xref: /plugin/combo/renderer/analytics.php (revision f3748b38fbeea8fd460b3c8b25641c56e84a356c)
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';
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    {
982c067407Sgerardnico        global $ID;
992c067407Sgerardnico        $this->page = new Page($ID);
1001c5862d3Sgerardnico        $analytics = $this->page->getAnalyticsFromDb();
1012c067407Sgerardnico        if (!empty($analytics)) {
1022c067407Sgerardnico            $this->internalLinkBefore = $analytics[Analytics::STATISTICS];
1032c067407Sgerardnico        }
1042c067407Sgerardnico
1052c067407Sgerardnico    }
106007225e5Sgerardnico
107007225e5Sgerardnico
108007225e5Sgerardnico    /**
109007225e5Sgerardnico     * Here the score is calculated
110007225e5Sgerardnico     */
111007225e5Sgerardnico    public function document_end() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
112007225e5Sgerardnico    {
113007225e5Sgerardnico        /**
114*f3748b38Sgerardnico         * The exported object
115*f3748b38Sgerardnico         */
116*f3748b38Sgerardnico        $statExport = $this->stats;
117*f3748b38Sgerardnico
118*f3748b38Sgerardnico        /**
119007225e5Sgerardnico         * The metadata
120007225e5Sgerardnico         */
121007225e5Sgerardnico        global $ID;
122007225e5Sgerardnico        $meta = p_get_metadata($ID);
123007225e5Sgerardnico
124007225e5Sgerardnico        /**
125*f3748b38Sgerardnico         * Edit author stats
126*f3748b38Sgerardnico         */
127*f3748b38Sgerardnico        $changelog = new PageChangeLog($ID);
128*f3748b38Sgerardnico        $revs = $changelog->getRevisions(0, 10000);
129*f3748b38Sgerardnico        array_push($revs, $meta['last_change']['date']);
130*f3748b38Sgerardnico        $statExport[Analytics::EDITS_COUNT] = count($revs);
131*f3748b38Sgerardnico        foreach ($revs as $rev) {
132*f3748b38Sgerardnico            $info = $changelog->getRevisionInfo($rev);
133*f3748b38Sgerardnico            if ($info['user']) {
134*f3748b38Sgerardnico                $statExport['authors'][$info['user']] += 1;
135*f3748b38Sgerardnico            } else {
136*f3748b38Sgerardnico                $statExport['authors']['*'] += 1;
137*f3748b38Sgerardnico            }
138*f3748b38Sgerardnico        }
139*f3748b38Sgerardnico
140*f3748b38Sgerardnico        /**
141007225e5Sgerardnico         * Word and chars count
142007225e5Sgerardnico         * The word count does not take into account
143007225e5Sgerardnico         * words with non-words characters such as < =
144007225e5Sgerardnico         * Therefore the node and attribute are not taken in the count
145007225e5Sgerardnico         */
146007225e5Sgerardnico        $text = rawWiki($ID);
147*f3748b38Sgerardnico        $statExport[Analytics::CHARS_COUNT] = strlen($text);
148*f3748b38Sgerardnico        $statExport[Analytics::WORDS_COUNT] = Text::getWordCount($text);
149007225e5Sgerardnico
150007225e5Sgerardnico
151007225e5Sgerardnico        /**
152007225e5Sgerardnico         * Internal link distance summary calculation
153007225e5Sgerardnico         */
154007225e5Sgerardnico        if (array_key_exists(Analytics::INTERNAL_LINK_DISTANCE, $statExport)) {
155007225e5Sgerardnico            $linkLengths = $statExport[Analytics::INTERNAL_LINK_DISTANCE];
156007225e5Sgerardnico            unset($statExport[Analytics::INTERNAL_LINK_DISTANCE]);
157007225e5Sgerardnico            $countBacklinks = count($linkLengths);
158007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = null;
159007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = null;
160007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = null;
161007225e5Sgerardnico            if ($countBacklinks > 0) {
162007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = array_sum($linkLengths) / $countBacklinks;
163007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = max($linkLengths);
164007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = min($linkLengths);
165007225e5Sgerardnico            }
166007225e5Sgerardnico        }
167007225e5Sgerardnico
168007225e5Sgerardnico        /**
169007225e5Sgerardnico         * Quality Report / Rules
170007225e5Sgerardnico         */
