xref: /plugin/combo/renderer/analytics.php (revision 71f916b90657b0ec64f620dd9cb434dd66a4bb62)
1007225e5Sgerardnico<?php
2007225e5Sgerardnico
3007225e5Sgerardnico
4007225e5Sgerardnicouse ComboStrap\Analytics;
5007225e5Sgerardnicouse ComboStrap\LinkUtility;
6007225e5Sgerardnicouse ComboStrap\LogUtility;
7007225e5Sgerardnicouse ComboStrap\LowQualityPage;
8007225e5Sgerardnicouse ComboStrap\Sqlite;
9007225e5Sgerardnicouse ComboStrap\Text;
10*71f916b9Sgerardnicouse 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;
91007225e5Sgerardnico
92007225e5Sgerardnico
93007225e5Sgerardnico    /**
94007225e5Sgerardnico     * Here the score is calculated
95007225e5Sgerardnico     */
96007225e5Sgerardnico    public function document_end() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
97007225e5Sgerardnico    {
98007225e5Sgerardnico        /**
99007225e5Sgerardnico         * The metadata
100007225e5Sgerardnico         */
101007225e5Sgerardnico        global $ID;
102007225e5Sgerardnico        $meta = p_get_metadata($ID);
103007225e5Sgerardnico
104007225e5Sgerardnico        /**
105007225e5Sgerardnico         * Word and chars count
106007225e5Sgerardnico         * The word count does not take into account
107007225e5Sgerardnico         * words with non-words characters such as < =
108007225e5Sgerardnico         * Therefore the node and attribute are not taken in the count
109007225e5Sgerardnico         */
110007225e5Sgerardnico        $text = rawWiki($ID);
111007225e5Sgerardnico        $this->stats[Analytics::CHARS_COUNT] = strlen($text);
112007225e5Sgerardnico        $this->stats[Analytics::WORDS_COUNT] = Text::getWordCount($text);
113007225e5Sgerardnico
114007225e5Sgerardnico        /**
115007225e5Sgerardnico         * The exported object
116007225e5Sgerardnico         */
117007225e5Sgerardnico        $statExport = $this->stats;
118007225e5Sgerardnico
119007225e5Sgerardnico
120007225e5Sgerardnico        /**
121007225e5Sgerardnico         * Internal link distance summary calculation
122007225e5Sgerardnico         */
123007225e5Sgerardnico        if (array_key_exists(Analytics::INTERNAL_LINK_DISTANCE, $statExport)) {
124007225e5Sgerardnico            $linkLengths = $statExport[Analytics::INTERNAL_LINK_DISTANCE];
125007225e5Sgerardnico            unset($statExport[Analytics::INTERNAL_LINK_DISTANCE]);
126007225e5Sgerardnico            $countBacklinks = count($linkLengths);
127007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = null;
128007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = null;
129007225e5Sgerardnico            $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = null;
130007225e5Sgerardnico            if ($countBacklinks > 0) {
131007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['avg'] = array_sum($linkLengths) / $countBacklinks;
132007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['max'] = max($linkLengths);
133007225e5Sgerardnico                $statExport[Analytics::INTERNAL_LINK_DISTANCE]['min'] = min($linkLengths);
134007225e5Sgerardnico            }
135007225e5Sgerardnico        }
136007225e5Sgerardnico
137007225e5Sgerardnico        /**
138007225e5Sgerardnico         * Quality Report / Rules
139007225e5Sgerardnico         */
140007225e5Sgerardnico        // The array that hold the results of the quality rules
141007225e5Sgerardnico        $ruleResults = array();
142007225e5Sgerardnico        // The array that hold the quality score details
143007225e5Sgerardnico        $qualityScores = array();
144007225e5Sgerardnico
145007225e5Sgerardnico
146007225e5Sgerardnico        /**
147007225e5Sgerardnico         * No fixme
148007225e5Sgerardnico         */
149007225e5Sgerardnico        $fixmeCount = $this->stats[self::FIXME];
150007225e5Sgerardnico        $statExport[self::FIXME] = $fixmeCount == null ? 0 : $fixmeCount;
