View Javadoc
1   ///////////////////////////////////////////////////////////////////////////////////////////////
2   // checkstyle: Checks Java source code and other text files for adherence to a set of rules.
3   // Copyright (C) 2001-2025 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 +(\\*)|^ +(\\*)");
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                     break;
215                 case JavadocCommentsTokenTypes.HTML_ELEMENT:
216                     isDefinedFirst = isHtmlTagWithoutText(currentAst);
217                     break;
218                 case JavadocCommentsTokenTypes.LEADING_ASTERISK:
219                 case JavadocCommentsTokenTypes.NEWLINE:
220                     // Ignore formatting tokens
221                     break;
222                 default:
223                     isDefinedFirst = false;
224                     break;
225             }
226             currentAst = currentAst.getPreviousSibling();
227         }
228         return isDefinedFirst;
229     }
230 
231     /**
232      * Whether some text is present inside the HTML element or tag.
233      *
234      * @param node DetailNode of type {@link JavadocCommentsTokenTypes#HTML_ELEMENT}
235      * @return {@code true} if some text is present inside the HTML element
236      */
237     public static boolean isHtmlTagWithoutText(DetailNode node) {
238         boolean isEmpty = true;
239         final DetailNode htmlContentToken =
240              JavadocUtil.findFirstToken(node, JavadocCommentsTokenTypes.HTML_CONTENT);
241 
242         if (htmlContentToken != null) {
243             final DetailNode child = htmlContentToken.getFirstChild();
244             isEmpty = child.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT
245                         && isHtmlTagWithoutText(child);
246         }
247         return isEmpty;
248     }
249 
250     /**
251      * Checks if the given node is an inline summary tag.
252      *
253      * @param javadocInlineTag node
254      * @return {@code true} if inline tag is of
255      *       type {@link JavadocCommentsTokenTypes#SUMMARY_INLINE_TAG}
256      */
257     private static boolean isSummaryTag(DetailNode javadocInlineTag) {
258         return javadocInlineTag.getType() == JavadocCommentsTokenTypes.SUMMARY_INLINE_TAG;
259     }
260 
261     /**
262      * Checks if the given node is an inline return node.
263      *
264      * @param javadocInlineTag node
265      * @return {@code true} if inline tag is of
266      *       type {@link JavadocCommentsTokenTypes#RETURN_INLINE_TAG}
267      */
268     private static boolean isInlineReturnTag(DetailNode javadocInlineTag) {
269         return javadocInlineTag.getType() == JavadocCommentsTokenTypes.RETURN_INLINE_TAG;
270     }
271 
272     /**
273      * Checks the inline summary (if present) for {@code period} at end and forbidden fragments.
274      *
275      * @param inlineSummaryTag node of type {@link JavadocCommentsTokenTypes#SUMMARY_INLINE_TAG}
276      */
277     private void validateSummaryTag(DetailNode inlineSummaryTag) {
278         final DetailNode descriptionNode = JavadocUtil.findFirstToken(
279                 inlineSummaryTag, JavadocCommentsTokenTypes.DESCRIPTION);
280         final String inlineSummary = getContentOfInlineCustomTag(descriptionNode);
281         final String summaryVisible = getVisibleContent(inlineSummary);
282         if (summaryVisible.isEmpty()) {
283             log(inlineSummaryTag.getLineNumber(), MSG_SUMMARY_JAVADOC_MISSING);
284         }
285         else if (!period.isEmpty()) {
286             final boolean isPeriodNotAtEnd =
287                     summaryVisible.lastIndexOf(period) != summaryVisible.length() - 1;
288             if (isPeriodNotAtEnd) {
289                 log(inlineSummaryTag.getLineNumber(), MSG_SUMMARY_MISSING_PERIOD);
290             }
291             else if (containsForbiddenFragment(inlineSummary)) {
292                 log(inlineSummaryTag.getLineNumber(), MSG_SUMMARY_JAVADOC);
293             }
294         }
295     }
296 
297     /**
298      * Checks the inline return for forbidden fragments.