171007225e5Sgerardnico        // The array that hold the results of the quality rules
172007225e5Sgerardnico        $ruleResults = array();
173007225e5Sgerardnico        // The array that hold the quality score details
174007225e5Sgerardnico        $qualityScores = array();
175007225e5Sgerardnico
176007225e5Sgerardnico
177007225e5Sgerardnico        /**
178007225e5Sgerardnico         * No fixme
179007225e5Sgerardnico         */
180007225e5Sgerardnico        $fixmeCount = $this->stats[self::FIXME];
181007225e5Sgerardnico        $statExport[self::FIXME] = $fixmeCount == null ? 0 : $fixmeCount;
182007225e5Sgerardnico        if ($fixmeCount != 0) {
183007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::FAILED;
184007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = 0;
185007225e5Sgerardnico        } else {
186007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::PASSED;
187007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = $this->getConf(self::CONF_QUALITY_SCORE_NO_FIXME, 1);;
188007225e5Sgerardnico        }
189007225e5Sgerardnico
190007225e5Sgerardnico        /**
191007225e5Sgerardnico         * A title should be present
192007225e5Sgerardnico         */
193007225e5Sgerardnico        if (empty($this->metadata[Analytics::TITLE])) {
194007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::FAILED;
195007225e5Sgerardnico            $ruleInfo[self::RULE_TITLE_PRESENT] = "A title is not present in the frontmatter";
196007225e5Sgerardnico            $this->metadata[Analytics::TITLE] = $meta[Analytics::TITLE];
197007225e5Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = 0;
198007225e5Sgerardnico        } else {
199007225e5Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = $this->getConf(self::CONF_QUALITY_SCORE_TITLE_PRESENT, 10);;
200007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::PASSED;
201007225e5Sgerardnico        }
202007225e5Sgerardnico
203007225e5Sgerardnico        /**
204007225e5Sgerardnico         * A description should be present
205007225e5Sgerardnico         */
206007225e5Sgerardnico        if (empty($this->metadata[self::DESCRIPTION])) {
207007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::FAILED;
208c25e802bSgerardnico            $ruleInfo[self::RULE_DESCRIPTION_PRESENT] = "A description is not present in the frontmatter";
209007225e5Sgerardnico            $this->metadata[self::DESCRIPTION] = $meta[self::DESCRIPTION]["abstract"];
210007225e5Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = 0;
211007225e5Sgerardnico        } else {
212007225e5Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = $this->getConf(self::CONF_QUALITY_SCORE_DESCRIPTION_PRESENT, 8);;
213007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::PASSED;
214007225e5Sgerardnico        }
215007225e5Sgerardnico
216007225e5Sgerardnico        /**
217007225e5Sgerardnico         * A canonical should be present
218007225e5Sgerardnico         */
21971f916b9Sgerardnico        if (empty($this->metadata[Page::CANONICAL_PROPERTY])) {
220*f3748b38Sgerardnico            global $conf;
221*f3748b38Sgerardnico            $root = $conf['start'];
222*f3748b38Sgerardnico            if ($ID!=$root) {
223007225e5Sgerardnico                $qualityScores[self::RULE_CANONICAL_PRESENT] = 0;
224007225e5Sgerardnico                $ruleResults[self::RULE_CANONICAL_PRESENT] = self::FAILED;
225007225e5Sgerardnico                $ruleInfo[self::RULE_CANONICAL_PRESENT] = "A canonical is not present in the frontmatter";
226*f3748b38Sgerardnico            }
227007225e5Sgerardnico        } else {
228007225e5Sgerardnico            $qualityScores[self::RULE_CANONICAL_PRESENT] = $this->getConf(self::CONF_QUALITY_SCORE_CANONICAL_PRESENT, 5);;
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        $correctContentLength = true;
267*f3748b38Sgerardnico        if ($statExport[Analytics::WORDS_COUNT] < $minimalWordCount) {
268007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::FAILED;
269007225e5Sgerardnico            $correctContentLength = false;
270007225e5Sgerardnico            $ruleInfo[self::RULE_WORDS_MINIMAL] = "The number of words is less than {$minimalWordCount}";