151007225e5Sgerardnico        if ($fixmeCount != 0) {
152007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::FAILED;
153007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = 0;
154007225e5Sgerardnico        } else {
155007225e5Sgerardnico            $ruleResults[self::RULE_FIXME] = self::PASSED;
156007225e5Sgerardnico            $qualityScores['no_' . self::FIXME] = $this->getConf(self::CONF_QUALITY_SCORE_NO_FIXME, 1);;
157007225e5Sgerardnico        }
158007225e5Sgerardnico
159007225e5Sgerardnico        /**
160007225e5Sgerardnico         * A title should be present
161007225e5Sgerardnico         */
162007225e5Sgerardnico        if (empty($this->metadata[Analytics::TITLE])) {
163007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::FAILED;
164007225e5Sgerardnico            $ruleInfo[self::RULE_TITLE_PRESENT] = "A title is not present in the frontmatter";
165007225e5Sgerardnico            $this->metadata[Analytics::TITLE] = $meta[Analytics::TITLE];
166007225e5Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = 0;
167007225e5Sgerardnico        } else {
168007225e5Sgerardnico            $qualityScores[self::RULE_TITLE_PRESENT] = $this->getConf(self::CONF_QUALITY_SCORE_TITLE_PRESENT, 10);;
169007225e5Sgerardnico            $ruleResults[self::RULE_TITLE_PRESENT] = self::PASSED;
170007225e5Sgerardnico        }
171007225e5Sgerardnico
172007225e5Sgerardnico        /**
173007225e5Sgerardnico         * A description should be present
174007225e5Sgerardnico         */
175007225e5Sgerardnico        if (empty($this->metadata[self::DESCRIPTION])) {
176007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::FAILED;
177007225e5Sgerardnico            $ruleInfo[self::RULE_CANONICAL_PRESENT] = "A description is not present in the frontmatter";
178007225e5Sgerardnico            $this->metadata[self::DESCRIPTION] = $meta[self::DESCRIPTION]["abstract"];
179007225e5Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = 0;
180007225e5Sgerardnico        } else {
181007225e5Sgerardnico            $qualityScores[self::RULE_DESCRIPTION_PRESENT] = $this->getConf(self::CONF_QUALITY_SCORE_DESCRIPTION_PRESENT, 8);;
182007225e5Sgerardnico            $ruleResults[self::RULE_DESCRIPTION_PRESENT] = self::PASSED;
183007225e5Sgerardnico        }
184007225e5Sgerardnico
185007225e5Sgerardnico        /**
186007225e5Sgerardnico         * A canonical should be present
187007225e5Sgerardnico         */
188*71f916b9Sgerardnico        if (empty($this->metadata[Page::CANONICAL_PROPERTY])) {
189007225e5Sgerardnico            $qualityScores[self::RULE_CANONICAL_PRESENT] = 0;
190007225e5Sgerardnico            $ruleResults[self::RULE_CANONICAL_PRESENT] = self::FAILED;
191007225e5Sgerardnico            $ruleInfo[self::RULE_CANONICAL_PRESENT] = "A canonical is not present in the frontmatter";
192007225e5Sgerardnico        } else {
193007225e5Sgerardnico            $qualityScores[self::RULE_CANONICAL_PRESENT] = $this->getConf(self::CONF_QUALITY_SCORE_CANONICAL_PRESENT, 5);;
194007225e5Sgerardnico            $ruleResults[self::RULE_CANONICAL_PRESENT] = self::PASSED;
195007225e5Sgerardnico        }
196007225e5Sgerardnico
197007225e5Sgerardnico        /**
198007225e5Sgerardnico         * Outline / Header structure
199007225e5Sgerardnico         */
200007225e5Sgerardnico        $treeError = 0;
201007225e5Sgerardnico        $headersCount = 0;
202007225e5Sgerardnico        if (array_key_exists(Analytics::HEADER_POSITION, $this->stats)) {
203007225e5Sgerardnico            $headersCount = count($this->stats[Analytics::HEADER_POSITION]);
204007225e5Sgerardnico            unset($statExport[Analytics::HEADER_POSITION]);
205007225e5Sgerardnico            for ($i = 1; $i < $headersCount; $i++) {
206007225e5Sgerardnico                $currentHeaderLevel = $this->stats['header_struct'][$i];
