xref: /plugin/combo/renderer/analytics.php (revision 722648ea66ebf5492a297f883b2890d60f7e9bfa)
1007225e5Sgerardnico<?php
2007225e5Sgerardnico
3007225e5Sgerardnico
4007225e5Sgerardnicouse ComboStrap\Analytics;
5007225e5Sgerardnicouse ComboStrap\LinkUtility;
67c33ecc6Sgerardnicouse ComboStrap\StringUtility;
77c33ecc6Sgerardnico
871f916b9Sgerardnicouse ComboStrap\Page;
9007225e5Sgerardnicouse dokuwiki\ChangeLog\PageChangeLog;
10007225e5Sgerardnico
11007225e5Sgerardnicorequire_once(__DIR__ . '/../class/LowQualityPage.php');
12007225e5Sgerardnicorequire_once(__DIR__ . '/../class/Analytics.php');
13007225e5Sgerardnico
14007225e5Sgerardnico
15007225e5Sgerardnico/**
16007225e5Sgerardnico * A analysis Renderer that exports stats/quality/metadata in a json format
17007225e5Sgerardnico * You can export the data with
18007225e5Sgerardnico * doku.php?id=somepage&do=export_combo_analytics
19007225e5Sgerardnico */
20007225e5Sgerardnicoclass renderer_plugin_combo_analytics extends Doku_Renderer
21007225e5Sgerardnico{
227c33ecc6Sgerardnico
23007225e5Sgerardnico    const DATE_CREATED = 'date_created';
24007225e5Sgerardnico    const PLAINTEXT = 'formatted';
25007225e5Sgerardnico    const RESULT = "result";
26007225e5Sgerardnico    const DESCRIPTION = "description";
27007225e5Sgerardnico    const PASSED = "Passed";
28007225e5Sgerardnico    const FAILED = "Failed";
29007225e5Sgerardnico    const FIXME = 'fixme';
30007225e5Sgerardnico
31007225e5Sgerardnico    /**
32007225e5Sgerardnico     * Rules key
33007225e5Sgerardnico     */
34007225e5Sgerardnico    const RULE_WORDS_MINIMAL = 'words_min';
35007225e5Sgerardnico    const RULE_OUTLINE_STRUCTURE = "outline_structure";
36007225e5Sgerardnico    const RULE_INTERNAL_BACKLINKS_MIN = 'internal_backlinks_min';
37007225e5Sgerardnico    const RULE_WORDS_MAXIMAL = "words_max";
38007225e5Sgerardnico    const RULE_AVERAGE_WORDS_BY_SECTION_MIN = 'words_by_section_avg_min';
39007225e5Sgerardnico    const RULE_AVERAGE_WORDS_BY_SECTION_MAX = 'words_by_section_avg_max';
40007225e5Sgerardnico    const RULE_INTERNAL_LINKS_MIN = 'internal_links_min';
41007225e5Sgerardnico    const RULE_INTERNAL_BROKEN_LINKS_MAX = 'internal_links_broken_max';
42007225e5Sgerardnico    const RULE_DESCRIPTION_PRESENT = 'description_present';
43007225e5Sgerardnico    const RULE_FIXME = "fixme_min";
44007225e5Sgerardnico    const RULE_TITLE_PRESENT = "title_present";
45007225e5Sgerardnico    const RULE_CANONICAL_PRESENT = "canonical_present";
46007225e5Sgerardnico
47007225e5Sgerardnico    /**
48007225e5Sgerardnico     * The default man
49007225e5Sgerardnico     */
50007225e5Sgerardnico    const CONF_MANDATORY_QUALITY_RULES_DEFAULT_VALUE = [
51007225e5Sgerardnico        self::RULE_WORDS_MINIMAL,
52007225e5Sgerardnico        self::RULE_INTERNAL_BACKLINKS_MIN,
53007225e5Sgerardnico        self::RULE_INTERNAL_LINKS_MIN
54007225e5Sgerardnico    ];
55007225e5Sgerardnico    const CONF_MANDATORY_QUALITY_RULES = "mandatoryQualityRules";
56007225e5Sgerardnico
57007225e5Sgerardnico    /**
58007225e5Sgerardnico     * Quality Score factors
59007225e5Sgerardnico     * They are used to calculate the score
60007225e5Sgerardnico     */
61007225e5Sgerardnico    const CONF_QUALITY_SCORE_INTERNAL_BACKLINK_FACTOR = 'qualityScoreInternalBacklinksFactor';
62007225e5Sgerardnico    const CONF_QUALITY_SCORE_INTERNAL_LINK_FACTOR = 'qualityScoreInternalLinksFactor';
63007225e5Sgerardnico    const CONF_QUALITY_SCORE_TITLE_PRESENT = 'qualityScoreTitlePresent';
64007225e5Sgerardnico    const CONF_QUALITY_SCORE_CORRECT_HEADER_STRUCTURE = 'qualityScoreCorrectOutline';
65007225e5Sgerardnico    const CONF_QUALITY_SCORE_CORRECT_CONTENT = 'qualityScoreCorrectContentLength';
66007225e5Sgerardnico    const CONF_QUALITY_SCORE_NO_FIXME = 'qualityScoreNoFixMe';
67007225e5Sgerardnico    const CONF_QUALITY_SCORE_CORRECT_WORD_SECTION_AVERAGE = 'qualityScoreCorrectWordSectionAvg';
68007225e5Sgerardnico    const CONF_QUALITY_SCORE_INTERNAL_LINK_BROKEN_FACTOR = 'qualityScoreNoBrokenLinks';
69007225e5Sgerardnico    const CONF_QUALITY_SCORE_CHANGES_FACTOR = 'qualityScoreChangesFactor';
70007225e5Sgerardnico    const CONF_QUALITY_SCORE_DESCRIPTION_PRESENT = 'qualityScoreDescriptionPresent';
71007225e5Sgerardnico    const CONF_QUALITY_SCORE_CANONICAL_PRESENT = 'qualityScoreCanonicalPresent';
7208ca4f85Sgerardnico    const SCORING = "scoring";
7308ca4f85Sgerardnico    const SCORE = "score";
74007225e5Sgerardnico
75007225e5Sgerardnico
76007225e5Sgerardnico    /**
77007225e5Sgerardnico     * The processing data
78007225e5Sgerardnico     * that should be {@link  renderer_plugin_combo_analysis::reset()}
79007225e5Sgerardnico     */
