xref: /plugin/combo/renderer/analytics.php (revision d262537c0e3e167c29541034d59d4a6edf2ffecf)
1007225e5Sgerardnico<?php
2007225e5Sgerardnico
3007225e5Sgerardnico
4007225e5Sgerardnicouse ComboStrap\Analytics;
5007225e5Sgerardnicouse ComboStrap\LinkUtility;
6007225e5Sgerardnicouse ComboStrap\LogUtility;
7007225e5Sgerardnicouse ComboStrap\LowQualityPage;
8007225e5Sgerardnicouse ComboStrap\Sqlite;
9007225e5Sgerardnicouse ComboStrap\Text;
1071f916b9Sgerardnicouse ComboStrap\Page;
11007225e5Sgerardnicouse dokuwiki\ChangeLog\PageChangeLog;
12007225e5Sgerardnico
13007225e5Sgerardnicorequire_once(__DIR__ . '/../class/Text.php');
14007225e5Sgerardnicorequire_once(__DIR__ . '/../class/LowQualityPage.php');
15007225e5Sgerardnicorequire_once(__DIR__ . '/../class/Analytics.php');
16007225e5Sgerardnico
17007225e5Sgerardnico
18007225e5Sgerardnico/**
19007225e5Sgerardnico * A analysis Renderer that exports stats/quality/metadata in a json format
20007225e5Sgerardnico * You can export the data with
21007225e5Sgerardnico * doku.php?id=somepage&do=export_combo_analytics
22007225e5Sgerardnico */
23007225e5Sgerardnicoclass renderer_plugin_combo_analytics extends Doku_Renderer
24007225e5Sgerardnico{
25007225e5Sgerardnico    const DATE_CREATED = 'date_created';
26007225e5Sgerardnico    const PLAINTEXT = 'formatted';
27007225e5Sgerardnico    const RESULT = "result";
28007225e5Sgerardnico    const DESCRIPTION = "description";
29007225e5Sgerardnico    const PASSED = "Passed";
30007225e5Sgerardnico    const FAILED = "Failed";
31007225e5Sgerardnico    const FIXME = 'fixme';
32007225e5Sgerardnico
33007225e5Sgerardnico    /**
34007225e5Sgerardnico     * Rules key
35007225e5Sgerardnico     */
36007225e5Sgerardnico    const RULE_WORDS_MINIMAL = 'words_min';
37007225e5Sgerardnico    const RULE_OUTLINE_STRUCTURE = "outline_structure";
38007225e5Sgerardnico    const RULE_INTERNAL_BACKLINKS_MIN = 'internal_backlinks_min';
39007225e5Sgerardnico    const RULE_WORDS_MAXIMAL = "words_max";
40007225e5Sgerardnico    const RULE_AVERAGE_WORDS_BY_SECTION_MIN = 'words_by_section_avg_min';
41007225e5Sgerardnico    const RULE_AVERAGE_WORDS_BY_SECTION_MAX = 'words_by_section_avg_max';
42007225e5Sgerardnico    const RULE_INTERNAL_LINKS_MIN = 'internal_links_min';
43007225e5Sgerardnico    const RULE_INTERNAL_BROKEN_LINKS_MAX = 'internal_links_broken_max';
44007225e5Sgerardnico    const RULE_DESCRIPTION_PRESENT = 'description_present';
45007225e5Sgerardnico    const RULE_FIXME = "fixme_min";
46007225e5Sgerardnico    const RULE_TITLE_PRESENT = "title_present";
47007225e5Sgerardnico    const RULE_CANONICAL_PRESENT = "canonical_present";
48007225e5Sgerardnico
49007225e5Sgerardnico    /**
50007225e5Sgerardnico     * The default man
51007225e5Sgerardnico     */
52007225e5Sgerardnico    const CONF_MANDATORY_QUALITY_RULES_DEFAULT_VALUE = [
53007225e5Sgerardnico        self::RULE_WORDS_MINIMAL,
54007225e5Sgerardnico        self::RULE_INTERNAL_BACKLINKS_MIN,
55007225e5Sgerardnico        self::RULE_INTERNAL_LINKS_MIN
56007225e5Sgerardnico    ];
57007225e5Sgerardnico    const CONF_MANDATORY_QUALITY_RULES = "mandatoryQualityRules";
58007225e5Sgerardnico
59007225e5Sgerardnico    /**
60007225e5Sgerardnico     * Quality Score factors
61007225e5Sgerardnico     * They are used to calculate the score
62007225e5Sgerardnico     */
63007225e5Sgerardnico    const CONF_QUALITY_SCORE_INTERNAL_BACKLINK_FACTOR = 'qualityScoreInternalBacklinksFactor';
64007225e5Sgerardnico    const CONF_QUALITY_SCORE_INTERNAL_LINK_FACTOR = 'qualityScoreInternalLinksFactor';
65007225e5Sgerardnico    const CONF_QUALITY_SCORE_TITLE_PRESENT = 'qualityScoreTitlePresent';
66007225e5Sgerardnico    const CONF_QUALITY_SCORE_CORRECT_HEADER_STRUCTURE = 'qualityScoreCorrectOutline';
67007225e5Sgerardnico    const CONF_QUALITY_SCORE_CORRECT_CONTENT = 'qualityScoreCorrectContentLength';
68007225e5Sgerardnico    const CONF_QUALITY_SCORE_NO_FIXME = 'qualityScoreNoFixMe';
69007225e5Sgerardnico    const CONF_QUALITY_SCORE_CORRECT_WORD_SECTION_AVERAGE = 'qualityScoreCorrectWordSectionAvg';
70007225e5Sgerardnico    const CONF_QUALITY_SCORE_INTERNAL_LINK_BROKEN_FACTOR = 'qualityScoreNoBrokenLinks';
71007225e5Sgerardnico    const CONF_QUALITY_SCORE_CHANGES_FACTOR = 'qualityScoreChangesFactor';
72007225e5Sgerardnico    const CONF_QUALITY_SCORE_DESCRIPTION_PRESENT = 'qualityScoreDescriptionPresent';
73007225e5Sgerardnico    const CONF_QUALITY_SCORE_CANONICAL_PRESENT = 'qualityScoreCanonicalPresent';
7408ca4f85Sgerardnico    const SCORING = "scoring";
7508ca4f85Sgerardnico    const SCORE = "score";
76007225e5Sgerardnico
77007225e5Sgerardnico
78007225e5Sgerardnico    /**
79007225e5Sgerardnico     * The processing data
