1 //===--- YAMLParser.h - Simple YAML parser --------------------------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This is a YAML 1.2 parser.
12 // See http://www.yaml.org/spec/1.2/spec.html for the full standard.
14 // This currently does not implement the following:
15 // * Multi-line literal folding.
18 // * BOMs anywhere other than the first Unicode scalar value in the file.
20 // The most important class here is Stream. This represents a YAML stream with
21 // 0, 1, or many documents.
24 // StringRef input = getInput();
25 // yaml::Stream stream(input, sm);
27 // for (yaml::document_iterator di = stream.begin(), de = stream.end();
29 // yaml::Node *n = di->getRoot();
31 // // Do something with n...
36 //===----------------------------------------------------------------------===//
38 #ifndef LLVM_SUPPORT_YAMLPARSER_H
39 #define LLVM_SUPPORT_YAMLPARSER_H
41 #include "llvm/ADT/StringRef.h"
42 #include "llvm/Support/Allocator.h"
43 #include "llvm/Support/SMLoc.h"
45 #include <system_error>
49 class MemoryBufferRef;
56 class document_iterator;
62 /// \brief Dump all the tokens in this stream to OS.
63 /// \returns true if there was an error, false otherwise.
64 bool dumpTokens(StringRef Input, raw_ostream &);
66 /// \brief Scans all tokens in input without outputting anything. This is used
67 /// for benchmarking the tokenizer.
68 /// \returns true if there was an error, false otherwise.
69 bool scanTokens(StringRef Input);
71 /// \brief Escape \a Input for a double quoted scalar.
72 std::string escape(StringRef Input);
74 /// \brief This class represents a YAML stream potentially containing multiple
78 /// \brief This keeps a reference to the string referenced by \p Input.
79 Stream(StringRef Input, SourceMgr &, bool ShowColors = true,
80 std::error_code *EC = nullptr);
82 Stream(MemoryBufferRef InputBuffer, SourceMgr &, bool ShowColors = true,
83 std::error_code *EC = nullptr);
86 document_iterator begin();
87 document_iterator end();
95 void printError(Node *N, const Twine &Msg);
98 std::unique_ptr<Scanner> scanner;
99 std::unique_ptr<Document> CurrentDoc;
101 friend class Document;
104 /// \brief Abstract base class for all Nodes.
106 virtual void anchor();
119 Node(unsigned int Type, std::unique_ptr<Document> &, StringRef Anchor,
122 /// \brief Get the value of the anchor attached to this node. If it does not
123 /// have one, getAnchor().size() will be 0.
124 StringRef getAnchor() const { return Anchor; }
126 /// \brief Get the tag as it was written in the document. This does not
127 /// perform tag resolution.
128 StringRef getRawTag() const { return Tag; }
130 /// \brief Get the verbatium tag for a given Node. This performs tag resoluton
131 /// and substitution.
132 std::string getVerbatimTag() const;
134 SMRange getSourceRange() const { return SourceRange; }
135 void setSourceRange(SMRange SR) { SourceRange = SR; }
137 // These functions forward to Document and Scanner.
140 Node *parseBlockNode();
141 BumpPtrAllocator &getAllocator();
142 void setError(const Twine &Message, Token &Location) const;
145 virtual void skip() {}
147 unsigned int getType() const { return TypeID; }
149 void *operator new(size_t Size, BumpPtrAllocator &Alloc,
150 size_t Alignment = 16) noexcept {
151 return Alloc.Allocate(Size, Alignment);
154 void operator delete(void *Ptr, BumpPtrAllocator &Alloc,
155 size_t Size) noexcept {
156 Alloc.Deallocate(Ptr, Size);
160 std::unique_ptr<Document> &Doc;
163 void operator delete(void *) noexcept = delete;
170 /// \brief The tag as typed in the document.
174 /// \brief A null value.
178 class NullNode final : public Node {
179 void anchor() override;
182 NullNode(std::unique_ptr<Document> &D)
183 : Node(NK_Null, D, StringRef(), StringRef()) {}
185 static inline bool classof(const Node *N) { return N->getType() == NK_Null; }
188 /// \brief A scalar node is an opaque datum that can be presented as a
189 /// series of zero or more Unicode scalar values.
193 class ScalarNode final : public Node {
194 void anchor() override;
197 ScalarNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
199 : Node(NK_Scalar, D, Anchor, Tag), Value(Val) {
200 SMLoc Start = SMLoc::getFromPointer(Val.begin());
201 SMLoc End = SMLoc::getFromPointer(Val.end());
202 SourceRange = SMRange(Start, End);
205 // Return Value without any escaping or folding or other fun YAML stuff. This
206 // is the exact bytes that are contained in the file (after conversion to
208 StringRef getRawValue() const { return Value; }
210 /// \brief Gets the value of this node as a StringRef.
