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.Set;
023
024import javax.annotation.Nullable;
025
026import com.puppycrawl.tools.checkstyle.StatelessCheck;
027import com.puppycrawl.tools.checkstyle.api.DetailNode;
028import com.puppycrawl.tools.checkstyle.api.JavadocCommentsTokenTypes;
029import com.puppycrawl.tools.checkstyle.utils.CommonUtil;
030import com.puppycrawl.tools.checkstyle.utils.JavadocUtil;
031
032/**
033 * <div>
034 * Checks the Javadoc paragraph.
035 * </div>
036 *
037 * <p>
038 * Checks that:
039 * </p>
040 * <ul>
041 * <li>There is one blank line between each of two paragraphs.</li>
042 * <li>Each paragraph but the first has &lt;p&gt; immediately
043 * before the first word, with no space after.</li>
044 * <li>The outer most paragraph tags should not precede
045 * <a href="https://www.w3schools.com/html/html_blocks.asp">HTML block-tag</a>.
046 * Nested paragraph tags are allowed to do that. This check only supports following block-tags:
047 * &lt;address&gt;,&lt;blockquote&gt;
048 * ,&lt;div&gt;,&lt;dl&gt;
049 * ,&lt;h1&gt;,&lt;h2&gt;,&lt;h3&gt;,&lt;h4&gt;,&lt;h5&gt;,&lt;h6&gt;,&lt;hr&gt;
050 * ,&lt;ol&gt;,&lt;p&gt;,&lt;pre&gt;
051 * ,&lt;table&gt;,&lt;ul&gt;.
052 * </li>
053 * </ul>
054 *
055 * <p><b>ATTENTION:</b></p>
056 *
057 * <p>This Check ignores HTML comments.</p>
058 *
059 * <p>The Check ignores all the nested paragraph tags,
060 * it will not give any kind of violation if the paragraph tag is nested.
061 * It also ignores paragraph tags inside block tags.</p>
062 *
063 * @since 6.0
064 */
065@StatelessCheck
066public class JavadocParagraphCheck extends AbstractJavadocCheck {
067
068    /**
069     * A key is pointing to the warning message text in "messages.properties"
070     * file.
071     */
072    public static final String MSG_TAG_AFTER = "javadoc.paragraph.tag.after";
073
074    /**
075     * A key is pointing to the warning message text in "messages.properties"
076     * file.
077     */
078    public static final String MSG_LINE_BEFORE = "javadoc.paragraph.line.before";
079
080    /**
081     * A key is pointing to the warning message text in "messages.properties"
082     * file.
083     */
084    public static final String MSG_REDUNDANT_PARAGRAPH = "javadoc.paragraph.redundant.paragraph";
085
086    /**
087     * A key is pointing to the warning message text in "messages.properties"
088     * file.
089     */
090    public static final String MSG_MISPLACED_TAG = "javadoc.paragraph.misplaced.tag";
091
092    /**
093     * A key is pointing to the warning message text in "messages.properties"
094     * file.
095     */
096    public static final String MSG_PRECEDED_BLOCK_TAG = "javadoc.paragraph.preceded.block.tag";
097
098    /**
099     * Constant for the paragraph tag name.
100     */
101    private static final String PARAGRAPH_TAG = "p";
102
103    /**
104     * Set of block tags supported by this check.
105     */
106    private static final Set<String> BLOCK_TAGS =
107            Set.of("address", "blockquote", "div", "dl",
108                   "h1", "h2", "h3", "h4", "h5", "h6", "hr",
109                   "ol", PARAGRAPH_TAG, "pre", "table", "ul");
110
111    /**
112     * Control whether the &lt;p&gt; tag should be placed immediately before the first word.
113     */
114    private boolean allowNewlineParagraph = true;
115
116    /**
117     * Setter to control whether the &lt;p&gt; tag should be placed
118     * immediately before the first word.
119     *
120     * @param value value to set.
