| OLD | NEW |
|---|
| (Empty) |
| 1 import 'package:analyzer/src/generated/ast.dart'; | |
| 2 import 'package:analyzer/src/generated/element.dart'; | |
| 3 import 'package:analyzer/src/generated/scanner.dart'; | |
| 4 import 'package:analyzer/src/generated/utilities_dart.dart'; | |
| 5 | |
| 6 int _computeArgIndex(AstNode containingNode, Object entity) { | |
| 7 var argList = containingNode; | |
| 8 if (argList is ArgumentList) { | |
| 9 NodeList<Expression> args = argList.arguments; | |
| 10 for (int index = 0; index < args.length; ++index) { | |
| 11 if (entity == args[index]) { | |
| 12 return index; | |
| 13 } | |
| 14 } | |
| 15 if (args.isEmpty) { | |
| 16 return 0; | |
| 17 } | |
| 18 } | |
| 19 return null; | |
| 20 } | |
| 21 | |
| 22 /** | |
| 23 * A CompletionTarget represents an edge in the parse tree which connects an | |
| 24 * AST node (the [containingNode] of the completion) to one of its children | |
| 25 * (the [entity], which represents the place in the parse tree where the newly | |
| 26 * completed text will be inserted). | |
| 27 * | |
| 28 * To illustrate, consider the following snippet of code, and its associated | |
| 29 * parse tree. (T's represent tokens, N's represent AST nodes. Some trivial | |
| 30 * AST nodes are not shown). | |
| 31 * | |
| 32 * ___N (function declaration) | |
| 33 * / \ | |
| 34 * / __N_______ (function body) | |
| 35 * / / |a \ | |
| 36 * / / N______ \ (statement) | |
| 37 * / / / \ \ | |
| 38 * / / N____ \ \ (assignment expression) | |
| 39 * | / /| \ \ \ | |
| 40 * . | / | _N___ \ | ("as" expression) | |
| 41 * . | | | / | \c | | | |
| 42 * . | N | N |b N | | (simple identifiers) | |
| 43 * | | | | | | | | | |
| 44 * T T T T T T T T | |
| 45 * m() { foo = bar as Baz; } | |
| 46 * | |
| 47 * The Completion target is usually placed as high in the tree as possible so | |
| 48 * that we can produce the most meaningful completions with minimal effort. | |
| 49 * For instance, if the cursor is inside the identifier "foo", the completion | |
| 50 * target will be the edge marked "a", so that we will produce all completions | |
| 51 * that could possibly start a statement, even those which would conflict with | |
| 52 * the current parse (such as the keyword "for", which begins a "for" | |
| 53 * statement). As a consequence of this, the [entity] will usually not be the | |
| 54 * first child of the [containingNode] node. | |
| 55 * | |
| 56 * Note that the [containingNode] is always an AST node, but the [entity] may | |
| 57 * not be. For instance, if the cursor is inside the keyword "as", the | |
| 58 * completion target will be the edge marked "b", so the [entity] is the token | |
| 59 * "as". | |
| 60 * | |
| 61 * If the cursor is between tokens, the completion target is usually associated | |
| 62 * with the token that follows the cursor (since that's the token that will be | |
| 63 * displaced when the new text is inserted). For example, if the cursor is | |
| 64 * after the "{" character, then the completion target will be the edge marked | |
| 65 * "a", just as it is if the cursor is inside the identifier "foo". However | |
| 66 * there is one exception: if the cursor is at the rightmost edge of a keyword | |
| 67 * or identifier, then the completion target is associated with that token, | |
| 68 * since any further letters typed will change the meaning of the identifier or | |
| 69 * keyword, rather than creating a new token. So for instance, if the cursor | |
| 70 * is just after the "s" of "as", the completion target will be the edge marked | |
| 71 * "b", but if the cursor target is after the first space following "as", then | |
| 72 * the completion target will be the edge marked "c". | |
| 73 * | |
| 74 * If the file is empty, or the cursor is after all the text in the file, then | |
| 75 * there may be no edge in the parse tree which is appropriate to act as the | |
| 76 * completion target; in this case, [entity] is set to null and | |
| 77 * [containingNode] is set to the CompilationUnit. | |
