View Javadoc
1   ///////////////////////////////////////////////////////////////////////////////////////////////
2   // checkstyle: Checks Java source code and other text files for adherence to a set of rules.
3   // Copyright (C) 2001-2026 the original author or authors.
4   //
5   // This library is free software; you can redistribute it and/or
6   // modify it under the terms of the GNU Lesser General Public
7   // License as published by the Free Software Foundation; either
8   // version 2.1 of the License, or (at your option) any later version.
9   //
10  // This library is distributed in the hope that it will be useful,
11  // but WITHOUT ANY WARRANTY; without even the implied warranty of
12  // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13  // Lesser General Public License for more details.
14  //
15  // You should have received a copy of the GNU Lesser General Public
16  // License along with this library; if not, write to the Free Software
17  // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
18  ///////////////////////////////////////////////////////////////////////////////////////////////
19  
20  package com.puppycrawl.tools.checkstyle.checks.javadoc;
21  
22  import java.util.ArrayList;
23  import java.util.List;
24  import java.util.Optional;
25  import java.util.function.Function;
26  import java.util.regex.Pattern;
27  import java.util.stream.Stream;
28  
29  import com.puppycrawl.tools.checkstyle.FileStatefulCheck;
30  import com.puppycrawl.tools.checkstyle.api.DetailNode;
31  import com.puppycrawl.tools.checkstyle.api.JavadocCommentsTokenTypes;
32  import com.puppycrawl.tools.checkstyle.utils.CommonUtil;
33  import com.puppycrawl.tools.checkstyle.utils.JavadocUtil;
34  
35  /**
36   * <div>
37   * Checks that
38   * <a href="https://www.oracle.com/technical-resources/articles/java/javadoc-tool.html#firstsentence">
39   * Javadoc summary sentence</a> does not contain phrases that are not recommended to use.
40   * Summaries that contain only the {@code {@inheritDoc}} tag are skipped.
41   * Summaries that contain a non-empty {@code {@return}} are allowed.
42   * Check also violate Javadoc that does not contain first sentence, though with {@code {@return}} a
43   * period is not required as the Javadoc tool adds it.
44   * </div>
45   *
46   * <p>
47   * Note: For defining a summary, both the first sentence and the @summary tag approaches
48   * are supported.
49   * </p>
50   *
51   * @since 6.0
52   */
53  @FileStatefulCheck
54  public class SummaryJavadocCheck extends AbstractJavadocCheck {
55  
56      /**
57       * A key is pointing to the warning message text in "messages.properties"
58       * file.
59       */
60      public static final String MSG_SUMMARY_FIRST_SENTENCE = "summary.first.sentence";
61  
62      /**
63       * A key is pointing to the warning message text in "messages.properties"
64       * file.
65       */
66      public static final String MSG_SUMMARY_JAVADOC = "summary.javaDoc";
67  
68      /**
69       * A key is pointing to the warning message text in "messages.properties"
70       * file.
71       */
72      public static final String MSG_SUMMARY_JAVADOC_MISSING = "summary.javaDoc.missing";
73  
74      /**
75       * A key is pointing to the warning message text in "messages.properties" file.
76       */
77      public static final String MSG_SUMMARY_MISSING_PERIOD = "summary.javaDoc.missing.period";
78  
79      /**
80       * This regexp is used to convert multiline javadoc to single-line without stars.
81       */
82      private static final Pattern JAVADOC_MULTILINE_TO_SINGLELINE_PATTERN =
83              Pattern.compile("\n[ \\t]+(\\*)|^[ \\t]+(\\*)");
84  
85      /**
86       * This regexp is used to remove html tags, whitespace, and asterisks from a string.
87       */
88      private static final Pattern HTML_ELEMENTS =
89              Pattern.compile("<[^>]*>");
90  
91      /** Default period literal. */
92      private static final String DEFAULT_PERIOD = ".";
93  
94      /**
95       * Specify the regexp for forbidden summary fragments.
96       */
97      private Pattern forbiddenSummaryFragments = CommonUtil.createPattern("^$");
98  
99      /**
100      * Specify the period symbol. Used to check the first sentence ends with a period. Periods that
101      * are not followed by a whitespace character are ignored (eg. the period in v1.0). Because some
102      * periods include whitespace built into the character, if this is set to a non-default value
103      * any period will end the sentence, whether it is followed by whitespace or not.
