
336    /// @returns The code unit after the b-break, or Position if it's not a
337    ///          b-break.
338    StringRef::iterator skip_b_break(StringRef::iterator Position);
339  
340    /// Skip a single s-space[31] starting at Position.
341    ///
342    /// An s-space is 0x20
343    ///
344    /// @returns The code unit after the s-space, or Position if it's not a
345    ///          s-space.
346    StringRef::iterator skip_s_space(StringRef::iterator Position);
347  
348    /// Skip a single s-white[33] starting at Position.
349    ///
350    /// A s-white is 0x20 | 0x9
351    ///
352    /// @returns The code unit after the s-white, or Position if it's not a
353    ///          s-white.
354    StringRef::iterator skip_s_white(StringRef::iterator Position);
355  
356    /// Skip a single ns-char[34] starting at Position.
357    ///
358    /// A ns-char is nb-char - s-white
359    ///
360    /// @returns The code unit after the ns-char, or Position if it's not a
361    ///          ns-char.
362    StringRef::iterator skip_ns_char(StringRef::iterator Position);
363  
364    using SkipWhileFunc = StringRef::iterator (Scanner::*)(StringRef::iterator);
365  
366    /// Skip minimal well-formed code unit subsequences until Func
367    ///        returns its input.
368    ///
369    /// @returns The code unit after the last minimal well-formed code unit
370    ///          subsequence that Func accepted.
371    StringRef::iterator skip_while( SkipWhileFunc Func
372                                  , StringRef::iterator Position);
373  
374    /// Skip minimal well-formed code unit subsequences until Func returns its
375    /// input.
376    void advanceWhile(SkipWhileFunc Func);
377  
378    /// Scan ns-uri-char[39]s starting at Cur.
379    ///
380    /// This updates Cur and Column while scanning.
381    void scan_ns_uri_char();
382  
383    /// Consume a minimal well-formed code unit subsequence starting at
384    ///        \a Cur. Return false if it is not the same Unicode scalar value as
385    ///        \a Expected. This updates \a Column.
386    bool consume(uint32_t Expected);
387  
388    /// Skip \a Distance UTF-8 code units. Updates \a Cur and \a Column.
389    void skip(uint32_t Distance);
390  
391    /// Return true if the minimal well-formed code unit subsequence at
392    ///        Pos is whitespace or a new line
393    bool isBlankOrBreak(StringRef::iterator Position);
394  
395    /// Return true if the minimal well-formed code unit subsequence at
396    ///        Pos is considered a "safe" character for plain scalars.
397    bool isPlainSafeNonBlank(StringRef::iterator Position);
398  
399    /// Return true if the line is a line break, false otherwise.
400    bool isLineEmpty(StringRef Line);
401  
402    /// Consume a single b-break[28] if it's present at the current position.
403    ///
404    /// Return false if the code unit at the current position isn't a line break.
405    bool consumeLineBreakIfPresent();
406  
407    /// If IsSimpleKeyAllowed, create and push_back a new SimpleKey.
408    void saveSimpleKeyCandidate( TokenQueueT::iterator Tok
409                               , unsigned AtColumn
410                               , bool IsRequired);
411  
412    /// Remove simple keys that can no longer be valid simple keys.
413    ///
414    /// Invalid simple keys are not on the current line or are further than 1024
415    /// columns back.
416    void removeStaleSimpleKeyCandidates();
417  
418    /// Remove all simple keys on FlowLevel \a Level.
419    void removeSimpleKeyCandidatesOnFlowLevel(unsigned Level);
420  
421    /// Unroll indentation in \a Indents back to \a Col. Creates BlockEnd
422    ///        tokens if needed.
423    bool unrollIndent(int ToColumn);
424  
425    /// Increase indent to \a Col. Creates \a Kind token at \a InsertPoint
426    ///        if needed.
427    bool rollIndent( int ToColumn
428                   , Token::TokenKind Kind
429                   , TokenQueueT::iterator InsertPoint);
430  
431    /// Skip a single-line comment when the comment starts at the current
432    /// position of the scanner.
433    void skipComment();
434  
435    /// Skip whitespace and comments until the start of the next token.
436    void scanToNextToken();
437  
438    /// Must be the first token generated.
439    bool scanStreamStart();
440  
441    /// Generate tokens needed to close out the stream.
442    bool scanStreamEnd();
443  
444    /// Scan a %BLAH directive.
445    bool scanDirective();
446  
447    /// Scan a ... or ---.
448    bool scanDocumentIndicator(bool IsStart);
449  
450    /// Scan a [ or { and generate the proper flow collection start token.
451    bool scanFlowCollectionStart(bool IsSequence);
452  
453    /// Scan a ] or } and generate the proper flow collection end token.
454    bool scanFlowCollectionEnd(bool IsSequence);
455  
456    /// Scan the , that separates entries in a flow collection.
457    bool scanFlowEntry();
458  
459    /// Scan the - that starts block sequence entries.
460    bool scanBlockEntry();
461  
462    /// Scan an explicit ? indicating a key.
463    bool scanKey();
464  
465    /// Scan an explicit : indicating a value.
466    bool scanValue();
467  
468    /// Scan a quoted scalar.
469    bool scanFlowScalar(bool IsDoubleQuoted);
470  
471    /// Scan an unquoted scalar.
472    bool scanPlainScalar();
473  
474    /// Scan an Alias or Anchor starting with * or &.
475    bool scanAliasOrAnchor(bool IsAlias);
476  
477    /// Scan a block scalar starting with | or >.
478    bool scanBlockScalar(bool IsLiteral);
479  
480    /// Scan a block scalar style indicator and header.
481    ///
482    /// Note: This is distinct from scanBlockScalarHeader to mirror the fact that
483    /// YAML does not consider the style indicator to be a part of the header.
484    ///
485    /// Return false if an error occurred.
486    bool scanBlockScalarIndicators(char &StyleIndicator, char &ChompingIndicator,
487                                   unsigned &IndentIndicator, bool &IsDone);
488  
489    /// Scan a style indicator in a block scalar header.
490    char scanBlockStyleIndicator();
491  
492    /// Scan a chomping indicator in a block scalar header.
493    char scanBlockChompingIndicator();
494  
495    /// Scan an indentation indicator in a block scalar header.
496    unsigned scanBlockIndentationIndicator();
497  
498    /// Scan a block scalar header.
499    ///
500    /// Return false if an error occurred.
501    bool scanBlockScalarHeader(char &ChompingIndicator, unsigned &IndentIndicator,
502                               bool &IsDone);
503  
504    /// Look for the indentation level of a block scalar.
505    ///
506    /// Return false if an error occurred.
507    bool findBlockScalarIndent(unsigned &BlockIndent, unsigned BlockExitIndent,
508                               unsigned &LineBreaks, bool &IsDone);
509  
510    /// Scan the indentation of a text line in a block scalar.
511    ///
512    /// Return false if an error occurred.
513    bool scanBlockScalarIndent(unsigned BlockIndent, unsigned BlockExitIndent,
514                               bool &IsDone);
515  
516    /// Scan a tag of the form !stuff.
517    bool scanTag();
518  
519    /// Dispatch to the next scanning function based on \a *Cur.
520    bool fetchMoreTokens();
521  
522    /// The SourceMgr used for diagnostics and buffer management.
523    SourceMgr &SM;
524  
525    /// The original input.
526    MemoryBufferRef InputBuffer;
527  
528    /// The current position of the scanner.
529    StringRef::iterator Current;
530  
531    /// The end of the input (one past the last character).
532    StringRef::iterator End;
533  
534    /// Current YAML indentation level in spaces.
535    int Indent;
536  
537    /// Current column number in Unicode code points.
538    unsigned Column;
539  
540    /// Current line number.
541    unsigned Line;
542  
543    /// How deep we are in flow style containers. 0 Means at block level.
544    unsigned FlowLevel;
545  
546    /// Are we at the start of the stream?
547    bool IsStartOfStream;
548  
549    /// Can the next token be the start of a simple key?
550    bool IsSimpleKeyAllowed;
551  
552    /// Can the next token be a value indicator even if it does not have a
553    /// trailing space?
554    bool IsAdjacentValueAllowedInFlow;
555  
556    /// True if an error has occurred.
557    bool Failed;
558  
559    /// Should colors be used when printing out the diagnostic messages?
560    bool ShowColors;
561  
562    /// Queue of tokens. This is required to queue up tokens while looking
563    ///        for the end of a simple key. And for cases where a single character
564    ///        can produce multiple tokens (e.g. BlockEnd).
565    TokenQueueT TokenQueue;
566  
567    /// Indentation levels.
568    SmallVector<int, 4> Indents;
569  
570    /// Potential simple keys.
571    SmallVector<SimpleKey, 4> SimpleKeys;
572  
573    std::error_code *EC;
574  };
575  
576  } // end namespace yaml
577  } // end namespace llvm
578  
579  /// encodeUTF8 - Encode \a UnicodeScalarValue in UTF-8 and append it to result.
