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 {@literal <p>} 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 * {@literal <address>},{@literal <blockquote>}
048 * ,{@literal <div>},{@literal <dl>}
049 * ,{@literal <h1>},{@literal <h2>},{@literal <h3>},{@literal <h4>},{@literal <h5>},
050 * {@literal <h6>},{@literal <hr>}
051 * ,{@literal <ol>},{@literal <p>},{@literal <pre>}
052 * ,{@literal <table>},{@literal <ul>}.
053 * </li>
054 * </ul>
055 *
056 * <p><b>ATTENTION:</b></p>
057 *
058 * <p>This Check ignores HTML comments.</p>
059 *
060 * <p>The Check ignores all the nested paragraph tags,
061 * it will not give any kind of violation if the paragraph tag is nested.
062 * It also ignores paragraph tags inside block tags.</p>
063 *
064 * @since 6.0
065 */
066@StatelessCheck
067public class JavadocParagraphCheck extends AbstractJavadocCheck {
068
069    /**
070     * A key is pointing to the warning message text in "messages.properties"
071     * file.
072     */
073    public static final String MSG_TAG_AFTER = "javadoc.paragraph.tag.after";
074
075    /**
076     * A key is pointing to the warning message text in "messages.properties"
077     * file.
078     */
079    public static final String MSG_LINE_BEFORE = "javadoc.paragraph.line.before";
080
081    /**
082     * A key is pointing to the warning message text in "messages.properties"
083     * file.
084     */
085    public static final String MSG_REDUNDANT_PARAGRAPH = "javadoc.paragraph.redundant.paragraph";
086
087    /**
088     * A key is pointing to the warning message text in "messages.properties"
089     * file.
090     */
091    public static final String MSG_MISPLACED_TAG = "javadoc.paragraph.misplaced.tag";
092
093    /**
094     * A key is pointing to the warning message text in "messages.properties"
095     * file.
096     */
097    public static final String MSG_PRECEDED_BLOCK_TAG = "javadoc.paragraph.preceded.block.tag";
098
099    /**
100     * Constant for the paragraph tag name.
101     */
102    private static final String PARAGRAPH_TAG = "p";
103
104    /**
105     * Set of block tags supported by this check.
106     */
107    private static final Set<String> BLOCK_TAGS =
108            Set.of("address", "blockquote", "div", "dl",
109                   "h1", "h2", "h3", "h4", "h5", "h6", "hr",
110                   "ol", PARAGRAPH_TAG, "pre", "table", "ul");
111
112    /**
113     * Control whether the {@literal <p>} tag should be placed immediately before the first word.
114     */
115    private boolean allowNewlineParagraph = true;
116
117    /**
118     * Creates a new {@code JavadocParagraphCheck} instance.
119     */
120    public JavadocParagraphCheck() {
121        // no code by default
122    }
123
124    /**
125     * Setter to control whether the {@literal <p>} tag should be placed
126     * immediately before the first word.
127     *
128     * @param value value to set.
129     * @since 6.9
130     */
131    public void setAllowNewlineParagraph(boolean value) {
132        allowNewlineParagraph = value;
133    }
134
135    @Override
136    public int[] getDefaultJavadocTokens() {
137        return new int[] {
138            JavadocCommentsTokenTypes.NEWLINE,
139            JavadocCommentsTokenTypes.HTML_ELEMENT,
140        };
141    }
142
143    @Override
144    public int[] getRequiredJavadocTokens() {
145        return getAcceptableJavadocTokens();
146    }
147
148    @Override
149    public void visitJavadocToken(DetailNode ast) {
150        if (ast.getType() == JavadocCommentsTokenTypes.NEWLINE && isEmptyLine(ast)) {
151            checkEmptyLine(ast);
152        }
153        else if (JavadocUtil.isTag(ast, PARAGRAPH_TAG)) {
154            checkParagraphTag(ast);
155        }
156    }
157
158    /**
159     * Determines whether or not the next line after empty line has paragraph tag in the beginning.
160     *
161     * @param newline NEWLINE node.
162     */
163    private void checkEmptyLine(DetailNode newline) {
164        final DetailNode nearestToken = getNearestNode(newline);
165        if (nearestToken != null && nearestToken.getType() == JavadocCommentsTokenTypes.TEXT
166                && !CommonUtil.isBlank(nearestToken.getText())) {
167            log(newline, MSG_TAG_AFTER);
168        }
169    }
170
171    /**
172     * Determines whether or not the line with paragraph tag has previous empty line.
173     *
174     * @param tag html tag.
