Support advanced characters for entity name.

See redhat-developer/vscode-xml#262

Signed-off-by: azerr <[email protected]>

org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/extensions/contentmodel/participants/DTDErrorCode.java

@@ -70,7 +70,7 @@ public enum DTDErrorCode implements IXMLErrorCode {
  QuoteRequiredInSystemID,
  SpaceRequiredAfterSYSTEM,
  dtd_not_found("dtd-not-found");
- 
+
  private final String code;
 
  private DTDErrorCode() {
@@ -132,7 +132,7 @@ public static Range toLSPRange(XMLLocator location, DTDErrorCode code, Object[]
  String attrName = getString(arguments[0]);
  return XMLPositionUtility.selectAttributeValueAt(attrName, offset, document);
  }
- 
+
  case MSG_ELEMENT_WITH_ID_REQUIRED: {
  DOMElement element = document.getDocumentElement();
  if (element != null) {
@@ -160,21 +160,7 @@ public static Range toLSPRange(XMLLocator location, DTDErrorCode code, Object[]
  return XMLPositionUtility.getLastValidDTDDeclParameter(offset, document, true);
  }
  case EntityNotDeclared: {
- try {
- Position position = document.positionAt(offset);
- int line = position.getLine();
- String text = document.lineText(line);
- String name = getString(arguments[0]);
- String subString = "&" + name + ";";
- int start = text.indexOf(subString);
- int end = start + subString.length();
- Position startPosition = new Position(line, start);
- Position endPosition = new Position(line, end);
- return new Range(startPosition, endPosition);
- } catch (BadLocationException e) {
-
- }
-
+ return XMLPositionUtility.selectEntityReference(offset - 1, document);
  }
  case QuoteRequiredInPublicID:
  case QuoteRequiredInSystemID:
@@ -206,7 +192,6 @@ public static Range toLSPRange(XMLLocator location, DTDErrorCode code, Object[]
  try {
  return new Range(new Position(0, 0), document.positionAt(document.getEnd()));
  } catch (BadLocationException e) {
-
  }
  }
  return null;

org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/extensions/entities/participants/EntitiesCompletionParticipant.java

@@ -42,7 +42,7 @@ public class EntitiesCompletionParticipant extends CompletionParticipantAdapter
 
  @Override
  public void onXMLContent(ICompletionRequest request, ICompletionResponse response) throws Exception {
- Range entityRange = XMLPositionUtility.selectEntity(request.getOffset(), request.getXMLDocument());
+ Range entityRange = XMLPositionUtility.selectEntityReference(request.getOffset(), request.getXMLDocument());
  if (entityRange == null) {
  return;
  }

org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/extensions/entities/participants/EntitiesDefinitionParticipant.java

@@ -12,12 +12,8 @@
  */
 package org.eclipse.lemminx.extensions.entities.participants;
 
-import static org.eclipse.lemminx.utils.StringUtils.findEndWord;
-import static org.eclipse.lemminx.utils.StringUtils.findStartWord;
-
 import java.util.Collection;
 import java.util.List;
-import java.util.function.Predicate;
 
 import org.eclipse.lemminx.dom.DOMDocument;
 import org.eclipse.lemminx.dom.DOMDocumentType;
@@ -41,8 +37,6 @@
  */
 public class EntitiesDefinitionParticipant extends AbstractDefinitionParticipant {
 
- private static final Predicate<Character> isValidChar = (c) -> Character.isJavaIdentifierPart(c) || c == '-';
-
  @Override
  protected boolean match(DOMDocument document) {
  return true;
@@ -60,13 +54,18 @@ protected void doFindDefinition(IDefinitionRequest request, List<LocationLink> l
  int offset = request.getOffset();
  String text = document.getText();
 
- int paramStart = findStartWord(text, offset, isValidChar);
- if (paramStart > 0 && text.charAt(paramStart - 1) == '&') {
- int paramEnd = findEndWord(text, offset, isValidChar);
+ int entityReferenceStart = XMLPositionUtility.getEntityReferenceStartOffset(text, offset);
+ if (entityReferenceStart != -1) {
+ // the start offset is the amp character, ignore it
+ entityReferenceStart++;
+ int entityReferenceEnd = XMLPositionUtility.getEntityReferenceEndOffset(text, offset);
  // Find definition of the entity
  // abcd &loc|al; --> in this case search local entity
- String entityName = text.substring(paramStart, paramEnd);
- Range entityRange = XMLPositionUtility.createRange(paramStart, paramEnd, document);
+ if (text.charAt(entityReferenceEnd - 1) == ';') {
+ entityReferenceEnd--;
+ }
+ String entityName = text.substring(entityReferenceStart, entityReferenceEnd);
+ Range entityRange = XMLPositionUtility.createRange(entityReferenceStart, entityReferenceEnd, document);
  searchInInternalEntities(entityName, entityRange, document, locations, cancelChecker);
  searchInExternalEntities(entityName, entityRange, document, locations, request, cancelChecker);
  }