80007225e5Sgerardnico     * that should be {@link  renderer_plugin_combo_analysis::reset()}
81007225e5Sgerardnico     */
82007225e5Sgerardnico    public $stats = array(); // the stats
83007225e5Sgerardnico    protected $metadata = array(); // the metadata
84007225e5Sgerardnico    protected $headerId = 0; // the id of the header on the page (first, second, ...)
85007225e5Sgerardnico
86007225e5Sgerardnico    /**
87007225e5Sgerardnico     * Don't known this variable ?
88007225e5Sgerardnico     */
89007225e5Sgerardnico    protected $quotelevel = 0;
90007225e5Sgerardnico    protected $formattingBracket = 0;
91007225e5Sgerardnico    protected $tableopen = false;
92007225e5Sgerardnico    private $plainTextId = 0;
932c067407Sgerardnico    /**
942c067407Sgerardnico     * @var Page
952c067407Sgerardnico     */
962c067407Sgerardnico    private $page;
972c067407Sgerardnico
982c067407Sgerardnico    public function document_start()
992c067407Sgerardnico    {
1002c067407Sgerardnico        global $ID;
1012c067407Sgerardnico        $this->page = new Page($ID);
1021c5862d3Sgerardnico        $analytics = $this->page->getAnalyticsFromDb();
1032c067407Sgerardnico        if (!empty($analytics)) {
1042c067407Sgerardnico            $this->internalLinkBefore = $analytics[Analytics::STATISTICS];
1052c067407Sgerardnico        }
1062c067407Sgerardnico
1072c067407Sgerardnico    }
108007225e5Sgerardnico
109007225e5Sgerardnico
110007225e5Sgerardnico    /**
111007225e5Sgerardnico     * Here the score is calculated
112007225e5Sgerardnico     */
113007225e5Sgerardnico    public function document_end() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
114007225e5Sgerardnico    {
115007225e5Sgerardnico        /**
116f3748b38Sgerardnico         * The exported object
117f3748b38Sgerardnico         */
118f3748b38Sgerardnico        $statExport = $this->stats;
119f3748b38Sgerardnico
120f3748b38Sgerardnico        /**
121007225e5Sgerardnico         * The metadata
122007225e5Sgerardnico         */
123007225e5Sgerardnico        global $ID;
124007225e5Sgerardnico        $meta = p_get_metadata($ID);
125007225e5Sgerardnico
126007225e5Sgerardnico        /**
127f3748b38Sgerardnico         * Edit author stats
128f3748b38Sgerardnico         */
129f3748b38Sgerardnico        $changelog = new PageChangeLog($ID);
130f3748b38Sgerardnico        $revs = $changelog->getRevisions(0, 10000);
131f3748b38Sgerardnico        array_push($revs, $meta['last_change']['date']);
132f3748b38Sgerardnico        $statExport[Analytics::EDITS_COUNT] = count($revs);
133f3748b38Sgerardnico        foreach ($revs as $rev) {
134f3748b38Sgerardnico            $info = $changelog->getRevisionInfo($rev);
135f3748b38Sgerardnico            if ($info['user']) {
136f3748b38Sgerardnico                $statExport['authors'][$info['user']] += 1;
137f3748b38Sgerardnico            } else {
138f3748b38Sgerardnico                $statExport['authors']['*'] += 1;
139f3748b38Sgerardnico            }
140f3748b38Sgerardnico        }
141f3748b38Sgerardnico
142f3748b38Sgerardnico        /**
143007225e5Sgerardnico         * Word and chars count
144007225e5Sgerardnico         * The word count does not take into account
145007225e5Sgerardnico         * words with non-words characters such as < =
146007225e5Sgerardnico         * Therefore the node and attribute are not taken in the count
147007225e5Sgerardnico         */
148007225e5Sgerardnico        $text = rawWiki($ID);
149f3748b38Sgerardnico        $statExport[Analytics::CHARS_COUNT] = strlen($text);
150f3748b38Sgerardnico        $statExport[Analytics::WORDS_COUNT] = Text::getWordCount($text);
151007225e5Sgerardnico
152007225e5Sgerardnico
153007225e5Sgerardnico        /**
154007225e5Sgerardnico         * Internal link distance summary calculation
155007225e5Sgerardnico         */
156007225e5Sgerardnico        if (array_key_exists(Analytics::INTERNAL_LINK_DISTANCE, $statExport)) {
157007225e5Sgerardnico            $linkLengths = $statExport[Analytics::INTERNAL_LINK_DISTANCE];
158007225e5Sgerardnico            unset($statExport[Analytics::INTERNAL_LINK_DISTANCE]);
159007225e5Sgerardnico            $countBacklinks = count($linkLengths);
160007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = null;
161007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = null;
162007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = null;
163007225e5Sgerardnico            if ($countBacklinks > 0) {
164007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = array_sum($linkLengths) / $countBacklinks;
165007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = max($linkLengths);
166007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = min($linkLengths);
167007225e5Sgerardnico            }
168007225e5Sgerardnico        }