271007225e5Sgerardnico        } else {
272007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::PASSED;
273007225e5Sgerardnico        }
274*f3748b38Sgerardnico        if ($statExport[Analytics::WORDS_COUNT] > $maximalWordCount) {
275007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::FAILED;
276007225e5Sgerardnico            $ruleInfo[self::RULE_WORDS_MAXIMAL] = "The number of words is more than {$maximalWordCount}";
277007225e5Sgerardnico            $correctContentLength = false;
278007225e5Sgerardnico        } else {
279007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::PASSED;
280007225e5Sgerardnico        }
281007225e5Sgerardnico        if ($correctContentLength) {
282007225e5Sgerardnico            $qualityScores['correct_content_length'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_CONTENT, 10);
283007225e5Sgerardnico        } else {
284007225e5Sgerardnico            $qualityScores['correct_content_length'] = 0;
285007225e5Sgerardnico        }
286007225e5Sgerardnico
287007225e5Sgerardnico
288007225e5Sgerardnico        /**
289007225e5Sgerardnico         * Average Number of words by header section to text ratio
290007225e5Sgerardnico         */
291007225e5Sgerardnico        $headers = $this->stats[Analytics::HEADERS_COUNT];
292007225e5Sgerardnico        if ($headers != null) {
293007225e5Sgerardnico            $headerCount = array_sum($headers);
294007225e5Sgerardnico            $headerCount--; // h1 is supposed to have no words
295007225e5Sgerardnico            if ($headerCount > 0) {
296007225e5Sgerardnico
297007225e5Sgerardnico                $avgWordsCountBySection = round($this->stats[Analytics::WORDS_COUNT] / $headerCount);
298007225e5Sgerardnico                $statExport['word_section_count']['avg'] = $avgWordsCountBySection;
299007225e5Sgerardnico
300007225e5Sgerardnico                /**
301007225e5Sgerardnico                 * Min words by header section
302007225e5Sgerardnico                 */
303007225e5Sgerardnico                $wordsByHeaderMin = 20;
304007225e5Sgerardnico                /**
305007225e5Sgerardnico                 * Max words by header section
306007225e5Sgerardnico                 */
307007225e5Sgerardnico                $wordsByHeaderMax = 300;
308007225e5Sgerardnico                $correctAverageWordsBySection = true;
309007225e5Sgerardnico                if ($avgWordsCountBySection < $wordsByHeaderMin) {
310007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::FAILED;
311007225e5Sgerardnico                    $correctAverageWordsBySection = false;
312007225e5Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = "The number of words by section is less than {$wordsByHeaderMin}";
313007225e5Sgerardnico                } else {
314007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::PASSED;
315007225e5Sgerardnico                }
316007225e5Sgerardnico                if ($avgWordsCountBySection > $wordsByHeaderMax) {
317007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::FAILED;
318007225e5Sgerardnico                    $correctAverageWordsBySection = false;
319007225e5Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = "The number of words by section is more than {$wordsByHeaderMax}";
320007225e5Sgerardnico                } else {
321007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::PASSED;
322007225e5Sgerardnico                }
323007225e5Sgerardnico                if ($correctAverageWordsBySection) {
324007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_WORD_SECTION_AVERAGE, 10);
325007225e5Sgerardnico                } else {
326007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = 0;
327007225e5Sgerardnico                }
328007225e5Sgerardnico
329007225e5Sgerardnico            }
330007225e5Sgerardnico        }
331007225e5Sgerardnico
332007225e5Sgerardnico        /**
333007225e5Sgerardnico         * Internal Backlinks rule
334007225e5Sgerardnico         *
335007225e5Sgerardnico         * If a page is a low quality page, if the process run
336007225e5Sgerardnico         * anonymous, we will not see all {@link ft_backlinks()}
337007225e5Sgerardnico         * we use then the index directly to avoid confusion
