001/////////////////////////////////////////////////////////////////////////////////////////////// 002// checkstyle: Checks Java source code and other text files for adherence to a set of rules. 003// Copyright (C) 2001-2026 the original author or authors. 004// 005// This library is free software; you can redistribute it and/or 006// modify it under the terms of the GNU Lesser General Public 007// License as published by the Free Software Foundation; either 008// version 2.1 of the License, or (at your option) any later version. 009// 010// This library is distributed in the hope that it will be useful, 011// but WITHOUT ANY WARRANTY; without even the implied warranty of 012// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 013// Lesser General Public License for more details. 014// 015// You should have received a copy of the GNU Lesser General Public 016// License along with this library; if not, write to the Free Software 017// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA 018/////////////////////////////////////////////////////////////////////////////////////////////// 019 020package com.puppycrawl.tools.checkstyle.checks.javadoc; 021 022import java.util.ArrayList; 023import java.util.List; 024import java.util.Optional; 025import java.util.function.Function; 026import java.util.regex.Pattern; 027import java.util.stream.Stream; 028 029import com.puppycrawl.tools.checkstyle.FileStatefulCheck; 030import com.puppycrawl.tools.checkstyle.api.DetailNode; 031import com.puppycrawl.tools.checkstyle.api.JavadocCommentsTokenTypes; 032import com.puppycrawl.tools.checkstyle.utils.CommonUtil; 033import com.puppycrawl.tools.checkstyle.utils.JavadocUtil; 034 035/** 036 * <div> 037 * Checks that 038 * <a href="https://www.oracle.com/technical-resources/articles/java/javadoc-tool.html#firstsentence"> 039 * Javadoc summary sentence</a> does not contain phrases that are not recommended to use. 040 * Summaries that contain only the {@code {@inheritDoc}} tag are skipped. 041 * Summaries that contain a non-empty {@code {@return}} are allowed. 042 * Check also violate Javadoc that does not contain first sentence, though with {@code {@return}} a 043 * period is not required as the Javadoc tool adds it. 044 * </div> 045 * 046 * <p> 047 * Note: For defining a summary, both the first sentence and the @summary tag approaches 048 * are supported. 049 * </p> 050 * 051 * @since 6.0 052 */ 053@FileStatefulCheck 054public class SummaryJavadocCheck extends AbstractJavadocCheck { 055 056 /** 057 * A key is pointing to the warning message text in "messages.properties" 058 * file. 059 */ 060 public static final String MSG_SUMMARY_FIRST_SENTENCE = "summary.first.sentence"; 061 062 /** 063 * A key is pointing to the warning message text in "messages.properties" 064 * file. 065 */ 066 public static final String MSG_SUMMARY_JAVADOC = "summary.javaDoc"; 067 068 /** 069 * A key is pointing to the warning message text in "messages.properties" 070 * file. 071 */ 072 public static final String MSG_SUMMARY_JAVADOC_MISSING = "summary.javaDoc.missing"; 073 074 /** 075 * A key is pointing to the warning message text in "messages.properties" file. 076 */ 077 public static final String MSG_SUMMARY_MISSING_PERIOD = "summary.javaDoc.missing.period"; 078 079 /** 080 * This regexp is used to convert multiline javadoc to single-line without stars. 081 */ 082 private static final Pattern JAVADOC_MULTILINE_TO_SINGLELINE_PATTERN = 083 Pattern.compile("\n[ \\t]+(\\*)|^[ \\t]+(\\*)"); 084 085 /** 086 * This regexp is used to remove html tags, whitespace, and asterisks from a string. 087 */ 088 private static final Pattern HTML_ELEMENTS = 089 Pattern.compile("<[^>]*>"); 090 091 /** Default period literal. */ 092 private static final String DEFAULT_PERIOD = "."; 093 094 /** 095 * Specify the regexp for forbidden summary fragments. 