| 78 */ | |
| 79 class CompletionTarget { | |
| 80 /** | |
| 81 * The context in which the completion is occurring. This is the AST node | |
| 82 * which is a direct parent of [entity]. | |
| 83 */ | |
| 84 final AstNode containingNode; | |
| 85 | |
| 86 /** | |
| 87 * The entity which the completed text will replace (or which will be | |
| 88 * displaced once the completed text is inserted). This may be an AstNode or | |
| 89 * a Token, or it may be null if the cursor is after all tokens in the file. | |
| 90 * | |
| 91 * Usually, the entity won't be the first child of the [containingNode] (this | |
| 92 * is a consequence of placing the completion target as high in the tree as | |
| 93 * possible). However, there is one exception: when the cursor is inside of | |
| 94 * a multi-character token which is not a keyword or identifier (e.g. a | |
| 95 * comment, or a token like "+=", the entity will be always be the token. | |
| 96 */ | |
| 97 final Object entity; | |
| 98 | |
| 99 /** | |
| 100 * If the target is an argument in an [ArgumentList], then this is the index | |
| 101 * of the argument in the list, otherwise this is `null`. | |
| 102 */ | |
| 103 final int argIndex; | |
| 104 | |
| 105 /** | |
| 106 * Compute the appropriate [CompletionTarget] for the given [offset] within | |
| 107 * the [compilationUnit]. | |
| 108 */ | |
| 109 factory CompletionTarget.forOffset( | |
| 110 CompilationUnit compilationUnit, int offset) { | |
| 111 // The precise algorithm is as follows. We perform a depth-first search of | |
| 112 // all edges in the parse tree (both those that point to AST nodes and | |
| 113 // those that point to tokens), visiting parents before children. The | |
| 114 // first edge which points to an entity satisfying either _isCandidateToken | |
| 115 // or _isCandidateNode is the completion target. If no edge is found that | |
| 116 // satisfies these two predicates, then we set the completion target entity | |
| 117 // to null and the containingNode to the compilationUnit. | |
| 118 // | |
| 119 // Note that if a token is not a candidate target, then none of the tokens | |
| 120 // that precede it are candidate targets either. Therefore any entity | |
| 121 // whose last token is not a candidate target can be skipped. This lets us | |
| 122 // prune the search to the point where no recursion is necessary; at each | |
| 123 // step in the process we know exactly which child node we need to proceed | |
| 124 // to. | |
| 125 AstNode containingNode = compilationUnit; | |
| 126 outerLoop: while (true) { | |
| 127 if (containingNode is Comment) { | |
| 128 // Comments are handled specially: we descend into any CommentReference | |
| 129 // child node that contains the cursor offset. | |
| 130 Comment comment = containingNode; | |
| 131 for (CommentReference commentReference in comment.references) { | |
| 132 if (commentReference.offset <= offset && | |
| 133 offset <= commentReference.end) { | |
| 134 containingNode = commentReference; | |
| 135 continue outerLoop; | |
| 136 } | |
| 137 } | |
| 138 } | |
| 139 for (var entity in containingNode.childEntities) { | |
| 140 if (entity is Token) { | |
| 141 if (_isCandidateToken(entity, offset)) { | |
| 142 // Target found. | |
| 143 return new CompletionTarget._(containingNode, entity); | |
| 144 } else { | |
| 145 // Since entity is a token, we don't need to look inside it; just | |
| 146 // proceed to the next entity. | |
| 147 continue; | |
| 148 } | |
| 149 } else if (entity is AstNode) { | |
| 150 // If the last token in the node isn't a candidate target, then | |
| 151 // neither the node nor any of its descendants can possibly be the | |
| 152 // completion target, so we can skip the node entirely. | |
| 153 if (!_isCandidateToken(entity.endToken, offset)) { | |
| 154 continue; | |
| 155 } | |
| 156 | |
| 157 // If the node is a candidate target, then we are done. | |
| 158 if (_isCandidateNode(entity, offset)) { | |
| 159 // Check to see if the offset is in a preceeding comment | |