175     */
176    private void checkParagraphTag(DetailNode tag) {
177        if (!isNestedParagraph(tag) && !isInsideBlockTag(tag)) {
178            final DetailNode newLine = getNearestEmptyLine(tag);
179            if (isFirstParagraph(tag)) {
180                log(tag, MSG_REDUNDANT_PARAGRAPH);
181            }
182            else if (newLine == null || tag.getLineNumber() - newLine.getLineNumber() != 1) {
183                log(tag, MSG_LINE_BEFORE);
184            }
185
186            final String blockTagName = findFollowedBlockTagName(tag);
187            if (blockTagName != null) {
188                log(tag, MSG_PRECEDED_BLOCK_TAG, blockTagName);
189            }
190
191            if (!allowNewlineParagraph && isImmediatelyFollowedByNewLine(tag)) {
192                log(tag, MSG_MISPLACED_TAG);
193            }
194            if (isImmediatelyFollowedByText(tag)) {
195                log(tag, MSG_MISPLACED_TAG);
196            }
197        }
198    }
199
200    /**
201     * Determines whether the paragraph tag is nested.
202     *
203     * @param tag html tag.
204     * @return true, if the paragraph tag is nested.
205     */
206    private static boolean isNestedParagraph(DetailNode tag) {
207        boolean nested = false;
208        DetailNode parent = tag.getParent();
209
210        while (parent != null) {
211            if (parent.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT) {
212                nested = true;
213                break;
214            }
215            parent = parent.getParent();
216        }
217
218        return nested;
219    }
220
221    /**
222     * Determines whether the paragraph tag is inside javadoc block tag.
223     *
224     * @param tag html tag.
225     * @return true, if the paragraph tag is inside javadoc block tag.
226     */
227    private static boolean isInsideBlockTag(DetailNode tag) {
228        boolean result = false;
229        DetailNode parent = tag;
230
231        while (parent != null) {
232            if (parent.getType() == JavadocCommentsTokenTypes.JAVADOC_BLOCK_TAG) {
233                result = true;
234                break;
235            }
236            parent = parent.getParent();
237        }
238
239        return result;
240    }
241
242    /**
243     * Determines whether or not the paragraph tag is followed by block tag.
244     *
245     * @param tag html tag.
246     * @return block tag if the paragraph tag is followed by block tag or null if not found.
247     */
248    @Nullable
249    private static String findFollowedBlockTagName(DetailNode tag) {
250        final DetailNode htmlElement = findFirstHtmlElementAfter(tag);
251        String blockTagName = null;
252
253        if (htmlElement != null) {
254            blockTagName = getHtmlElementName(htmlElement);
255        }
256
257        return blockTagName;
258    }
259
260    /**
261     * Finds and returns first html element after the tag.
262     *
263     * @param tag html tag.
264     * @return first html element after the paragraph tag or null if not found.
265     */
266    @Nullable
267    private static DetailNode findFirstHtmlElementAfter(DetailNode tag) {
268        DetailNode htmlElement = getNextSibling(tag);
269
270        while (htmlElement != null
271                && htmlElement.getType() != JavadocCommentsTokenTypes.HTML_ELEMENT) {
272            if (htmlElement.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
273                htmlElement = htmlElement.getFirstChild();
274            }
275            else if (htmlElement.getType() == JavadocCommentsTokenTypes.TEXT
276                    && !CommonUtil.isBlank(htmlElement.getText())) {
277                htmlElement = null;
278                break;
279            }
280            else {
281                htmlElement = htmlElement.getNextSibling();
282            }
283        }
284        if (htmlElement != null
285                && JavadocUtil.findFirstToken(htmlElement,
286                        JavadocCommentsTokenTypes.HTML_TAG_END) == null) {
287            htmlElement = null;
288        }
289
290        return htmlElement;
291    }
292
293    /**
294     * Finds and returns first block-level html element name.
295     *
296     * @param htmlElement block-level html tag.
297     * @return block-level html element name or null if not found.
298     */
299    @Nullable
300    private static String getHtmlElementName(DetailNode htmlElement) {
301        final DetailNode htmlTagStart = htmlElement.getFirstChild();
302        final DetailNode htmlTagName =
303                JavadocUtil.findFirstToken(htmlTagStart, JavadocCommentsTokenTypes.TAG_NAME);
304        String blockTagName = null;
305        if (BLOCK_TAGS.contains(htmlTagName.getText())) {
306            blockTagName = htmlTagName.getText();
307        }
308
309        return blockTagName;
310    }
311
312    /**
313     * Returns nearest node.
314     *
315     * @param node DetailNode node.
316     * @return nearest node.
317     */
318    private static DetailNode getNearestNode(DetailNode node) {
319        DetailNode currentNode = node;
320        while (currentNode != null
321                && (currentNode.getType() == JavadocCommentsTokenTypes.LEADING_ASTERISK
322                    || currentNode.getType() == JavadocCommentsTokenTypes.NEWLINE)) {
323            currentNode = currentNode.getNextSibling();
324        }
325        if (currentNode != null
326                && currentNode.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
327            currentNode = currentNode.getFirstChild();
328        }
329        return currentNode;
330    }
331
332    /**
333     * Determines whether or not the line is empty line.
334     *
335     * @param newLine NEWLINE node.
336     * @return true, if line is empty line.