207007225e5Sgerardnico                $previousHeaderLevel = $this->stats['header_struct'][$i - 1];
208007225e5Sgerardnico                if ($currentHeaderLevel - $previousHeaderLevel > 1) {
209007225e5Sgerardnico                    $treeError += 1;
210007225e5Sgerardnico                    $ruleInfo[self::RULE_OUTLINE_STRUCTURE] = "The " . $i . " header (h" . $currentHeaderLevel . ") has a level bigger than its precedent (" . $previousHeaderLevel . ")";
211007225e5Sgerardnico                }
212007225e5Sgerardnico            }
213007225e5Sgerardnico        }
214007225e5Sgerardnico        if ($treeError > 0 || $headersCount == 0) {
215007225e5Sgerardnico            $qualityScores['correct_outline'] = 0;
216007225e5Sgerardnico            $ruleResults[self::RULE_OUTLINE_STRUCTURE] = self::FAILED;
217007225e5Sgerardnico            if ($headersCount==0){
218007225e5Sgerardnico                $ruleInfo[self::RULE_OUTLINE_STRUCTURE] = "There is no header";
219007225e5Sgerardnico            }
220007225e5Sgerardnico        } else {
221007225e5Sgerardnico            $qualityScores['correct_outline'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_HEADER_STRUCTURE, 3);
222007225e5Sgerardnico            $ruleResults[self::RULE_OUTLINE_STRUCTURE] = self::PASSED;
223007225e5Sgerardnico        }
224007225e5Sgerardnico
225007225e5Sgerardnico
226007225e5Sgerardnico        /**
227007225e5Sgerardnico         * Document length
228007225e5Sgerardnico         */
229007225e5Sgerardnico        $minimalWordCount = 50;
230007225e5Sgerardnico        $maximalWordCount = 1500;
231007225e5Sgerardnico        $correctContentLength = true;
232007225e5Sgerardnico        if ($this->stats[Analytics::WORDS_COUNT] < $minimalWordCount) {
233007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::FAILED;
234007225e5Sgerardnico            $correctContentLength = false;
235007225e5Sgerardnico            $ruleInfo[self::RULE_WORDS_MINIMAL] = "The number of words is less than {$minimalWordCount}";
236007225e5Sgerardnico        } else {
237007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MINIMAL] = self::PASSED;
238007225e5Sgerardnico        }
239007225e5Sgerardnico        if ($this->stats[Analytics::WORDS_COUNT] > $maximalWordCount) {
240007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::FAILED;
241007225e5Sgerardnico            $ruleInfo[self::RULE_WORDS_MAXIMAL] = "The number of words is more than {$maximalWordCount}";
242007225e5Sgerardnico            $correctContentLength = false;
243007225e5Sgerardnico        } else {
244007225e5Sgerardnico            $ruleResults[self::RULE_WORDS_MAXIMAL] = self::PASSED;
245007225e5Sgerardnico        }
246007225e5Sgerardnico        if ($correctContentLength) {
247007225e5Sgerardnico            $qualityScores['correct_content_length'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_CONTENT, 10);
248007225e5Sgerardnico        } else {
249007225e5Sgerardnico            $qualityScores['correct_content_length'] = 0;
250007225e5Sgerardnico        }
251007225e5Sgerardnico
252007225e5Sgerardnico
253007225e5Sgerardnico        /**
254007225e5Sgerardnico         * Average Number of words by header section to text ratio
255007225e5Sgerardnico         */
256007225e5Sgerardnico        $headers = $this->stats[Analytics::HEADERS_COUNT];
257007225e5Sgerardnico        if ($headers != null) {
258007225e5Sgerardnico            $headerCount = array_sum($headers);
259007225e5Sgerardnico            $headerCount--; // h1 is supposed to have no words
260007225e5Sgerardnico            if ($headerCount > 0) {
261007225e5Sgerardnico
262007225e5Sgerardnico                $avgWordsCountBySection = round($this->stats[Analytics::WORDS_COUNT] / $headerCount);
263007225e5Sgerardnico                $statExport['word_section_count']['avg'] = $avgWordsCountBySection;