580  static void encodeUTF8( uint32_t UnicodeScalarValue
581                        , SmallVectorImpl<char> &Result) {
582    if (UnicodeScalarValue <= 0x7F) {
583      Result.push_back(UnicodeScalarValue & 0x7F);
584    } else if (UnicodeScalarValue <= 0x7FF) {
585      uint8_t FirstByte = 0xC0 | ((UnicodeScalarValue & 0x7C0) >> 6);
586      uint8_t SecondByte = 0x80 | (UnicodeScalarValue & 0x3F);
587      Result.push_back(FirstByte);
588      Result.push_back(SecondByte);
589    } else if (UnicodeScalarValue <= 0xFFFF) {
590      uint8_t FirstByte = 0xE0 | ((UnicodeScalarValue & 0xF000) >> 12);
591      uint8_t SecondByte = 0x80 | ((UnicodeScalarValue & 0xFC0) >> 6);
592      uint8_t ThirdByte = 0x80 | (UnicodeScalarValue & 0x3F);
593      Result.push_back(FirstByte);
594      Result.push_back(SecondByte);
595      Result.push_back(ThirdByte);
596    } else if (UnicodeScalarValue <= 0x10FFFF) {
597      uint8_t FirstByte = 0xF0 | ((UnicodeScalarValue & 0x1F0000) >> 18);
598      uint8_t SecondByte = 0x80 | ((UnicodeScalarValue & 0x3F000) >> 12);
599      uint8_t ThirdByte = 0x80 | ((UnicodeScalarValue & 0xFC0) >> 6);
600      uint8_t FourthByte = 0x80 | (UnicodeScalarValue & 0x3F);
601      Result.push_back(FirstByte);
602      Result.push_back(SecondByte);
603      Result.push_back(ThirdByte);
604      Result.push_back(FourthByte);
605    }
606  }
607  
608  bool yaml::dumpTokens(StringRef Input, raw_ostream &OS) {
609    SourceMgr SM;
610    Scanner scanner(Input, SM);
611    while (true) {
612      Token T = scanner.getNext();
613      switch (T.Kind) {
614      case Token::TK_StreamStart:
615        OS << "Stream-Start: ";
616        break;
617      case Token::TK_StreamEnd:
618        OS << "Stream-End: ";
619        break;
620      case Token::TK_VersionDirective:
621        OS << "Version-Directive: ";
622        break;
623      case Token::TK_TagDirective:
624        OS << "Tag-Directive: ";
625        break;
626      case Token::TK_DocumentStart:
627        OS << "Document-Start: ";
628        break;
629      case Token::TK_DocumentEnd:
630        OS << "Document-End: ";
631        break;
632      case Token::TK_BlockEntry:
633        OS << "Block-Entry: ";
634        break;
635      case Token::TK_BlockEnd:
636        OS << "Block-End: ";
637        break;
638      case Token::TK_BlockSequenceStart:
639        OS << "Block-Sequence-Start: ";
640        break;
641      case Token::TK_BlockMappingStart:
642        OS << "Block-Mapping-Start: ";
643        break;
644      case Token::TK_FlowEntry:
645        OS << "Flow-Entry: ";
646        break;
647      case Token::TK_FlowSequenceStart:
648        OS << "Flow-Sequence-Start: ";
649        break;
650      case Token::TK_FlowSequenceEnd:
651        OS << "Flow-Sequence-End: ";
652        break;
653      case Token::TK_FlowMappingStart:
654        OS << "Flow-Mapping-Start: ";
655        break;
656      case Token::TK_FlowMappingEnd:
657        OS << "Flow-Mapping-End: ";
658        break;
659      case Token::TK_Key:
660        OS << "Key: ";
661        break;
662      case Token::TK_Value:
663        OS << "Value: ";
664        break;
665      case Token::TK_Scalar:
666        OS << "Scalar: ";
667        break;
668      case Token::TK_BlockScalar:
669        OS << "Block Scalar: ";
670        break;
671      case Token::TK_Alias:
672        OS << "Alias: ";
673        break;
674      case Token::TK_Anchor:
675        OS << "Anchor: ";
676        break;
677      case Token::TK_Tag:
678        OS << "Tag: ";
679        break;
680      case Token::TK_Error:
681        break;
682      }
683      OS << T.Range << "\n";
684      if (T.Kind == Token::TK_StreamEnd)
685        break;
686      else if (T.Kind == Token::TK_Error)
687        return false;
688    }
689    return true;
690  }
691  
692  bool yaml::scanTokens(StringRef Input) {
693    SourceMgr SM;
694    Scanner scanner(Input, SM);
695    while (true) {
696      Token T = scanner.getNext();
697      if (T.Kind == Token::TK_StreamEnd)
698        break;
699      else if (T.Kind == Token::TK_Error)
700        return false;
701    }
702    return true;
703  }
704  
705  std::string yaml::escape(StringRef Input, bool EscapePrintable) {
706    std::string EscapedInput;
707    for (StringRef::iterator i = Input.begin(), e = Input.end(); i != e; ++i) {
708      if (*i == '\\')
709        EscapedInput += "\\\\";
710      else if (*i == '"')
711        EscapedInput += "\\\"";
712      else if (*i == 0)
713        EscapedInput += "\\0";
714      else if (*i == 0x07)
715        EscapedInput += "\\a";
716      else if (*i == 0x08)
717        EscapedInput += "\\b";
718      else if (*i == 0x09)
719        EscapedInput += "\\t";
720      else if (*i == 0x0A)
721        EscapedInput += "\\n";
722      else if (*i == 0x0B)
723        EscapedInput += "\\v";
724      else if (*i == 0x0C)
725        EscapedInput += "\\f";
726      else if (*i == 0x0D)
727        EscapedInput += "\\r";
728      else if (*i == 0x1B)
729        EscapedInput += "\\e";
730      else if ((unsigned char)*i < 0x20) { // Control characters not handled above.
731        std::string HexStr = utohexstr(*i);
732        EscapedInput += "\\x" + std::string(2 - HexStr.size(), '0') + HexStr;
733      } else if (*i & 0x80) { // UTF-8 multiple code unit subsequence.
734        UTF8Decoded UnicodeScalarValue
735          = decodeUTF8(StringRef(i, Input.end() - i));
736        if (UnicodeScalarValue.second == 0) {
737          // Found invalid char.
738          SmallString<4> Val;
739          encodeUTF8(0xFFFD, Val);
740          llvm::append_range(EscapedInput, Val);
741          // FIXME: Error reporting.
742          return EscapedInput;
743        }
744        if (UnicodeScalarValue.first == 0x85)
745          EscapedInput += "\\N";
746        else if (UnicodeScalarValue.first == 0xA0)
747          EscapedInput += "\\_";
748        else if (UnicodeScalarValue.first == 0x2028)
749          EscapedInput += "\\L";
750        else if (UnicodeScalarValue.first == 0x2029)
751          EscapedInput += "\\P";
752        else if (!EscapePrintable &&
753                 sys::unicode::isPrintable(UnicodeScalarValue.first))
754          EscapedInput += StringRef(i, UnicodeScalarValue.second);
755        else {
756          std::string HexStr = utohexstr(UnicodeScalarValue.first);
757          if (HexStr.size() <= 2)
758            EscapedInput += "\\x" + std::string(2 - HexStr.size(), '0') + HexStr;
759          else if (HexStr.size() <= 4)
760            EscapedInput += "\\u" + std::string(4 - HexStr.size(), '0') + HexStr;
761          else if (HexStr.size() <= 8)
762            EscapedInput += "\\U" + std::string(8 - HexStr.size(), '0') + HexStr;
763        }
764        i += UnicodeScalarValue.second - 1;
765      } else
766        EscapedInput.push_back(*i);
767    }
768    return EscapedInput;
769  }
770  
771  std::optional<bool> yaml::parseBool(StringRef S) {
772    switch (S.size()) {
773    case 1:
774      switch (S.front()) {
775      case 'y':
776      case 'Y':
777        return true;
778      case 'n':
779      case 'N':
780        return false;
781      default:
782        return std::nullopt;
783      }
784    case 2:
785      switch (S.front()) {
786      case 'O':
787        if (S[1] == 'N') // ON
788          return true;
789        [[fallthrough]];
790      case 'o':
791        if (S[1] == 'n') //[Oo]n
792          return true;
793        return std::nullopt;
794      case 'N':
795        if (S[1] == 'O') // NO
796          return false;
797        [[fallthrough]];
798      case 'n':
799        if (S[1] == 'o') //[Nn]o
800          return false;
801        return std::nullopt;
802      default:
803        return std::nullopt;
804      }
805    case 3:
806      switch (S.front()) {
807      case 'O':
808        if (S.drop_front() == "FF") // OFF
809          return false;
810        [[fallthrough]];
811      case 'o':
812        if (S.drop_front() == "ff") //[Oo]ff
813          return false;
814        return std::nullopt;
815      case 'Y':
816        if (S.drop_front() == "ES") // YES
817          return true;
818        [[fallthrough]];
819      case 'y':
820        if (S.drop_front() == "es") //[Yy]es
821          return true;
822        return std::nullopt;
823      default:
824        return std::nullopt;
825      }
826    case 4:
827      switch (S.front()) {
828      case 'T':
829        if (S.drop_front() == "RUE") // TRUE
830          return true;
831        [[fallthrough]];
832      case 't':
833        if (S.drop_front() == "rue") //[Tt]rue
834          return true;
835        return std::nullopt;
836      default:
837        return std::nullopt;
838      }
839    case 5:
840      switch (S.front()) {
841      case 'F':
842        if (S.drop_front() == "ALSE") // FALSE
843          return false;
844        [[fallthrough]];
845      case 'f':
846        if (S.drop_front() == "alse") //[Ff]alse
847          return false;
848        return std::nullopt;
849      default:
850        return std::nullopt;
851      }
852    default:
853      return std::nullopt;
854    }
855  }
856  
857  Scanner::Scanner(StringRef Input, SourceMgr &sm, bool ShowColors,
858                   std::error_code *EC)
859      : SM(sm), ShowColors(ShowColors), EC(EC) {
860    init(MemoryBufferRef(Input, "YAML"));
861  }
862  
863  Scanner::Scanner(MemoryBufferRef Buffer, SourceMgr &SM_, bool ShowColors,
864                   std::error_code *EC)
865      : SM(SM_), ShowColors(ShowColors), EC(EC) {
866    init(Buffer);
867  }
868  
869  void Scanner::init(MemoryBufferRef Buffer) {
870    InputBuffer = Buffer;
871    Current = InputBuffer.getBufferStart();
872    End = InputBuffer.getBufferEnd();
873    Indent = -1;
874    Column = 0;
875    Line = 0;
876    FlowLevel = 0;
877    IsStartOfStream = true;
878    IsSimpleKeyAllowed = true;
879    IsAdjacentValueAllowedInFlow = false;
880    Failed = false;
881    std::unique_ptr<MemoryBuffer> InputBufferOwner =
882        MemoryBuffer::getMemBuffer(Buffer, /*RequiresNullTerminator=*/false);
883    SM.AddNewSourceBuffer(std::move(InputBufferOwner), SMLoc());
884  }
885  
886  Token &Scanner::peekNext() {
887    // If the current token is a possible simple key, keep parsing until we
888    // can confirm.
889    bool NeedMore = false;
890    while (true) {
891      if (TokenQueue.empty() || NeedMore) {
892        if (!fetchMoreTokens()) {
893          TokenQueue.clear();
894          SimpleKeys.clear();
895          TokenQueue.push_back(Token());
896          return TokenQueue.front();
897        }
898      }
899      assert(!TokenQueue.empty() &&
900              "fetchMoreTokens lied about getting tokens!");
901  
902      removeStaleSimpleKeyCandidates();
903      SimpleKey SK;
904      SK.Tok = TokenQueue.begin();
905      if (!is_contained(SimpleKeys, SK))
906        break;
907      else
908        NeedMore = true;
909    }
910    return TokenQueue.front();
911  }
912  
913  Token Scanner::getNext() {
914    Token Ret = peekNext();
915    // TokenQueue can be empty if there was an error getting the next token.
916    if (!TokenQueue.empty())
917      TokenQueue.pop_front();
918  
919    // There cannot be any referenced Token's if the TokenQueue is empty. So do a
920    // quick deallocation of them all.