| 160 Token commentToken = _getContainingCommentToken(entity, offset); | |
| 161 if (commentToken != null) { | |
| 162 entity = commentToken; | |
| 163 // If the preceeding comment is dartdoc token then update | |
| 164 // the containing node to be the dartdoc comment | |
| 165 Comment docComment = | |
| 166 _getContainingDocComment(containingNode, commentToken); | |
| 167 if (docComment != null) { | |
| 168 containingNode = docComment; | |
| 169 } | |
| 170 } | |
| 171 return new CompletionTarget._(containingNode, entity); | |
| 172 } | |
| 173 | |
| 174 // Otherwise, the completion target is somewhere inside the entity, | |
| 175 // so we need to jump to the start of the outer loop to examine its | |
| 176 // contents. | |
| 177 containingNode = entity; | |
| 178 continue outerLoop; | |
| 179 } else { | |
| 180 // Unexpected entity found (all entities in a parse tree should be | |
| 181 // AST nodes or tokens). | |
| 182 assert(false); | |
| 183 } | |
| 184 } | |
| 185 | |
| 186 // No completion target found. It should only be possible to reach here | |
| 187 // the first time through the outer loop (since we only jump to the start | |
| 188 // of the outer loop after determining that the completion target is | |
| 189 // inside an entity). We can check that assumption by verifying that | |
| 190 // containingNode is still the compilationUnit. | |
| 191 assert(identical(containingNode, compilationUnit)); | |
| 192 | |
| 193 // Since no completion target was found, we set the completion target | |
| 194 // entity to null and use the compilationUnit as the parent. | |
| 195 return new CompletionTarget._(compilationUnit, null); | |
| 196 } | |
| 197 } | |
| 198 | |
| 199 /** | |
| 200 * Create a [CompletionTarget] holding the given [containingNode] and | |
| 201 * [entity]. | |
| 202 */ | |
| 203 CompletionTarget._(AstNode containingNode, Object entity) | |
| 204 : this.containingNode = containingNode, | |
| 205 this.entity = entity, | |
| 206 this.argIndex = _computeArgIndex(containingNode, entity); | |
| 207 | |
| 208 /** | |
| 209 * Return `true` if the target is a functional argument in an argument list. | |
| 210 * The target [AstNode] hierarchy *must* be resolved for this to work. | |
| 211 */ | |
| 212 bool isFunctionalArgument() { | |
| 213 if (argIndex == null) { | |
| 214 return false; | |
| 215 } | |
| 216 AstNode argList = containingNode; | |
| 217 if (argList is! ArgumentList) { | |
| 218 return false; | |
| 219 } | |
| 220 AstNode parent = argList.parent; | |
| 221 if (parent is InstanceCreationExpression) { | |
| 222 DartType instType = parent.bestType; | |
| 223 if (instType != null) { | |
| 224 Element intTypeElem = instType.element; | |
| 225 if (intTypeElem is ClassElement) { | |
| 226 SimpleIdentifier constructorName = parent.constructorName.name; | |
| 227 ConstructorElement constructor = constructorName != null | |
| 228 ? intTypeElem.getNamedConstructor(constructorName.name) | |
| 229 : intTypeElem.unnamedConstructor; | |
| 230 return constructor != null && | |
| 231 _isFunctionalParameter(constructor.parameters, argIndex); | |
| 232 } | |
| 233 } | |
| 234 } else if (parent is MethodInvocation) { | |
| 235 SimpleIdentifier methodName = parent.methodName; | |
| 236 if (methodName != null) { | |
| 237 Element methodElem = methodName.bestElement; | |
| 238 if (methodElem is MethodElement) { | |
| 239 return _isFunctionalParameter(methodElem.parameters, argIndex); | |
| 240 } else if (methodElem is FunctionElement) { | |
| 241 return _isFunctionalParameter(methodElem.parameters, argIndex); | |
| 242 } | |
| 243 } | |
| 244 } | |
| 245 return false; | |
| 246 } | |
| 247 | |
| 248 /** | |
| 249 * Determine if the offset is contained in a preceeding comment token | |
| 250 * and return that token, otherwise return `null`. | |
| 251 */ | |
| 252 static Token _getContainingCommentToken(AstNode node, int offset) { | |
| 253 if (offset >= node.offset) { | |
| 254 return null; | |
| 255 } | |
| 256 Token token = node.beginToken; | |
| 257 if (token == null) { | |