264007225e5Sgerardnico
265007225e5Sgerardnico                /**
266007225e5Sgerardnico                 * Min words by header section
267007225e5Sgerardnico                 */
268007225e5Sgerardnico                $wordsByHeaderMin = 20;
269007225e5Sgerardnico                /**
270007225e5Sgerardnico                 * Max words by header section
271007225e5Sgerardnico                 */
272007225e5Sgerardnico                $wordsByHeaderMax = 300;
273007225e5Sgerardnico                $correctAverageWordsBySection = true;
274007225e5Sgerardnico                if ($avgWordsCountBySection < $wordsByHeaderMin) {
275007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::FAILED;
276007225e5Sgerardnico                    $correctAverageWordsBySection = false;
277007225e5Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = "The number of words by section is less than {$wordsByHeaderMin}";
278007225e5Sgerardnico                } else {
279007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MIN] = self::PASSED;
280007225e5Sgerardnico                }
281007225e5Sgerardnico                if ($avgWordsCountBySection > $wordsByHeaderMax) {
282007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::FAILED;
283007225e5Sgerardnico                    $correctAverageWordsBySection = false;
284007225e5Sgerardnico                    $ruleInfo[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = "The number of words by section is more than {$wordsByHeaderMax}";
285007225e5Sgerardnico                } else {
286007225e5Sgerardnico                    $ruleResults[self::RULE_AVERAGE_WORDS_BY_SECTION_MAX] = self::PASSED;
287007225e5Sgerardnico                }
288007225e5Sgerardnico                if ($correctAverageWordsBySection) {
289007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = $this->getConf(self::CONF_QUALITY_SCORE_CORRECT_WORD_SECTION_AVERAGE, 10);
290007225e5Sgerardnico                } else {
291007225e5Sgerardnico                    $qualityScores['correct_word_avg_by_section'] = 0;
292007225e5Sgerardnico                }
293007225e5Sgerardnico
294007225e5Sgerardnico            }
295007225e5Sgerardnico        }
296007225e5Sgerardnico
297007225e5Sgerardnico        /**
298007225e5Sgerardnico         * Internal Backlinks rule
299007225e5Sgerardnico         *
300007225e5Sgerardnico         * If a page is a low quality page, if the process run
301007225e5Sgerardnico         * anonymous, we will not see all {@link ft_backlinks()}
302007225e5Sgerardnico         * we use then the index directly to avoid confusion
303007225e5Sgerardnico         */
304007225e5Sgerardnico        $backlinks = idx_get_indexer()->lookupKey('relation_references', $ID);
305007225e5Sgerardnico        $countBacklinks = count($backlinks);
306007225e5Sgerardnico        $statExport[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks;
307007225e5Sgerardnico        if ($countBacklinks == 0) {
308007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = 0;
309007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::FAILED;
310007225e5Sgerardnico            $ruleInfo[self::RULE_INTERNAL_BACKLINKS_MIN] = "There is no backlinks";
311007225e5Sgerardnico        } else {
312007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_BACKLINKS_COUNT] = $countBacklinks * $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_BACKLINK_FACTOR, 1);
313007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BACKLINKS_MIN] = self::PASSED;
314007225e5Sgerardnico        }
315007225e5Sgerardnico
316007225e5Sgerardnico        /**
317007225e5Sgerardnico         * Internal links
318007225e5Sgerardnico         */
319007225e5Sgerardnico        $internalLinksCount = $this->stats[Analytics::INTERNAL_LINKS_COUNT];