212 /// \param Storage is used to store the content of the returned StringRef iff
213 /// it requires any modification from how it appeared in the source.
214 /// This happens with escaped characters and multi-line literals.
215 StringRef getValue(SmallVectorImpl<char> &Storage) const;
217 static inline bool classof(const Node *N) {
218 return N->getType() == NK_Scalar;
224 StringRef unescapeDoubleQuoted(StringRef UnquotedValue,
225 StringRef::size_type Start,
226 SmallVectorImpl<char> &Storage) const;
229 /// \brief A block scalar node is an opaque datum that can be presented as a
230 /// series of zero or more Unicode scalar values.
236 class BlockScalarNode final : public Node {
237 void anchor() override;
240 BlockScalarNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
241 StringRef Value, StringRef RawVal)
242 : Node(NK_BlockScalar, D, Anchor, Tag), Value(Value) {
243 SMLoc Start = SMLoc::getFromPointer(RawVal.begin());
244 SMLoc End = SMLoc::getFromPointer(RawVal.end());
245 SourceRange = SMRange(Start, End);
248 /// \brief Gets the value of this node as a StringRef.
249 StringRef getValue() const { return Value; }
251 static inline bool classof(const Node *N) {
252 return N->getType() == NK_BlockScalar;
259 /// \brief A key and value pair. While not technically a Node under the YAML
260 /// representation graph, it is easier to treat them this way.
262 /// TODO: Consider making this not a child of Node.
266 class KeyValueNode final : public Node {
267 void anchor() override;
270 KeyValueNode(std::unique_ptr<Document> &D)
271 : Node(NK_KeyValue, D, StringRef(), StringRef()), Key(nullptr),
274 /// \brief Parse and return the key.
276 /// This may be called multiple times.
278 /// \returns The key, or nullptr if failed() == true.
281 /// \brief Parse and return the value.
283 /// This may be called multiple times.
285 /// \returns The value, or nullptr if failed() == true.
288 void skip() override {
290 if (Node *Val = getValue())
294 static inline bool classof(const Node *N) {
295 return N->getType() == NK_KeyValue;
303 /// \brief This is an iterator abstraction over YAML collections shared by both
304 /// sequences and maps.
306 /// BaseT must have a ValueT* member named CurrentEntry and a member function
307 /// increment() which must set CurrentEntry to 0 to create an end iterator.
308 template <class BaseT, class ValueT>
309 class basic_collection_iterator
310 : public std::iterator<std::input_iterator_tag, ValueT> {
312 basic_collection_iterator() : Base(nullptr) {}
313 basic_collection_iterator(BaseT *B) : Base(B) {}
315 ValueT *operator->() const {
316 assert(Base && Base->CurrentEntry && "Attempted to access end iterator!");
317 return Base->CurrentEntry;
320 ValueT &operator*() const {
321 assert(Base && Base->CurrentEntry &&
322 "Attempted to dereference end iterator!");
323 return *Base->CurrentEntry;
326 operator ValueT *() const {
327 assert(Base && Base->CurrentEntry && "Attempted to access end iterator!");
328 return Base->CurrentEntry;
331 /// Note on EqualityComparable:
333 /// The iterator is not re-entrant,
334 /// it is meant to be used for parsing YAML on-demand
335 /// Once iteration started - it can point only to one entry at a time
336 /// hence Base.CurrentEntry and Other.Base.CurrentEntry are equal
337 /// iff Base and Other.Base are equal.
338 bool operator==(const basic_collection_iterator &Other) const {
339 if (Base && (Base == Other.Base)) {
340 assert((Base->CurrentEntry == Other.Base->CurrentEntry)
341 && "Equal Bases expected to point to equal Entries");
344 return Base == Other.Base;
347 bool operator!=(const basic_collection_iterator &Other) const {
348 return !(Base == Other.Base);
351 basic_collection_iterator &operator++() {
352 assert(Base && "Attempted to advance iterator past end!");
354 // Create an end iterator.
355 if (!Base->CurrentEntry)
364 // The following two templates are used for both MappingNode and Sequence Node.
365 template <class CollectionType>
366 typename CollectionType::iterator begin(CollectionType &C) {
367 assert(C.IsAtBeginning && "You may only iterate over a collection once!");
368 C.IsAtBeginning = false;
369 typename CollectionType::iterator ret(&C);
374 template <class CollectionType> void skip(CollectionType &C) {
375 // TODO: support skipping from the middle of a parsed collection ;/
376 assert((C.IsAtBeginning || C.IsAtEnd) && "Cannot skip mid parse!");
378 for (typename CollectionType::iterator i = begin(C), e = C.end(); i != e;
383 /// \brief Represents a YAML map created from either a block map for a flow map.
385 /// This parses the YAML stream as increment() is called.