299      *
300      * @param inlineReturnTag node of type {@link JavadocCommentsTokenTypes#RETURN_INLINE_TAG}
301      */
302     private void validateInlineReturnTag(DetailNode inlineReturnTag) {
303         final DetailNode descriptionNode = JavadocUtil.findFirstToken(
304                 inlineReturnTag, JavadocCommentsTokenTypes.DESCRIPTION);
305         final String inlineReturn = getContentOfInlineCustomTag(descriptionNode);
306         final String returnVisible = getVisibleContent(inlineReturn);
307         if (returnVisible.isEmpty()) {
308             log(inlineReturnTag.getLineNumber(), MSG_SUMMARY_JAVADOC_MISSING);
309         }
310         else if (containsForbiddenFragment(inlineReturn)) {
311             log(inlineReturnTag.getLineNumber(), MSG_SUMMARY_JAVADOC);
312         }
313     }
314 
315     /**
316      * Gets the content of inline custom tag.
317      *
318      * @param descriptionNode node of type {@link JavadocCommentsTokenTypes#DESCRIPTION}
319      * @return String consisting of the content of inline custom tag.
320      */
321     public static String getContentOfInlineCustomTag(DetailNode descriptionNode) {
322         final StringBuilder customTagContent = new StringBuilder(256);
323         DetailNode curNode = descriptionNode;
324         while (curNode != null) {
325             if (curNode.getFirstChild() == null
326                 && curNode.getType() != JavadocCommentsTokenTypes.LEADING_ASTERISK) {
327                 customTagContent.append(curNode.getText());
328             }
329 
330             DetailNode toVisit = curNode.getFirstChild();
331             while (curNode != descriptionNode && toVisit == null) {
332                 toVisit = curNode.getNextSibling();
333                 curNode = curNode.getParent();
334             }
335 
336             curNode = toVisit;
337         }
338         return customTagContent.toString();
339     }
340 
341     /**
342      * Gets the string that is visible to user in javadoc.
343      *
344      * @param summary entire content of summary javadoc.
345      * @return string that is visible to user in javadoc.
346      */
347     private static String getVisibleContent(String summary) {
348         final String visibleSummary = HTML_ELEMENTS.matcher(summary).replaceAll("");
349         return visibleSummary.trim();
350     }
351 
352     /**
353      * Tests if first sentence contains forbidden summary fragment.
354      *
355      * @param firstSentence string with first sentence.
356      * @return {@code true} if first sentence contains forbidden summary fragment.
357      */
358     private boolean containsForbiddenFragment(String firstSentence) {
359         final String javadocText = JAVADOC_MULTILINE_TO_SINGLELINE_PATTERN
360                 .matcher(firstSentence).replaceAll(" ");
361         return forbiddenSummaryFragments.matcher(trimExcessWhitespaces(javadocText)).find();
362     }
363 
364     /**
365      * Trims the given {@code text} of duplicate whitespaces.
366      *
367      * @param text the text to transform.
368      * @return the finalized form of the text.
369      */
370     private static String trimExcessWhitespaces(String text) {
371         final StringBuilder result = new StringBuilder(256);
372         boolean previousWhitespace = true;
373 
374         for (char letter : text.toCharArray()) {
375             final char print;
376             if (Character.isWhitespace(letter)) {
377                 if (previousWhitespace) {
378                     continue;
379                 }
380 
381                 previousWhitespace = true;
382                 print = ' ';
383             }
384             else {
385                 previousWhitespace = false;
386                 print = letter;
387             }
388 
389             result.append(print);
390         }
391 
392         return result.toString();
393     }
394 
395     /**
396      * Checks if the node starts with an {&#64;inheritDoc}.
397      *
398      * @param root the root node to examine.
399      * @return {@code true} if the javadoc starts with an {&#64;inheritDoc}.
400      */
401     private static boolean startsWithInheritDoc(DetailNode root) {
402         boolean found = false;
403         DetailNode node = root.getFirstChild();
404 
405         while (node != null) {
406             if (node.getType() == JavadocCommentsTokenTypes.JAVADOC_INLINE_TAG
407                     && node.getFirstChild().getType()
408                             == JavadocCommentsTokenTypes.INHERIT_DOC_INLINE_TAG) {
409                 found = true;
410             }
411             if ((node.getType() == JavadocCommentsTokenTypes.TEXT
412                     || node.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT)
413                     && !CommonUtil.isBlank(node.getText())) {
414                 break;
415             }
416             node = node.getNextSibling();
417         }
418 
419         return found;
420     }
421 
422     /**
423      * Finds and returns summary sentence.
424      *
425      * @param ast javadoc root node.