320007225e5Sgerardnico        if ($internalLinksCount == 0) {
321007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = 0;
322007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::FAILED;
323007225e5Sgerardnico            $ruleInfo[self::RULE_INTERNAL_BACKLINKS_MIN] = "There is no internal links";
324007225e5Sgerardnico        } else {
325007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_LINKS_MIN] = self::PASSED;
326007225e5Sgerardnico            $qualityScores[Analytics::INTERNAL_LINKS_COUNT] = $countBacklinks * $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_FACTOR, 1);;
327007225e5Sgerardnico        }
328007225e5Sgerardnico
329007225e5Sgerardnico        /**
330007225e5Sgerardnico         * Broken Links
331007225e5Sgerardnico         */
332007225e5Sgerardnico        $brokenLinksCount = $this->stats[Analytics::INTERNAL_LINKS_BROKEN_COUNT];
333007225e5Sgerardnico        if ($brokenLinksCount > 2) {
334007225e5Sgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = 0;
335007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::FAILED;
336007225e5Sgerardnico            $ruleInfo[self::RULE_INTERNAL_BACKLINKS_MIN] = "There is {$brokenLinksCount} broken links";
337007225e5Sgerardnico        } else {
338007225e5Sgerardnico            $qualityScores['no_' . Analytics::INTERNAL_LINKS_BROKEN_COUNT] = $this->getConf(self::CONF_QUALITY_SCORE_INTERNAL_LINK_BROKEN_FACTOR, 2);;;
339007225e5Sgerardnico            $ruleResults[self::RULE_INTERNAL_BROKEN_LINKS_MAX] = self::PASSED;
340007225e5Sgerardnico        }
341007225e5Sgerardnico
342007225e5Sgerardnico        /**
343007225e5Sgerardnico         * Changes, the more changes the better
344007225e5Sgerardnico         */
345007225e5Sgerardnico        $qualityScores[Analytics::EDITS_COUNT] = $this->stats[Analytics::EDITS_COUNT] * $this->getConf(self::CONF_QUALITY_SCORE_CHANGES_FACTOR, 0.25);;;
346007225e5Sgerardnico
347007225e5Sgerardnico
348007225e5Sgerardnico        /**
349007225e5Sgerardnico         * Rules that comes from the qc plugin
350007225e5Sgerardnico         * but are not yet fully implemented
351007225e5Sgerardnico         */
352007225e5Sgerardnico
353007225e5Sgerardnico//        // 2 points for lot's of formatting
354007225e5Sgerardnico//        if ($this->stats[self::PLAINTEXT] && $this->stats['chars'] / $this->stats[self::PLAINTEXT] < 3) {
355007225e5Sgerardnico//            $ruleResults['manyformat'] = 2;
356007225e5Sgerardnico//        }
357007225e5Sgerardnico//
358007225e5Sgerardnico//        // 1/2 points for deeply nested quotations
359007225e5Sgerardnico//        if ($this->stats['quote_nest'] > 2) {
360007225e5Sgerardnico//            $ruleResults['deepquote'] += $this->stats['quote_nest'] / 2;
361007225e5Sgerardnico//        }
362007225e5Sgerardnico//
363007225e5Sgerardnico//        // 1/2 points for too many hr
364007225e5Sgerardnico//        if ($this->stats['hr'] > 2) {
365007225e5Sgerardnico//            $ruleResults['manyhr'] = ($this->stats['hr'] - 2) / 2;
366007225e5Sgerardnico//        }
367007225e5Sgerardnico//
368007225e5Sgerardnico//        // 1 point for too many line breaks
369007225e5Sgerardnico//        if ($this->stats['linebreak'] > 2) {
370007225e5Sgerardnico//            $ruleResults['manybr'] = $this->stats['linebreak'] - 2;
371007225e5Sgerardnico//        }
372007225e5Sgerardnico//
373007225e5Sgerardnico//        // 1 point for single author only
374007225e5Sgerardnico//        if (!$this->getConf('single_author_only') && count($this->stats['authors']) == 1) {
375007225e5Sgerardnico//            $ruleResults['singleauthor'] = 1;
376007225e5Sgerardnico//        }
377007225e5Sgerardnico
378007225e5Sgerardnico        // Too much cdata (plaintext), see cdata
379007225e5Sgerardnico        // if ($len > 500) $statExport[self::QUALITY][self::ERROR]['plaintext']++;