921    if (TokenQueue.empty())
922      TokenQueue.resetAlloc();
923  
924    return Ret;
925  }
926  
927  StringRef::iterator Scanner::skip_nb_char(StringRef::iterator Position) {
928    if (Position == End)
929      return Position;
930    // Check 7 bit c-printable - b-char.
931    if (   *Position == 0x09
932        || (*Position >= 0x20 && *Position <= 0x7E))
933      return Position + 1;
934  
935    // Check for valid UTF-8.
936    if (uint8_t(*Position) & 0x80) {
937      UTF8Decoded u8d = decodeUTF8(Position);
938      if (   u8d.second != 0
939          && u8d.first != 0xFEFF
940          && ( u8d.first == 0x85
941            || ( u8d.first >= 0xA0
942              && u8d.first <= 0xD7FF)
943            || ( u8d.first >= 0xE000
944              && u8d.first <= 0xFFFD)
945            || ( u8d.first >= 0x10000
946              && u8d.first <= 0x10FFFF)))
947        return Position + u8d.second;
948    }
949    return Position;
950  }
951  
952  StringRef::iterator Scanner::skip_b_break(StringRef::iterator Position) {
953    if (Position == End)
954      return Position;
955    if (*Position == 0x0D) {
956      if (Position + 1 != End && *(Position + 1) == 0x0A)
957        return Position + 2;
958      return Position + 1;
959    }
960  
961    if (*Position == 0x0A)
962      return Position + 1;
963    return Position;
964  }
965  
966  StringRef::iterator Scanner::skip_s_space(StringRef::iterator Position) {
967    if (Position == End)
968      return Position;
969    if (*Position == ' ')
970      return Position + 1;
971    return Position;
972  }
973  
974  StringRef::iterator Scanner::skip_s_white(StringRef::iterator Position) {
975    if (Position == End)
976      return Position;
977    if (*Position == ' ' || *Position == '\t')
978      return Position + 1;
979    return Position;
980  }
981  
982  StringRef::iterator Scanner::skip_ns_char(StringRef::iterator Position) {
983    if (Position == End)
984      return Position;
985    if (*Position == ' ' || *Position == '\t')
986      return Position;
987    return skip_nb_char(Position);
988  }
989  
990  StringRef::iterator Scanner::skip_while( SkipWhileFunc Func
991                                         , StringRef::iterator Position) {
992    while (true) {
993      StringRef::iterator i = (this->*Func)(Position);
994      if (i == Position)
995        break;
996      Position = i;
997    }
998    return Position;
999  }
1000  
1001  void Scanner::advanceWhile(SkipWhileFunc Func) {
1002    auto Final = skip_while(Func, Current);
1003    Column += Final - Current;
1004    Current = Final;
1005  }
1006  
1007  static bool is_ns_hex_digit(const char C) { return isAlnum(C); }
1008  
1009  static bool is_ns_word_char(const char C) { return C == '-' || isAlpha(C); }
1010  
1011  void Scanner::scan_ns_uri_char() {
1012    while (true) {
1013      if (Current == End)
1014        break;
1015      if ((   *Current == '%'
1016            && Current + 2 < End
1017            && is_ns_hex_digit(*(Current + 1))
1018            && is_ns_hex_digit(*(Current + 2)))
1019          || is_ns_word_char(*Current)
1020          || StringRef(Current, 1).find_first_of("#;/?:@&=+$,_.!~*'()[]")
1021            != StringRef::npos) {
1022        ++Current;
1023        ++Column;
1024      } else
1025        break;
1026    }
1027  }
1028  
1029  bool Scanner::consume(uint32_t Expected) {
1030    if (Expected >= 0x80) {
1031      setError("Cannot consume non-ascii characters", Current);
1032      return false;
1033    }
1034    if (Current == End)
1035      return false;
1036    if (uint8_t(*Current) >= 0x80) {
1037      setError("Cannot consume non-ascii characters", Current);
1038      return false;
1039    }
1040    if (uint8_t(*Current) == Expected) {
1041      ++Current;
1042      ++Column;
1043      return true;
1044    }
1045    return false;
1046  }
1047  
1048  void Scanner::skip(uint32_t Distance) {
1049    Current += Distance;
1050    Column += Distance;
1051    assert(Current <= End && "Skipped past the end");
1052  }
1053  
1054  bool Scanner::isBlankOrBreak(StringRef::iterator Position) {
1055    if (Position == End)
1056      return false;
1057    return *Position == ' ' || *Position == '\t' || *Position == '\r' ||
1058           *Position == '\n';
1059  }
1060  
1061  bool Scanner::isPlainSafeNonBlank(StringRef::iterator Position) {
1062    if (Position == End || isBlankOrBreak(Position))
1063      return false;
1064    if (FlowLevel &&
1065        StringRef(Position, 1).find_first_of(",[]{}") != StringRef::npos)
1066      return false;
1067    return true;
1068  }
1069  
1070  bool Scanner::isLineEmpty(StringRef Line) {
1071    for (const auto *Position = Line.begin(); Position != Line.end(); ++Position)
1072      if (!isBlankOrBreak(Position))
1073        return false;
1074    return true;
1075  }
1076  
1077  bool Scanner::consumeLineBreakIfPresent() {
1078    auto Next = skip_b_break(Current);
1079    if (Next == Current)
1080      return false;
1081    Column = 0;
1082    ++Line;
1083    Current = Next;
1084    return true;
1085  }
1086  
1087  void Scanner::saveSimpleKeyCandidate( TokenQueueT::iterator Tok
1088                                      , unsigned AtColumn
1089                                      , bool IsRequired) {
1090    if (IsSimpleKeyAllowed) {
1091      SimpleKey SK;
1092      SK.Tok = Tok;
1093      SK.Line = Line;
1094      SK.Column = AtColumn;
1095      SK.IsRequired = IsRequired;
1096      SK.FlowLevel = FlowLevel;
1097      SimpleKeys.push_back(SK);
1098    }
1099  }
1100  
1101  void Scanner::removeStaleSimpleKeyCandidates() {
1102    for (SmallVectorImpl<SimpleKey>::iterator i = SimpleKeys.begin();
1103                                              i != SimpleKeys.end();) {
1104      if (i->Line != Line || i->Column + 1024 < Column) {
1105        if (i->IsRequired)
1106          setError( "Could not find expected : for simple key"
1107                  , i->Tok->Range.begin());
1108        i = SimpleKeys.erase(i);
1109      } else
1110        ++i;
1111    }
1112  }
1113  
1114  void Scanner::removeSimpleKeyCandidatesOnFlowLevel(unsigned Level) {
1115    if (!SimpleKeys.empty() && (SimpleKeys.end() - 1)->FlowLevel == Level)
1116      SimpleKeys.pop_back();
1117  }
1118  
1119  bool Scanner::unrollIndent(int ToColumn) {
1120    Token T;
1121    // Indentation is ignored in flow.
1122    if (FlowLevel != 0)
1123      return true;
1124  
1125    while (Indent > ToColumn) {
1126      T.Kind = Token::TK_BlockEnd;
1127      T.Range = StringRef(Current, 1);
1128      TokenQueue.push_back(T);
1129      Indent = Indents.pop_back_val();
1130    }
1131  
1132    return true;
1133  }
1134  
1135  bool Scanner::rollIndent( int ToColumn
1136                          , Token::TokenKind Kind
1137                          , TokenQueueT::iterator InsertPoint) {
1138    if (FlowLevel)
1139      return true;
1140    if (Indent < ToColumn) {
1141      Indents.push_back(Indent);
1142      Indent = ToColumn;
1143  
1144      Token T;
1145      T.Kind = Kind;
1146      T.Range = StringRef(Current, 0);
1147      TokenQueue.insert(InsertPoint, T);
1148    }
1149    return true;
1150  }
1151  
1152  void Scanner::skipComment() {
1153    if (Current == End || *Current != '#')
1154      return;
1155    while (true) {
1156      // This may skip more than one byte, thus Column is only incremented
1157      // for code points.
1158      StringRef::iterator I = skip_nb_char(Current);
1159      if (I == Current)
1160        break;
1161      Current = I;
1162      ++Column;
1163    }
1164  }
1165  
1166  void Scanner::scanToNextToken() {
1167    while (true) {
1168      while (Current != End && (*Current == ' ' || *Current == '\t')) {
1169        skip(1);
1170      }
1171  
1172      skipComment();
1173  
1174      // Skip EOL.
1175      StringRef::iterator i = skip_b_break(Current);
1176      if (i == Current)
1177        break;
1178      Current = i;
1179      ++Line;
1180      Column = 0;
1181      // New lines may start a simple key.
1182      if (!FlowLevel)
1183        IsSimpleKeyAllowed = true;
1184    }
1185  }
1186  
1187  bool Scanner::scanStreamStart() {
1188    IsStartOfStream = false;
1189  
1190    EncodingInfo EI = getUnicodeEncoding(currentInput());
1191  
1192    Token T;
1193    T.Kind = Token::TK_StreamStart;
1194    T.Range = StringRef(Current, EI.second);
1195    TokenQueue.push_back(T);
1196    Current += EI.second;
1197    return true;
1198  }
1199  
1200  bool Scanner::scanStreamEnd() {
1201    // Force an ending new line if one isn't present.
1202    if (Column != 0) {
1203      Column = 0;
1204      ++Line;
1205    }
1206  
1207    unrollIndent(-1);
1208    SimpleKeys.clear();
1209    IsSimpleKeyAllowed = false;
1210    IsAdjacentValueAllowedInFlow = false;
1211  
1212    Token T;
1213    T.Kind = Token::TK_StreamEnd;
1214    T.Range = StringRef(Current, 0);
1215    TokenQueue.push_back(T);
1216    return true;
1217  }
1218  
1219  bool Scanner::scanDirective() {
1220    // Reset the indentation level.