104      */
105     private String period = DEFAULT_PERIOD;
106 
107     /**
108      * Whether to validate untagged summary text in Javadoc.
109      */
110     private boolean shouldValidateUntaggedSummary = true;
111 
112     /**
113      * Setter to specify the regexp for forbidden summary fragments.
114      *
115      * @param pattern a pattern.
116      * @since 6.0
117      */
118     public void setForbiddenSummaryFragments(Pattern pattern) {
119         forbiddenSummaryFragments = pattern;
120     }
121 
122     /**
123      * Setter to specify the period symbol. Used to check the first sentence ends with a period.
124      * Periods that are not followed by a whitespace character are ignored (eg. the period in v1.0).
125      * Because some periods include whitespace built into the character, if this is set to a
126      * non-default value any period will end the sentence, whether it is followed by whitespace or
127      * not.
128      *
129      * @param period period's value.
130      * @since 6.2
131      */
132     public void setPeriod(String period) {
133         this.period = period;
134     }
135 
136     @Override
137     public int[] getDefaultJavadocTokens() {
138         return new int[] {
139             JavadocCommentsTokenTypes.JAVADOC_CONTENT,
140             JavadocCommentsTokenTypes.SUMMARY_INLINE_TAG,
141             JavadocCommentsTokenTypes.RETURN_INLINE_TAG,
142         };
143     }
144 
145     @Override
146     public int[] getRequiredJavadocTokens() {
147         return getAcceptableJavadocTokens();
148     }
149 
150     @Override
151     public void visitJavadocToken(DetailNode ast) {
152         if (isSummaryTag(ast) && isDefinedFirst(ast.getParent())) {
153             shouldValidateUntaggedSummary = false;
154             validateSummaryTag(ast);
155         }
156         else if (isInlineReturnTag(ast)) {
157             shouldValidateUntaggedSummary = false;
158             validateInlineReturnTag(ast);
159         }
160     }
161 
162     @Override
163     public void leaveJavadocToken(DetailNode ast) {
164         if (ast.getType() == JavadocCommentsTokenTypes.JAVADOC_CONTENT) {
165             if (shouldValidateUntaggedSummary && !startsWithInheritDoc(ast)) {
166                 validateUntaggedSummary(ast);
167             }
168             shouldValidateUntaggedSummary = true;
169         }
170     }
171 
172     /**
173      * Checks the javadoc text for {@code period} at end and forbidden fragments.
174      *
175      * @param ast the javadoc text node
176      */
177     private void validateUntaggedSummary(DetailNode ast) {
178         final String summaryDoc = getSummarySentence(ast);
179         if (summaryDoc.isEmpty()) {
180             log(ast.getLineNumber(), MSG_SUMMARY_JAVADOC_MISSING);
181         }
182         else if (!period.isEmpty()) {
183             if (summaryDoc.contains(period)) {
184                 final Optional<String> firstSentence = getFirstSentence(ast, period);
185 
186                 if (firstSentence.isPresent()) {
187                     if (containsForbiddenFragment(firstSentence.get())) {
188                         log(ast.getLineNumber(), MSG_SUMMARY_JAVADOC);
189                     }
190                 }
191                 else {
192                     log(ast.getLineNumber(), MSG_SUMMARY_FIRST_SENTENCE);
193                 }
194             }
195             else {
196                 log(ast.getLineNumber(), MSG_SUMMARY_FIRST_SENTENCE);
197             }
198         }
199     }
200 
201     /**
202      * Whether the {@code {@summary}} tag is defined first in the javadoc.