80007225e5Sgerardnico    public $stats = array(); // the stats
81007225e5Sgerardnico    protected $metadata = array(); // the metadata
82007225e5Sgerardnico    protected $headerId = 0; // the id of the header on the page (first, second, ...)
83007225e5Sgerardnico
84007225e5Sgerardnico    /**
85007225e5Sgerardnico     * Don't known this variable ?
86007225e5Sgerardnico     */
87007225e5Sgerardnico    protected $quotelevel = 0;
88007225e5Sgerardnico    protected $formattingBracket = 0;
89007225e5Sgerardnico    protected $tableopen = false;
90007225e5Sgerardnico    private $plainTextId = 0;
912c067407Sgerardnico    /**
922c067407Sgerardnico     * @var Page
932c067407Sgerardnico     */
942c067407Sgerardnico    private $page;
952c067407Sgerardnico
962c067407Sgerardnico    public function document_start()
972c067407Sgerardnico    {
987c33ecc6Sgerardnico        $this->reset();
992c067407Sgerardnico        global $ID;
1002c067407Sgerardnico        $this->page = new Page($ID);
1012c067407Sgerardnico
1022c067407Sgerardnico    }
103007225e5Sgerardnico
104007225e5Sgerardnico
105007225e5Sgerardnico    /**
106007225e5Sgerardnico     * Here the score is calculated
107007225e5Sgerardnico     */
108007225e5Sgerardnico    public function document_end() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
109007225e5Sgerardnico    {
110007225e5Sgerardnico        /**
111f3748b38Sgerardnico         * The exported object
112f3748b38Sgerardnico         */
113f3748b38Sgerardnico        $statExport = $this->stats;
114f3748b38Sgerardnico
115f3748b38Sgerardnico        /**
116007225e5Sgerardnico         * The metadata
117007225e5Sgerardnico         */
118007225e5Sgerardnico        global $ID;
119007225e5Sgerardnico        $meta = p_get_metadata($ID);
120007225e5Sgerardnico
121007225e5Sgerardnico        /**
122f3748b38Sgerardnico         * Edit author stats
123f3748b38Sgerardnico         */
124f3748b38Sgerardnico        $changelog = new PageChangeLog($ID);
125f3748b38Sgerardnico        $revs = $changelog->getRevisions(0, 10000);
126f3748b38Sgerardnico        array_push($revs, $meta['last_change']['date']);
127f3748b38Sgerardnico        $statExport[Analytics::EDITS_COUNT] = count($revs);
128f3748b38Sgerardnico        foreach ($revs as $rev) {
129f3748b38Sgerardnico            $info = $changelog->getRevisionInfo($rev);
130f3748b38Sgerardnico            if ($info['user']) {
131f3748b38Sgerardnico                $statExport['authors'][$info['user']] += 1;
132f3748b38Sgerardnico            } else {
133f3748b38Sgerardnico                $statExport['authors']['*'] += 1;
134f3748b38Sgerardnico            }
135f3748b38Sgerardnico        }
136f3748b38Sgerardnico
137f3748b38Sgerardnico        /**
138007225e5Sgerardnico         * Word and chars count
139007225e5Sgerardnico         * The word count does not take into account
140007225e5Sgerardnico         * words with non-words characters such as < =
141007225e5Sgerardnico         * Therefore the node and attribute are not taken in the count
142007225e5Sgerardnico         */
143007225e5Sgerardnico        $text = rawWiki($ID);
144f3748b38Sgerardnico        $statExport[Analytics::CHARS_COUNT] = strlen($text);
1457c33ecc6Sgerardnico        $statExport[Analytics::WORDS_COUNT] = StringUtility::getWordCount($text);
146007225e5Sgerardnico
147007225e5Sgerardnico
148007225e5Sgerardnico        /**
149007225e5Sgerardnico         * Internal link distance summary calculation
150007225e5Sgerardnico         */
151007225e5Sgerardnico        if (array_key_exists(Analytics::INTERNAL_LINK_DISTANCE, $statExport)) {
152007225e5Sgerardnico            $linkLengths = $statExport[Analytics::INTERNAL_LINK_DISTANCE];
153007225e5Sgerardnico            unset($statExport[Analytics::INTERNAL_LINK_DISTANCE]);
154007225e5Sgerardnico            $countBacklinks = count($linkLengths);
155007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = null;
156007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = null;
157007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = null;
158007225e5Sgerardnico            if ($countBacklinks > 0) {
159007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = array_sum($linkLengths) / $countBacklinks;
160007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = max($linkLengths);
161007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = min($linkLengths);
162007225e5Sgerardnico            }
163007225e5Sgerardnico        }
164007225e5Sgerardnico
165007225e5Sgerardnico        /**
166007225e5Sgerardnico         * Quality Report / Rules
167007225e5Sgerardnico         */
168007225e5Sgerardnico        // The array that hold the results of the quality rules
169007225e5Sgerardnico        $ruleResults = array();