1221    unrollIndent(-1);
1222    SimpleKeys.clear();
1223    IsSimpleKeyAllowed = false;
1224    IsAdjacentValueAllowedInFlow = false;
1225  
1226    StringRef::iterator Start = Current;
1227    consume('%');
1228    StringRef::iterator NameStart = Current;
1229    Current = skip_while(&Scanner::skip_ns_char, Current);
1230    StringRef Name(NameStart, Current - NameStart);
1231    Current = skip_while(&Scanner::skip_s_white, Current);
1232  
1233    Token T;
1234    if (Name == "YAML") {
1235      Current = skip_while(&Scanner::skip_ns_char, Current);
1236      T.Kind = Token::TK_VersionDirective;
1237      T.Range = StringRef(Start, Current - Start);
1238      TokenQueue.push_back(T);
1239      return true;
1240    } else if(Name == "TAG") {
1241      Current = skip_while(&Scanner::skip_ns_char, Current);
1242      Current = skip_while(&Scanner::skip_s_white, Current);
1243      Current = skip_while(&Scanner::skip_ns_char, Current);
1244      T.Kind = Token::TK_TagDirective;
1245      T.Range = StringRef(Start, Current - Start);
1246      TokenQueue.push_back(T);
1247      return true;
1248    }
1249    return false;
1250  }
1251  
1252  bool Scanner::scanDocumentIndicator(bool IsStart) {
1253    unrollIndent(-1);
1254    SimpleKeys.clear();
1255    IsSimpleKeyAllowed = false;
1256    IsAdjacentValueAllowedInFlow = false;
1257  
1258    Token T;
1259    T.Kind = IsStart ? Token::TK_DocumentStart : Token::TK_DocumentEnd;
1260    T.Range = StringRef(Current, 3);
1261    skip(3);
1262    TokenQueue.push_back(T);
1263    return true;
1264  }
1265  
1266  bool Scanner::scanFlowCollectionStart(bool IsSequence) {
1267    Token T;
1268    T.Kind = IsSequence ? Token::TK_FlowSequenceStart
1269                        : Token::TK_FlowMappingStart;
1270    T.Range = StringRef(Current, 1);
1271    skip(1);
1272    TokenQueue.push_back(T);
1273  
1274    // [ and { may begin a simple key.
1275    saveSimpleKeyCandidate(--TokenQueue.end(), Column - 1, false);
1276  
1277    // And may also be followed by a simple key.
1278    IsSimpleKeyAllowed = true;
1279    // Adjacent values are allowed in flows only after JSON-style keys.
1280    IsAdjacentValueAllowedInFlow = false;
1281    ++FlowLevel;
1282    return true;
1283  }
1284  
1285  bool Scanner::scanFlowCollectionEnd(bool IsSequence) {
1286    removeSimpleKeyCandidatesOnFlowLevel(FlowLevel);
1287    IsSimpleKeyAllowed = false;
1288    IsAdjacentValueAllowedInFlow = true;
1289    Token T;
1290    T.Kind = IsSequence ? Token::TK_FlowSequenceEnd
1291                        : Token::TK_FlowMappingEnd;
1292    T.Range = StringRef(Current, 1);
1293    skip(1);
1294    TokenQueue.push_back(T);
1295    if (FlowLevel)
1296      --FlowLevel;
1297    return true;
1298  }
1299  
1300  bool Scanner::scanFlowEntry() {
1301    removeSimpleKeyCandidatesOnFlowLevel(FlowLevel);
1302    IsSimpleKeyAllowed = true;
1303    IsAdjacentValueAllowedInFlow = false;
1304    Token T;
1305    T.Kind = Token::TK_FlowEntry;
1306    T.Range = StringRef(Current, 1);
1307    skip(1);
1308    TokenQueue.push_back(T);
1309    return true;
1310  }
1311  
1312  bool Scanner::scanBlockEntry() {
1313    rollIndent(Column, Token::TK_BlockSequenceStart, TokenQueue.end());
1314    removeSimpleKeyCandidatesOnFlowLevel(FlowLevel);
1315    IsSimpleKeyAllowed = true;
1316    IsAdjacentValueAllowedInFlow = false;
1317    Token T;
1318    T.Kind = Token::TK_BlockEntry;
1319    T.Range = StringRef(Current, 1);
1320    skip(1);
1321    TokenQueue.push_back(T);
1322    return true;
1323  }
1324  
1325  bool Scanner::scanKey() {
1326    if (!FlowLevel)
1327      rollIndent(Column, Token::TK_BlockMappingStart, TokenQueue.end());
1328  
1329    removeSimpleKeyCandidatesOnFlowLevel(FlowLevel);
1330    IsSimpleKeyAllowed = !FlowLevel;
1331    IsAdjacentValueAllowedInFlow = false;
1332  
1333    Token T;
1334    T.Kind = Token::TK_Key;
1335    T.Range = StringRef(Current, 1);
1336    skip(1);
1337    TokenQueue.push_back(T);
1338    return true;
1339  }
1340  
1341  bool Scanner::scanValue() {
1342    // If the previous token could have been a simple key, insert the key token
1343    // into the token queue.
1344    if (!SimpleKeys.empty()) {
1345      SimpleKey SK = SimpleKeys.pop_back_val();
1346      Token T;
1347      T.Kind = Token::TK_Key;
1348      T.Range = SK.Tok->Range;
1349      TokenQueueT::iterator i, e;
1350      for (i = TokenQueue.begin(), e = TokenQueue.end(); i != e; ++i) {
1351        if (i == SK.Tok)
1352          break;
1353      }
1354      if (i == e) {
1355        Failed = true;
1356        return false;
1357      }
1358      i = TokenQueue.insert(i, T);
1359  
1360      // We may also need to add a Block-Mapping-Start token.
1361      rollIndent(SK.Column, Token::TK_BlockMappingStart, i);
1362  
1363      IsSimpleKeyAllowed = false;
1364    } else {
1365      if (!FlowLevel)
1366        rollIndent(Column, Token::TK_BlockMappingStart, TokenQueue.end());
1367      IsSimpleKeyAllowed = !FlowLevel;
1368    }
1369    IsAdjacentValueAllowedInFlow = false;
1370  
1371    Token T;
1372    T.Kind = Token::TK_Value;
1373    T.Range = StringRef(Current, 1);
1374    skip(1);
1375    TokenQueue.push_back(T);
1376    return true;
1377  }
1378  
1379  // Forbidding inlining improves performance by roughly 20%.
1380  // FIXME: Remove once llvm optimizes this to the faster version without hints.
1381  LLVM_ATTRIBUTE_NOINLINE static bool
1382  wasEscaped(StringRef::iterator First, StringRef::iterator Position);
1383  
1384  // Returns whether a character at 'Position' was escaped with a leading '\'.
1385  // 'First' specifies the position of the first character in the string.
1386  static bool wasEscaped(StringRef::iterator First,
1387                         StringRef::iterator Position) {
1388    assert(Position - 1 >= First);
1389    StringRef::iterator I = Position - 1;
1390    // We calculate the number of consecutive '\'s before the current position
1391    // by iterating backwards through our string.
1392    while (I >= First && *I == '\\') --I;
1393    // (Position - 1 - I) now contains the number of '\'s before the current
1394    // position. If it is odd, the character at 'Position' was escaped.
1395    return (Position - 1 - I) % 2 == 1;
1396  }
1397  
1398  bool Scanner::scanFlowScalar(bool IsDoubleQuoted) {
1399    StringRef::iterator Start = Current;
1400    unsigned ColStart = Column;
1401    if (IsDoubleQuoted) {
1402      do {
1403        ++Current;
1404        while (Current != End && *Current != '"')
1405          ++Current;
1406        // Repeat until the previous character was not a '\' or was an escaped
1407        // backslash.
1408      } while (   Current != End
1409               && *(Current - 1) == '\\'
1410               && wasEscaped(Start + 1, Current));
1411    } else {
1412      skip(1);
1413      while (Current != End) {
1414        // Skip a ' followed by another '.
1415        if (Current + 1 < End && *Current == '\'' && *(Current + 1) == '\'') {
1416          skip(2);
1417          continue;
1418        } else if (*Current == '\'')
1419          break;
1420        StringRef::iterator i = skip_nb_char(Current);
1421        if (i == Current) {
1422          i = skip_b_break(Current);
1423          if (i == Current)
1424            break;
1425          Current = i;
1426          Column = 0;
1427          ++Line;
1428        } else {
1429          if (i == End)
1430            break;
1431          Current = i;
1432          ++Column;
1433        }
1434      }
1435    }
1436  
1437    if (Current == End) {
1438      setError("Expected quote at end of scalar", Current);
1439      return false;
1440    }
1441  
1442    skip(1); // Skip ending quote.
1443    Token T;
1444    T.Kind = Token::TK_Scalar;
1445    T.Range = StringRef(Start, Current - Start);
1446    TokenQueue.push_back(T);
1447  
1448    saveSimpleKeyCandidate(--TokenQueue.end(), ColStart, false);
1449  
1450    IsSimpleKeyAllowed = false;
1451    IsAdjacentValueAllowedInFlow = true;
1452  
1453    return true;
1454  }
1455  
1456  bool Scanner::scanPlainScalar() {
1457    StringRef::iterator Start = Current;
1458    unsigned ColStart = Column;
1459    unsigned LeadingBlanks = 0;
1460    assert(Indent >= -1 && "Indent must be >= -1 !");
1461    unsigned indent = static_cast<unsigned>(Indent + 1);
1462    while (Current != End) {
1463      if (*Current == '#')
1464        break;
1465  
1466      while (Current != End &&
1467             ((*Current != ':' && isPlainSafeNonBlank(Current)) ||
1468              (*Current == ':' && isPlainSafeNonBlank(Current + 1)))) {
1469        StringRef::iterator i = skip_nb_char(Current);
1470        if (i == Current)
1471          break;
1472        Current = i;
1473        ++Column;
1474      }
1475  
1476      // Are we at the end?
1477      if (!isBlankOrBreak(Current))
1478        break;
1479  
1480      // Eat blanks.
1481      StringRef::iterator Tmp = Current;
1482      while (isBlankOrBreak(Tmp)) {
1483        StringRef::iterator i = skip_s_white(Tmp);
1484        if (i != Tmp) {
1485          if (LeadingBlanks && (Column < indent) && *Tmp == '\t') {
1486            setError("Found invalid tab character in indentation", Tmp);
1487            return false;
1488          }
1489          Tmp = i;
1490          ++Column;
1491        } else {
1492          i = skip_b_break(Tmp);
1493          if (!LeadingBlanks)
1494            LeadingBlanks = 1;
1495          Tmp = i;
1496          Column = 0;
1497          ++Line;
1498        }
1499      }
1500  
1501      if (!FlowLevel && Column < indent)
1502        break;
1503  
1504      Current = Tmp;
1505    }
1506    if (Start == Current) {
1507      setError("Got empty plain scalar", Start);
1508      return false;
1509    }
1510    Token T;
1511    T.Kind = Token::TK_Scalar;
1512    T.Range = StringRef(Start, Current - Start);
1513    TokenQueue.push_back(T);
1514  
1515    // Plain scalars can be simple keys.