203      *
204      * @param inlineTagNode node of type {@link JavadocCommentsTokenTypes#JAVADOC_INLINE_TAG}
205      * @return {@code true} if the {@code {@summary}} tag is defined first in the javadoc
206      */
207     private static boolean isDefinedFirst(DetailNode inlineTagNode) {
208         boolean isDefinedFirst = true;
209         DetailNode currentAst = inlineTagNode.getPreviousSibling();
210         while (currentAst != null && isDefinedFirst) {
211             switch (currentAst.getType()) {
212                 case JavadocCommentsTokenTypes.TEXT ->
213                     isDefinedFirst = currentAst.getText().isBlank();
214                 case JavadocCommentsTokenTypes.HTML_ELEMENT ->
215                     isDefinedFirst = isHtmlTagWithoutText(currentAst);
216                 case JavadocCommentsTokenTypes.LEADING_ASTERISK,
217                      JavadocCommentsTokenTypes.NEWLINE -> {
218                     // Ignore formatting tokens
219                 }
220                 default -> isDefinedFirst = false;
221             }
222             currentAst = currentAst.getPreviousSibling();
223         }
224         return isDefinedFirst;
225     }
226 
227     /**
228      * Whether some text is present inside the HTML element or tag.
229      *
230      * @param node DetailNode of type {@link JavadocCommentsTokenTypes#HTML_ELEMENT}
231      * @return {@code true} if some text is present inside the HTML element
232      */
233     public static boolean isHtmlTagWithoutText(DetailNode node) {
234         boolean isEmpty = true;
235         final DetailNode htmlContentToken =
236              JavadocUtil.findFirstToken(node, JavadocCommentsTokenTypes.HTML_CONTENT);
237 
238         if (htmlContentToken != null) {
239             final DetailNode child = htmlContentToken.getFirstChild();
240             isEmpty = child.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT
241                         && isHtmlTagWithoutText(child);
242         }
243         return isEmpty;
244     }
245 
246     /**
247      * Checks if the given node is an inline summary tag.
248      *
249      * @param javadocInlineTag node
250      * @return {@code true} if inline tag is of
251      *       type {@link JavadocCommentsTokenTypes#SUMMARY_INLINE_TAG}
252      */
253     private static boolean isSummaryTag(DetailNode javadocInlineTag) {
254         return javadocInlineTag.getType() == JavadocCommentsTokenTypes.SUMMARY_INLINE_TAG;
255     }
256 
257     /**
258      * Checks if the given node is an inline return node.
259      *
260      * @param javadocInlineTag node
261      * @return {@code true} if inline tag is of
262      *       type {@link JavadocCommentsTokenTypes#RETURN_INLINE_TAG}
263      */
264     private static boolean isInlineReturnTag(DetailNode javadocInlineTag) {
265         return javadocInlineTag.getType() == JavadocCommentsTokenTypes.RETURN_INLINE_TAG;
266     }
267 
268     /**
269      * Checks the inline summary (if present) for {@code period} at end and forbidden fragments.
270      *
271      * @param inlineSummaryTag node of type {@link JavadocCommentsTokenTypes#SUMMARY_INLINE_TAG}
272      */
273     private void validateSummaryTag(DetailNode inlineSummaryTag) {
274         final DetailNode descriptionNode = JavadocUtil.findFirstToken(
275                 inlineSummaryTag, JavadocCommentsTokenTypes.DESCRIPTION);
276         final String inlineSummary = getContentOfInlineCustomTag(descriptionNode);
277         final String summaryVisible = getVisibleContent(inlineSummary);
278         if (summaryVisible.isEmpty()) {
279             log(inlineSummaryTag.getLineNumber(), MSG_SUMMARY_JAVADOC_MISSING);
280         }
281         else if (!period.isEmpty()) {
282             final boolean isPeriodNotAtEnd =
283                     summaryVisible.lastIndexOf(period) != summaryVisible.length() - 1;
284             if (isPeriodNotAtEnd) {
285                 log(inlineSummaryTag.getLineNumber(), MSG_SUMMARY_MISSING_PERIOD);
286             }
287             else if (containsForbiddenFragment(inlineSummary)) {
288                 log(inlineSummaryTag.getLineNumber(), MSG_SUMMARY_JAVADOC);
289             }
290         }
291     }
292 
293     /**
294      * Checks the inline return for forbidden fragments.
295      *
296      * @param inlineReturnTag node of type {@link JavadocCommentsTokenTypes#RETURN_INLINE_TAG}
297      */
298     private void validateInlineReturnTag(DetailNode inlineReturnTag) {
299         final DetailNode descriptionNode = JavadocUtil.findFirstToken(
300                 inlineReturnTag, JavadocCommentsTokenTypes.DESCRIPTION);
301         final String inlineReturn = getContentOfInlineCustomTag(descriptionNode);
302         final String returnVisible = getVisibleContent(inlineReturn);
303         if (returnVisible.isEmpty()) {
304             log(inlineReturnTag.getLineNumber(), MSG_SUMMARY_JAVADOC_MISSING);
305         }
306         else if (containsForbiddenFragment(inlineReturn)) {
307             log(inlineReturnTag.getLineNumber(), MSG_SUMMARY_JAVADOC);
308         }
309     }
310 
311     /**
312      * Gets the content of inline custom tag.