170007225e5Sgerardnico        // The array that hold the quality score details
171007225e5Sgerardnico        $qualityScores = array();
172007225e5Sgerardnico
173007225e5Sgerardnico
174007225e5Sgerardnico        /**
175007225e5Sgerardnico         * No fixme
176007225e5Sgerardnico         */
177007225e5Sgerardnico        $fixmeCount = $this->stats[self::FIXME];
178007225e5Sgerardnico        $statExport[self::FIXME] = $fixmeCount == null ? 0 : $fixmeCount;
179007225e5Sgerardnico        if ($fixmeCount != 0) {
180007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::FAILED;
181007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = 0;
182007225e5Sgerardnico        } else {
183007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::PASSED;
1847c33ecc6Sgerardnico            $qualityScores['no_' . self::FIXME] = $this->getConf(self::CONF_QUALITY_SCORE_NO_FIXME, 1);
185007225e5Sgerardnico        }
186007225e5Sgerardnico
187007225e5Sgerardnico        /**
188007225e5Sgerardnico         * A title should be present
189007225e5Sgerardnico         */
19008ca4f85Sgerardnico        $titleScore = $this->getConf(self::CONF_QUALITY_SCORE_TITLE_PRESENT, 10);
191007225e5Sgerardnico        if (empty($this->metadata[Analytics::TITLE])) {
192007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::FAILED;
19308ca4f85Sgerardnico            $ruleInfo[self::RULE_TITLE_PRESENT] = "Add a title in the frontmatter for {$titleScore} points";
194007225e5Sgerardnico            $this->metadata[Analytics::TITLE] = $meta[Analytics::TITLE];
195007225e5Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = 0;
196007225e5Sgerardnico        } else {
1977c33ecc6Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = $titleScore;
198007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::PASSED;
199007225e5Sgerardnico        }
200007225e5Sgerardnico
201007225e5Sgerardnico        /**
202007225e5Sgerardnico         * A description should be present
203007225e5Sgerardnico         */
20408ca4f85Sgerardnico        $descScore = $this->getConf(self::CONF_QUALITY_SCORE_DESCRIPTION_PRESENT, 8);
205007225e5Sgerardnico        if (empty($this->metadata[self::DESCRIPTION])) {
206007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::FAILED;
20708ca4f85Sgerardnico            $ruleInfo[self::RULE_DESCRIPTION_PRESENT] = "Add a description in the frontmatter for {$descScore} points";
208007225e5Sgerardnico            $this->metadata[self::DESCRIPTION] = $meta[self::DESCRIPTION]["abstract"];
209007225e5Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = 0;
210007225e5Sgerardnico        } else {
2117c33ecc6Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = $descScore;
212007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::PASSED;
213007225e5Sgerardnico        }
214007225e5Sgerardnico
215007225e5Sgerardnico        /**
216007225e5Sgerardnico         * A canonical should be present
217007225e5Sgerardnico         */
21808ca4f85Sgerardnico        $canonicalScore = $this->getConf(self::CONF_QUALITY_SCORE_CANONICAL_PRESENT, 5);
21971f916b9Sgerardnico        if (empty($this->metadata[Page::CANONICAL_PROPERTY])) {
220f3748b38Sgerardnico            global $conf;
221f3748b38Sgerardnico            $root = $conf['start'];
222f3748b38Sgerardnico            if ($ID != $root) {
223007225e5Sgerardnico                $qualityScores[self::RULE_CANONICAL_PRESENT] = 0;
224007225e5Sgerardnico                $ruleResults[self::RULE_CANONICAL_PRESENT] = self::FAILED;
22508ca4f85Sgerardnico                $ruleInfo[self::RULE_CANONICAL_PRESENT] = "Add a canonical in the frontmatter for {$canonicalScore} points";
226f3748b38Sgerardnico            }
227007225e5Sgerardnico        } else {
2287c33ecc6Sgerardnico            $qualityScores[self::RULE_CANONICAL_PRESENT] = $canonicalScore;
229007225e5Sgerardnico            $ruleResults[self::RULE_CANONICAL_PRESENT] = self::PASSED;
230007225e5Sgerardnico        }
231007225e5Sgerardnico
232007225e5Sgerardnico        /**
233007225e5Sgerardnico         * Outline / Header structure
234007225e5Sgerardnico         */
235007225e5Sgerardnico        $treeError = 0;
236007225e5Sgerardnico        $headersCount = 0;
237007225e5Sgerardnico        if (array_key_exists(Analytics::HEADER_POSITION, $this->stats)) {
238007225e5Sgerardnico            $headersCount = count($this->stats[Analytics::HEADER_POSITION]);
239007225e5Sgerardnico            unset($statExport[Analytics::HEADER_POSITION]);
240007225e5Sgerardnico            for ($i = 1; $i < $headersCount; $i++) {
241007225e5Sgerardnico                $currentHeaderLevel = $this->stats['header_struct'][$i];
242007225e5Sgerardnico                $previousHeaderLevel = $this->stats['header_struct'][$i - 1];