org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/utils/StringUtils.java

@@ -405,15 +405,15 @@ public static String getString(Object obj) {
  * and -1 if no word.
  */
  public static int findStartWord(String text, int offset, Predicate<Character> isValidChar) {
- if (!isValidChar.test(text.charAt(offset))) {
+ if (offset < 0 || offset >= text.length() || !isValidChar.test(text.charAt(offset))) {
  return -1;
  }
  for (int i = offset - 1; i >= 0; i--) {
  if (!isValidChar.test(text.charAt(i))) {
  return i + 1;
  }
  }
- return -1;
+ return 0;
  }
 
  /**
@@ -428,7 +428,7 @@ public static int findStartWord(String text, int offset, Predicate<Character> is
  * and -1 if no word.
  */
  public static int findEndWord(String text, int offset, Predicate<Character> isValidChar) {
- if (!isValidChar.test(text.charAt(offset))) {
+ if (offset < 0 || offset >= text.length() || !isValidChar.test(text.charAt(offset))) {
  return -1;
  }
  for (int i = offset + 1; i < text.length(); i++) {

org.eclipse.lemminx/src/main/java/org/eclipse/lemminx/utils/XMLPositionUtility.java

@@ -13,6 +13,7 @@
 package org.eclipse.lemminx.utils;
 
 import java.util.List;
+import java.util.function.Predicate;
 import java.util.logging.Level;
 import java.util.logging.Logger;
 
@@ -47,6 +48,11 @@ public class XMLPositionUtility {
 
  private static final Logger LOGGER = Logger.getLogger(XMLPositionUtility.class.getName());
 
+ private static final Predicate<Character> ENTITY_NAME_PREDICATE = ch -> {
+ // [_:\w-.\d]*
+ return ch == '_' || ch == ':' || ch == '.' || ch == '-' || Character.isLetterOrDigit(ch);
+ };
+
  private XMLPositionUtility() {
  }
 
@@ -476,33 +482,79 @@ public static Range selectPreviousNodesEndTag(int offset, DOMDocument document)
  // ------------ Entities selection
 
  /**
- * Returns the range of the used entity in a text node (ex : &amp;) and null
- * otherwise.
+ * Returns the range of the entity reference in a text node (ex : &amp;) and
+ * null otherwise.
  * 
  * @param offset the offset
  * @param document the document
- * @return the range of the used entity in a text node (ex : &amp;) and null
- * otherwise.
+ * @return the range of the entity reference in a text node (ex : &amp;) and
+ * null otherwise.
  */
- public static Range selectEntity(int offset, DOMDocument document) {
+ public static Range selectEntityReference(int offset, DOMDocument document) {
  String text = document.getText();
  // Search '&' character on the left of the offset
- int startEntityOffset = offset - 1;
- while (startEntityOffset > -1 && Character.isLetterOrDigit(text.charAt(startEntityOffset))) {
- startEntityOffset--;
- }
- if (startEntityOffset == -1 || text.charAt(startEntityOffset) != '&') {
+ int entityReferenceStart = getEntityReferenceStartOffset(text, offset);
+ if (entityReferenceStart == -1) {
  return null;
  }
  // Search ';' (or character on the right of the offset
- int endEntityOffset = offset;
- while (endEntityOffset < text.length() && Character.isLetterOrDigit(text.charAt(endEntityOffset))) {
- endEntityOffset++;
+ int entityReferenceEnd = getEntityReferenceEndOffset(text, offset);
+ return createRange(entityReferenceStart, entityReferenceEnd, document);
+ }
+
+ /**
+ * Returns the start offset of the entity reference (ex : &am|p;) from the left
+ * of the given offset and -1 if no entity reference.
+ * 
+ * @param text the XML content.
+ * @param offset the offset.
+ * @return the start offset of the entity reference (ex : &am|p;) from the left
+ * of the given offset and -1 if no entity reference.
+ */
+ public static int getEntityReferenceStartOffset(String text, int offset) {
+ // adjust offset to get the left character of the offset
+ offset--;
+ if (offset < 0) {
+ // case where offset is on the first character
+ return -1;
+ }
+ if (text.charAt(offset) == '&') {
+ // case with &|abcd
+ return offset;
+ }
+ if (offset == 0) {
+ // case with a|bcd -> there are no '&'
+ return -1;
+ }
+ int startEntityOffset = StringUtils.findStartWord(text, offset, ENTITY_NAME_PREDICATE);
+ if (startEntityOffset <= 0) {
+ return -1;
+ }
+ // check if the left character is '&'
+ if (text.charAt(startEntityOffset - 1) != '&') {
+ return -1;
+ }
+ return startEntityOffset - 1;
+ }
+
+ /**
+ * Returns the end offset of the entity reference (ex : &am|p;) from the right
+ * of the given offset.
+ * 
+ * @param text the XML content.
+ * @param offset the offset.
+ * @return the end offset of the entity reference (ex : &am|p;) from the right
+ * of the given offset.
+ */
+ public static int getEntityReferenceEndOffset(String text, int offset) {
+ int endEntityOffset = StringUtils.findEndWord(text, offset, ENTITY_NAME_PREDICATE);
+ if (endEntityOffset == -1) {
+ return offset;
  }
- if (endEntityOffset + 1 < text.length() && text.charAt(endEntityOffset) == ';') {
+ if (endEntityOffset < text.length() && text.charAt(endEntityOffset) == ';') {
  endEntityOffset++;
  }
- return createRange(startEntityOffset, endEntityOffset, document);
+ return endEntityOffset;
  }
 