426      * @return violation string.
427      */
428     private static String getSummarySentence(DetailNode ast) {
429         final StringBuilder result = new StringBuilder(256);
430         DetailNode node = ast.getFirstChild();
431         while (node != null) {
432             if (node.getType() == JavadocCommentsTokenTypes.TEXT) {
433                 result.append(node.getText());
434             }
435             else {
436                 final String summary = result.toString();
437                 if (CommonUtil.isBlank(summary)
438                         && node.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT) {
439                     final DetailNode htmlContentToken = JavadocUtil.findFirstToken(
440                             node, JavadocCommentsTokenTypes.HTML_CONTENT);
441                     result.append(getStringInsideHtmlTag(summary, htmlContentToken));
442                 }
443             }
444             node = node.getNextSibling();
445         }
446         return result.toString().trim();
447     }
448 
449     /**
450      * Get concatenated string within text of html tags.
451      *
452      * @param result javadoc string
453      * @param detailNode htmlContent node
454      * @return java doc tag content appended in result
455      */
456     private static String getStringInsideHtmlTag(String result, DetailNode detailNode) {
457         final StringBuilder contents = new StringBuilder(result);
458         if (detailNode != null) {
459             DetailNode tempNode = detailNode.getFirstChild();
460             while (tempNode != null) {
461                 if (tempNode.getType() == JavadocCommentsTokenTypes.TEXT) {
462                     contents.append(tempNode.getText());
463                 }
464                 tempNode = tempNode.getNextSibling();
465             }
466         }
467         return contents.toString();
468     }
469 
470     /**
471      * Finds the first sentence.
472      *
473      * @param ast The Javadoc root node.
474      * @param period The configured period symbol.
475      * @return An Optional containing the first sentence
476      *     up to and excluding the period, or an empty
477      *     Optional if no ending was found.
478      */
479     private static Optional<String> getFirstSentence(DetailNode ast, String period) {
480         final List<String> sentenceParts = new ArrayList<>();
481         Optional<String> result = Optional.empty();
482         for (String text : (Iterable<String>) streamTextParts(ast)::iterator) {
483             final Optional<String> sentenceEnding = findSentenceEnding(text, period);
484 
485             if (sentenceEnding.isPresent()) {
486                 sentenceParts.add(sentenceEnding.get());
487                 result = Optional.of(String.join("", sentenceParts));
488                 break;
489             }
490             sentenceParts.add(text);
491         }
492         return result;
493     }
494 
495     /**
496      * Streams through all the text under the given node.
497      *
498      * @param node The Javadoc node to examine.
499      * @return All the text in all nodes that have no child nodes.
500      */
501     private static Stream<String> streamTextParts(DetailNode node) {
502         final Stream<String> result;
503         if (node.getFirstChild() == null) {
504             result = Stream.of(node.getText());
505         }
506         else {
507             final List<Stream<String>> childStreams = new ArrayList<>();
508             DetailNode child = node.getFirstChild();
509             while (child != null) {
510                 childStreams.add(streamTextParts(child));
511                 child = child.getNextSibling();
512             }
513             result = childStreams.stream().flatMap(Function.identity());
514         }
515         return result;
516     }
517 
518     /**
519      * Finds the end of a sentence. The end of sentence detection here could be replaced in the
520      * future by Java's built-in BreakIterator class.
521      *
522      * @param text The string to search.
523      * @param period The period character to find.
524      * @return An Optional containing the string up to and excluding the period,
525      *     or empty Optional if no ending was found.
526      */
527     private static Optional<String> findSentenceEnding(String text, String period) {
528         int periodIndex = text.indexOf(period);
529         Optional<String> result = Optional.empty();
530         while (periodIndex >= 0) {
531             final int afterPeriodIndex = periodIndex + period.length();
532 
533             // Handle western period separately as it is only the end of a sentence if followed
534             // by whitespace. Other period characters often include whitespace in the character.
535             if (!DEFAULT_PERIOD.equals(period)
536                 || afterPeriodIndex >= text.length()
537                 || Character.isWhitespace(text.charAt(afterPeriodIndex))) {
538                 final String resultStr = text.substring(0, periodIndex);
539                 result = Optional.of(resultStr);
540                 break;
541             }
542             periodIndex = text.indexOf(period, afterPeriodIndex);
543         }
544         return result;
545     }
546 }