243007225e5Sgerardnico                if ($currentHeaderLevel - $previousHeaderLevel > 1) {
244007225e5Sgerardnico                    $treeError += 1;
245007225e5Sgerardnico                    $ruleInfo[self::RULE_OUTLINE_STRUCTURE] = "The " . $i . " header (h" . $currentHeaderLevel . ") has a level bigger than its precedent (" . $previousHeaderLevel . ")";
246007225e5Sgerardnico                }
247007225e5Sgerardnico            }
248007225e5Sgerardnico        }
249007225e5Sgerardnico        if ($treeError > 0 || $headersCount == 0) {
250007225e5Sgerardnico            $qualityScores['correct_outline'] = 0;
251007225e5Sgerardnico            $ruleResults[self::RULE_OUTLINE_STRUCTURE] = self::FAILED;
252007225e5Sgerardnico            if ($headersCount == 0) {
253007225e5Sgerardnico                $ruleInfo[self::RULE_OUTLINE_STRUCTURE] = "There is no header";
254007225e5Sgerardnico            }
255007225e5Sgerardnico        } else {
256007225e5Sgerardnico            $qualityScores['correct_outline'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_HEADER_STRUCTURE, 3);
257007225e5Sgerardnico            $ruleResults[self::RULE_OUTLINE_STRUCTURE] = self::PASSED;
258007225e5Sgerardnico        }
259007225e5Sgerardnico
260007225e5Sgerardnico
261007225e5Sgerardnico        /**
262007225e5Sgerardnico         * Document length
263007225e5Sgerardnico         */
264007225e5Sgerardnico        $minimalWordCount = 50;
265007225e5Sgerardnico        $maximalWordCount = 1500;
266007225e5Sgerardnico        $correctContentLength = true;
26708ca4f85Sgerardnico        $correctLengthScore = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_CONTENT, 10);
26808ca4f85Sgerardnico        $missingWords = $minimalWordCount - $statExport[Analytics::WORDS_COUNT];
26908ca4f85Sgerardnico        if ($missingWords > 0) {
270007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::FAILED;
271007225e5Sgerardnico            $correctContentLength = false;
27208ca4f85Sgerardnico            $ruleInfo[self::RULE_WORDS_MINIMAL] = "Add {$missingWords} words to get {$correctLengthScore} points";
273007225e5Sgerardnico        } else {
274007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::PASSED;
275007225e5Sgerardnico        }
27608ca4f85Sgerardnico        $tooMuchWords = $statExport[Analytics::WORDS_COUNT] - $maximalWordCount;
27708ca4f85Sgerardnico        if ($tooMuchWords > 0) {
278007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::FAILED;
27908ca4f85Sgerardnico            $ruleInfo[self::RULE_WORDS_MAXIMAL] = "Delete {$tooMuchWords} words to get {$correctLengthScore} points";
280007225e5Sgerardnico            $correctContentLength = false;
281007225e5Sgerardnico        } else {
282007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::PASSED;
283007225e5Sgerardnico        }
284007225e5Sgerardnico        if ($correctContentLength) {
28508ca4f85Sgerardnico            $qualityScores['correct_content_length'] = $correctLengthScore;
286007225e5Sgerardnico        } else {
287007225e5Sgerardnico            $qualityScores['correct_content_length'] = 0;
288007225e5Sgerardnico        }
289007225e5Sgerardnico
290007225e5Sgerardnico
291007225e5Sgerardnico        /**
292007225e5Sgerardnico         * Average Number of words by header section to text ratio
293007225e5Sgerardnico         */
294007225e5Sgerardnico        $headers = $this->stats[Analytics::HEADERS_COUNT];
295007225e5Sgerardnico        if ($headers != null) {
296007225e5Sgerardnico            $headerCount = array_sum($headers);
297007225e5Sgerardnico            $headerCount--; // h1 is supposed to have no words
298007225e5Sgerardnico            if ($headerCount > 0) {
299007225e5Sgerardnico
300007225e5Sgerardnico                $avgWordsCountBySection = round($this->stats[Analytics::WORDS_COUNT] / $headerCount);
301007225e5Sgerardnico                $statExport['word_section_count']['avg'] = $avgWordsCountBySection;
302007225e5Sgerardnico
303007225e5Sgerardnico                /**
304007225e5Sgerardnico                 * Min words by header section
305007225e5Sgerardnico                 */
306007225e5Sgerardnico                $wordsByHeaderMin = 20;
307007225e5Sgerardnico                /**
308007225e5Sgerardnico                 * Max words by header section
309007225e5Sgerardnico                 */
310007225e5Sgerardnico                $wordsByHeaderMax = 300;
311007225e5Sgerardnico                $correctAverageWordsBySection = true;
312007225e5Sgerardnico                if ($avgWordsCountBySection < $wordsByHeaderMin) {
313007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::FAILED;
314007225e5Sgerardnico                    $correctAverageWordsBySection = false;