| 258 return null; | |
| 259 } | |
| 260 token = token.precedingComments; | |
| 261 while (token != null) { | |
| 262 if (offset <= token.offset) { | |
| 263 return null; | |
| 264 } | |
| 265 if (offset <= token.end) { | |
| 266 if (token.type == TokenType.SINGLE_LINE_COMMENT || offset < token.end) { | |
| 267 return token; | |
| 268 } | |
| 269 } | |
| 270 token = token.next; | |
| 271 } | |
| 272 return null; | |
| 273 } | |
| 274 | |
| 275 /** | |
| 276 * Determine if the given token is part of the given node's dart doc. | |
| 277 */ | |
| 278 static Comment _getContainingDocComment(AstNode node, Token token) { | |
| 279 if (node is AnnotatedNode) { | |
| 280 Comment docComment = node.documentationComment; | |
| 281 if (docComment != null && docComment.tokens.contains(token)) { | |
| 282 return docComment; | |
| 283 } | |
| 284 } | |
| 285 return null; | |
| 286 } | |
| 287 | |
| 288 /** | |
| 289 * Determine whether [node] could possibly be the [entity] for a | |
| 290 * [CompletionTarget] associated with the given [offset]. | |
| 291 */ | |
| 292 static bool _isCandidateNode(AstNode node, int offset) { | |
| 293 // If the node's first token is a keyword or identifier, then the node is a | |
| 294 // candidate entity if its first token is. | |
| 295 Token beginToken = node.beginToken; | |
| 296 if (beginToken.type == TokenType.KEYWORD || | |
| 297 beginToken.type == TokenType.IDENTIFIER) { | |
| 298 return _isCandidateToken(beginToken, offset); | |
| 299 } | |
| 300 | |
| 301 // Otherwise, the node is a candidate entity only if the offset is before | |
| 302 // the beginning of the node. This ensures that completions within a token | |
| 303 // (e.g. inside a literal string or inside a comment) are evaluated within | |
| 304 // the context of the token itself. | |
| 305 return offset <= node.offset; | |
| 306 } | |
| 307 | |
| 308 /** | |
| 309 * Determine whether [token] could possibly be the [entity] for a | |
| 310 * [CompletionTarget] associated with the given [offset]. | |
| 311 */ | |
| 312 static bool _isCandidateToken(Token token, int offset) { | |
| 313 // A token is considered a candidate entity if the cursor offset is (a) | |
| 314 // before the start of the token, (b) within the token, (c) at the end of | |
| 315 // the token and the token is a keyword or identifier, or (d) at the | |
| 316 // location of the token and the token is zero length. | |
| 317 if (offset < token.end) { | |
| 318 return true; | |
| 319 } else if (offset == token.end) { | |
| 320 return token.type == TokenType.KEYWORD || | |
| 321 token.type == TokenType.IDENTIFIER || | |
| 322 token.length == 0; | |
| 323 } else if (!token.isSynthetic) { | |
| 324 return false; | |
| 325 } | |
| 326 // If the current token is synthetic, then check the previous token | |
| 327 // because it may have been dropped from the parse tree | |
| 328 Token previous = token.previous; | |
| 329 if (offset < previous.end) { | |
| 330 return true; | |
| 331 } else if (offset == previous.end) { | |
| 332 return token.type == TokenType.KEYWORD || | |
| 333 previous.type == TokenType.IDENTIFIER; | |
| 334 } else { | |
| 335 return false; | |
| 336 } | |
| 337 } | |
| 338 | |
| 339 /** | |
| 340 * Return `true` if the parameter is a functional parameter. | |
| 341 */ | |
| 342 static bool _isFunctionalParameter( | |
| 343 List<ParameterElement> parameters, int paramIndex) { | |
| 344 if (paramIndex < parameters.length) { | |
| 345 ParameterElement param = parameters[paramIndex]; | |
| 346 DartType paramType = param.type; | |
| 347 if (param.parameterKind == ParameterKind.NAMED) { | |
| 348 // TODO(danrubel) handle named parameters | |
| 349 return false; | |
| 350 } else { | |
| 351 return paramType is FunctionType || paramType is FunctionTypeAlias; | |
| 352 } | |
| 353 } | |
| 354 return false; | |
| 355 } | |
| 356 } | |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 1383313002:
Move completion APIs to a private location (Closed)
Base URL: https://github.com/dart-lang/sdk.git@master