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