390 class MappingNode final : public Node {
391 void anchor() override;
397 MT_Inline ///< An inline mapping node is used for "[key: value]".
400 MappingNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
402 : Node(NK_Mapping, D, Anchor, Tag), Type(MT), IsAtBeginning(true),
403 IsAtEnd(false), CurrentEntry(nullptr) {}
405 friend class basic_collection_iterator<MappingNode, KeyValueNode>;
406 typedef basic_collection_iterator<MappingNode, KeyValueNode> iterator;
407 template <class T> friend typename T::iterator yaml::begin(T &);
408 template <class T> friend void yaml::skip(T &);
410 iterator begin() { return yaml::begin(*this); }
412 iterator end() { return iterator(); }
414 void skip() override { yaml::skip(*this); }
416 static inline bool classof(const Node *N) {
417 return N->getType() == NK_Mapping;
424 KeyValueNode *CurrentEntry;
429 /// \brief Represents a YAML sequence created from either a block sequence for a
432 /// This parses the YAML stream as increment() is called.
437 class SequenceNode final : public Node {
438 void anchor() override;
450 // As a BlockMappingEntry and BlockEnd are not created in this case.
454 SequenceNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
456 : Node(NK_Sequence, D, Anchor, Tag), SeqType(ST), IsAtBeginning(true),
458 WasPreviousTokenFlowEntry(true), // Start with an imaginary ','.
459 CurrentEntry(nullptr) {}
461 friend class basic_collection_iterator<SequenceNode, Node>;
462 typedef basic_collection_iterator<SequenceNode, Node> iterator;
463 template <class T> friend typename T::iterator yaml::begin(T &);
464 template <class T> friend void yaml::skip(T &);
468 iterator begin() { return yaml::begin(*this); }
470 iterator end() { return iterator(); }
472 void skip() override { yaml::skip(*this); }
474 static inline bool classof(const Node *N) {
475 return N->getType() == NK_Sequence;
479 SequenceType SeqType;
482 bool WasPreviousTokenFlowEntry;
486 /// \brief Represents an alias to a Node with an anchor.
490 class AliasNode final : public Node {
491 void anchor() override;
494 AliasNode(std::unique_ptr<Document> &D, StringRef Val)
495 : Node(NK_Alias, D, StringRef(), StringRef()), Name(Val) {}
497 StringRef getName() const { return Name; }
500 static inline bool classof(const Node *N) { return N->getType() == NK_Alias; }
506 /// \brief A YAML Stream is a sequence of Documents. A document contains a root
510 /// \brief Root for parsing a node. Returns a single node.
511 Node *parseBlockNode();
513 Document(Stream &ParentStream);
515 /// \brief Finish parsing the current document and return true if there are
516 /// more. Return false otherwise.
519 /// \brief Parse and return the root level node.
523 return Root = parseBlockNode();
526 const std::map<StringRef, StringRef> &getTagMap() const { return TagMap; }
530 friend class document_iterator;
532 /// \brief Stream to read tokens from.
535 /// \brief Used to allocate nodes to. All are destroyed without calling their
536 /// destructor when the document is destroyed.
537 BumpPtrAllocator NodeAllocator;
539 /// \brief The root node. Used to support skipping a partially parsed
543 /// \brief Maps tag prefixes to their expansion.
544 std::map<StringRef, StringRef> TagMap;
548 void setError(const Twine &Message, Token &Location) const;
551 /// \brief Parse %BLAH directives and return true if any were encountered.
552 bool parseDirectives();
554 /// \brief Parse %YAML
555 void parseYAMLDirective();
557 /// \brief Parse %TAG
558 void parseTAGDirective();
560 /// \brief Consume the next token and error if it is not \a TK.
561 bool expectToken(int TK);
564 /// \brief Iterator abstraction for Documents over a Stream.
565 class document_iterator {
567 document_iterator() : Doc(nullptr) {}
568 document_iterator(std::unique_ptr<Document> &D) : Doc(&D) {}
570 bool operator==(const document_iterator &Other) {
571 if (isAtEnd() || Other.isAtEnd())
572 return isAtEnd() && Other.isAtEnd();
574 return Doc == Other.Doc;
576 bool operator!=(const document_iterator &Other) { return !(*this == Other); }
578 document_iterator operator++() {
579 assert(Doc && "incrementing iterator past the end.");
580 if (!(*Doc)->skip()) {
583 Stream &S = (*Doc)->stream;
584 Doc->reset(new Document(S));
589 Document &operator*() { return *Doc->get(); }
591 std::unique_ptr<Document> &operator->() { return *Doc; }
594 bool isAtEnd() const { return !Doc || !*Doc; }
596 std::unique_ptr<Document> *Doc;
599 } // End namespace yaml.
601 } // End namespace llvm.