31508ca4f85Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = "The number of words by section is less than {$wordsByHeaderMin}";
316007225e5Sgerardnico                } else {
317007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::PASSED;
318007225e5Sgerardnico                }
319007225e5Sgerardnico                if ($avgWordsCountBySection > $wordsByHeaderMax) {
320007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::FAILED;
321007225e5Sgerardnico                    $correctAverageWordsBySection = false;
322007225e5Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = "The number of words by section is more than {$wordsByHeaderMax}";
323007225e5Sgerardnico                } else {
324007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::PASSED;
325007225e5Sgerardnico                }
326007225e5Sgerardnico                if ($correctAverageWordsBySection) {
327007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_WORD_SECTION_AVERAGE, 10);
328007225e5Sgerardnico                } else {
329007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = 0;
330007225e5Sgerardnico                }
331007225e5Sgerardnico
332007225e5Sgerardnico            }
333007225e5Sgerardnico        }
334007225e5Sgerardnico
335007225e5Sgerardnico        /**
336007225e5Sgerardnico         * Internal Backlinks rule
337007225e5Sgerardnico         *
338007225e5Sgerardnico         * If a page is a low quality page, if the process run
339007225e5Sgerardnico         * anonymous, we will not see all {@link ft_backlinks()}
340007225e5Sgerardnico         * we use then the index directly to avoid confusion
341007225e5Sgerardnico         */
342007225e5Sgerardnico        $backlinks = idx_get_indexer()->lookupKey('relation_references', $ID);
343007225e5Sgerardnico        $countBacklinks = count($backlinks);
344007225e5Sgerardnico        $statExport[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks;
345d262537cSgerardnico        $backlinkScore = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_BACKLINK_FACTOR, 1);
346007225e5Sgerardnico        if ($countBacklinks == 0) {
347007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = 0;
348007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::FAILED;
349d262537cSgerardnico            $ruleInfo[self::RULE_INTERNAL_BACKLINKS_MIN] = "Add backlinks for {$backlinkScore} point each";
350007225e5Sgerardnico        } else {
351d262537cSgerardnico
352d262537cSgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks * $backlinkScore;
353007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::PASSED;
354007225e5Sgerardnico        }
355007225e5Sgerardnico
356007225e5Sgerardnico        /**
357007225e5Sgerardnico         * Internal links
358007225e5Sgerardnico         */
359007225e5Sgerardnico        $internalLinksCount = $this->stats[Analytics::INTERNAL_LINKS_COUNT];
360d262537cSgerardnico        $internalLinkScore = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_FACTOR, 1);
361007225e5Sgerardnico        if ($internalLinksCount == 0) {
362007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = 0;
363007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::FAILED;
364d262537cSgerardnico            $ruleInfo[self::RULE_INTERNAL_LINKS_MIN] = "Add internal links for {$internalLinkScore} point each";
365007225e5Sgerardnico        } else {
366007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::PASSED;
367d262537cSgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = $countBacklinks * $internalLinkScore;
368007225e5Sgerardnico        }
369007225e5Sgerardnico
370007225e5Sgerardnico        /**
371007225e5Sgerardnico         * Broken Links
372007225e5Sgerardnico         */
373d262537cSgerardnico        $brokenLinkScore = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_BROKEN_FACTOR, 2);
374007225e5Sgerardnico        $brokenLinksCount = $this->stats[Analytics::INTERNAL_LINKS_BROKEN_COUNT];
375007225e5Sgerardnico        if ($brokenLinksCount > 2) {
376007225e5Sgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = 0;
377007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::FAILED;
378d262537cSgerardnico            $ruleInfo[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = "Delete the {$brokenLinksCount} broken links and add {$brokenLinkScore} points";
379007225e5Sgerardnico        } else {
380d262537cSgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = $brokenLinkScore;
381007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::PASSED;
382007225e5Sgerardnico        }
383007225e5Sgerardnico
384007225e5Sgerardnico        /**
385007225e5Sgerardnico         * Changes, the more changes the better