380007225e5Sgerardnico        // if ($len > 500) $statExport[self::QUALITY][self::ERROR]['plaintext']++;
381007225e5Sgerardnico        //
382007225e5Sgerardnico        // // 1 point for formattings longer than 500 chars
383007225e5Sgerardnico        // $statExport[self::QUALITY][self::ERROR]['multiformat']
384007225e5Sgerardnico
385007225e5Sgerardnico        /**
386007225e5Sgerardnico         * Quality Score
387007225e5Sgerardnico         */
388007225e5Sgerardnico        ksort($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         */
429007225e5Sgerardnico        $quality["low"] = $lowLevel;
430007225e5Sgerardnico        if (sizeof($mandatoryRulesBroken) > 0) {
431007225e5Sgerardnico            ksort($mandatoryRulesBroken);
432007225e5Sgerardnico            $quality['failed_mandatory_rules'] = $mandatoryRulesBroken;
433007225e5Sgerardnico        }
434007225e5Sgerardnico        $quality["scoring"] = $qualityScoring;
435007225e5Sgerardnico        $quality["rules"][self::RESULT] = $qualityResult;
436007225e5Sgerardnico        if (!empty($ruleInfo)) {
437007225e5Sgerardnico            $quality["rules"]["info"] = $ruleInfo;
438007225e5Sgerardnico        }
439007225e5Sgerardnico
440007225e5Sgerardnico        ksort($ruleResults);
441007225e5Sgerardnico        $quality["rules"]['details'] = $ruleResults;
442007225e5Sgerardnico
443007225e5Sgerardnico        /**
444007225e5Sgerardnico         * Metadata
445007225e5Sgerardnico         */
446007225e5Sgerardnico        $this->metadata[Analytics::TITLE] = $meta['title'];
447007225e5Sgerardnico        $timestampCreation = $meta['date']['created'];
448007225e5Sgerardnico        $this->metadata[self::DATE_CREATED] = date('Y-m-d h:i:s', $timestampCreation);
449007225e5Sgerardnico        $timestampModification = $meta['date']['modified'];
450007225e5Sgerardnico        $this->metadata[Analytics::DATE_MODIFIED] = date('Y-m-d h:i:s', $timestampModification);
451007225e5Sgerardnico        $this->metadata['age_creation'] = round((time() - $timestampCreation) / 60 / 60 / 24);
452007225e5Sgerardnico        $this->metadata['age_modification'] = round((time() - $timestampModification) / 60 / 60 / 24);
453007225e5Sgerardnico
454007225e5Sgerardnico
455007225e5Sgerardnico        // get author info
456007225e5Sgerardnico        $changelog = new PageChangeLog($ID);
457007225e5Sgerardnico        $revs = $changelog->getRevisions(0, 10000);
458007225e5Sgerardnico        array_push($revs, $meta['last_change']['date']);
459007225e5Sgerardnico        $this->stats[Analytics::EDITS_COUNT] = count($revs);
460007225e5Sgerardnico        foreach ($revs as $rev) {
461007225e5Sgerardnico            $info = $changelog->getRevisionInfo($rev);
462007225e5Sgerardnico            if ($info['user']) {
463007225e5Sgerardnico                $this->stats['authors'][$info['user']] += 1;
464007225e5Sgerardnico            } else {
465007225e5Sgerardnico                $this->stats['authors']['*'] += 1;
466007225e5Sgerardnico            }
467007225e5Sgerardnico        }
468007225e5Sgerardnico
469007225e5Sgerardnico        /**
470007225e5Sgerardnico         * Building the Top JSON in order
471007225e5Sgerardnico         */
472007225e5Sgerardnico        global $ID;
473007225e5Sgerardnico        $json = array();
474007225e5Sgerardnico        $json["id"] = $ID;
475007225e5Sgerardnico        $json['metadata'] = $this->metadata;
476007225e5Sgerardnico        ksort($statExport);
477007225e5Sgerardnico        $json[Analytics::STATISTICS] = $statExport;
478007225e5Sgerardnico        $json[Analytics::QUALITY] = $quality; // Quality after the sort to get them at the end
479007225e5Sgerardnico
480007225e5Sgerardnico
481007225e5Sgerardnico        /**
482007225e5Sgerardnico         * The result can be seen with