169007225e5Sgerardnico
170007225e5Sgerardnico        /**
171007225e5Sgerardnico         * Quality Report / Rules
172007225e5Sgerardnico         */
173007225e5Sgerardnico        // The array that hold the results of the quality rules
174007225e5Sgerardnico        $ruleResults = array();
175007225e5Sgerardnico        // The array that hold the quality score details
176007225e5Sgerardnico        $qualityScores = array();
177007225e5Sgerardnico
178007225e5Sgerardnico
179007225e5Sgerardnico        /**
180007225e5Sgerardnico         * No fixme
181007225e5Sgerardnico         */
182007225e5Sgerardnico        $fixmeCount = $this->stats[self::FIXME];
183007225e5Sgerardnico        $statExport[self::FIXME] = $fixmeCount == null ? 0 : $fixmeCount;
184007225e5Sgerardnico        if ($fixmeCount != 0) {
185007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::FAILED;
186007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = 0;
187007225e5Sgerardnico        } else {
188007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::PASSED;
189007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = $this->getConf(self::CONF_QUALITY_SCORE_NO_FIXME, 1);;
190007225e5Sgerardnico        }
191007225e5Sgerardnico
192007225e5Sgerardnico        /**
193007225e5Sgerardnico         * A title should be present
194007225e5Sgerardnico         */
19508ca4f85Sgerardnico        $titleScore = $this->getConf(self::CONF_QUALITY_SCORE_TITLE_PRESENT, 10);
196007225e5Sgerardnico        if (empty($this->metadata[Analytics::TITLE])) {
197007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::FAILED;
19808ca4f85Sgerardnico            $ruleInfo[self::RULE_TITLE_PRESENT] = "Add a title in the frontmatter for {$titleScore} points";
199007225e5Sgerardnico            $this->metadata[Analytics::TITLE] = $meta[Analytics::TITLE];
200007225e5Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = 0;
201007225e5Sgerardnico        } else {
20208ca4f85Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = $titleScore;;
203007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::PASSED;
204007225e5Sgerardnico        }
205007225e5Sgerardnico
206007225e5Sgerardnico        /**
207007225e5Sgerardnico         * A description should be present
208007225e5Sgerardnico         */
20908ca4f85Sgerardnico        $descScore = $this->getConf(self::CONF_QUALITY_SCORE_DESCRIPTION_PRESENT, 8);
210007225e5Sgerardnico        if (empty($this->metadata[self::DESCRIPTION])) {
211007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::FAILED;
21208ca4f85Sgerardnico            $ruleInfo[self::RULE_DESCRIPTION_PRESENT] = "Add a description in the frontmatter for {$descScore} points";
213007225e5Sgerardnico            $this->metadata[self::DESCRIPTION] = $meta[self::DESCRIPTION]["abstract"];
214007225e5Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = 0;
215007225e5Sgerardnico        } else {
21608ca4f85Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = $descScore;;
217007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::PASSED;
218007225e5Sgerardnico        }
219007225e5Sgerardnico
220007225e5Sgerardnico        /**
221007225e5Sgerardnico         * A canonical should be present
222007225e5Sgerardnico         */
22308ca4f85Sgerardnico        $canonicalScore = $this->getConf(self::CONF_QUALITY_SCORE_CANONICAL_PRESENT, 5);
22471f916b9Sgerardnico        if (empty($this->metadata[Page::CANONICAL_PROPERTY])) {
225f3748b38Sgerardnico            global $conf;
226f3748b38Sgerardnico            $root = $conf['start'];
227f3748b38Sgerardnico            if ($ID!=$root) {
228007225e5Sgerardnico                $qualityScores[self::RULE_CANONICAL_PRESENT] = 0;
229007225e5Sgerardnico                $ruleResults[self::RULE_CANONICAL_PRESENT] = self::FAILED;
23008ca4f85Sgerardnico                $ruleInfo[self::RULE_CANONICAL_PRESENT] = "Add a canonical in the frontmatter for {$canonicalScore} points";
231f3748b38Sgerardnico            }
232007225e5Sgerardnico        } else {
23308ca4f85Sgerardnico            $qualityScores[self::RULE_CANONICAL_PRESENT] = $canonicalScore;;
234007225e5Sgerardnico            $ruleResults[self::RULE_CANONICAL_PRESENT] = self::PASSED;
235007225e5Sgerardnico        }
236007225e5Sgerardnico
237007225e5Sgerardnico        /**
238007225e5Sgerardnico         * Outline / Header structure
239007225e5Sgerardnico         */
240007225e5Sgerardnico        $treeError = 0;
241007225e5Sgerardnico        $headersCount = 0;
242007225e5Sgerardnico        if (array_key_exists(Analytics::HEADER_POSITION, $this->stats)) {
243007225e5Sgerardnico            $headersCount = count($this->stats[Analytics::HEADER_POSITION]);
244007225e5Sgerardnico            unset($statExport[Analytics::HEADER_POSITION]);