338007225e5Sgerardnico         */
339007225e5Sgerardnico        $backlinks = idx_get_indexer()->lookupKey('relation_references', $ID);
340007225e5Sgerardnico        $countBacklinks = count($backlinks);
341007225e5Sgerardnico        $statExport[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks;
342007225e5Sgerardnico        if ($countBacklinks == 0) {
343007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = 0;
344007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::FAILED;
345007225e5Sgerardnico            $ruleInfo[self::RULE_INTERNAL_BACKLINKS_MIN] = "There is no backlinks";
346007225e5Sgerardnico        } else {
347007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks * $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_BACKLINK_FACTOR, 1);
348007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::PASSED;
349007225e5Sgerardnico        }
350007225e5Sgerardnico
351007225e5Sgerardnico        /**
352007225e5Sgerardnico         * Internal links
353007225e5Sgerardnico         */
354007225e5Sgerardnico        $internalLinksCount = $this->stats[Analytics::INTERNAL_LINKS_COUNT];
355007225e5Sgerardnico        if ($internalLinksCount == 0) {
356007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = 0;
357007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::FAILED;
3581c5862d3Sgerardnico            $ruleInfo[self::RULE_INTERNAL_LINKS_MIN] = "There is no internal links";
359007225e5Sgerardnico        } else {
360007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::PASSED;
361007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = $countBacklinks * $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_FACTOR, 1);;
362007225e5Sgerardnico        }
363007225e5Sgerardnico
364007225e5Sgerardnico        /**
365007225e5Sgerardnico         * Broken Links
366007225e5Sgerardnico         */
367007225e5Sgerardnico        $brokenLinksCount = $this->stats[Analytics::INTERNAL_LINKS_BROKEN_COUNT];
368007225e5Sgerardnico        if ($brokenLinksCount > 2) {
369007225e5Sgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = 0;
370007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::FAILED;
3711c5862d3Sgerardnico            $ruleInfo[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = "There is {$brokenLinksCount} broken links";
372007225e5Sgerardnico        } else {
373007225e5Sgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_BROKEN_FACTOR, 2);;;
374007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::PASSED;
375007225e5Sgerardnico        }
376007225e5Sgerardnico
377007225e5Sgerardnico        /**
378007225e5Sgerardnico         * Changes, the more changes the better
379007225e5Sgerardnico         */
380007225e5Sgerardnico        $qualityScores[Analytics::EDITS_COUNT] = $this->stats[Analytics::EDITS_COUNT] * $this->getConf(self::CONF_QUALITY_SCORE_CHANGES_FACTOR, 0.25);;;
381007225e5Sgerardnico
382007225e5Sgerardnico
383007225e5Sgerardnico
384007225e5Sgerardnico
385007225e5Sgerardnico        /**
386007225e5Sgerardnico         * Quality Score
387007225e5Sgerardnico         */
388007225e5Sgerardnico        ksort($qualityScores);
389007225e5Sgerardnico        $qualityScoring = array();
390007225e5Sgerardnico        $qualityScoring["score"] = array_sum($qualityScores);
391007225e5Sgerardnico        $qualityScoring["scores"] = $qualityScores;
392007225e5Sgerardnico
393007225e5Sgerardnico
394007225e5Sgerardnico        /**
395007225e5Sgerardnico         * The rule that if broken will set the quality level to low
396007225e5Sgerardnico         */
397007225e5Sgerardnico        $brokenRules = array();
398007225e5Sgerardnico        foreach ($ruleResults as $ruleName => $ruleResult) {
399007225e5Sgerardnico            if ($ruleResult == self::FAILED) {
400007225e5Sgerardnico                $brokenRules[] = $ruleName;
401007225e5Sgerardnico            }
402007225e5Sgerardnico        }
403007225e5Sgerardnico        $ruleErrorCount = sizeof($brokenRules);
404007225e5Sgerardnico        if ($ruleErrorCount > 0) {
405007225e5Sgerardnico            $qualityResult = $ruleErrorCount . " quality rules errors";
406007225e5Sgerardnico        } else {