313      *
314      * @param descriptionNode node of type {@link JavadocCommentsTokenTypes#DESCRIPTION}
315      * @return String consisting of the content of inline custom tag.
316      */
317     public static String getContentOfInlineCustomTag(DetailNode descriptionNode) {
318         final StringBuilder customTagContent = new StringBuilder(256);
319         DetailNode curNode = descriptionNode;
320         while (curNode != null) {
321             if (curNode.getFirstChild() == null
322                 && curNode.getType() != JavadocCommentsTokenTypes.LEADING_ASTERISK) {
323                 customTagContent.append(curNode.getText());
324             }
325 
326             DetailNode toVisit = curNode.getFirstChild();
327             while (curNode != descriptionNode && toVisit == null) {
328                 toVisit = curNode.getNextSibling();
329                 curNode = curNode.getParent();
330             }
331 
332             curNode = toVisit;
333         }
334         return customTagContent.toString();
335     }
336 
337     /**
338      * Gets the string that is visible to user in javadoc.
339      *
340      * @param summary entire content of summary javadoc.
341      * @return string that is visible to user in javadoc.
342      */
343     private static String getVisibleContent(String summary) {
344         final String visibleSummary = HTML_ELEMENTS.matcher(summary).replaceAll("");
345         return visibleSummary.trim();
346     }
347 
348     /**
349      * Tests if first sentence contains forbidden summary fragment.
350      *
351      * @param firstSentence string with first sentence.
352      * @return {@code true} if first sentence contains forbidden summary fragment.
353      */
354     private boolean containsForbiddenFragment(String firstSentence) {
355         final String javadocText = JAVADOC_MULTILINE_TO_SINGLELINE_PATTERN
356                 .matcher(firstSentence).replaceAll(" ");
357         return forbiddenSummaryFragments.matcher(trimExcessWhitespaces(javadocText)).find();
358     }
359 
360     /**
361      * Trims the given {@code text} of duplicate whitespaces.
362      *
363      * @param text the text to transform.
364      * @return the finalized form of the text.
365      */
366     private static String trimExcessWhitespaces(String text) {
367         final StringBuilder result = new StringBuilder(256);
368         boolean previousWhitespace = true;
369 
370         for (int index = 0; index < text.length(); index++) {
371             final char letter = text.charAt(index);
372             final char print;
373             if (Character.isWhitespace(letter)) {
374                 if (previousWhitespace) {
375                     continue;
376                 }
377 
378                 previousWhitespace = true;
379                 print = ' ';
380             }
381             else {
382                 previousWhitespace = false;
383                 print = letter;
384             }
385 
386             result.append(print);
387         }
388 
389         return result.toString();
390     }
391 
392     /**
393      * Checks if the node starts with an {&#64;inheritDoc}.
394      *
395      * @param root the root node to examine.
396      * @return {@code true} if the javadoc starts with an {&#64;inheritDoc}.
397      */
398     private static boolean startsWithInheritDoc(DetailNode root) {
399         boolean found = false;
400         DetailNode node = root.getFirstChild();
401 
402         while (node != null) {
403             if (node.getType() == JavadocCommentsTokenTypes.JAVADOC_INLINE_TAG
404                     && node.getFirstChild().getType()
405                             == JavadocCommentsTokenTypes.INHERIT_DOC_INLINE_TAG) {
406                 found = true;
407             }
408             if ((node.getType() == JavadocCommentsTokenTypes.TEXT
409                     || node.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT)
410                     && !CommonUtil.isBlank(node.getText())) {
411                 break;
412             }
413             node = node.getNextSibling();
414         }
415 
416         return found;
417     }
418 
419     /**
420      * Finds and returns summary sentence.
421      *
422      * @param ast javadoc root node.
423      * @return violation string.