245007225e5Sgerardnico            for ($i = 1; $i < $headersCount; $i++) {
246007225e5Sgerardnico                $currentHeaderLevel = $this->stats['header_struct'][$i];
247007225e5Sgerardnico                $previousHeaderLevel = $this->stats['header_struct'][$i - 1];
248007225e5Sgerardnico                if ($currentHeaderLevel - $previousHeaderLevel > 1) {
249007225e5Sgerardnico                    $treeError += 1;
250007225e5Sgerardnico                    $ruleInfo[self::RULE_OUTLINE_STRUCTURE] = "The " . $i . " header (h" . $currentHeaderLevel . ") has a level bigger than its precedent (" . $previousHeaderLevel . ")";
251007225e5Sgerardnico                }
252007225e5Sgerardnico            }
253007225e5Sgerardnico        }
254007225e5Sgerardnico        if ($treeError > 0 || $headersCount == 0) {
255007225e5Sgerardnico            $qualityScores['correct_outline'] = 0;
256007225e5Sgerardnico            $ruleResults[self::RULE_OUTLINE_STRUCTURE] = self::FAILED;
257007225e5Sgerardnico            if ($headersCount == 0) {
258007225e5Sgerardnico                $ruleInfo[self::RULE_OUTLINE_STRUCTURE] = "There is no header";
259007225e5Sgerardnico            }
260007225e5Sgerardnico        } else {
261007225e5Sgerardnico            $qualityScores['correct_outline'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_HEADER_STRUCTURE, 3);
262007225e5Sgerardnico            $ruleResults[self::RULE_OUTLINE_STRUCTURE] = self::PASSED;
263007225e5Sgerardnico        }
264007225e5Sgerardnico
265007225e5Sgerardnico
266007225e5Sgerardnico        /**
267007225e5Sgerardnico         * Document length
268007225e5Sgerardnico         */
269007225e5Sgerardnico        $minimalWordCount = 50;
270007225e5Sgerardnico        $maximalWordCount = 1500;
271007225e5Sgerardnico        $correctContentLength = true;
27208ca4f85Sgerardnico        $correctLengthScore = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_CONTENT, 10);
27308ca4f85Sgerardnico        $missingWords = $minimalWordCount - $statExport[Analytics::WORDS_COUNT];
27408ca4f85Sgerardnico        if ($missingWords>0) {
275007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::FAILED;
276007225e5Sgerardnico            $correctContentLength = false;
27708ca4f85Sgerardnico            $ruleInfo[self::RULE_WORDS_MINIMAL] = "Add {$missingWords} words to get {$correctLengthScore} points";
278007225e5Sgerardnico        } else {
279007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::PASSED;
280007225e5Sgerardnico        }
28108ca4f85Sgerardnico        $tooMuchWords = $statExport[Analytics::WORDS_COUNT] - $maximalWordCount;
28208ca4f85Sgerardnico        if ($tooMuchWords>0) {
283007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::FAILED;
28408ca4f85Sgerardnico            $ruleInfo[self::RULE_WORDS_MAXIMAL] = "Delete {$tooMuchWords} words to get {$correctLengthScore} points";
285007225e5Sgerardnico            $correctContentLength = false;
286007225e5Sgerardnico        } else {
287007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::PASSED;
288007225e5Sgerardnico        }
289007225e5Sgerardnico        if ($correctContentLength) {
29008ca4f85Sgerardnico            $qualityScores['correct_content_length'] = $correctLengthScore;
291007225e5Sgerardnico        } else {
292007225e5Sgerardnico            $qualityScores['correct_content_length'] = 0;
293007225e5Sgerardnico        }
294007225e5Sgerardnico
295007225e5Sgerardnico
296007225e5Sgerardnico        /**
297007225e5Sgerardnico         * Average Number of words by header section to text ratio
298007225e5Sgerardnico         */
299007225e5Sgerardnico        $headers = $this->stats[Analytics::HEADERS_COUNT];
300007225e5Sgerardnico        if ($headers != null) {
301007225e5Sgerardnico            $headerCount = array_sum($headers);
302007225e5Sgerardnico            $headerCount--; // h1 is supposed to have no words
303007225e5Sgerardnico            if ($headerCount > 0) {
304007225e5Sgerardnico
305007225e5Sgerardnico                $avgWordsCountBySection = round($this->stats[Analytics::WORDS_COUNT] / $headerCount);
306007225e5Sgerardnico                $statExport['word_section_count']['avg'] = $avgWordsCountBySection;
307007225e5Sgerardnico
308007225e5Sgerardnico                /**
309007225e5Sgerardnico                 * Min words by header section
310007225e5Sgerardnico                 */
311007225e5Sgerardnico                $wordsByHeaderMin = 20;
312007225e5Sgerardnico                /**
313007225e5Sgerardnico                 * Max words by header section
314007225e5Sgerardnico                 */
315007225e5Sgerardnico                $wordsByHeaderMax = 300;
316007225e5Sgerardnico                $correctAverageWordsBySection = true;