407007225e5Sgerardnico            $qualityResult = "All quality rules passed";
408007225e5Sgerardnico        }
409007225e5Sgerardnico
410007225e5Sgerardnico        /**
411007225e5Sgerardnico         * Low level
412007225e5Sgerardnico         */
413007225e5Sgerardnico        $mandatoryRules = preg_split("/,/", $this->getConf(self::CONF_MANDATORY_QUALITY_RULES));
414007225e5Sgerardnico        $mandatoryRulesBroken = [];
415007225e5Sgerardnico        foreach ($mandatoryRules as $lowLevelRule) {
416007225e5Sgerardnico            if (in_array($lowLevelRule, $brokenRules)) {
417007225e5Sgerardnico                $mandatoryRulesBroken[] = $lowLevelRule;
418007225e5Sgerardnico            }
419007225e5Sgerardnico        }
420007225e5Sgerardnico        $lowLevel = false;
421007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
422007225e5Sgerardnico            $lowLevel = true;
423007225e5Sgerardnico        }
424007225e5Sgerardnico        LowQualityPage::setLowQualityPage($ID, $lowLevel);
425007225e5Sgerardnico
426007225e5Sgerardnico        /**
427007225e5Sgerardnico         * Building the quality object in order
428007225e5Sgerardnico         */
429*f3748b38Sgerardnico        $quality[Analytics::LOW] = $lowLevel;
430007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
431007225e5Sgerardnico            ksort($mandatoryRulesBroken);
432007225e5Sgerardnico            $quality['failed_mandatory_rules'] = $mandatoryRulesBroken;
433007225e5Sgerardnico        }
434007225e5Sgerardnico        $quality["scoring"] = $qualityScoring;
435*f3748b38Sgerardnico        $quality[Analytics::RULES][self::RESULT] = $qualityResult;
436007225e5Sgerardnico        if (!empty($ruleInfo)) {
437*f3748b38Sgerardnico            $quality[Analytics::RULES]["info"] = $ruleInfo;
438007225e5Sgerardnico        }
439007225e5Sgerardnico
440007225e5Sgerardnico        ksort($ruleResults);
441*f3748b38Sgerardnico        $quality[Analytics::RULES][Analytics::DETAILS] = $ruleResults;
442007225e5Sgerardnico
443007225e5Sgerardnico        /**
444007225e5Sgerardnico         * Metadata
445007225e5Sgerardnico         */
446c42a1196Sgerardnico        $title = $meta['title'];
447c42a1196Sgerardnico        $this->metadata[Analytics::TITLE] = $title;
448c42a1196Sgerardnico        if ($title != $meta['h1']) {
449c42a1196Sgerardnico            $this->metadata[Analytics::H1] = $meta['h1'];
450c42a1196Sgerardnico        }
451007225e5Sgerardnico        $timestampCreation = $meta['date']['created'];
452007225e5Sgerardnico        $this->metadata[self::DATE_CREATED] = date('Y-m-d h:i:s', $timestampCreation);
453007225e5Sgerardnico        $timestampModification = $meta['date']['modified'];
454007225e5Sgerardnico        $this->metadata[Analytics::DATE_MODIFIED] = date('Y-m-d h:i:s', $timestampModification);
455007225e5Sgerardnico        $this->metadata['age_creation'] = round((time() - $timestampCreation) / 60 / 60 / 24);
456007225e5Sgerardnico        $this->metadata['age_modification'] = round((time() - $timestampModification) / 60 / 60 / 24);
457007225e5Sgerardnico
458007225e5Sgerardnico
459*f3748b38Sgerardnico
460007225e5Sgerardnico
461007225e5Sgerardnico        /**
462007225e5Sgerardnico         * Building the Top JSON in order
463007225e5Sgerardnico         */
464007225e5Sgerardnico        global $ID;
4652c067407Sgerardnico        $finalStats = array();
4662c067407Sgerardnico        $finalStats["id"] = $ID;
467c42a1196Sgerardnico        $finalStats["date"] = date('Y-m-d H:i:s', time());
4682c067407Sgerardnico        $finalStats['metadata'] = $this->metadata;
469007225e5Sgerardnico        ksort($statExport);
4702c067407Sgerardnico        $finalStats[Analytics::STATISTICS] = $statExport;
4712c067407Sgerardnico        $finalStats[Analytics::QUALITY] = $quality; // Quality after the sort to get them at the end
472007225e5Sgerardnico
473007225e5Sgerardnico
474007225e5Sgerardnico        /**
475007225e5Sgerardnico         * The result can be seen with
476007225e5Sgerardnico         * doku.php?id=somepage&do=export_combo_analysis
477007225e5Sgerardnico         */
478007225e5Sgerardnico        /**
479007225e5Sgerardnico         * Set the header for the export.php file
480007225e5Sgerardnico         */