096 */ 097 private Pattern forbiddenSummaryFragments = CommonUtil.createPattern("^$"); 098 099 /** 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 * Creates a new {@code SummaryJavadocCheck} instance. 114 */ 115 public SummaryJavadocCheck() { 116 // no code by default 117 } 118 119 /** 120 * Setter to specify the regexp for forbidden summary fragments. 121 * 122 * @param pattern a pattern. 123 * @since 6.0 124 */ 125 public void setForbiddenSummaryFragments(Pattern pattern) { 126 forbiddenSummaryFragments = pattern; 127 } 128 129 /** 130 * Setter to specify the period symbol. Used to check the first sentence ends with a period. 131 * Periods that are not followed by a whitespace character are ignored (eg. the period in v1.0). 132 * Because some periods include whitespace built into the character, if this is set to a 133 * non-default value any period will end the sentence, whether it is followed by whitespace or 134 * not. 135 * 136 * @param period period's value. 137 * @since 6.2 138 */ 139 public void setPeriod(String period) { 140 this.period = period; 141 } 142 143 @Override 144 public int[] getDefaultJavadocTokens() { 145 return new int[] { 146 JavadocCommentsTokenTypes.JAVADOC_CONTENT, 147 JavadocCommentsTokenTypes.SUMMARY_INLINE_TAG, 148 JavadocCommentsTokenTypes.RETURN_INLINE_TAG, 149 }; 150 } 151 152 @Override 153 public int[] getRequiredJavadocTokens() { 154 return getAcceptableJavadocTokens(); 155 } 156 157 @Override 158 public void visitJavadocToken(DetailNode ast) { 159 if (isSummaryTag(ast) && isDefinedFirst(ast.getParent())) { 160 shouldValidateUntaggedSummary = false; 161 validateSummaryTag(ast); 162 } 163 else if (isInlineReturnTag(ast)) { 164 shouldValidateUntaggedSummary = false; 165 validateInlineReturnTag(ast); 166 } 167 } 168 169 @Override 170 public void leaveJavadocToken(DetailNode ast) { 171 if (ast.getType() == JavadocCommentsTokenTypes.JAVADOC_CONTENT) { 172 if (shouldValidateUntaggedSummary && !startsWithInheritDoc(ast)) { 173 validateUntaggedSummary(ast); 174 } 175 shouldValidateUntaggedSummary = true; 176 } 177 } 178 179 /** 180 * Checks the javadoc text for {@code period} at end and forbidden fragments. 181 * 182 * @param ast the javadoc text node 183 */ 184 private void validateUntaggedSummary(DetailNode ast) { 185 final String summaryDoc = getSummarySentence(ast); 186 if (summaryDoc.isEmpty()) { 187 log(ast, MSG_SUMMARY_JAVADOC_MISSING); 188 } 189 else if (!period.isEmpty()) { 190 if (summaryDoc.contains(period)) { 191 final Optional<String> firstSentence = getFirstSentence(ast, period); 192 193 if (firstSentence.isPresent()) { 194 if (containsForbiddenFragment(firstSentence.get())) { 195 log(ast, MSG_SUMMARY_JAVADOC); 196 } 197 } 198 else { 199 log(ast, MSG_SUMMARY_FIRST_SENTENCE); 200 } 201 } 202 else { 203 log(ast, MSG_SUMMARY_FIRST_SENTENCE); 204 } 205 } 206 } 207 208 /** 209 * Whether the {@code {@summary}} tag is defined first in the javadoc. 210 * 211 * @param inlineTagNode node of type {@link JavadocCommentsTokenTypes#JAVADOC_INLINE_TAG} 212 * @return {@code true} if the {@code {@summary}} tag is defined first in the javadoc 213 */ 214 private static boolean isDefinedFirst(DetailNode inlineTagNode) { 215 boolean isDefinedFirst = true; 216 DetailNode currentAst = inlineTagNode.getPreviousSibling(); 217 while (currentAst != null && isDefinedFirst) { 218 switch (currentAst.getType()) { 219 case JavadocCommentsTokenTypes.TEXT -> 220 isDefinedFirst = currentAst.getText().isBlank(); 221 case JavadocCommentsTokenTypes.HTML_ELEMENT -> 222 isDefinedFirst = isHtmlTagWithoutText(currentAst); 223 case JavadocCommentsTokenTypes.LEADING_ASTERISK, 224 JavadocCommentsTokenTypes.NEWLINE -> { 225 // Ignore formatting tokens 226 } 227 default -> isDefinedFirst = false; 228 } 229 currentAst = currentAst.getPreviousSibling(); 230 } 231 return isDefinedFirst; 232 } 233 234 /** 235 * Whether some text is present inside the HTML element or tag. 236 * 237 * @param node DetailNode of type {@link JavadocCommentsTokenTypes#HTML_ELEMENT} 238 * @return {@code true} if some text is present inside the HTML element 239 */ 240 public static boolean isHtmlTagWithoutText(DetailNode node) { 241 boolean isEmpty = true; 242 final DetailNode htmlContentToken = 243 JavadocUtil.findFirstToken(node, JavadocCommentsTokenTypes.HTML_CONTENT); 244 245 if (htmlContentToken != null) { 246 final DetailNode child = htmlContentToken.getFirstChild(); 247 isEmpty = child.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT 248 && isHtmlTagWithoutText(child); 249 } 250 return isEmpty; 251 } 252 253 /** 254 * Checks if the given node is an inline summary tag. 255 * 256 * @param javadocInlineTag node 257 * @return {@code true} if inline tag is of 258 * type {@link JavadocCommentsTokenTypes#SUMMARY_INLINE_TAG} 259 */ 260 private static boolean isSummaryTag(DetailNode javadocInlineTag) { 261 return javadocInlineTag.getType() == JavadocCommentsTokenTypes.SUMMARY_INLINE_TAG; 262 } 263 264 /** 265 * Checks if the given node is an inline return node. 266 * 267 * @param javadocInlineTag node 268 * @return {@code true} if inline tag is of 269 * type {@link JavadocCommentsTokenTypes#RETURN_INLINE_TAG} 270 */ 271 private static boolean isInlineReturnTag(DetailNode javadocInlineTag) { 272 return javadocInlineTag.getType() == JavadocCommentsTokenTypes.RETURN_INLINE_TAG; 273 } 274 275 /** 276 * Checks the inline summary (if present) for {@code period} at end and forbidden fragments. 277 * 278 * @param inlineSummaryTag node of type {@link JavadocCommentsTokenTypes#SUMMARY_INLINE_TAG} 279 */ 280 private void validateSummaryTag(DetailNode inlineSummaryTag) { 281 final DetailNode descriptionNode = JavadocUtil.findFirstToken( 282 inlineSummaryTag, JavadocCommentsTokenTypes.DESCRIPTION); 283 final String inlineSummary = getContentOfInlineCustomTag(descriptionNode); 284 final String summaryVisible = getVisibleContent(inlineSummary); 285 if (summaryVisible.isEmpty()) { 286 log(inlineSummaryTag, MSG_SUMMARY_JAVADOC_MISSING); 287 } 288 else if (!period.isEmpty()) { 289 final boolean isPeriodNotAtEnd = 290 summaryVisible.lastIndexOf(period) != summaryVisible.length() - 1; 291 if (isPeriodNotAtEnd) { 292 log(inlineSummaryTag, MSG_SUMMARY_MISSING_PERIOD); 293 } 294 else if (containsForbiddenFragment(inlineSummary)) { 295 log(inlineSummaryTag, MSG_SUMMARY_JAVADOC); 296 } 297 } 298 } 299 300 /** 301 * Checks the inline return for forbidden fragments. 302 * 303 * @param inlineReturnTag node of type {@link JavadocCommentsTokenTypes#RETURN_INLINE_TAG} 304 */ 305 private void validateInlineReturnTag(DetailNode inlineReturnTag) { 306 final DetailNode descriptionNode = JavadocUtil.findFirstToken( 307 inlineReturnTag, JavadocCommentsTokenTypes.DESCRIPTION); 308 final String inlineReturn = getContentOfInlineCustomTag(descriptionNode); 309 final String returnVisible = getVisibleContent(inlineReturn); 310 if (returnVisible.isEmpty()) { 311 log(inlineReturnTag, MSG_SUMMARY_JAVADOC_MISSING); 312 } 313 else if (containsForbiddenFragment(inlineReturn)) { 314 log(inlineReturnTag, MSG_SUMMARY_JAVADOC); 315 } 316 } 317 318 /** 319 * Gets the content of inline custom tag. 320 * 321 * @param descriptionNode node of type {@link JavadocCommentsTokenTypes#DESCRIPTION} 322 * @return String consisting of the content of inline custom tag. 323 */ 324 public static String getContentOfInlineCustomTag(DetailNode descriptionNode) { 325 final StringBuilder customTagContent = new StringBuilder(256); 326 DetailNode curNode = descriptionNode; 327 while (curNode != null) { 328 if (curNode.getFirstChild() == null 329 && curNode.getType() != JavadocCommentsTokenTypes.LEADING_ASTERISK) { 330 customTagContent.append(curNode.getText()); 331 } 332 333 DetailNode toVisit = curNode.getFirstChild(); 334 while (curNode != descriptionNode && toVisit == null) { 335 toVisit = curNode.getNextSibling(); 336 curNode = curNode.getParent(); 337 } 338 339 curNode = toVisit; 340 } 341 return customTagContent.toString(); 342 } 343 344 /** 345 * Gets the string that is visible to user in javadoc. 346 * 347 * @param summary entire content of summary javadoc. 348 * @return string that is visible to user in javadoc. 349 */ 350 private static String getVisibleContent(String summary) { 351 final String visibleSummary = HTML_ELEMENTS.matcher(summary).replaceAll(""); 352 return visibleSummary.trim(); 353 } 354 355 /** 356 * Tests if first sentence contains forbidden summary fragment. 357 * 358 * @param firstSentence string with first sentence. 359 * @return {@code true} if first sentence contains forbidden summary fragment. 360 */ 361 private boolean containsForbiddenFragment(String firstSentence) { 362 final String javadocText = JAVADOC_MULTILINE_TO_SINGLELINE_PATTERN 363 .matcher(firstSentence).replaceAll(" "); 364 return forbiddenSummaryFragments.matcher(trimExcessWhitespaces(javadocText)).find(); 365 } 366 367 /** 368 * Trims the given {@code text} of duplicate whitespaces. 369 * 370 * @param text the text to transform. 371 * @return the finalized form of the text. 372 */ 373 private static String trimExcessWhitespaces(String text) { 374 final StringBuilder result = new StringBuilder(256); 375 boolean previousWhitespace = true; 376 377 for (int index = 0; index < text.length(); index++) { 378 final char letter = text.charAt(index); 379 final char print; 380 if (Character.isWhitespace(letter)) { 381 if (previousWhitespace) { 382 continue; 383 } 384 385 previousWhitespace = true; 386 print = ' '; 387 } 388 else { 389 previousWhitespace = false; 390 print = letter; 391 } 392 393 result.append(print); 394 } 395 396 return result.toString(); 397 } 398 399 /** 400 * Checks if the node starts with an {@inheritDoc}. 401 * 402 * @param root the root node to examine. 403 * @return {@code true} if the javadoc starts with an {@inheritDoc}. 404 */ 405 private static boolean startsWithInheritDoc(DetailNode root) { 406 boolean found = false; 407 DetailNode node = root.getFirstChild(); 408 409 while (node != null) { 410 if (node.getType() == JavadocCommentsTokenTypes.JAVADOC_INLINE_TAG 411 && node.getFirstChild().getType() 412 == JavadocCommentsTokenTypes.INHERIT_DOC_INLINE_TAG) { 413 found = true; 414 } 415 if ((node.getType() == JavadocCommentsTokenTypes.TEXT 416 || node.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT) 417 && !CommonUtil.isBlank(node.getText())) { 418 break; 419 } 420 node = node.getNextSibling(); 421 } 422 423 return found; 424 } 425 426 /** 427 * Finds and returns summary sentence. 428 * 429 * @param ast javadoc root node. 430 * @return violation string. 431 */ 432 private static String getSummarySentence(DetailNode ast) { 433 final StringBuilder result = new StringBuilder(256); 434 DetailNode node = ast.getFirstChild(); 435 while (node != null) { 436 if (node.getType() == JavadocCommentsTokenTypes.TEXT) { 437 result.append(node.getText()); 438 } 439 else { 440 final String summary = result.toString(); 441 if (CommonUtil.isBlank(summary) 442 && node.