386007225e5Sgerardnico         */
3877c33ecc6Sgerardnico        $qualityScores[Analytics::EDITS_COUNT] = $this->stats[Analytics::EDITS_COUNT] * $this->getConf(self::CONF_QUALITY_SCORE_CHANGES_FACTOR, 0.25);
388007225e5Sgerardnico
389007225e5Sgerardnico
390007225e5Sgerardnico        /**
391007225e5Sgerardnico         * Quality Score
392007225e5Sgerardnico         */
393007225e5Sgerardnico        ksort($qualityScores);
394007225e5Sgerardnico        $qualityScoring = array();
39508ca4f85Sgerardnico        $qualityScoring[self::SCORE] = array_sum($qualityScores);
396007225e5Sgerardnico        $qualityScoring["scores"] = $qualityScores;
397007225e5Sgerardnico
398007225e5Sgerardnico
399007225e5Sgerardnico        /**
400007225e5Sgerardnico         * The rule that if broken will set the quality level to low
401007225e5Sgerardnico         */
402007225e5Sgerardnico        $brokenRules = array();
403007225e5Sgerardnico        foreach ($ruleResults as $ruleName => $ruleResult) {
404007225e5Sgerardnico            if ($ruleResult == self::FAILED) {
405007225e5Sgerardnico                $brokenRules[] = $ruleName;
406007225e5Sgerardnico            }
407007225e5Sgerardnico        }
408007225e5Sgerardnico        $ruleErrorCount = sizeof($brokenRules);
409007225e5Sgerardnico        if ($ruleErrorCount > 0) {
410007225e5Sgerardnico            $qualityResult = $ruleErrorCount . " quality rules errors";
411007225e5Sgerardnico        } else {
412007225e5Sgerardnico            $qualityResult = "All quality rules passed";
413007225e5Sgerardnico        }
414007225e5Sgerardnico
415007225e5Sgerardnico        /**
416007225e5Sgerardnico         * Low level
417007225e5Sgerardnico         */
418007225e5Sgerardnico        $mandatoryRules = preg_split("/,/", $this->getConf(self::CONF_MANDATORY_QUALITY_RULES));
419007225e5Sgerardnico        $mandatoryRulesBroken = [];
420007225e5Sgerardnico        foreach ($mandatoryRules as $lowLevelRule) {
421007225e5Sgerardnico            if (in_array($lowLevelRule, $brokenRules)) {
422007225e5Sgerardnico                $mandatoryRulesBroken[] = $lowLevelRule;
423007225e5Sgerardnico            }
424007225e5Sgerardnico        }
425007225e5Sgerardnico        $lowLevel = false;
426007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
427007225e5Sgerardnico            $lowLevel = true;
428007225e5Sgerardnico        }
4297c33ecc6Sgerardnico        $this->page->setLowQualityIndicator($lowLevel);
430007225e5Sgerardnico
431007225e5Sgerardnico        /**
432007225e5Sgerardnico         * Building the quality object in order
433007225e5Sgerardnico         */
434f3748b38Sgerardnico        $quality[Analytics::LOW] = $lowLevel;
435007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
436007225e5Sgerardnico            ksort($mandatoryRulesBroken);
437*722648eaSgerardnico            $quality[Analytics::FAILED_MANDATORY_RULES] = $mandatoryRulesBroken;
438007225e5Sgerardnico        }
43908ca4f85Sgerardnico        $quality[self::SCORING] = $qualityScoring;
440f3748b38Sgerardnico        $quality[Analytics::RULES][self::RESULT] = $qualityResult;
441007225e5Sgerardnico        if (!empty($ruleInfo)) {
442f3748b38Sgerardnico            $quality[Analytics::RULES]["info"] = $ruleInfo;
443007225e5Sgerardnico        }
444007225e5Sgerardnico
445007225e5Sgerardnico        ksort($ruleResults);
446f3748b38Sgerardnico        $quality[Analytics::RULES][Analytics::DETAILS] = $ruleResults;
447007225e5Sgerardnico
448007225e5Sgerardnico        /**
449007225e5Sgerardnico         * Metadata
450007225e5Sgerardnico         */
451c42a1196Sgerardnico        $title = $meta['title'];
452c42a1196Sgerardnico        $this->metadata[Analytics::TITLE] = $title;
453c42a1196Sgerardnico        if ($title != $meta['h1']) {
454c42a1196Sgerardnico            $this->metadata[Analytics::H1] = $meta['h1'];
455c42a1196Sgerardnico        }
456007225e5Sgerardnico        $timestampCreation = $meta['date']['created'];
457007225e5Sgerardnico        $this->metadata[self::DATE_CREATED] = date('Y-m-d h:i:s', $timestampCreation);
458007225e5Sgerardnico        $timestampModification = $meta['date']['modified'];
459007225e5Sgerardnico        $this->metadata[Analytics::DATE_MODIFIED] = date('Y-m-d h:i:s', $timestampModification);
460007225e5Sgerardnico        $this->metadata['age_creation'] = round((time() - $timestampCreation) / 60 / 60 / 24);
461007225e5Sgerardnico        $this->metadata['age_modification'] = round((time() - $timestampModification) / 60 / 60 / 24);
462007225e5Sgerardnico
463007225e5Sgerardnico
464007225e5Sgerardnico        /**
465007225e5Sgerardnico         * Building the Top JSON in order
466007225e5Sgerardnico         */