481007225e5Sgerardnico        p_set_metadata($ID, array("format" =>
482007225e5Sgerardnico            array("combo_" . $this->getPluginComponent() => array("Content-Type" => 'application/json'))
483007225e5Sgerardnico        ));
4842c067407Sgerardnico        $json_encoded = json_encode($finalStats, JSON_PRETTY_PRINT);
485007225e5Sgerardnico
4862c067407Sgerardnico        $page = new Page($ID);
4872c067407Sgerardnico        $page->saveAnalytics($finalStats);
488007225e5Sgerardnico        $this->doc .= $json_encoded;
489007225e5Sgerardnico
490007225e5Sgerardnico    }
491007225e5Sgerardnico
492007225e5Sgerardnico    /**
493007225e5Sgerardnico     */
494007225e5Sgerardnico    public function getFormat()
495007225e5Sgerardnico    {
496007225e5Sgerardnico        return Analytics::RENDERER_FORMAT;
497007225e5Sgerardnico    }
498007225e5Sgerardnico
499007225e5Sgerardnico    public function internallink($id, $name = null, $search = null, $returnonly = false, $linktype = 'content')
500007225e5Sgerardnico    {
501007225e5Sgerardnico
502*f3748b38Sgerardnico        $attribute = array(
503*f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_ID => $id,
504*f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TYPE => LinkUtility::TYPE_INTERNAL
505*f3748b38Sgerardnico        );
506*f3748b38Sgerardnico        LinkUtility::processLinkStats($attribute, $this->stats);
507007225e5Sgerardnico
508007225e5Sgerardnico    }
509007225e5Sgerardnico
510007225e5Sgerardnico    public function externallink($url, $name = null)
511007225e5Sgerardnico    {
512*f3748b38Sgerardnico        $attribute = array(
513*f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_ID => $url,
514*f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TYPE => LinkUtility::TYPE_EXTERNAL,
515*f3748b38Sgerardnico            LinkUtility::ATTRIBUTE_TITLE => $name
516*f3748b38Sgerardnico        );
517*f3748b38Sgerardnico        LinkUtility::processLinkStats($attribute, $this->stats);
518007225e5Sgerardnico    }
519007225e5Sgerardnico
520007225e5Sgerardnico    public function header($text, $level, $pos)
521007225e5Sgerardnico    {
522007225e5Sgerardnico        $this->stats[Analytics::HEADERS_COUNT]['h' . $level]++;
523007225e5Sgerardnico        $this->headerId++;
524007225e5Sgerardnico        $this->stats[Analytics::HEADER_POSITION][$this->headerId] = 'h' . $level;
525007225e5Sgerardnico
526007225e5Sgerardnico    }
527007225e5Sgerardnico
528007225e5Sgerardnico    public function smiley($smiley)
529007225e5Sgerardnico    {
530007225e5Sgerardnico        if ($smiley == 'FIXME') $this->stats[self::FIXME]++;
531007225e5Sgerardnico    }
532007225e5Sgerardnico
533007225e5Sgerardnico    public function linebreak()
534007225e5Sgerardnico    {
535007225e5Sgerardnico        if (!$this->tableopen) {
536007225e5Sgerardnico            $this->stats['linebreak']++;
537007225e5Sgerardnico        }
538007225e5Sgerardnico    }
539007225e5Sgerardnico
540007225e5Sgerardnico    public function table_open($maxcols = null, $numrows = null, $pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
541007225e5Sgerardnico    {
542007225e5Sgerardnico        $this->tableopen = true;
543007225e5Sgerardnico    }
544007225e5Sgerardnico
545007225e5Sgerardnico    public function table_close($pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
546007225e5Sgerardnico    {
547007225e5Sgerardnico        $this->tableopen = false;
548007225e5Sgerardnico    }
549007225e5Sgerardnico
550007225e5Sgerardnico    public function hr()
551007225e5Sgerardnico    {
552007225e5Sgerardnico        $this->stats['hr']++;
553007225e5Sgerardnico    }
554007225e5Sgerardnico
555007225e5Sgerardnico    public function quote_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
556007225e5Sgerardnico    {
557007225e5Sgerardnico        $this->stats['quote_count']++;
558007225e5Sgerardnico        $this->quotelevel++;
559007225e5Sgerardnico        $this->stats['quote_nest'] = max($this->quotelevel, $this->stats['quote_nest']);
560007225e5Sgerardnico    }
561007225e5Sgerardnico
562007225e5Sgerardnico    public function quote_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
563007225e5Sgerardnico    {