337     */
338    private static boolean isEmptyLine(DetailNode newLine) {
339        boolean result = false;
340        DetailNode previousSibling = newLine.getPreviousSibling();
341        if (previousSibling != null && (previousSibling.getParent().getType()
342                == JavadocCommentsTokenTypes.JAVADOC_CONTENT
343                || insideNonTightHtml(previousSibling))) {
344            if (previousSibling.getType() == JavadocCommentsTokenTypes.TEXT
345                    && CommonUtil.isBlank(previousSibling.getText())) {
346                previousSibling = previousSibling.getPreviousSibling();
347            }
348            result = previousSibling != null
349                    && previousSibling.getType() == JavadocCommentsTokenTypes.LEADING_ASTERISK;
350        }
351        return result;
352    }
353
354    /**
355     * Checks whether the given node is inside a non-tight HTML element.
356     *
357     * @param previousSibling the node to check
358     * @return true if inside non-tight HTML, false otherwise
359     */
360    private static boolean insideNonTightHtml(DetailNode previousSibling) {
361        final DetailNode parent = previousSibling.getParent();
362        DetailNode htmlElement = parent;
363        if (parent.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
364            htmlElement = parent.getParent();
365        }
366        return htmlElement.getType() == JavadocCommentsTokenTypes.HTML_ELEMENT
367                && JavadocUtil.findFirstToken(htmlElement,
368                    JavadocCommentsTokenTypes.HTML_TAG_END) == null;
369    }
370
371    /**
372     * Determines whether or not the line with paragraph tag is first line in javadoc.
373     *
374     * @param paragraphTag paragraph tag.
375     * @return true, if line with paragraph tag is first line in javadoc.
376     */
377    private static boolean isFirstParagraph(DetailNode paragraphTag) {
378        boolean result = true;
379        DetailNode previousNode = paragraphTag.getPreviousSibling();
380        while (previousNode != null) {
381            if (previousNode.getType() == JavadocCommentsTokenTypes.TEXT
382                    && !CommonUtil.isBlank(previousNode.getText())
383                || previousNode.getType() != JavadocCommentsTokenTypes.LEADING_ASTERISK
384                    && previousNode.getType() != JavadocCommentsTokenTypes.NEWLINE
385                    && previousNode.getType() != JavadocCommentsTokenTypes.TEXT) {
386                result = false;
387                break;
388            }
389            previousNode = previousNode.getPreviousSibling();
390        }
391        return result;
392    }
393
394    /**
395     * Finds and returns nearest empty line in javadoc.
396     *
397     * @param node DetailNode node.
398     * @return Some nearest empty line in javadoc.
399     */
400    private static DetailNode getNearestEmptyLine(DetailNode node) {
401        DetailNode newLine = node;
402        while (newLine != null) {
403            final DetailNode previousSibling = newLine.getPreviousSibling();
404            if (newLine.getType() == JavadocCommentsTokenTypes.NEWLINE && isEmptyLine(newLine)) {
405                break;
406            }
407            newLine = previousSibling;
408        }
409        return newLine;
410    }
411
412    /**
413     * Tests whether the paragraph tag is immediately followed by the text.
414     *
415     * @param tag html tag.
416     * @return true, if the paragraph tag is immediately followed by the text.
417     */
418    private static boolean isImmediatelyFollowedByText(DetailNode tag) {
419        final DetailNode nextSibling = getNextSibling(tag);
420
421        return nextSibling == null || nextSibling.getText().startsWith(" ");
422    }
423
424    /**
425     * Tests whether the paragraph tag is immediately followed by the new line.
426     *
427     * @param tag html tag.
428     * @return true, if the paragraph tag is immediately followed by the new line.
429     */
430    private static boolean isImmediatelyFollowedByNewLine(DetailNode tag) {
431        final DetailNode sibling = getNextSibling(tag);
432        return sibling != null && sibling.getType() == JavadocCommentsTokenTypes.NEWLINE;
433    }
434
435    /**
436     * Custom getNextSibling method to handle different types of paragraph tag.
437     * It works for both {@code <p>} and {@code <p></p>} tags.
438     *
439     * @param tag HTML_ELEMENT tag.
440     * @return next sibling of the tag.
441     */
442    private static DetailNode getNextSibling(DetailNode tag) {
443        DetailNode nextSibling;
444        final DetailNode paragraphStartTagToken = tag.getFirstChild();
445        final DetailNode nextNode = paragraphStartTagToken.getNextSibling();
446
447        if (nextNode == null) {
448            nextSibling = tag.getNextSibling();
449        }
450        else if (nextNode.getType() == JavadocCommentsTokenTypes.HTML_CONTENT) {
451            nextSibling = nextNode.getFirstChild();
452        }
453        else {
454            nextSibling = nextNode;
455        }
456
457        if (nextSibling != null
458                && nextSibling.getType() == JavadocCommentsTokenTypes.HTML_COMMENT) {
459            nextSibling = nextSibling.getNextSibling();
460        }
461        return nextSibling;
462    }
463
464}