121     * @since 6.9
122     */
123    public void setAllowNewlineParagraph(boolean value) {
124        allowNewlineParagraph = value;
125    }
126
127    @Override
128    public int[] getDefaultJavadocTokens() {
129        return new int[] {
130            JavadocCommentsTokenTypes.NEWLINE,
131            JavadocCommentsTokenTypes.HTML_ELEMENT,
132        };
133    }
134
135    @Override
136    public int[] getRequiredJavadocTokens() {
137        return getAcceptableJavadocTokens();
138    }
139
140    @Override
141    public void visitJavadocToken(DetailNode ast) {
142        if (ast.getType() == JavadocCommentsTokenTypes.NEWLINE && isEmptyLine(ast)) {
143            checkEmptyLine(ast);
144        }
145        else if (JavadocUtil.isTag(ast, PARAGRAPH_TAG)) {
146            checkParagraphTag(ast);
147        }
148    }
149
150    /**
151     * Determines whether or not the next line after empty line has paragraph tag in the beginning.
152     *
153     * @param newline NEWLINE node.
154     */
155    private void checkEmptyLine(DetailNode newline) {
156        final DetailNode nearestToken = getNearestNode(newline);
157        if (nearestToken != null && nearestToken.getType() == JavadocCommentsTokenTypes.TEXT
158                && !CommonUtil.isBlank(nearestToken.getText())) {
159            log(newline.getLineNumber(), newline.getColumnNumber(), MSG_TAG_AFTER);
160        }
161    }
162
163    /**
164     * Determines whether or not the line with paragraph tag has previous empty line.
165     *
166     * @param tag html tag.
167     */
168    private void checkParagraphTag(DetailNode tag) {
169        if (!isNestedParagraph(tag) && !isInsideBlockTag(tag)) {
170            final DetailNode newLine = getNearestEmptyLine(tag);
171            if (isFirstParagraph(tag)) {
172                log(tag.getLineNumber(), tag.getColumnNumber(), MSG_REDUNDANT_PARAGRAPH);
173            }
174            else if (newLine == null || tag.getLineNumber() - newLine.getLineNumber() != 1) {
175                log(tag.getLineNumber(), tag.getColumnNumber(), MSG_LINE_BEFORE);
176            }
177
178            final String blockTagName = findFollowedBlockTagName(tag);
179            if (blockTagName != null) {
180                log(tag.getLineNumber(), tag.getColumnNumber(),
181                        MSG_PRECEDED_BLOCK_TAG, blockTagName);
182            }
183
184            if (!allowNewlineParagraph && isImmediatelyFollowedByNewLine(tag)) {
185                log(tag.getLineNumber(), tag.getColumnNumber(), MSG_MISPLACED_TAG);
186            }
187            if (isImmediatelyFollowedByText(tag)) {
188                log(tag.getLineNumber(), tag.getColumnNumber(), MSG_MISPLACED_TAG);
189            }
190        }
191    }
192
193    /**
194     * Determines whether the paragraph tag is nested.
195     *
196     * @param tag html tag.
197     * @return true, if the paragraph tag is nested.
198     */
199    private static boolean isNestedParagraph(DetailNode tag) {
200        boolean nested = false;
201        DetailNode parent = tag.getParent();
202
203        while (parent != null) {
204            if (parent.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT) {
205                nested = true;
206                break;
207            }
208            parent = parent.getParent();
209        }
210
211        return nested;
212    }
213
214    /**
215     * Determines whether the paragraph tag is inside javadoc block tag.
216     *
217     * @param tag html tag.
218     * @return true, if the paragraph tag is inside javadoc block tag.
219     */
220    private static boolean isInsideBlockTag(DetailNode tag) {
221        boolean result = false;
222        DetailNode parent = tag;
223
224        while (parent != null) {
225            if (parent.getType() == JavadocCommentsTokenTypes.JAVADOC_BLOCK_TAG) {
226                result = true;
227                break;
228            }
229            parent = parent.getParent();
230        }
231
232        return result;
233    }
234
235    /**
236     * Determines whether or not the paragraph tag is followed by block tag.