  // ------------ Text selection

org.eclipse.lemminx/src/test/java/org/eclipse/lemminx/extensions/entities/EntitiesCompletionExtensionsTest.java

@@ -32,7 +32,7 @@ public class EntitiesCompletionExtensionsTest {
  public void afterAmp() throws BadLocationException {
  // &|
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -49,7 +49,7 @@ public void afterAmp() throws BadLocationException {
  public void afterCharacter() throws BadLocationException {
  // &m|
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -66,7 +66,7 @@ public void afterCharacter() throws BadLocationException {
  public void inside() throws BadLocationException {
  // &m|dblablabla
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -78,11 +78,29 @@ public void inside() throws BadLocationException {
  c("&mdash;", "&mdash;", r(5, 2, 5, 14), "&mdash;"));
  }
 
+ @Test
+ public void underscoreEntityName() throws BadLocationException {
+ // &foo_b|
+ String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
+ " <!ENTITY foo_bar \"&#x2014;\">\r\n" + //
+ " <!ENTITY foo_baz \"&#x2014;\">\r\n" + //
+ "]>\r\n" + //
+ "<root>\r\n" + //
+ " &foo_b|\r\n" + // <- here completion shows mdash entity
+ "</root>";
+ testCompletionFor(xml, 2 + //
+ 2 /* CDATA and Comments */ + //
+ PredefinedEntity.values().length /* predefined entities */,
+ c("&foo_bar;", "&foo_bar;", r(6, 2, 6, 8), "&foo_bar;"), //
+ c("&foo_baz;", "&foo_baz;", r(6, 2, 6, 8), "&foo_baz;"));
+ }
+
  @Test
  public void insideWithAmp() throws BadLocationException {
  // &m|d;blablabla
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -97,7 +115,7 @@ public void insideWithAmp() throws BadLocationException {
  @Test
  public void none() throws BadLocationException {
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //

org.eclipse.lemminx/src/test/java/org/eclipse/lemminx/extensions/entities/EntitiesDefinitionExtensionsTest.java

@@ -30,7 +30,7 @@ public class EntitiesDefinitionExtensionsTest {
  @Test
  public void local() throws BadLocationException {
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -42,7 +42,7 @@ public void local() throws BadLocationException {
  @Test
  public void beforeAmp() throws BadLocationException {
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -54,7 +54,7 @@ public void beforeAmp() throws BadLocationException {
  @Test
  public void unknownEntity() throws BadLocationException {
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -66,7 +66,7 @@ public void unknownEntity() throws BadLocationException {
  @Test
  public void insideText() throws BadLocationException {
  String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
- "<!DOCTYPE article [\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
  " <!ENTITY mdash \"&#x2014;\">\r\n" + //
  "]>\r\n" + //
  "<root>\r\n" + //
@@ -75,6 +75,19 @@ public void insideText() throws BadLocationException {
  testDefinitionFor(xml, "test.xml", ll("test.xml", r(5, 7, 5, 12), r(2, 11, 2, 16)));
  }
 
+ @Test
+ public void underscoreEntityName() throws BadLocationException {
+ String xml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n" + //
+ "<!DOCTYPE root [\r\n" + //
+ " <!ENTITY foo_bar \"&#x2014;\">\r\n" + //
+ " <!ENTITY foo_baz \"&#x2014;\">\r\n" + //
+ "]>\r\n" + //
+ "<root>\r\n" + //
+ " &foo_b|ar;efgh\r\n" + // <- here definition for foo_bar
+ "</root>";
+ testDefinitionFor(xml, "test.xml", ll("test.xml", r(6, 2, 6, 9), r(2, 11, 2, 18)));
+ }
+
  // Test for external entities
 
  public void external() throws BadLocationException, MalformedURIException {