483007225e5Sgerardnico         * doku.php?id=somepage&do=export_combo_analysis
484007225e5Sgerardnico         */
485007225e5Sgerardnico        /**
486007225e5Sgerardnico         * Set the header for the export.php file
487007225e5Sgerardnico         */
488007225e5Sgerardnico        p_set_metadata($ID, array("format" =>
489007225e5Sgerardnico            array("combo_" . $this->getPluginComponent() => array("Content-Type" => 'application/json'))
490007225e5Sgerardnico        ));
491007225e5Sgerardnico        $json_encoded = json_encode($json, JSON_PRETTY_PRINT);
492007225e5Sgerardnico
493007225e5Sgerardnico        $sqlite = Sqlite::getSqlite();
494007225e5Sgerardnico        if ($sqlite != null) {
495007225e5Sgerardnico            /**
496007225e5Sgerardnico             * Sqlite Plugin installed
497007225e5Sgerardnico             */
498*71f916b9Sgerardnico            $canonical = $this->metadata[Page::CANONICAL_PROPERTY];
499007225e5Sgerardnico            if (empty($canonical)) {
500007225e5Sgerardnico                $canonical = $ID; // not null constraint unfortunately
501007225e5Sgerardnico            }
502007225e5Sgerardnico            $entry = array(
503007225e5Sgerardnico                'CANONICAL' => $canonical,
504007225e5Sgerardnico                'ANALYTICS' => $json_encoded,
505007225e5Sgerardnico                'ID' => $ID
506007225e5Sgerardnico            );
507007225e5Sgerardnico            $res = $sqlite->query("SELECT count(*) FROM PAGES where ID = ?", $ID);
508007225e5Sgerardnico            if ($sqlite->res2single($res) == 1) {
509007225e5Sgerardnico                // Upset not supported on all version
510007225e5Sgerardnico                //$upsert = 'insert into PAGES (ID,CANONICAL,ANALYTICS) values (?,?,?) on conflict (ID,CANONICAL) do update set ANALYTICS = EXCLUDED.ANALYTICS';
511007225e5Sgerardnico                $update = 'update PAGES SET CANONICAL = ?, ANALYTICS = ? where ID=?';
512007225e5Sgerardnico                $res = $sqlite->query($update, $entry);
513007225e5Sgerardnico            } else {
514007225e5Sgerardnico                $res = $sqlite->storeEntry('PAGES', $entry);
515007225e5Sgerardnico            }
516007225e5Sgerardnico            if (!$res) {
517007225e5Sgerardnico                LogUtility::msg("There was a problem during the upsert: {$sqlite->getAdapter()->getDb()->errorInfo()}");
518007225e5Sgerardnico            }
519007225e5Sgerardnico            $sqlite->res_close($res);
520007225e5Sgerardnico        }
521007225e5Sgerardnico        $this->doc .= $json_encoded;
522007225e5Sgerardnico
523007225e5Sgerardnico    }
524007225e5Sgerardnico
525007225e5Sgerardnico    /**
526007225e5Sgerardnico     */
527007225e5Sgerardnico    public function getFormat()
528007225e5Sgerardnico    {
529007225e5Sgerardnico        return Analytics::RENDERER_FORMAT;
530007225e5Sgerardnico    }
531007225e5Sgerardnico
532007225e5Sgerardnico    public function internallink($id, $name = null, $search = null, $returnonly = false, $linktype = 'content')
533007225e5Sgerardnico    {
534007225e5Sgerardnico
535007225e5Sgerardnico        LinkUtility::processInternalLinkStats($id, $this->stats);
536007225e5Sgerardnico
537007225e5Sgerardnico    }
538007225e5Sgerardnico
539007225e5Sgerardnico    public function externallink($url, $name = null)
540007225e5Sgerardnico    {
541007225e5Sgerardnico        $this->stats[Analytics::EXTERNAL_LINKS_COUNT]++;
542007225e5Sgerardnico    }
543007225e5Sgerardnico
544007225e5Sgerardnico    public function header($text, $level, $pos)
545007225e5Sgerardnico    {
546007225e5Sgerardnico        $this->stats[Analytics::HEADERS_COUNT]['h' . $level]++;
547007225e5Sgerardnico        $this->headerId++;
548007225e5Sgerardnico        $this->stats[Analytics::HEADER_POSITION][$this->headerId] = 'h' . $level;
549007225e5Sgerardnico
550007225e5Sgerardnico    }