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT) { 443 final DetailNode htmlContentToken = JavadocUtil.findFirstToken( 444 node, JavadocCommentsTokenTypes.HTML_CONTENT); 445 result.append(getStringInsideHtmlTag(summary, htmlContentToken)); 446 } 447 } 448 node = node.getNextSibling(); 449 } 450 return result.toString().trim(); 451 } 452 453 /** 454 * Get concatenated string within text of html tags. 455 * 456 * @param result javadoc string 457 * @param detailNode htmlContent node 458 * @return java doc tag content appended in result 459 */ 460 private static String getStringInsideHtmlTag(String result, DetailNode detailNode) { 461 final StringBuilder contents = new StringBuilder(result); 462 if (detailNode != null) { 463 DetailNode tempNode = detailNode.getFirstChild(); 464 while (tempNode != null) { 465 if (tempNode.getType() == JavadocCommentsTokenTypes.TEXT) { 466 contents.append(tempNode.getText()); 467 } 468 else { 469 final DetailNode htmlContentToken = JavadocUtil.findFirstToken( 470 tempNode, JavadocCommentsTokenTypes.HTML_CONTENT); 471 contents.append(getStringInsideHtmlTag("", htmlContentToken)); 472 } 473 tempNode = tempNode.getNextSibling(); 474 } 475 } 476 return contents.toString(); 477 } 478 479 /** 480 * Finds the first sentence. 481 * 482 * @param ast The Javadoc root node. 483 * @param period The configured period symbol. 484 * @return An Optional containing the first sentence 485 * up to and excluding the period, or an empty 486 * Optional if no ending was found. 487 */ 488 private static Optional<String> getFirstSentence(DetailNode ast, String period) { 489 final List<String> sentenceParts = new ArrayList<>(); 490 Optional<String> result = Optional.empty(); 491 for (String text : (Iterable<String>) streamTextParts(ast)::iterator) { 492 final Optional<String> sentenceEnding = findSentenceEnding(text, period); 493 494 if (sentenceEnding.isPresent()) { 495 sentenceParts.add(sentenceEnding.get()); 496 result = Optional.of(String.join("", sentenceParts)); 497 break; 498 } 499 sentenceParts.add(text); 500 } 501 return result; 502 } 503 504 /** 505 * Streams through all the text under the given node. 506 * 507 * @param node The Javadoc node to examine. 508 * @return All the text in all nodes that have no child nodes. 509 */ 510 private static Stream<String> streamTextParts(DetailNode node) { 511 final Stream<String> result; 512 if (node.getFirstChild() == null) { 513 result = Stream.of(node.getText()); 514 } 515 else { 516 final List<Stream<String>> childStreams = new ArrayList<>(); 517 DetailNode child = node.getFirstChild(); 518 while (child != null) { 519 childStreams.add(streamTextParts(child)); 520 child = child.getNextSibling(); 521 } 522 result = childStreams.stream().flatMap(Function.identity()); 523 } 524 return result; 525 } 526 527 /** 528 * Finds the end of a sentence. The end of sentence detection here could be replaced in the 529 * future by Java's built-in BreakIterator class. 530 * 531 * @param text The string to search. 532 * @param period The period character to find. 533 * @return An Optional containing the string up to and excluding the period, 534 * or empty Optional if no ending was found. 535 */ 536 private static Optional<String> findSentenceEnding(String text, String period) { 537 int periodIndex = text.indexOf(period); 538 Optional<String> result = Optional.empty(); 539 while (periodIndex >= 0) { 540 final int afterPeriodIndex = periodIndex + period.length(); 541 542 // Handle western period separately as it is only the end of a sentence if followed 543 // by whitespace. Other period characters often include whitespace in the character. 544 if (!DEFAULT_PERIOD.equals(period) 545 || afterPeriodIndex >= text.length() 546 || Character.isWhitespace(text.charAt(afterPeriodIndex))) { 547 final String resultStr = text.substring(0, periodIndex); 548 result = Optional.of(resultStr); 549 break; 550 } 551 periodIndex = text.indexOf(period, afterPeriodIndex); 552 } 553 return result; 554 } 555 556}