424      */
425     private static String getSummarySentence(DetailNode ast) {
426         final StringBuilder result = new StringBuilder(256);
427         DetailNode node = ast.getFirstChild();
428         while (node != null) {
429             if (node.getType() == JavadocCommentsTokenTypes.TEXT) {
430                 result.append(node.getText());
431             }
432             else {
433                 final String summary = result.toString();
434                 if (CommonUtil.isBlank(summary)
435                         && node.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT) {
436                     final DetailNode htmlContentToken = JavadocUtil.findFirstToken(
437                             node, JavadocCommentsTokenTypes.HTML_CONTENT);
438                     result.append(getStringInsideHtmlTag(summary, htmlContentToken));
439                 }
440             }
441             node = node.getNextSibling();
442         }
443         return result.toString().trim();
444     }
445 
446     /**
447      * Get concatenated string within text of html tags.
448      *
449      * @param result javadoc string
450      * @param detailNode htmlContent node
451      * @return java doc tag content appended in result
452      */
453     private static String getStringInsideHtmlTag(String result, DetailNode detailNode) {
454         final StringBuilder contents = new StringBuilder(result);
455         if (detailNode != null) {
456             DetailNode tempNode = detailNode.getFirstChild();
457             while (tempNode != null) {
458                 if (tempNode.getType() == JavadocCommentsTokenTypes.TEXT) {
459                     contents.append(tempNode.getText());
460                 }
461                 tempNode = tempNode.getNextSibling();
462             }
463         }
464         return contents.toString();
465     }
466 
467     /**
468      * Finds the first sentence.
469      *
470      * @param ast The Javadoc root node.
471      * @param period The configured period symbol.
472      * @return An Optional containing the first sentence
473      *     up to and excluding the period, or an empty
474      *     Optional if no ending was found.
475      */
476     private static Optional<String> getFirstSentence(DetailNode ast, String period) {
477         final List<String> sentenceParts = new ArrayList<>();
478         Optional<String> result = Optional.empty();
479         for (String text : (Iterable<String>) streamTextParts(ast)::iterator) {
480             final Optional<String> sentenceEnding = findSentenceEnding(text, period);
481 
482             if (sentenceEnding.isPresent()) {
483                 sentenceParts.add(sentenceEnding.get());
484                 result = Optional.of(String.join("", sentenceParts));
485                 break;
486             }
487             sentenceParts.add(text);
488         }
489         return result;
490     }
491 
492     /**
493      * Streams through all the text under the given node.
494      *
495      * @param node The Javadoc node to examine.
496      * @return All the text in all nodes that have no child nodes.
497      */
498     private static Stream<String> streamTextParts(DetailNode node) {
499         final Stream<String> result;
500         if (node.getFirstChild() == null) {
501             result = Stream.of(node.getText());
502         }
503         else {
504             final List<Stream<String>> childStreams = new ArrayList<>();
505             DetailNode child = node.getFirstChild();
506             while (child != null) {
507                 childStreams.add(streamTextParts(child));
508                 child = child.getNextSibling();
509             }
510             result = childStreams.stream().flatMap(Function.identity());
511         }
512         return result;
513     }
514 
515     /**
516      * Finds the end of a sentence. The end of sentence detection here could be replaced in the
517      * future by Java's built-in BreakIterator class.
518      *
519      * @param text The string to search.
520      * @param period The period character to find.
521      * @return An Optional containing the string up to and excluding the period,
522      *     or empty Optional if no ending was found.
523      */
524     private static Optional<String> findSentenceEnding(String text, String period) {
525         int periodIndex = text.indexOf(period);
526         Optional<String> result = Optional.empty();
527         while (periodIndex >= 0) {
528             final int afterPeriodIndex = periodIndex + period.length();
529 
530             // Handle western period separately as it is only the end of a sentence if followed
531             // by whitespace. Other period characters often include whitespace in the character.
532             if (!DEFAULT_PERIOD.equals(period)
533                 || afterPeriodIndex >= text.length()
534                 || Character.isWhitespace(text.charAt(afterPeriodIndex))) {
535                 final String resultStr = text.substring(0, periodIndex);
536                 result = Optional.of(resultStr);
537                 break;
538             }
539             periodIndex = text.indexOf(period, afterPeriodIndex);
540         }
541         return result;
542     }
543 }