317007225e5Sgerardnico                if ($avgWordsCountBySection < $wordsByHeaderMin) {
318007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::FAILED;
319007225e5Sgerardnico                    $correctAverageWordsBySection = false;
32008ca4f85Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = "The number of words by section is less than {$wordsByHeaderMin}";
321007225e5Sgerardnico                } else {
322007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::PASSED;
323007225e5Sgerardnico                }
324007225e5Sgerardnico                if ($avgWordsCountBySection > $wordsByHeaderMax) {
325007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::FAILED;
326007225e5Sgerardnico                    $correctAverageWordsBySection = false;
327007225e5Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = "The number of words by section is more than {$wordsByHeaderMax}";
328007225e5Sgerardnico                } else {
329007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::PASSED;
330007225e5Sgerardnico                }
331007225e5Sgerardnico                if ($correctAverageWordsBySection) {
332007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_WORD_SECTION_AVERAGE, 10);
333007225e5Sgerardnico                } else {
334007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = 0;
335007225e5Sgerardnico                }
336007225e5Sgerardnico
337007225e5Sgerardnico            }
338007225e5Sgerardnico        }
339007225e5Sgerardnico
340007225e5Sgerardnico        /**
341007225e5Sgerardnico         * Internal Backlinks rule
342007225e5Sgerardnico         *
343007225e5Sgerardnico         * If a page is a low quality page, if the process run
344007225e5Sgerardnico         * anonymous, we will not see all {@link ft_backlinks()}
345007225e5Sgerardnico         * we use then the index directly to avoid confusion
346007225e5Sgerardnico         */
347007225e5Sgerardnico        $backlinks = idx_get_indexer()->lookupKey('relation_references', $ID);
348007225e5Sgerardnico        $countBacklinks = count($backlinks);
349007225e5Sgerardnico        $statExport[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks;
350*d262537cSgerardnico        $backlinkScore = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_BACKLINK_FACTOR, 1);
351007225e5Sgerardnico        if ($countBacklinks == 0) {
352007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = 0;
353007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::FAILED;
354*d262537cSgerardnico            $ruleInfo[self::RULE_INTERNAL_BACKLINKS_MIN] = "Add backlinks for {$backlinkScore} point each";
355007225e5Sgerardnico        } else {
356*d262537cSgerardnico
357*d262537cSgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks * $backlinkScore;
358007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::PASSED;
359007225e5Sgerardnico        }
360007225e5Sgerardnico
361007225e5Sgerardnico        /**
362007225e5Sgerardnico         * Internal links
363007225e5Sgerardnico         */
364007225e5Sgerardnico        $internalLinksCount = $this->stats[Analytics::INTERNAL_LINKS_COUNT];
365*d262537cSgerardnico        $internalLinkScore = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_FACTOR, 1);
366007225e5Sgerardnico        if ($internalLinksCount == 0) {
367007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = 0;
368007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::FAILED;
369*d262537cSgerardnico            $ruleInfo[self::RULE_INTERNAL_LINKS_MIN] = "Add internal links for {$internalLinkScore} point each";
370007225e5Sgerardnico        } else {
371007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::PASSED;
372*d262537cSgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = $countBacklinks * $internalLinkScore;
373007225e5Sgerardnico        }
374007225e5Sgerardnico
375007225e5Sgerardnico        /**
376007225e5Sgerardnico         * Broken Links
377007225e5Sgerardnico         */
378*d262537cSgerardnico        $brokenLinkScore = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_BROKEN_FACTOR, 2);
379007225e5Sgerardnico        $brokenLinksCount = $this->stats[Analytics::INTERNAL_LINKS_BROKEN_COUNT];
380007225e5Sgerardnico        if ($brokenLinksCount > 2) {
381007225e5Sgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = 0;
382007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::FAILED;
383*d262537cSgerardnico            $ruleInfo[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = "Delete the {$brokenLinksCount} broken links and add {$brokenLinkScore} points";
384007225e5Sgerardnico        } else {
385*d262537cSgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = $brokenLinkScore;
386007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::PASSED;