564007225e5Sgerardnico        $this->quotelevel--;
565007225e5Sgerardnico    }
566007225e5Sgerardnico
567007225e5Sgerardnico    public function strong_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
568007225e5Sgerardnico    {
569007225e5Sgerardnico        $this->formattingBracket++;
570007225e5Sgerardnico    }
571007225e5Sgerardnico
572007225e5Sgerardnico    public function strong_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
573007225e5Sgerardnico    {
574007225e5Sgerardnico        $this->formattingBracket--;
575007225e5Sgerardnico    }
576007225e5Sgerardnico
577007225e5Sgerardnico    public function emphasis_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
578007225e5Sgerardnico    {
579007225e5Sgerardnico        $this->formattingBracket++;
580007225e5Sgerardnico    }
581007225e5Sgerardnico
582007225e5Sgerardnico    public function emphasis_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
583007225e5Sgerardnico    {
584007225e5Sgerardnico        $this->formattingBracket--;
585007225e5Sgerardnico    }
586007225e5Sgerardnico
587007225e5Sgerardnico    public function underline_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
588007225e5Sgerardnico    {
589007225e5Sgerardnico        $this->formattingBracket++;
590007225e5Sgerardnico    }
591007225e5Sgerardnico
592007225e5Sgerardnico    public function underline_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
593007225e5Sgerardnico    {
594007225e5Sgerardnico        $this->formattingBracket--;
595007225e5Sgerardnico    }
596007225e5Sgerardnico
597007225e5Sgerardnico    public function cdata($text)
598007225e5Sgerardnico    {
599007225e5Sgerardnico
600007225e5Sgerardnico        /**
601007225e5Sgerardnico         * It seems that you receive cdata
602007225e5Sgerardnico         * when emphasis_open / underline_open / strong_open
603007225e5Sgerardnico         * Stats are not for them
604007225e5Sgerardnico         */
605007225e5Sgerardnico        if (!$this->formattingBracket) return;
606007225e5Sgerardnico
607007225e5Sgerardnico        $this->plainTextId++;
608007225e5Sgerardnico
609007225e5Sgerardnico        /**
610007225e5Sgerardnico         * Length
611007225e5Sgerardnico         */
612007225e5Sgerardnico        $len = strlen($text);
613007225e5Sgerardnico        $this->stats[self::PLAINTEXT][$this->plainTextId]['len'] = $len;
614007225e5Sgerardnico
615007225e5Sgerardnico
616007225e5Sgerardnico        /**
617007225e5Sgerardnico         * Multi-formatting
618007225e5Sgerardnico         */
619007225e5Sgerardnico        if ($this->formattingBracket > 1) {
620007225e5Sgerardnico            $numberOfFormats = 1 * ($this->formattingBracket - 1);
621007225e5Sgerardnico            $this->stats[self::PLAINTEXT][$this->plainTextId]['multiformat'] += $numberOfFormats;
622007225e5Sgerardnico        }
623007225e5Sgerardnico
624007225e5Sgerardnico        /**
625007225e5Sgerardnico         * Total
626007225e5Sgerardnico         */
627007225e5Sgerardnico        $this->stats[self::PLAINTEXT][0] += $len;
628007225e5Sgerardnico    }
629007225e5Sgerardnico
630007225e5Sgerardnico    public function internalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
631007225e5Sgerardnico    {
632007225e5Sgerardnico        $this->stats[Analytics::INTERNAL_MEDIAS_COUNT]++;
633007225e5Sgerardnico    }
634007225e5Sgerardnico
635007225e5Sgerardnico    public function externalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
636007225e5Sgerardnico    {
637007225e5Sgerardnico        $this->stats[Analytics::EXTERNAL_MEDIAS]++;
638007225e5Sgerardnico    }
639007225e5Sgerardnico
640007225e5Sgerardnico    public function reset()
641007225e5Sgerardnico    {
642007225e5Sgerardnico        $this->stats = array();
643007225e5Sgerardnico        $this->metadata = array();
644007225e5Sgerardnico        $this->headerId = 0;
645007225e5Sgerardnico    }
646007225e5Sgerardnico
647007225e5Sgerardnico    public function setMeta($key, $value)
648007225e5Sgerardnico    {
649007225e5Sgerardnico        $this->metadata[$key] = $value;
650007225e5Sgerardnico    }
651007225e5Sgerardnico
652007225e5Sgerardnico
653007225e5Sgerardnico}
654007225e5Sgerardnico
655