467007225e5Sgerardnico        global $ID;
4682c067407Sgerardnico        $finalStats = array();
4692c067407Sgerardnico        $finalStats["id"] = $ID;
470c42a1196Sgerardnico        $finalStats["date"] = date('Y-m-d H:i:s', time());
4712c067407Sgerardnico        $finalStats['metadata'] = $this->metadata;
472007225e5Sgerardnico        ksort($statExport);
4732c067407Sgerardnico        $finalStats[Analytics::STATISTICS] = $statExport;
4742c067407Sgerardnico        $finalStats[Analytics::QUALITY] = $quality; // Quality after the sort to get them at the end
475007225e5Sgerardnico
476007225e5Sgerardnico
477007225e5Sgerardnico        /**
478007225e5Sgerardnico         * The result can be seen with
479007225e5Sgerardnico         * doku.php?id=somepage&do=export_combo_analysis
4807c33ecc6Sgerardnico         *
4817c33ecc6Sgerardnico         * Set the header temporarily for the export.php file
482007225e5Sgerardnico         */
4837c33ecc6Sgerardnico        p_set_metadata(
4847c33ecc6Sgerardnico            $ID,
4857c33ecc6Sgerardnico            array("format" => array("combo_" . $this->getPluginComponent() => array("Content-Type" => 'application/json'))),
4867c33ecc6Sgerardnico            false,
4877c33ecc6Sgerardnico            false // Persistence is not needed, this is just in case this is an export
4887c33ecc6Sgerardnico        );
4892c067407Sgerardnico        $json_encoded = json_encode($finalStats, JSON_PRETTY_PRINT);
490007225e5Sgerardnico
4917c33ecc6Sgerardnico        $this->page->saveAnalytics($finalStats);
492007225e5Sgerardnico        $this->doc .= $json_encoded;
493007225e5Sgerardnico
494007225e5Sgerardnico    }
495007225e5Sgerardnico
496007225e5Sgerardnico    /**
497007225e5Sgerardnico     */
498007225e5Sgerardnico    public function getFormat()
499007225e5Sgerardnico    {
500007225e5Sgerardnico        return Analytics::RENDERER_FORMAT;
501007225e5Sgerardnico    }
502007225e5Sgerardnico
503007225e5Sgerardnico    public function internallink($id, $name = null, $search = null, $returnonly = false, $linktype = 'content')
504007225e5Sgerardnico    {
505007225e5Sgerardnico
506f3748b38Sgerardnico        $attribute = array(
507f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_ID => $id,
508f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TYPE => LinkUtility::TYPE_INTERNAL
509f3748b38Sgerardnico        );
510f3748b38Sgerardnico        LinkUtility::processLinkStats($attribute, $this->stats);
511007225e5Sgerardnico
512007225e5Sgerardnico    }
513007225e5Sgerardnico
514007225e5Sgerardnico    public function externallink($url, $name = null)
515007225e5Sgerardnico    {
516f3748b38Sgerardnico        $attribute = array(
517f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_ID => $url,
518f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TYPE => LinkUtility::TYPE_EXTERNAL,
519*722648eaSgerardnico            LinkUtility::ATTRIBUTE_NAME => $name
520f3748b38Sgerardnico        );
521f3748b38Sgerardnico        LinkUtility::processLinkStats($attribute, $this->stats);
522007225e5Sgerardnico    }
523007225e5Sgerardnico
524007225e5Sgerardnico    public function header($text, $level, $pos)
525007225e5Sgerardnico    {
526007225e5Sgerardnico        $this->stats[Analytics::HEADERS_COUNT]['h' . $level]++;
527007225e5Sgerardnico        $this->headerId++;
528007225e5Sgerardnico        $this->stats[Analytics::HEADER_POSITION][$this->headerId] = 'h' . $level;
529007225e5Sgerardnico
530007225e5Sgerardnico    }
531007225e5Sgerardnico
532007225e5Sgerardnico    public function smiley($smiley)
533007225e5Sgerardnico    {
534007225e5Sgerardnico        if ($smiley == 'FIXME') $this->stats[self::FIXME]++;
535007225e5Sgerardnico    }
536007225e5Sgerardnico
537007225e5Sgerardnico    public function linebreak()
538007225e5Sgerardnico    {
539007225e5Sgerardnico        if (!$this->tableopen) {
540007225e5Sgerardnico            $this->stats['linebreak']++;
541007225e5Sgerardnico        }
542007225e5Sgerardnico    }
543007225e5Sgerardnico
544007225e5Sgerardnico    public function table_open($maxcols = null, $numrows = null, $pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
545007225e5Sgerardnico    {
546007225e5Sgerardnico        $this->tableopen = true;
547007225e5Sgerardnico    }
548007225e5Sgerardnico
549007225e5Sgerardnico    public function table_close($pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
550007225e5Sgerardnico    {
551007225e5Sgerardnico        $this->tableopen = false;
552007225e5Sgerardnico    }
553007225e5Sgerardnico
554007225e5Sgerardnico    public function hr()
555007225e5Sgerardnico    {
556007225e5Sgerardnico        $this->stats['hr']++;
557007225e5Sgerardnico    }
558007225e5Sgerardnico