387007225e5Sgerardnico        }
388007225e5Sgerardnico
389007225e5Sgerardnico        /**
390007225e5Sgerardnico         * Changes, the more changes the better
391007225e5Sgerardnico         */
392007225e5Sgerardnico        $qualityScores[Analytics::EDITS_COUNT] = $this->stats[Analytics::EDITS_COUNT] * $this->getConf(self::CONF_QUALITY_SCORE_CHANGES_FACTOR, 0.25);;;
393007225e5Sgerardnico
394007225e5Sgerardnico
395007225e5Sgerardnico
396007225e5Sgerardnico
397007225e5Sgerardnico        /**
398007225e5Sgerardnico         * Quality Score
399007225e5Sgerardnico         */
400007225e5Sgerardnico        ksort($qualityScores);
401007225e5Sgerardnico        $qualityScoring = array();
40208ca4f85Sgerardnico        $qualityScoring[self::SCORE] = array_sum($qualityScores);
403007225e5Sgerardnico        $qualityScoring["scores"] = $qualityScores;
404007225e5Sgerardnico
405007225e5Sgerardnico
406007225e5Sgerardnico        /**
407007225e5Sgerardnico         * The rule that if broken will set the quality level to low
408007225e5Sgerardnico         */
409007225e5Sgerardnico        $brokenRules = array();
410007225e5Sgerardnico        foreach ($ruleResults as $ruleName => $ruleResult) {
411007225e5Sgerardnico            if ($ruleResult == self::FAILED) {
412007225e5Sgerardnico                $brokenRules[] = $ruleName;
413007225e5Sgerardnico            }
414007225e5Sgerardnico        }
415007225e5Sgerardnico        $ruleErrorCount = sizeof($brokenRules);
416007225e5Sgerardnico        if ($ruleErrorCount > 0) {
417007225e5Sgerardnico            $qualityResult = $ruleErrorCount . " quality rules errors";
418007225e5Sgerardnico        } else {
419007225e5Sgerardnico            $qualityResult = "All quality rules passed";
420007225e5Sgerardnico        }
421007225e5Sgerardnico
422007225e5Sgerardnico        /**
423007225e5Sgerardnico         * Low level
424007225e5Sgerardnico         */
425007225e5Sgerardnico        $mandatoryRules = preg_split("/,/", $this->getConf(self::CONF_MANDATORY_QUALITY_RULES));
426007225e5Sgerardnico        $mandatoryRulesBroken = [];
427007225e5Sgerardnico        foreach ($mandatoryRules as $lowLevelRule) {
428007225e5Sgerardnico            if (in_array($lowLevelRule, $brokenRules)) {
429007225e5Sgerardnico                $mandatoryRulesBroken[] = $lowLevelRule;
430007225e5Sgerardnico            }
431007225e5Sgerardnico        }
432007225e5Sgerardnico        $lowLevel = false;
433007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
434007225e5Sgerardnico            $lowLevel = true;
435007225e5Sgerardnico        }
436007225e5Sgerardnico        LowQualityPage::setLowQualityPage($ID, $lowLevel);
437007225e5Sgerardnico
438007225e5Sgerardnico        /**
439007225e5Sgerardnico         * Building the quality object in order
440007225e5Sgerardnico         */
441f3748b38Sgerardnico        $quality[Analytics::LOW] = $lowLevel;
442007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
443007225e5Sgerardnico            ksort($mandatoryRulesBroken);
444007225e5Sgerardnico            $quality['failed_mandatory_rules'] = $mandatoryRulesBroken;
445007225e5Sgerardnico        }
44608ca4f85Sgerardnico        $quality[self::SCORING] = $qualityScoring;
447f3748b38Sgerardnico        $quality[Analytics::RULES][self::RESULT] = $qualityResult;
448007225e5Sgerardnico        if (!empty($ruleInfo)) {
449f3748b38Sgerardnico            $quality[Analytics::RULES]["info"] = $ruleInfo;
450007225e5Sgerardnico        }
451007225e5Sgerardnico
452007225e5Sgerardnico        ksort($ruleResults);
453f3748b38Sgerardnico        $quality[Analytics::RULES][Analytics::DETAILS] = $ruleResults;
454007225e5Sgerardnico
455007225e5Sgerardnico        /**
456007225e5Sgerardnico         * Metadata
457007225e5Sgerardnico         */
458c42a1196Sgerardnico        $title = $meta['title'];
459c42a1196Sgerardnico        $this->metadata[Analytics::TITLE] = $title;
460c42a1196Sgerardnico        if ($title != $meta['h1']) {
461c42a1196Sgerardnico            $this->metadata[Analytics::H1] = $meta['h1'];
462c42a1196Sgerardnico        }
463007225e5Sgerardnico        $timestampCreation = $meta['date']['created'];
464007225e5Sgerardnico        $this->metadata[self::DATE_CREATED] = date('Y-m-d h:i:s', $timestampCreation);
465007225e5Sgerardnico        $timestampModification = $meta['date']['modified'];
466007225e5Sgerardnico        $this->metadata[Analytics::DATE_MODIFIED] = date('Y-m-d h:i:s', $timestampModification);
467007225e5Sgerardnico        $this->metadata['age_creation'] = round((time() - $timestampCreation) / 60 / 60 / 24);
468007225e5Sgerardnico        $this->metadata['age_modification'] = round((time() - $timestampModification) / 60 / 60 / 24);
469007225e5Sgerardnico
470007225e5Sgerardnico