1516    saveSimpleKeyCandidate(--TokenQueue.end(), ColStart, false);
1517  
1518    IsSimpleKeyAllowed = false;
1519    IsAdjacentValueAllowedInFlow = false;
1520  
1521    return true;
1522  }
1523  
1524  bool Scanner::scanAliasOrAnchor(bool IsAlias) {
1525    StringRef::iterator Start = Current;
1526    unsigned ColStart = Column;
1527    skip(1);
1528    while (Current != End) {
1529      if (   *Current == '[' || *Current == ']'
1530          || *Current == '{' || *Current == '}'
1531          || *Current == ','
1532          || *Current == ':')
1533        break;
1534      StringRef::iterator i = skip_ns_char(Current);
1535      if (i == Current)
1536        break;
1537      Current = i;
1538      ++Column;
1539    }
1540  
1541    if (Start + 1 == Current) {
1542      setError("Got empty alias or anchor", Start);
1543      return false;
1544    }
1545  
1546    Token T;
1547    T.Kind = IsAlias ? Token::TK_Alias : Token::TK_Anchor;
1548    T.Range = StringRef(Start, Current - Start);
1549    TokenQueue.push_back(T);
1550  
1551    // Alias and anchors can be simple keys.
1552    saveSimpleKeyCandidate(--TokenQueue.end(), ColStart, false);
1553  
1554    IsSimpleKeyAllowed = false;
1555    IsAdjacentValueAllowedInFlow = false;
1556  
1557    return true;
1558  }
1559  
1560  bool Scanner::scanBlockScalarIndicators(char &StyleIndicator,
1561                                          char &ChompingIndicator,
1562                                          unsigned &IndentIndicator,
1563                                          bool &IsDone) {
1564    StyleIndicator = scanBlockStyleIndicator();
1565    if (!scanBlockScalarHeader(ChompingIndicator, IndentIndicator, IsDone))
1566      return false;
1567    return true;
1568  }
1569  
1570  char Scanner::scanBlockStyleIndicator() {
1571    char Indicator = ' ';
1572    if (Current != End && (*Current == '>' || *Current == '|')) {
1573      Indicator = *Current;
1574      skip(1);
1575    }
1576    return Indicator;
1577  }
1578  
1579  char Scanner::scanBlockChompingIndicator() {
1580    char Indicator = ' ';
1581    if (Current != End && (*Current == '+' || *Current == '-')) {
1582      Indicator = *Current;
1583      skip(1);
1584    }
1585    return Indicator;
1586  }
1587  
1588  /// Get the number of line breaks after chomping.
1589  ///
1590  /// Return the number of trailing line breaks to emit, depending on
1591  /// \p ChompingIndicator.
1592  static unsigned getChompedLineBreaks(char ChompingIndicator,
1593                                       unsigned LineBreaks, StringRef Str) {
1594    if (ChompingIndicator == '-') // Strip all line breaks.
1595      return 0;
1596    if (ChompingIndicator == '+') // Keep all line breaks.
1597      return LineBreaks;
1598    // Clip trailing lines.
1599    return Str.empty() ? 0 : 1;
1600  }
1601  
1602  unsigned Scanner::scanBlockIndentationIndicator() {
1603    unsigned Indent = 0;
1604    if (Current != End && (*Current >= '1' && *Current <= '9')) {
1605      Indent = unsigned(*Current - '0');
1606      skip(1);
1607    }
1608    return Indent;
1609  }
1610  
1611  bool Scanner::scanBlockScalarHeader(char &ChompingIndicator,
1612                                      unsigned &IndentIndicator, bool &IsDone) {
1613    auto Start = Current;
1614  
1615    ChompingIndicator = scanBlockChompingIndicator();
1616    IndentIndicator = scanBlockIndentationIndicator();
1617    // Check for the chomping indicator once again.
1618    if (ChompingIndicator == ' ')
1619      ChompingIndicator = scanBlockChompingIndicator();
1620    Current = skip_while(&Scanner::skip_s_white, Current);
1621    skipComment();
1622  
1623    if (Current == End) { // EOF, we have an empty scalar.
1624      Token T;
1625      T.Kind = Token::TK_BlockScalar;
1626      T.Range = StringRef(Start, Current - Start);
1627      TokenQueue.push_back(T);
1628      IsDone = true;
1629      return true;
1630    }
1631  
1632    if (!consumeLineBreakIfPresent()) {
1633      setError("Expected a line break after block scalar header", Current);
1634      return false;
1635    }
1636    return true;
1637  }
1638  
1639  bool Scanner::findBlockScalarIndent(unsigned &BlockIndent,
1640                                      unsigned BlockExitIndent,
1641                                      unsigned &LineBreaks, bool &IsDone) {
1642    unsigned MaxAllSpaceLineCharacters = 0;
1643    StringRef::iterator LongestAllSpaceLine;
1644  
1645    while (true) {
1646      advanceWhile(&Scanner::skip_s_space);
1647      if (skip_nb_char(Current) != Current) {
1648        // This line isn't empty, so try and find the indentation.
1649        if (Column <= BlockExitIndent) { // End of the block literal.
1650          IsDone = true;
1651          return true;
1652        }
1653        // We found the block's indentation.
1654        BlockIndent = Column;
1655        if (MaxAllSpaceLineCharacters > BlockIndent) {
1656          setError(
1657              "Leading all-spaces line must be smaller than the block indent",
1658              LongestAllSpaceLine);
1659          return false;
1660        }
1661        return true;
1662      }
1663      if (skip_b_break(Current) != Current &&
1664          Column > MaxAllSpaceLineCharacters) {
1665        // Record the longest all-space line in case it's longer than the
1666        // discovered block indent.
1667        MaxAllSpaceLineCharacters = Column;
1668        LongestAllSpaceLine = Current;
1669      }
1670  
1671      // Check for EOF.
1672      if (Current == End) {
1673        IsDone = true;
1674        return true;
1675      }
1676  
1677      if (!consumeLineBreakIfPresent()) {
1678        IsDone = true;
1679        return true;
1680      }
1681      ++LineBreaks;
1682    }
1683    return true;
1684  }
1685  
1686  bool Scanner::scanBlockScalarIndent(unsigned BlockIndent,
1687                                      unsigned BlockExitIndent, bool &IsDone) {
1688    // Skip the indentation.
1689    while (Column < BlockIndent) {
1690      auto I = skip_s_space(Current);
1691      if (I == Current)
1692        break;
1693      Current = I;
1694      ++Column;
1695    }
1696  
1697    if (skip_nb_char(Current) == Current)
1698      return true;
1699  
1700    if (Column <= BlockExitIndent) { // End of the block literal.
1701      IsDone = true;
1702      return true;
1703    }
1704  
1705    if (Column < BlockIndent) {
1706      if (Current != End && *Current == '#') { // Trailing comment.
1707        IsDone = true;
1708        return true;
1709      }
1710      setError("A text line is less indented than the block scalar", Current);
1711      return false;
1712    }
1713    return true; // A normal text line.
1714  }
1715  
1716  bool Scanner::scanBlockScalar(bool IsLiteral) {
1717    assert(*Current == '|' || *Current == '>');
1718    char StyleIndicator;
1719    char ChompingIndicator;
1720    unsigned BlockIndent;
1721    bool IsDone = false;
1722    if (!scanBlockScalarIndicators(StyleIndicator, ChompingIndicator, BlockIndent,
1723                                   IsDone))
1724      return false;
1725    if (IsDone)
1726      return true;
1727    bool IsFolded = StyleIndicator == '>';
1728  
1729    const auto *Start = Current;
1730    unsigned BlockExitIndent = Indent < 0 ? 0 : (unsigned)Indent;
1731    unsigned LineBreaks = 0;
1732    if (BlockIndent == 0) {
1733      if (!findBlockScalarIndent(BlockIndent, BlockExitIndent, LineBreaks,
1734                                 IsDone))
1735        return false;
1736    }
1737  
1738    // Scan the block's scalars body.
1739    SmallString<256> Str;
1740    while (!IsDone) {
1741      if (!scanBlockScalarIndent(BlockIndent, BlockExitIndent, IsDone))
1742        return false;
1743      if (IsDone)
1744        break;
1745  
1746      // Parse the current line.
1747      auto LineStart = Current;
1748      advanceWhile(&Scanner::skip_nb_char);
1749      if (LineStart != Current) {
1750        if (LineBreaks && IsFolded && !Scanner::isLineEmpty(Str)) {
1751          // The folded style "folds" any single line break between content into a
1752          // single space, except when that content is "empty" (only contains
1753          // whitespace) in which case the line break is left as-is.
1754          if (LineBreaks == 1) {
1755            Str.append(LineBreaks,
1756                       isLineEmpty(StringRef(LineStart, Current - LineStart))
1757                           ? '\n'
1758                           : ' ');
1759          }
1760          // If we saw a single line break, we are completely replacing it and so
1761          // want `LineBreaks == 0`. Otherwise this decrement accounts for the
1762          // fact that the first line break is "trimmed", only being used to
1763          // signal a sequence of line breaks which should not be folded.
1764          LineBreaks--;
1765        }
1766        Str.append(LineBreaks, '\n');
1767        Str.append(StringRef(LineStart, Current - LineStart));
1768        LineBreaks = 0;
1769      }
1770  
1771      // Check for EOF.
1772      if (Current == End)
1773        break;
1774  
1775      if (!consumeLineBreakIfPresent())
1776        break;
1777      ++LineBreaks;
1778    }
1779  
1780    if (Current == End && !LineBreaks)
1781      // Ensure that there is at least one line break before the end of file.
1782      LineBreaks = 1;
1783    Str.append(getChompedLineBreaks(ChompingIndicator, LineBreaks, Str), '\n');
1784  
1785    // New lines may start a simple key.
1786    if (!FlowLevel)
1787      IsSimpleKeyAllowed = true;
1788    IsAdjacentValueAllowedInFlow = false;
1789  
1790    Token T;
1791    T.Kind = Token::TK_BlockScalar;
1792    T.Range = StringRef(Start, Current - Start);
1793    T.Value = std::string(Str);
1794    TokenQueue.push_back(T);
1795    return true;
1796  }
1797  
1798  bool Scanner::scanTag() {
1799    StringRef::iterator Start = Current;
1800    unsigned ColStart = Column;
1801    skip(1); // Eat !.
1802    if (Current == End || isBlankOrBreak(Current)); // An empty tag.
1803    else if (*Current == '<') {
1804      skip(1);
1805      scan_ns_uri_char();
1806      if (!consume('>'))
1807        return false;
1808    } else {
1809      // FIXME: Actually parse the c-ns-shorthand-tag rule.