237     *
238     * @param tag html tag.
239     * @return block tag if the paragraph tag is followed by block tag or null if not found.
240     */
241    @Nullable
242    private static String findFollowedBlockTagName(DetailNode tag) {
243        final DetailNode htmlElement = findFirstHtmlElementAfter(tag);
244        String blockTagName = null;
245
246        if (htmlElement != null) {
247            blockTagName = getHtmlElementName(htmlElement);
248        }
249
250        return blockTagName;
251    }
252
253    /**
254     * Finds and returns first html element after the tag.
255     *
256     * @param tag html tag.
257     * @return first html element after the paragraph tag or null if not found.
258     */
259    @Nullable
260    private static DetailNode findFirstHtmlElementAfter(DetailNode tag) {
261        DetailNode htmlElement = getNextSibling(tag);
262
263        while (htmlElement != null
264                && htmlElement.getType() != JavadocCommentsTokenTypes.HTML_ELEMENT) {
265            if (htmlElement.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
266                htmlElement = htmlElement.getFirstChild();
267            }
268            else if (htmlElement.getType() == JavadocCommentsTokenTypes.TEXT
269                    && !CommonUtil.isBlank(htmlElement.getText())) {
270                htmlElement = null;
271                break;
272            }
273            else {
274                htmlElement = htmlElement.getNextSibling();
275            }
276        }
277        if (htmlElement != null
278                && JavadocUtil.findFirstToken(htmlElement,
279                        JavadocCommentsTokenTypes.HTML_TAG_END) == null) {
280            htmlElement = null;
281        }
282
283        return htmlElement;
284    }
285
286    /**
287     * Finds and returns first block-level html element name.
288     *
289     * @param htmlElement block-level html tag.
290     * @return block-level html element name or null if not found.
291     */
292    @Nullable
293    private static String getHtmlElementName(DetailNode htmlElement) {
294        final DetailNode htmlTagStart = htmlElement.getFirstChild();
295        final DetailNode htmlTagName =
296                JavadocUtil.findFirstToken(htmlTagStart, JavadocCommentsTokenTypes.TAG_NAME);
297        String blockTagName = null;
298        if (BLOCK_TAGS.contains(htmlTagName.getText())) {
299            blockTagName = htmlTagName.getText();
300        }
301
302        return blockTagName;
303    }
304
305    /**
306     * Returns nearest node.
307     *
308     * @param node DetailNode node.
309     * @return nearest node.
310     */
311    private static DetailNode getNearestNode(DetailNode node) {
312        DetailNode currentNode = node;
313        while (currentNode != null
314                && (currentNode.getType() == JavadocCommentsTokenTypes.LEADING_ASTERISK
315                    || currentNode.getType() == JavadocCommentsTokenTypes.NEWLINE)) {
316            currentNode = currentNode.getNextSibling();
317        }
318        if (currentNode != null
319                && currentNode.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
320            currentNode = currentNode.getFirstChild();
321        }
322        return currentNode;
323    }
324
325    /**
326     * Determines whether or not the line is empty line.
327     *
328     * @param newLine NEWLINE node.
329     * @return true, if line is empty line.
330     */
331    private static boolean isEmptyLine(DetailNode newLine) {
332        boolean result = false;
333        DetailNode previousSibling = newLine.getPreviousSibling();
334        if (previousSibling != null && (previousSibling.getParent().getType()
335                == JavadocCommentsTokenTypes.JAVADOC_CONTENT
336                || insideNonTightHtml(previousSibling))) {
337            if (previousSibling.getType() == JavadocCommentsTokenTypes.TEXT
338                    && CommonUtil.isBlank(previousSibling.getText())) {
339                previousSibling = previousSibling.getPreviousSibling();
340            }
341            result = previousSibling != null
342                    && previousSibling.getType() == JavadocCommentsTokenTypes.LEADING_ASTERISK;
343        }
344        return result;
345    }
346
347    /**
348     * Checks whether the given node is inside a non-tight HTML element.