471f3748b38Sgerardnico
472007225e5Sgerardnico
473007225e5Sgerardnico        /**
474007225e5Sgerardnico         * Building the Top JSON in order
475007225e5Sgerardnico         */
476007225e5Sgerardnico        global $ID;
4772c067407Sgerardnico        $finalStats = array();
4782c067407Sgerardnico        $finalStats["id"] = $ID;
479c42a1196Sgerardnico        $finalStats["date"] = date('Y-m-d H:i:s', time());
4802c067407Sgerardnico        $finalStats['metadata'] = $this->metadata;
481007225e5Sgerardnico        ksort($statExport);
4822c067407Sgerardnico        $finalStats[Analytics::STATISTICS] = $statExport;
4832c067407Sgerardnico        $finalStats[Analytics::QUALITY] = $quality; // Quality after the sort to get them at the end
484007225e5Sgerardnico
485007225e5Sgerardnico
486007225e5Sgerardnico        /**
487007225e5Sgerardnico         * The result can be seen with
488007225e5Sgerardnico         * doku.php?id=somepage&do=export_combo_analysis
489007225e5Sgerardnico         */
490007225e5Sgerardnico        /**
491007225e5Sgerardnico         * Set the header for the export.php file
492007225e5Sgerardnico         */
493007225e5Sgerardnico        p_set_metadata($ID, array("format" =>
494007225e5Sgerardnico            array("combo_" . $this->getPluginComponent() => array("Content-Type" => 'application/json'))
495007225e5Sgerardnico        ));
4962c067407Sgerardnico        $json_encoded = json_encode($finalStats, JSON_PRETTY_PRINT);
497007225e5Sgerardnico
4982c067407Sgerardnico        $page = new Page($ID);
4992c067407Sgerardnico        $page->saveAnalytics($finalStats);
500007225e5Sgerardnico        $this->doc .= $json_encoded;
501007225e5Sgerardnico
502007225e5Sgerardnico    }
503007225e5Sgerardnico
504007225e5Sgerardnico    /**
505007225e5Sgerardnico     */
506007225e5Sgerardnico    public function getFormat()
507007225e5Sgerardnico    {
508007225e5Sgerardnico        return Analytics::RENDERER_FORMAT;
509007225e5Sgerardnico    }
510007225e5Sgerardnico
511007225e5Sgerardnico    public function internallink($id, $name = null, $search = null, $returnonly = false, $linktype = 'content')
512007225e5Sgerardnico    {
513007225e5Sgerardnico
514f3748b38Sgerardnico        $attribute = array(
515f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_ID => $id,
516f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TYPE => LinkUtility::TYPE_INTERNAL
517f3748b38Sgerardnico        );
518f3748b38Sgerardnico        LinkUtility::processLinkStats($attribute, $this->stats);
519007225e5Sgerardnico
520007225e5Sgerardnico    }
521007225e5Sgerardnico
522007225e5Sgerardnico    public function externallink($url, $name = null)
523007225e5Sgerardnico    {
524f3748b38Sgerardnico        $attribute = array(
525f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_ID => $url,
526f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TYPE => LinkUtility::TYPE_EXTERNAL,
527f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TITLE => $name
528f3748b38Sgerardnico        );
529f3748b38Sgerardnico        LinkUtility::processLinkStats($attribute, $this->stats);
530007225e5Sgerardnico    }
531007225e5Sgerardnico
532007225e5Sgerardnico    public function header($text, $level, $pos)
533007225e5Sgerardnico    {
534007225e5Sgerardnico        $this->stats[Analytics::HEADERS_COUNT]['h' . $level]++;
535007225e5Sgerardnico        $this->headerId++;
536007225e5Sgerardnico        $this->stats[Analytics::HEADER_POSITION][$this->headerId] = 'h' . $level;
537007225e5Sgerardnico
538007225e5Sgerardnico    }
539007225e5Sgerardnico
540007225e5Sgerardnico    public function smiley($smiley)
541007225e5Sgerardnico    {
542007225e5Sgerardnico        if ($smiley == 'FIXME') $this->stats[self::FIXME]++;
543007225e5Sgerardnico    }
544007225e5Sgerardnico
545007225e5Sgerardnico    public function linebreak()
546007225e5Sgerardnico    {
547007225e5Sgerardnico        if (!$this->tableopen) {
548007225e5Sgerardnico            $this->stats['linebreak']++;
549007225e5Sgerardnico        }
550007225e5Sgerardnico    }
551007225e5Sgerardnico
552007225e5Sgerardnico    public function table_open($maxcols = null, $numrows = null, $pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
553007225e5Sgerardnico    {
554007225e5Sgerardnico        $this->tableopen = true;
555007225e5Sgerardnico    }
556007225e5Sgerardnico
557007225e5Sgerardnico    public function table_close($pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
558007225e5Sgerardnico    {
559007225e5Sgerardnico        $this->tableopen = false;
560007225e5Sgerardnico    }
561007225e5Sgerardnico
562007225e5Sgerardnico    public function hr()
563007225e5Sgerardnico    {
564007225e5Sgerardnico        $this->stats['hr']++;
565007225e5Sgerardnico    }