1810      Current = skip_while(&Scanner::skip_ns_char, Current);
1811    }
1812  
1813    Token T;
1814    T.Kind = Token::TK_Tag;
1815    T.Range = StringRef(Start, Current - Start);
1816    TokenQueue.push_back(T);
1817  
1818    // Tags can be simple keys.
1819    saveSimpleKeyCandidate(--TokenQueue.end(), ColStart, false);
1820  
1821    IsSimpleKeyAllowed = false;
1822    IsAdjacentValueAllowedInFlow = false;
1823  
1824    return true;
1825  }
1826  
1827  bool Scanner::fetchMoreTokens() {
1828    if (IsStartOfStream)
1829      return scanStreamStart();
1830  
1831    scanToNextToken();
1832  
1833    if (Current == End)
1834      return scanStreamEnd();
1835  
1836    removeStaleSimpleKeyCandidates();
1837  
1838    unrollIndent(Column);
1839  
1840    if (Column == 0 && *Current == '%')
1841      return scanDirective();
1842  
1843    if (Column == 0 && Current + 4 <= End
1844        && *Current == '-'
1845        && *(Current + 1) == '-'
1846        && *(Current + 2) == '-'
1847        && (Current + 3 == End || isBlankOrBreak(Current + 3)))
1848      return scanDocumentIndicator(true);
1849  
1850    if (Column == 0 && Current + 4 <= End
1851        && *Current == '.'
1852        && *(Current + 1) == '.'
1853        && *(Current + 2) == '.'
1854        && (Current + 3 == End || isBlankOrBreak(Current + 3)))
1855      return scanDocumentIndicator(false);
1856  
1857    if (*Current == '[')
1858      return scanFlowCollectionStart(true);
1859  
1860    if (*Current == '{')
1861      return scanFlowCollectionStart(false);
1862  
1863    if (*Current == ']')
1864      return scanFlowCollectionEnd(true);
1865  
1866    if (*Current == '}')
1867      return scanFlowCollectionEnd(false);
1868  
1869    if (*Current == ',')
1870      return scanFlowEntry();
1871  
1872    if (*Current == '-' && (isBlankOrBreak(Current + 1) || Current + 1 == End))
1873      return scanBlockEntry();
1874  
1875    if (*Current == '?' && (Current + 1 == End || isBlankOrBreak(Current + 1)))
1876      return scanKey();
1877  
1878    if (*Current == ':' &&
1879        (!isPlainSafeNonBlank(Current + 1) || IsAdjacentValueAllowedInFlow))
1880      return scanValue();
1881  
1882    if (*Current == '*')
1883      return scanAliasOrAnchor(true);
1884  
1885    if (*Current == '&')
1886      return scanAliasOrAnchor(false);
1887  
1888    if (*Current == '!')
1889      return scanTag();
1890  
1891    if (*Current == '|' && !FlowLevel)
1892      return scanBlockScalar(true);
1893  
1894    if (*Current == '>' && !FlowLevel)
1895      return scanBlockScalar(false);
1896  
1897    if (*Current == '\'')
1898      return scanFlowScalar(false);
1899  
1900    if (*Current == '"')
1901      return scanFlowScalar(true);
1902  
1903    // Get a plain scalar.
1904    StringRef FirstChar(Current, 1);
1905    if ((!isBlankOrBreak(Current) &&
1906         FirstChar.find_first_of("-?:,[]{}#&*!|>'\"%@`") == StringRef::npos) ||
1907        (FirstChar.find_first_of("?:-") != StringRef::npos &&
1908         isPlainSafeNonBlank(Current + 1)))
1909      return scanPlainScalar();
1910  
1911    setError("Unrecognized character while tokenizing.", Current);
1912    return false;
1913  }
1914  
1915  Stream::Stream(StringRef Input, SourceMgr &SM, bool ShowColors,
1916                 std::error_code *EC)
1917      : scanner(new Scanner(Input, SM, ShowColors, EC)) {}
1918  
1919  Stream::Stream(MemoryBufferRef InputBuffer, SourceMgr &SM, bool ShowColors,
1920                 std::error_code *EC)
1921      : scanner(new Scanner(InputBuffer, SM, ShowColors, EC)) {}
1922  
1923  Stream::~Stream() = default;
1924  
1925  bool Stream::failed() { return scanner->failed(); }
1926  
1927  void Stream::printError(Node *N, const Twine &Msg, SourceMgr::DiagKind Kind) {
1928    printError(N ? N->getSourceRange() : SMRange(), Msg, Kind);
1929  }
1930  
1931  void Stream::printError(const SMRange &Range, const Twine &Msg,
1932                          SourceMgr::DiagKind Kind) {
1933    scanner->printError(Range.Start, Kind, Msg, Range);
1934  }
1935  
1936  document_iterator Stream::begin() {
1937    if (CurrentDoc)
1938      report_fatal_error("Can only iterate over the stream once");
1939  
1940    // Skip Stream-Start.
1941    scanner->getNext();
1942  
1943    CurrentDoc.reset(new Document(*this));
1944    return document_iterator(CurrentDoc);
1945  }
1946  
1947  document_iterator Stream::end() {
1948    return document_iterator();
1949  }
1950  
1951  void Stream::skip() {
1952    for (Document &Doc : *this)
1953      Doc.skip();
1954  }
1955  
1956  Node::Node(unsigned int Type, std::unique_ptr<Document> &D, StringRef A,
1957             StringRef T)
1958      : Doc(D), TypeID(Type), Anchor(A), Tag(T) {
1959    SMLoc Start = SMLoc::getFromPointer(peekNext().Range.begin());
1960    SourceRange = SMRange(Start, Start);
1961  }
1962  
1963  std::string Node::getVerbatimTag() const {
1964    StringRef Raw = getRawTag();
1965    if (!Raw.empty() && Raw != "!") {
1966      std::string Ret;
1967      if (Raw.find_last_of('!') == 0) {
1968        Ret = std::string(Doc->getTagMap().find("!")->second);
1969        Ret += Raw.substr(1);
1970        return Ret;
1971      } else if (Raw.starts_with("!!")) {
1972        Ret = std::string(Doc->getTagMap().find("!!")->second);
1973        Ret += Raw.substr(2);
1974        return Ret;
1975      } else {
1976        StringRef TagHandle = Raw.substr(0, Raw.find_last_of('!') + 1);
1977        std::map<StringRef, StringRef>::const_iterator It =
1978            Doc->getTagMap().find(TagHandle);
1979        if (It != Doc->getTagMap().end())
1980          Ret = std::string(It->second);
1981        else {
1982          Token T;
1983          T.Kind = Token::TK_Tag;
1984          T.Range = TagHandle;
1985          setError(Twine("Unknown tag handle ") + TagHandle, T);
1986        }
1987        Ret += Raw.substr(Raw.find_last_of('!') + 1);
1988        return Ret;
1989      }
1990    }
1991  
1992    switch (getType()) {
1993    case NK_Null:
1994      return "tag:yaml.org,2002:null";
1995    case NK_Scalar:
1996    case NK_BlockScalar:
1997      // TODO: Tag resolution.
1998      return "tag:yaml.org,2002:str";
1999    case NK_Mapping:
2000      return "tag:yaml.org,2002:map";
2001    case NK_Sequence:
2002      return "tag:yaml.org,2002:seq";
2003    }
2004  
2005    return "";
2006  }
2007  
2008  Token &Node::peekNext() {
2009    return Doc->peekNext();
2010  }
2011  
2012  Token Node::getNext() {
2013    return Doc->getNext();
2014  }
2015  
2016  Node *Node::parseBlockNode() {
2017    return Doc->parseBlockNode();
2018  }
2019  
2020  BumpPtrAllocator &Node::getAllocator() {
2021    return Doc->NodeAllocator;
2022  }
2023  
2024  void Node::setError(const Twine &Msg, Token &Tok) const {
2025    Doc->setError(Msg, Tok);
2026  }
2027  
2028  bool Node::failed() const {
2029    return Doc->failed();
2030  }
2031  
2032  StringRef ScalarNode::getValue(SmallVectorImpl<char> &Storage) const {
2033    if (Value[0] == '"')
2034      return getDoubleQuotedValue(Value, Storage);
2035    if (Value[0] == '\'')
2036      return getSingleQuotedValue(Value, Storage);
2037    return getPlainValue(Value, Storage);
2038  }
2039  
2040  /// parseScalarValue - A common parsing routine for all flow scalar styles.
2041  /// It handles line break characters by itself, adds regular content characters
2042  /// to the result, and forwards escaped sequences to the provided routine for
2043  /// the style-specific processing.
2044  ///
2045  /// \param UnquotedValue - An input value without quotation marks.
2046  /// \param Storage - A storage for the result if the input value is multiline or
2047  /// contains escaped characters.
2048  /// \param LookupChars - A set of special characters to search in the input
2049  /// string. Should include line break characters and the escape character
2050  /// specific for the processing scalar style, if any.
2051  /// \param UnescapeCallback - This is called when the escape character is found
2052  /// in the input.
2053  /// \returns - The unfolded and unescaped value.
2054  static StringRef
2055  parseScalarValue(StringRef UnquotedValue, SmallVectorImpl<char> &Storage,
2056                   StringRef LookupChars,
2057                   std::function<StringRef(StringRef, SmallVectorImpl<char> &)>
2058                       UnescapeCallback) {
2059    size_t I = UnquotedValue.find_first_of(LookupChars);
2060    if (I == StringRef::npos)
2061      return UnquotedValue;
2062  
2063    Storage.clear();
2064    Storage.reserve(UnquotedValue.size());
2065    char LastNewLineAddedAs = '\0';
2066    for (; I != StringRef::npos; I = UnquotedValue.find_first_of(LookupChars)) {
2067      if (UnquotedValue[I] != '\r' && UnquotedValue[I] != '\n') {
2068        llvm::append_range(Storage, UnquotedValue.take_front(I));
2069        UnquotedValue = UnescapeCallback(UnquotedValue.drop_front(I), Storage);
2070        LastNewLineAddedAs = '\0';
2071        continue;
2072      }
2073      if (size_t LastNonSWhite = UnquotedValue.find_last_not_of(" \t", I);
2074          LastNonSWhite != StringRef::npos) {
2075        llvm::append_range(Storage, UnquotedValue.take_front(LastNonSWhite + 1));
2076        Storage.push_back(' ');
2077        LastNewLineAddedAs = ' ';
2078      } else {
2079        // Note: we can't just check if the last character in Storage is ' ',
2080        // '\n', or something else; that would give a wrong result for double
2081        // quoted values containing an escaped space character before a new-line
2082        // character.
