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"
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 &);
81 Stream(MemoryBufferRef InputBuffer, SourceMgr &);
84 document_iterator begin();
85 document_iterator end();
93 void printError(Node *N, const Twine &Msg);
96 std::unique_ptr<Scanner> scanner;
97 std::unique_ptr<Document> CurrentDoc;
99 friend class Document;
102 /// \brief Abstract base class for all Nodes.
104 virtual void anchor();
116 Node(unsigned int Type, std::unique_ptr<Document> &, StringRef Anchor,
119 /// \brief Get the value of the anchor attached to this node. If it does not
120 /// have one, getAnchor().size() will be 0.
121 StringRef getAnchor() const { return Anchor; }
123 /// \brief Get the tag as it was written in the document. This does not
124 /// perform tag resolution.
125 StringRef getRawTag() const { return Tag; }
127 /// \brief Get the verbatium tag for a given Node. This performs tag resoluton
128 /// and substitution.
129 std::string getVerbatimTag() const;
131 SMRange getSourceRange() const { return SourceRange; }
132 void setSourceRange(SMRange SR) { SourceRange = SR; }
134 // These functions forward to Document and Scanner.
137 Node *parseBlockNode();
138 BumpPtrAllocator &getAllocator();
139 void setError(const Twine &Message, Token &Location) const;
142 virtual void skip() {}
144 unsigned int getType() const { return TypeID; }
146 void *operator new(size_t Size, BumpPtrAllocator &Alloc,
147 size_t Alignment = 16) throw() {
148 return Alloc.Allocate(Size, Alignment);
151 void operator delete(void *Ptr, BumpPtrAllocator &Alloc, size_t Size) throw() {
152 Alloc.Deallocate(Ptr, Size);
156 std::unique_ptr<Document> &Doc;
159 void operator delete(void *) throw() {}
166 /// \brief The tag as typed in the document.
170 /// \brief A null value.
174 class NullNode : public Node {
175 void anchor() override;
178 NullNode(std::unique_ptr<Document> &D)
179 : Node(NK_Null, D, StringRef(), StringRef()) {}
181 static inline bool classof(const Node *N) { return N->getType() == NK_Null; }
184 /// \brief A scalar node is an opaque datum that can be presented as a
185 /// series of zero or more Unicode scalar values.
189 class ScalarNode : public Node {
190 void anchor() override;
193 ScalarNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
195 : Node(NK_Scalar, D, Anchor, Tag), Value(Val) {
196 SMLoc Start = SMLoc::getFromPointer(Val.begin());
197 SMLoc End = SMLoc::getFromPointer(Val.end());
198 SourceRange = SMRange(Start, End);
201 // Return Value without any escaping or folding or other fun YAML stuff. This
202 // is the exact bytes that are contained in the file (after conversion to
204 StringRef getRawValue() const { return Value; }
206 /// \brief Gets the value of this node as a StringRef.
208 /// \param Storage is used to store the content of the returned StringRef iff
209 /// it requires any modification from how it appeared in the source.
210 /// This happens with escaped characters and multi-line literals.
211 StringRef getValue(SmallVectorImpl<char> &Storage) const;
213 static inline bool classof(const Node *N) {
214 return N->getType() == NK_Scalar;
220 StringRef unescapeDoubleQuoted(StringRef UnquotedValue,
221 StringRef::size_type Start,
222 SmallVectorImpl<char> &Storage) const;
225 /// \brief A key and value pair. While not technically a Node under the YAML
226 /// representation graph, it is easier to treat them this way.
228 /// TODO: Consider making this not a child of Node.
232 class KeyValueNode : public Node {
233 void anchor() override;
236 KeyValueNode(std::unique_ptr<Document> &D)
237 : Node(NK_KeyValue, D, StringRef(), StringRef()), Key(nullptr),
240 /// \brief Parse and return the key.
242 /// This may be called multiple times.
244 /// \returns The key, or nullptr if failed() == true.
247 /// \brief Parse and return the value.
249 /// This may be called multiple times.
251 /// \returns The value, or nullptr if failed() == true.
254 void skip() override {
259 static inline bool classof(const Node *N) {
260 return N->getType() == NK_KeyValue;
268 /// \brief This is an iterator abstraction over YAML collections shared by both
269 /// sequences and maps.
271 /// BaseT must have a ValueT* member named CurrentEntry and a member function
272 /// increment() which must set CurrentEntry to 0 to create an end iterator.
273 template <class BaseT, class ValueT>
274 class basic_collection_iterator
275 : public std::iterator<std::forward_iterator_tag, ValueT> {
277 basic_collection_iterator() : Base(nullptr) {}
278 basic_collection_iterator(BaseT *B) : Base(B) {}
280 ValueT *operator->() const {
281 assert(Base && Base->CurrentEntry && "Attempted to access end iterator!");
282 return Base->CurrentEntry;
285 ValueT &operator*() const {
286 assert(Base && Base->CurrentEntry &&
287 "Attempted to dereference end iterator!");
288 return *Base->CurrentEntry;
291 operator ValueT *() const {
292 assert(Base && Base->CurrentEntry && "Attempted to access end iterator!");
293 return Base->CurrentEntry;
296 bool operator!=(const basic_collection_iterator &Other) const {
297 if (Base != Other.Base)
299 return (Base && Other.Base) &&
300 Base->CurrentEntry != Other.Base->CurrentEntry;
303 basic_collection_iterator &operator++() {
304 assert(Base && "Attempted to advance iterator past end!");
306 // Create an end iterator.