566007225e5Sgerardnico
567007225e5Sgerardnico    public function quote_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
568007225e5Sgerardnico    {
569007225e5Sgerardnico        $this->stats['quote_count']++;
570007225e5Sgerardnico        $this->quotelevel++;
571007225e5Sgerardnico        $this->stats['quote_nest'] = max($this->quotelevel, $this->stats['quote_nest']);
572007225e5Sgerardnico    }
573007225e5Sgerardnico
574007225e5Sgerardnico    public function quote_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
575007225e5Sgerardnico    {
576007225e5Sgerardnico        $this->quotelevel--;
577007225e5Sgerardnico    }
578007225e5Sgerardnico
579007225e5Sgerardnico    public function strong_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
580007225e5Sgerardnico    {
581007225e5Sgerardnico        $this->formattingBracket++;
582007225e5Sgerardnico    }
583007225e5Sgerardnico
584007225e5Sgerardnico    public function strong_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
585007225e5Sgerardnico    {
586007225e5Sgerardnico        $this->formattingBracket--;
587007225e5Sgerardnico    }
588007225e5Sgerardnico
589007225e5Sgerardnico    public function emphasis_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
590007225e5Sgerardnico    {
591007225e5Sgerardnico        $this->formattingBracket++;
592007225e5Sgerardnico    }
593007225e5Sgerardnico
594007225e5Sgerardnico    public function emphasis_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
595007225e5Sgerardnico    {
596007225e5Sgerardnico        $this->formattingBracket--;
597007225e5Sgerardnico    }
598007225e5Sgerardnico
599007225e5Sgerardnico    public function underline_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
600007225e5Sgerardnico    {
601007225e5Sgerardnico        $this->formattingBracket++;
602007225e5Sgerardnico    }
603007225e5Sgerardnico
604007225e5Sgerardnico    public function underline_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
605007225e5Sgerardnico    {
606007225e5Sgerardnico        $this->formattingBracket--;
607007225e5Sgerardnico    }
608007225e5Sgerardnico
609007225e5Sgerardnico    public function cdata($text)
610007225e5Sgerardnico    {
611007225e5Sgerardnico
612007225e5Sgerardnico        /**
613007225e5Sgerardnico         * It seems that you receive cdata
614007225e5Sgerardnico         * when emphasis_open / underline_open / strong_open
615007225e5Sgerardnico         * Stats are not for them
616007225e5Sgerardnico         */
617007225e5Sgerardnico        if (!$this->formattingBracket) return;
618007225e5Sgerardnico
619007225e5Sgerardnico        $this->plainTextId++;
620007225e5Sgerardnico
621007225e5Sgerardnico        /**
622007225e5Sgerardnico         * Length
623007225e5Sgerardnico         */
624007225e5Sgerardnico        $len = strlen($text);
625007225e5Sgerardnico        $this->stats[self::PLAINTEXT][$this->plainTextId]['len'] = $len;
626007225e5Sgerardnico
627007225e5Sgerardnico
628007225e5Sgerardnico        /**
629007225e5Sgerardnico         * Multi-formatting
630007225e5Sgerardnico         */
631007225e5Sgerardnico        if ($this->formattingBracket > 1) {
632007225e5Sgerardnico            $numberOfFormats = 1 * ($this->formattingBracket - 1);
633007225e5Sgerardnico            $this->stats[self::PLAINTEXT][$this->plainTextId]['multiformat'] += $numberOfFormats;
634007225e5Sgerardnico        }
635007225e5Sgerardnico
636007225e5Sgerardnico        /**
637007225e5Sgerardnico         * Total
638007225e5Sgerardnico         */
639007225e5Sgerardnico        $this->stats[self::PLAINTEXT][0] += $len;
640007225e5Sgerardnico    }
641007225e5Sgerardnico
642007225e5Sgerardnico    public function internalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
643007225e5Sgerardnico    {
644007225e5Sgerardnico        $this->stats[Analytics::INTERNAL_MEDIAS_COUNT]++;
645007225e5Sgerardnico    }
646007225e5Sgerardnico
647007225e5Sgerardnico    public function externalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
648007225e5Sgerardnico    {
649007225e5Sgerardnico        $this->stats[Analytics::EXTERNAL_MEDIAS]++;
650007225e5Sgerardnico    }
651007225e5Sgerardnico
652007225e5Sgerardnico    public function reset()
653007225e5Sgerardnico    {
654007225e5Sgerardnico        $this->stats = array();
655007225e5Sgerardnico        $this->metadata = array();
656007225e5Sgerardnico        $this->headerId = 0;
657007225e5Sgerardnico    }
658007225e5Sgerardnico
659007225e5Sgerardnico    public function setMeta($key, $value)
660007225e5Sgerardnico    {
661007225e5Sgerardnico        $this->metadata[$key] = $value;
662007225e5Sgerardnico    }
663007225e5Sgerardnico
664007225e5Sgerardnico
665007225e5Sgerardnico}
666007225e5Sgerardnico
667