349     *
350     * @param previousSibling the node to check
351     * @return true if inside non-tight HTML, false otherwise
352     */
353    private static boolean insideNonTightHtml(DetailNode previousSibling) {
354        final DetailNode parent = previousSibling.getParent();
355        DetailNode htmlElement = parent;
356        if (parent.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
357            htmlElement = parent.getParent();
358        }
359        return htmlElement.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT
360                && JavadocUtil.findFirstToken(htmlElement,
361                    JavadocCommentsTokenTypes.HTML_TAG_END) == null;
362    }
363
364    /**
365     * Determines whether or not the line with paragraph tag is first line in javadoc.
366     *
367     * @param paragraphTag paragraph tag.
368     * @return true, if line with paragraph tag is first line in javadoc.
369     */
370    private static boolean isFirstParagraph(DetailNode paragraphTag) {
371        boolean result = true;
372        DetailNode previousNode = paragraphTag.getPreviousSibling();
373        while (previousNode != null) {
374            if (previousNode.getType() == JavadocCommentsTokenTypes.TEXT
375                    && !CommonUtil.isBlank(previousNode.getText())
376                || previousNode.getType() != JavadocCommentsTokenTypes.LEADING_ASTERISK
377                    && previousNode.getType() != JavadocCommentsTokenTypes.NEWLINE
378                    && previousNode.getType() != JavadocCommentsTokenTypes.TEXT) {
379                result = false;
380                break;
381            }
382            previousNode = previousNode.getPreviousSibling();
383        }
384        return result;
385    }
386
387    /**
388     * Finds and returns nearest empty line in javadoc.
389     *
390     * @param node DetailNode node.
391     * @return Some nearest empty line in javadoc.
392     */
393    private static DetailNode getNearestEmptyLine(DetailNode node) {
394        DetailNode newLine = node;
395        while (newLine != null) {
396            final DetailNode previousSibling = newLine.getPreviousSibling();
397            if (newLine.getType() == JavadocCommentsTokenTypes.NEWLINE && isEmptyLine(newLine)) {
398                break;
399            }
400            newLine = previousSibling;
401        }
402        return newLine;
403    }
404
405    /**
406     * Tests whether the paragraph tag is immediately followed by the text.
407     *
408     * @param tag html tag.
409     * @return true, if the paragraph tag is immediately followed by the text.
410     */
411    private static boolean isImmediatelyFollowedByText(DetailNode tag) {
412        final DetailNode nextSibling = getNextSibling(tag);
413
414        return nextSibling == null || nextSibling.getText().startsWith(" ");
415    }
416
417    /**
418     * Tests whether the paragraph tag is immediately followed by the new line.
419     *
420     * @param tag html tag.
421     * @return true, if the paragraph tag is immediately followed by the new line.
422     */
423    private static boolean isImmediatelyFollowedByNewLine(DetailNode tag) {
424        final DetailNode sibling = getNextSibling(tag);
425        return sibling != null && sibling.getType() == JavadocCommentsTokenTypes.NEWLINE;
426    }
427
428    /**
429     * Custom getNextSibling method to handle different types of paragraph tag.
430     * It works for both {@code <p>} and {@code <p></p>} tags.
431     *
432     * @param tag HTML_ELEMENT tag.
433     * @return next sibling of the tag.
434     */
435    private static DetailNode getNextSibling(DetailNode tag) {
436        DetailNode nextSibling;
437        final DetailNode paragraphStartTagToken = tag.getFirstChild();
438        final DetailNode nextNode = paragraphStartTagToken.getNextSibling();
439
440        if (nextNode == null) {
441            nextSibling = tag.getNextSibling();
442        }
443        else if (nextNode.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
444            nextSibling = nextNode.getFirstChild();
445        }
446        else {
447            nextSibling = nextNode;
448        }
449
450        if (nextSibling != null
451                && nextSibling.getType() == JavadocCommentsTokenTypes.HTML_COMMENT) {
452            nextSibling = nextSibling.getNextSibling();
453        }
454        return nextSibling;
455    }
456}