307 if (!Base->CurrentEntry)
316 // The following two templates are used for both MappingNode and Sequence Node.
317 template <class CollectionType>
318 typename CollectionType::iterator begin(CollectionType &C) {
319 assert(C.IsAtBeginning && "You may only iterate over a collection once!");
320 C.IsAtBeginning = false;
321 typename CollectionType::iterator ret(&C);
326 template <class CollectionType> void skip(CollectionType &C) {
327 // TODO: support skipping from the middle of a parsed collection ;/
328 assert((C.IsAtBeginning || C.IsAtEnd) && "Cannot skip mid parse!");
330 for (typename CollectionType::iterator i = begin(C), e = C.end(); i != e;
335 /// \brief Represents a YAML map created from either a block map for a flow map.
337 /// This parses the YAML stream as increment() is called.
342 class MappingNode : public Node {
343 void anchor() override;
349 MT_Inline ///< An inline mapping node is used for "[key: value]".
352 MappingNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
354 : Node(NK_Mapping, D, Anchor, Tag), Type(MT), IsAtBeginning(true),
355 IsAtEnd(false), CurrentEntry(nullptr) {}
357 friend class basic_collection_iterator<MappingNode, KeyValueNode>;
358 typedef basic_collection_iterator<MappingNode, KeyValueNode> iterator;
359 template <class T> friend typename T::iterator yaml::begin(T &);
360 template <class T> friend void yaml::skip(T &);
362 iterator begin() { return yaml::begin(*this); }
364 iterator end() { return iterator(); }
366 void skip() override { yaml::skip(*this); }
368 static inline bool classof(const Node *N) {
369 return N->getType() == NK_Mapping;
376 KeyValueNode *CurrentEntry;
381 /// \brief Represents a YAML sequence created from either a block sequence for a
384 /// This parses the YAML stream as increment() is called.
389 class SequenceNode : public Node {
390 void anchor() override;
402 // As a BlockMappingEntry and BlockEnd are not created in this case.
406 SequenceNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
408 : Node(NK_Sequence, D, Anchor, Tag), SeqType(ST), IsAtBeginning(true),
410 WasPreviousTokenFlowEntry(true), // Start with an imaginary ','.
411 CurrentEntry(nullptr) {}
413 friend class basic_collection_iterator<SequenceNode, Node>;
414 typedef basic_collection_iterator<SequenceNode, Node> iterator;
415 template <class T> friend typename T::iterator yaml::begin(T &);
416 template <class T> friend void yaml::skip(T &);
420 iterator begin() { return yaml::begin(*this); }
422 iterator end() { return iterator(); }
424 void skip() override { yaml::skip(*this); }
426 static inline bool classof(const Node *N) {
427 return N->getType() == NK_Sequence;
431 SequenceType SeqType;
434 bool WasPreviousTokenFlowEntry;
438 /// \brief Represents an alias to a Node with an anchor.
442 class AliasNode : public Node {
443 void anchor() override;
446 AliasNode(std::unique_ptr<Document> &D, StringRef Val)
447 : Node(NK_Alias, D, StringRef(), StringRef()), Name(Val) {}
449 StringRef getName() const { return Name; }
452 static inline bool classof(const Node *N) { return N->getType() == NK_Alias; }
458 /// \brief A YAML Stream is a sequence of Documents. A document contains a root
462 /// \brief Root for parsing a node. Returns a single node.
463 Node *parseBlockNode();
465 Document(Stream &ParentStream);
467 /// \brief Finish parsing the current document and return true if there are
468 /// more. Return false otherwise.
471 /// \brief Parse and return the root level node.
475 return Root = parseBlockNode();
478 const std::map<StringRef, StringRef> &getTagMap() const { return TagMap; }
482 friend class document_iterator;
484 /// \brief Stream to read tokens from.
487 /// \brief Used to allocate nodes to. All are destroyed without calling their
488 /// destructor when the document is destroyed.
489 BumpPtrAllocator NodeAllocator;
491 /// \brief The root node. Used to support skipping a partially parsed
495 /// \brief Maps tag prefixes to their expansion.
496 std::map<StringRef, StringRef> TagMap;
500 void setError(const Twine &Message, Token &Location) const;
503 /// \brief Parse %BLAH directives and return true if any were encountered.
504 bool parseDirectives();
506 /// \brief Parse %YAML
507 void parseYAMLDirective();
509 /// \brief Parse %TAG
510 void parseTAGDirective();
512 /// \brief Consume the next token and error if it is not \a TK.
513 bool expectToken(int TK);
516 /// \brief Iterator abstraction for Documents over a Stream.
517 class document_iterator {
519 document_iterator() : Doc(nullptr) {}
520 document_iterator(std::unique_ptr<Document> &D) : Doc(&D) {}
522 bool operator==(const document_iterator &Other) {
523 if (isAtEnd() || Other.isAtEnd())
524 return isAtEnd() && Other.isAtEnd();
526 return Doc == Other.Doc;
528 bool operator!=(const document_iterator &Other) { return !(*this == Other); }
530 document_iterator operator++() {
531 assert(Doc && "incrementing iterator past the end.");
532 if (!(*Doc)->skip()) {
535 Stream &S = (*Doc)->stream;
536 Doc->reset(new Document(S));
541 Document &operator*() { return *Doc->get(); }
543 std::unique_ptr<Document> &operator->() { return *Doc; }
546 bool isAtEnd() const { return !Doc || !*Doc; }
548 std::unique_ptr<Document> *Doc;
551 } // End namespace yaml.
553 } // End namespace llvm.