2083        switch (LastNewLineAddedAs) {
2084        case ' ':
2085          assert(!Storage.empty() && Storage.back() == ' ');
2086          Storage.back() = '\n';
2087          LastNewLineAddedAs = '\n';
2088          break;
2089        case '\n':
2090          assert(!Storage.empty() && Storage.back() == '\n');
2091          Storage.push_back('\n');
2092          break;
2093        default:
2094          Storage.push_back(' ');
2095          LastNewLineAddedAs = ' ';
2096          break;
2097        }
2098      }
2099      // Handle Windows-style EOL
2100      if (UnquotedValue.substr(I, 2) == "\r\n")
2101        I++;
2102      UnquotedValue = UnquotedValue.drop_front(I + 1).ltrim(" \t");
2103    }
2104    llvm::append_range(Storage, UnquotedValue);
2105    return StringRef(Storage.begin(), Storage.size());
2106  }
2107  
2108  StringRef
2109  ScalarNode::getDoubleQuotedValue(StringRef RawValue,
2110                                   SmallVectorImpl<char> &Storage) const {
2111    assert(RawValue.size() >= 2 && RawValue.front() == '"' &&
2112           RawValue.back() == '"');
2113    StringRef UnquotedValue = RawValue.substr(1, RawValue.size() - 2);
2114  
2115    auto UnescapeFunc = [this](StringRef UnquotedValue,
2116                               SmallVectorImpl<char> &Storage) {
2117      assert(UnquotedValue.take_front(1) == "\\");
2118      if (UnquotedValue.size() == 1) {
2119        Token T;
2120        T.Range = UnquotedValue;
2121        setError("Unrecognized escape code", T);
2122        Storage.clear();
2123        return StringRef();
2124      }
2125      UnquotedValue = UnquotedValue.drop_front(1);
2126      switch (UnquotedValue[0]) {
2127      default: {
2128        Token T;
2129        T.Range = UnquotedValue.take_front(1);
2130        setError("Unrecognized escape code", T);
2131        Storage.clear();
2132        return StringRef();
2133      }
2134      case '\r':
2135        // Shrink the Windows-style EOL.
2136        if (UnquotedValue.size() >= 2 && UnquotedValue[1] == '\n')
2137          UnquotedValue = UnquotedValue.drop_front(1);
2138        [[fallthrough]];
2139      case '\n':
2140        return UnquotedValue.drop_front(1).ltrim(" \t");
2141      case '0':
2142        Storage.push_back(0x00);
2143        break;
2144      case 'a':
2145        Storage.push_back(0x07);
2146        break;
2147      case 'b':
2148        Storage.push_back(0x08);
2149        break;
2150      case 't':
2151      case 0x09:
2152        Storage.push_back(0x09);
2153        break;
2154      case 'n':
2155        Storage.push_back(0x0A);
2156        break;
2157      case 'v':
2158        Storage.push_back(0x0B);
2159        break;
2160      case 'f':
2161        Storage.push_back(0x0C);
2162        break;
2163      case 'r':
2164        Storage.push_back(0x0D);
2165        break;
2166      case 'e':
2167        Storage.push_back(0x1B);
2168        break;
2169      case ' ':
2170        Storage.push_back(0x20);
2171        break;
2172      case '"':
2173        Storage.push_back(0x22);
2174        break;
2175      case '/':
2176        Storage.push_back(0x2F);
2177        break;
2178      case '\\':
2179        Storage.push_back(0x5C);
2180        break;
2181      case 'N':
2182        encodeUTF8(0x85, Storage);
2183        break;
2184      case '_':
2185        encodeUTF8(0xA0, Storage);
2186        break;
2187      case 'L':
2188        encodeUTF8(0x2028, Storage);
2189        break;
2190      case 'P':
2191        encodeUTF8(0x2029, Storage);
2192        break;
2193      case 'x': {
2194        if (UnquotedValue.size() < 3)
2195          // TODO: Report error.
2196          break;
2197        unsigned int UnicodeScalarValue;
2198        if (UnquotedValue.substr(1, 2).getAsInteger(16, UnicodeScalarValue))
2199          // TODO: Report error.
2200          UnicodeScalarValue = 0xFFFD;
2201        encodeUTF8(UnicodeScalarValue, Storage);
2202        return UnquotedValue.drop_front(3);
2203      }
2204      case 'u': {
2205        if (UnquotedValue.size() < 5)
2206          // TODO: Report error.
2207          break;
2208        unsigned int UnicodeScalarValue;
2209        if (UnquotedValue.substr(1, 4).getAsInteger(16, UnicodeScalarValue))
2210          // TODO: Report error.
2211          UnicodeScalarValue = 0xFFFD;
2212        encodeUTF8(UnicodeScalarValue, Storage);
2213        return UnquotedValue.drop_front(5);
2214      }
2215      case 'U': {
2216        if (UnquotedValue.size() < 9)
2217          // TODO: Report error.
2218          break;
2219        unsigned int UnicodeScalarValue;
2220        if (UnquotedValue.substr(1, 8).getAsInteger(16, UnicodeScalarValue))
2221          // TODO: Report error.
2222          UnicodeScalarValue = 0xFFFD;
2223        encodeUTF8(UnicodeScalarValue, Storage);
2224        return UnquotedValue.drop_front(9);
2225      }
2226      }
2227      return UnquotedValue.drop_front(1);
2228    };
2229  
2230    return parseScalarValue(UnquotedValue, Storage, "\\\r\n", UnescapeFunc);
2231  }
2232  
2233  StringRef ScalarNode::getSingleQuotedValue(StringRef RawValue,
2234                                             SmallVectorImpl<char> &Storage) {
2235    assert(RawValue.size() >= 2 && RawValue.front() == '\'' &&
2236           RawValue.back() == '\'');
2237    StringRef UnquotedValue = RawValue.substr(1, RawValue.size() - 2);
2238  
2239    auto UnescapeFunc = [](StringRef UnquotedValue,
2240                           SmallVectorImpl<char> &Storage) {
2241      assert(UnquotedValue.take_front(2) == "''");
2242      Storage.push_back('\'');
2243      return UnquotedValue.drop_front(2);
2244    };
2245  
2246    return parseScalarValue(UnquotedValue, Storage, "'\r\n", UnescapeFunc);
2247  }
2248  
2249  StringRef ScalarNode::getPlainValue(StringRef RawValue,
2250                                      SmallVectorImpl<char> &Storage) {
2251    // Trim trailing whitespace ('b-char' and 's-white').
2252    // NOTE: Alternatively we could change the scanner to not include whitespace
2253    //       here in the first place.
2254    RawValue = RawValue.rtrim("\r\n \t");
2255    return parseScalarValue(RawValue, Storage, "\r\n", nullptr);
2256  }
2257  
2258  Node *KeyValueNode::getKey() {
2259    if (Key)
2260      return Key;
2261    // Handle implicit null keys.
2262    {
2263      Token &t = peekNext();
2264      if (   t.Kind == Token::TK_BlockEnd
2265          || t.Kind == Token::TK_Value
2266          || t.Kind == Token::TK_Error) {
2267        return Key = new (getAllocator()) NullNode(Doc);
2268      }
2269      if (t.Kind == Token::TK_Key)
2270        getNext(); // skip TK_Key.
2271    }
2272  
2273    // Handle explicit null keys.
2274    Token &t = peekNext();
2275    if (t.Kind == Token::TK_BlockEnd || t.Kind == Token::TK_Value) {
2276      return Key = new (getAllocator()) NullNode(Doc);
2277    }
2278  
2279    // We've got a normal key.
2280    return Key = parseBlockNode();
2281  }
2282  
2283  Node *KeyValueNode::getValue() {
2284    if (Value)
2285      return Value;
2286  
2287    if (Node* Key = getKey())
2288      Key->skip();
2289    else {
2290      setError("Null key in Key Value.", peekNext());
2291      return Value = new (getAllocator()) NullNode(Doc);
2292    }
2293  
2294    if (failed())
2295      return Value = new (getAllocator()) NullNode(Doc);
2296  
2297    // Handle implicit null values.
2298    {
2299      Token &t = peekNext();
2300      if (   t.Kind == Token::TK_BlockEnd
2301          || t.Kind == Token::TK_FlowMappingEnd
2302          || t.Kind == Token::TK_Key
2303          || t.Kind == Token::TK_FlowEntry
2304          || t.Kind == Token::TK_Error) {
2305        return Value = new (getAllocator()) NullNode(Doc);
2306      }
2307  
2308      if (t.Kind != Token::TK_Value) {
2309        setError("Unexpected token in Key Value.", t);
2310        return Value = new (getAllocator()) NullNode(Doc);
2311      }
2312      getNext(); // skip TK_Value.
2313    }
2314  
2315    // Handle explicit null values.
2316    Token &t = peekNext();
2317    if (t.Kind == Token::TK_BlockEnd || t.Kind == Token::TK_Key) {
2318      return Value = new (getAllocator()) NullNode(Doc);
2319    }
2320  
2321    // We got a normal value.
2322    return Value = parseBlockNode();
2323  }
2324  
2325  void MappingNode::increment() {
2326    if (failed()) {
2327      IsAtEnd = true;
2328      CurrentEntry = nullptr;
2329      return;
2330    }
2331    if (CurrentEntry) {
2332      CurrentEntry->skip();
2333      if (Type == MT_Inline) {
2334        IsAtEnd = true;
2335        CurrentEntry = nullptr;
2336        return;
2337      }
2338    }
2339    Token T = peekNext();
2340    if (T.Kind == Token::TK_Key || T.Kind == Token::TK_Scalar) {
2341      // KeyValueNode eats the TK_Key. That way it can detect null keys.
2342      CurrentEntry = new (getAllocator()) KeyValueNode(Doc);
2343    } else if (Type == MT_Block) {
2344      switch (T.Kind) {
2345      case Token::TK_BlockEnd:
2346        getNext();
2347        IsAtEnd = true;
2348        CurrentEntry = nullptr;
2349        break;
2350      default:
2351        setError("Unexpected token. Expected Key or Block End", T);
2352        [[fallthrough]];
2353      case Token::TK_Error:
2354        IsAtEnd = true;
2355        CurrentEntry = nullptr;
2356      }
2357    } else {
2358      switch (T.Kind) {
2359      case Token::TK_FlowEntry:
2360        // Eat the flow entry and recurse.
2361        getNext();
2362        return increment();
2363      case Token::TK_FlowMappingEnd:
2364        getNext();
2365        [[fallthrough]];
2366      case Token::TK_Error:
2367        // Set this to end iterator.
2368        IsAtEnd = true;
2369        CurrentEntry = nullptr;
2370        break;
2371      default:
2372        setError( "Unexpected token. Expected Key, Flow Entry, or Flow "
2373                  "Mapping End."