559007225e5Sgerardnico    public function quote_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
560007225e5Sgerardnico    {
561007225e5Sgerardnico        $this->stats['quote_count']++;
562007225e5Sgerardnico        $this->quotelevel++;
563007225e5Sgerardnico        $this->stats['quote_nest'] = max($this->quotelevel, $this->stats['quote_nest']);
564007225e5Sgerardnico    }
565007225e5Sgerardnico
566007225e5Sgerardnico    public function quote_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
567007225e5Sgerardnico    {
568007225e5Sgerardnico        $this->quotelevel--;
569007225e5Sgerardnico    }
570007225e5Sgerardnico
571007225e5Sgerardnico    public function strong_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
572007225e5Sgerardnico    {
573007225e5Sgerardnico        $this->formattingBracket++;
574007225e5Sgerardnico    }
575007225e5Sgerardnico
576007225e5Sgerardnico    public function strong_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
577007225e5Sgerardnico    {
578007225e5Sgerardnico        $this->formattingBracket--;
579007225e5Sgerardnico    }
580007225e5Sgerardnico
581007225e5Sgerardnico    public function emphasis_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
582007225e5Sgerardnico    {
583007225e5Sgerardnico        $this->formattingBracket++;
584007225e5Sgerardnico    }
585007225e5Sgerardnico
586007225e5Sgerardnico    public function emphasis_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
587007225e5Sgerardnico    {
588007225e5Sgerardnico        $this->formattingBracket--;
589007225e5Sgerardnico    }
590007225e5Sgerardnico
591007225e5Sgerardnico    public function underline_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
592007225e5Sgerardnico    {
593007225e5Sgerardnico        $this->formattingBracket++;
594007225e5Sgerardnico    }
595007225e5Sgerardnico
596007225e5Sgerardnico    public function underline_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
597007225e5Sgerardnico    {
598007225e5Sgerardnico        $this->formattingBracket--;
599007225e5Sgerardnico    }
600007225e5Sgerardnico
601007225e5Sgerardnico    public function cdata($text)
602007225e5Sgerardnico    {
603007225e5Sgerardnico
604007225e5Sgerardnico        /**
605007225e5Sgerardnico         * It seems that you receive cdata
606007225e5Sgerardnico         * when emphasis_open / underline_open / strong_open
607007225e5Sgerardnico         * Stats are not for them
608007225e5Sgerardnico         */
609007225e5Sgerardnico        if (!$this->formattingBracket) return;
610007225e5Sgerardnico
611007225e5Sgerardnico        $this->plainTextId++;
612007225e5Sgerardnico
613007225e5Sgerardnico        /**
614007225e5Sgerardnico         * Length
615007225e5Sgerardnico         */
616007225e5Sgerardnico        $len = strlen($text);
617007225e5Sgerardnico        $this->stats[self::PLAINTEXT][$this->plainTextId]['len'] = $len;
618007225e5Sgerardnico
619007225e5Sgerardnico
620007225e5Sgerardnico        /**
621007225e5Sgerardnico         * Multi-formatting
622007225e5Sgerardnico         */
623007225e5Sgerardnico        if ($this->formattingBracket > 1) {
624007225e5Sgerardnico            $numberOfFormats = 1 * ($this->formattingBracket - 1);
625007225e5Sgerardnico            $this->stats[self::PLAINTEXT][$this->plainTextId]['multiformat'] += $numberOfFormats;
626007225e5Sgerardnico        }
627007225e5Sgerardnico
628007225e5Sgerardnico        /**
629007225e5Sgerardnico         * Total
630007225e5Sgerardnico         */
631007225e5Sgerardnico        $this->stats[self::PLAINTEXT][0] += $len;
632007225e5Sgerardnico    }
633007225e5Sgerardnico
634007225e5Sgerardnico    public function internalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
635007225e5Sgerardnico    {
636007225e5Sgerardnico        $this->stats[Analytics::INTERNAL_MEDIAS_COUNT]++;
637007225e5Sgerardnico    }
638007225e5Sgerardnico
639007225e5Sgerardnico    public function externalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
640007225e5Sgerardnico    {
641007225e5Sgerardnico        $this->stats[Analytics::EXTERNAL_MEDIAS]++;
642007225e5Sgerardnico    }
643007225e5Sgerardnico
644007225e5Sgerardnico    public function reset()
645007225e5Sgerardnico    {
646007225e5Sgerardnico        $this->stats = array();
647007225e5Sgerardnico        $this->metadata = array();
648007225e5Sgerardnico        $this->headerId = 0;
649007225e5Sgerardnico    }
650007225e5Sgerardnico
651007225e5Sgerardnico    public function setMeta($key, $value)
652007225e5Sgerardnico    {
653007225e5Sgerardnico        $this->metadata[$key] = $value;
654007225e5Sgerardnico    }
655007225e5Sgerardnico
656007225e5Sgerardnico
657007225e5Sgerardnico}
658007225e5Sgerardnico
659