2374                , T);
2375        IsAtEnd = true;
2376        CurrentEntry = nullptr;
2377      }
2378    }
2379  }
2380  
2381  void SequenceNode::increment() {
2382    if (failed()) {
2383      IsAtEnd = true;
2384      CurrentEntry = nullptr;
2385      return;
2386    }
2387    if (CurrentEntry)
2388      CurrentEntry->skip();
2389    Token T = peekNext();
2390    if (SeqType == ST_Block) {
2391      switch (T.Kind) {
2392      case Token::TK_BlockEntry:
2393        getNext();
2394        CurrentEntry = parseBlockNode();
2395        if (!CurrentEntry) { // An error occurred.
2396          IsAtEnd = true;
2397          CurrentEntry = nullptr;
2398        }
2399        break;
2400      case Token::TK_BlockEnd:
2401        getNext();
2402        IsAtEnd = true;
2403        CurrentEntry = nullptr;
2404        break;
2405      default:
2406        setError( "Unexpected token. Expected Block Entry or Block End."
2407                , T);
2408        [[fallthrough]];
2409      case Token::TK_Error:
2410        IsAtEnd = true;
2411        CurrentEntry = nullptr;
2412      }
2413    } else if (SeqType == ST_Indentless) {
2414      switch (T.Kind) {
2415      case Token::TK_BlockEntry:
2416        getNext();
2417        CurrentEntry = parseBlockNode();
2418        if (!CurrentEntry) { // An error occurred.
2419          IsAtEnd = true;
2420          CurrentEntry = nullptr;
2421        }
2422        break;
2423      default:
2424      case Token::TK_Error:
2425        IsAtEnd = true;
2426        CurrentEntry = nullptr;
2427      }
2428    } else if (SeqType == ST_Flow) {
2429      switch (T.Kind) {
2430      case Token::TK_FlowEntry:
2431        // Eat the flow entry and recurse.
2432        getNext();
2433        WasPreviousTokenFlowEntry = true;
2434        return increment();
2435      case Token::TK_FlowSequenceEnd:
2436        getNext();
2437        [[fallthrough]];
2438      case Token::TK_Error:
2439        // Set this to end iterator.
2440        IsAtEnd = true;
2441        CurrentEntry = nullptr;
2442        break;
2443      case Token::TK_StreamEnd:
2444      case Token::TK_DocumentEnd:
2445      case Token::TK_DocumentStart:
2446        setError("Could not find closing ]!", T);
2447        // Set this to end iterator.
2448        IsAtEnd = true;
2449        CurrentEntry = nullptr;
2450        break;
2451      default:
2452        if (!WasPreviousTokenFlowEntry) {
2453          setError("Expected , between entries!", T);
2454          IsAtEnd = true;
2455          CurrentEntry = nullptr;
2456          break;
2457        }
2458        // Otherwise it must be a flow entry.
2459        CurrentEntry = parseBlockNode();
2460        if (!CurrentEntry) {
2461          IsAtEnd = true;
2462        }
2463        WasPreviousTokenFlowEntry = false;
2464        break;
2465      }
2466    }
2467  }
2468  
2469  Document::Document(Stream &S) : stream(S), Root(nullptr) {
2470    // Tag maps starts with two default mappings.
2471    TagMap["!"] = "!";
2472    TagMap["!!"] = "tag:yaml.org,2002:";
2473  
2474    if (parseDirectives())
2475      expectToken(Token::TK_DocumentStart);
2476    Token &T = peekNext();
2477    if (T.Kind == Token::TK_DocumentStart)
2478      getNext();
2479  }
2480  
2481  bool Document::skip()  {
2482    if (stream.scanner->failed())
2483      return false;
2484    if (!Root && !getRoot())
2485      return false;
2486    Root->skip();
2487    Token &T = peekNext();
2488    if (T.Kind == Token::TK_StreamEnd)
2489      return false;
2490    if (T.Kind == Token::TK_DocumentEnd) {
2491      getNext();
2492      return skip();
2493    }
2494    return true;
2495  }
2496  
2497  Token &Document::peekNext() {
2498    return stream.scanner->peekNext();
2499  }
2500  
2501  Token Document::getNext() {
2502    return stream.scanner->getNext();
2503  }
2504  
2505  void Document::setError(const Twine &Message, Token &Location) const {
2506    stream.scanner->setError(Message, Location.Range.begin());
2507  }
2508  
2509  bool Document::failed() const {
2510    return stream.scanner->failed();
2511  }
2512  
2513  Node *Document::parseBlockNode() {
2514    Token T = peekNext();
2515    // Handle properties.
2516    Token AnchorInfo;
2517    Token TagInfo;
2518  parse_property:
2519    switch (T.Kind) {
2520    case Token::TK_Alias:
2521      getNext();
2522      return new (NodeAllocator) AliasNode(stream.CurrentDoc, T.Range.substr(1));
2523    case Token::TK_Anchor:
2524      if (AnchorInfo.Kind == Token::TK_Anchor) {
2525        setError("Already encountered an anchor for this node!", T);
2526        return nullptr;
2527      }
2528      AnchorInfo = getNext(); // Consume TK_Anchor.
2529      T = peekNext();
2530      goto parse_property;
2531    case Token::TK_Tag:
2532      if (TagInfo.Kind == Token::TK_Tag) {
2533        setError("Already encountered a tag for this node!", T);
2534        return nullptr;
2535      }
2536      TagInfo = getNext(); // Consume TK_Tag.
2537      T = peekNext();
2538      goto parse_property;
2539    default:
2540      break;
2541    }
2542  
2543    switch (T.Kind) {
2544    case Token::TK_BlockEntry:
2545      // We got an unindented BlockEntry sequence. This is not terminated with
2546      // a BlockEnd.
2547      // Don't eat the TK_BlockEntry, SequenceNode needs it.
2548      return new (NodeAllocator) SequenceNode( stream.CurrentDoc
2549                                             , AnchorInfo.Range.substr(1)
2550                                             , TagInfo.Range
2551                                             , SequenceNode::ST_Indentless);
2552    case Token::TK_BlockSequenceStart:
2553      getNext();
2554      return new (NodeAllocator)
2555        SequenceNode( stream.CurrentDoc
2556                    , AnchorInfo.Range.substr(1)
2557                    , TagInfo.Range
2558                    , SequenceNode::ST_Block);
2559    case Token::TK_BlockMappingStart:
2560      getNext();
2561      return new (NodeAllocator)
2562        MappingNode( stream.CurrentDoc
2563                   , AnchorInfo.Range.substr(1)
2564                   , TagInfo.Range
2565                   , MappingNode::MT_Block);
2566    case Token::TK_FlowSequenceStart:
2567      getNext();
2568      return new (NodeAllocator)
2569        SequenceNode( stream.CurrentDoc
2570                    , AnchorInfo.Range.substr(1)
2571                    , TagInfo.Range
2572                    , SequenceNode::ST_Flow);
2573    case Token::TK_FlowMappingStart:
2574      getNext();
2575      return new (NodeAllocator)
2576        MappingNode( stream.CurrentDoc
2577                   , AnchorInfo.Range.substr(1)
2578                   , TagInfo.Range
2579                   , MappingNode::MT_Flow);
2580    case Token::TK_Scalar:
2581      getNext();
2582      return new (NodeAllocator)
2583        ScalarNode( stream.CurrentDoc
2584                  , AnchorInfo.Range.substr(1)
2585                  , TagInfo.Range
2586                  , T.Range);
2587    case Token::TK_BlockScalar: {
2588      getNext();
2589      StringRef NullTerminatedStr(T.Value.c_str(), T.Value.length() + 1);
2590      StringRef StrCopy = NullTerminatedStr.copy(NodeAllocator).drop_back();
2591      return new (NodeAllocator)
2592          BlockScalarNode(stream.CurrentDoc, AnchorInfo.Range.substr(1),
2593                          TagInfo.Range, StrCopy, T.Range);
2594    }
2595    case Token::TK_Key:
2596      // Don't eat the TK_Key, KeyValueNode expects it.
2597      return new (NodeAllocator)
2598        MappingNode( stream.CurrentDoc
2599                   , AnchorInfo.Range.substr(1)
2600                   , TagInfo.Range
2601                   , MappingNode::MT_Inline);
2602    case Token::TK_DocumentStart:
2603    case Token::TK_DocumentEnd:
2604    case Token::TK_StreamEnd:
2605    default:
2606      // TODO: Properly handle tags. "[!!str ]" should resolve to !!str "", not
2607      //       !!null null.
2608      return new (NodeAllocator) NullNode(stream.CurrentDoc);
2609    case Token::TK_FlowMappingEnd:
2610    case Token::TK_FlowSequenceEnd:
2611    case Token::TK_FlowEntry: {
2612      if (Root && (isa<MappingNode>(Root) || isa<SequenceNode>(Root)))
2613        return new (NodeAllocator) NullNode(stream.CurrentDoc);
2614  
2615      setError("Unexpected token", T);
2616      return nullptr;
2617    }
2618    case Token::TK_Error:
2619      return nullptr;
2620    }
2621    llvm_unreachable("Control flow shouldn't reach here.");
2622    return nullptr;
2623  }
2624  
2625  bool Document::parseDirectives() {
2626    bool isDirective = false;
2627    while (true) {
2628      Token T = peekNext();
2629      if (T.Kind == Token::TK_TagDirective) {
2630        parseTAGDirective();
2631        isDirective = true;
2632      } else if (T.Kind == Token::TK_VersionDirective) {
2633        parseYAMLDirective();
2634        isDirective = true;
2635      } else
2636        break;
2637    }
2638    return isDirective;
2639  }
2640  
2641  void Document::parseYAMLDirective() {
2642    getNext(); // Eat %YAML <version>
2643  }
2644  
2645  void Document::parseTAGDirective() {
2646    Token Tag = getNext(); // %TAG <handle> <prefix>
2647    StringRef T = Tag.Range;
2648    // Strip %TAG
2649    T = T.substr(T.find_first_of(" \t")).ltrim(" \t");
2650    std::size_t HandleEnd = T.find_first_of(" \t");
2651    StringRef TagHandle = T.substr(0, HandleEnd);
2652    StringRef TagPrefix = T.substr(HandleEnd).ltrim(" \t");
2653    TagMap[TagHandle] = TagPrefix;
2654  }
2655  
2656  bool Document::expectToken(int TK) {
2657    Token T = getNext();
2658    if (T.Kind != TK) {
2659      setError("Unexpected token", T);
2660      return false;
2661    }
2662    return true;
2663  }
2664