551007225e5Sgerardnico
552007225e5Sgerardnico    public function smiley($smiley)
553007225e5Sgerardnico    {
554007225e5Sgerardnico        if ($smiley == 'FIXME') $this->stats[self::FIXME]++;
555007225e5Sgerardnico    }
556007225e5Sgerardnico
557007225e5Sgerardnico    public function linebreak()
558007225e5Sgerardnico    {
559007225e5Sgerardnico        if (!$this->tableopen) {
560007225e5Sgerardnico            $this->stats['linebreak']++;
561007225e5Sgerardnico        }
562007225e5Sgerardnico    }
563007225e5Sgerardnico
564007225e5Sgerardnico    public function table_open($maxcols = null, $numrows = null, $pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
565007225e5Sgerardnico    {
566007225e5Sgerardnico        $this->tableopen = true;
567007225e5Sgerardnico    }
568007225e5Sgerardnico
569007225e5Sgerardnico    public function table_close($pos = null) // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
570007225e5Sgerardnico    {
571007225e5Sgerardnico        $this->tableopen = false;
572007225e5Sgerardnico    }
573007225e5Sgerardnico
574007225e5Sgerardnico    public function hr()
575007225e5Sgerardnico    {
576007225e5Sgerardnico        $this->stats['hr']++;
577007225e5Sgerardnico    }
578007225e5Sgerardnico
579007225e5Sgerardnico    public function quote_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
580007225e5Sgerardnico    {
581007225e5Sgerardnico        $this->stats['quote_count']++;
582007225e5Sgerardnico        $this->quotelevel++;
583007225e5Sgerardnico        $this->stats['quote_nest'] = max($this->quotelevel, $this->stats['quote_nest']);
584007225e5Sgerardnico    }
585007225e5Sgerardnico
586007225e5Sgerardnico    public function quote_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
587007225e5Sgerardnico    {
588007225e5Sgerardnico        $this->quotelevel--;
589007225e5Sgerardnico    }
590007225e5Sgerardnico
591007225e5Sgerardnico    public function strong_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
592007225e5Sgerardnico    {
593007225e5Sgerardnico        $this->formattingBracket++;
594007225e5Sgerardnico    }
595007225e5Sgerardnico
596007225e5Sgerardnico    public function strong_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
597007225e5Sgerardnico    {
598007225e5Sgerardnico        $this->formattingBracket--;
599007225e5Sgerardnico    }
600007225e5Sgerardnico
601007225e5Sgerardnico    public function emphasis_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
602007225e5Sgerardnico    {
603007225e5Sgerardnico        $this->formattingBracket++;
604007225e5Sgerardnico    }
605007225e5Sgerardnico
606007225e5Sgerardnico    public function emphasis_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
607007225e5Sgerardnico    {
608007225e5Sgerardnico        $this->formattingBracket--;
609007225e5Sgerardnico    }
610007225e5Sgerardnico
611007225e5Sgerardnico    public function underline_open() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
612007225e5Sgerardnico    {
613007225e5Sgerardnico        $this->formattingBracket++;
614007225e5Sgerardnico    }
615007225e5Sgerardnico
616007225e5Sgerardnico    public function underline_close() // phpcs:ignore PSR1.Methods.CamelCapsMethodName.NotCamelCaps
617007225e5Sgerardnico    {
618007225e5Sgerardnico        $this->formattingBracket--;
619007225e5Sgerardnico    }
620007225e5Sgerardnico
621007225e5Sgerardnico    public function cdata($text)
622007225e5Sgerardnico    {
623007225e5Sgerardnico
624007225e5Sgerardnico        /**
625007225e5Sgerardnico         * It seems that you receive cdata
626007225e5Sgerardnico         * when emphasis_open / underline_open / strong_open
627007225e5Sgerardnico         * Stats are not for them
628007225e5Sgerardnico         */
629007225e5Sgerardnico        if (!$this->formattingBracket) return;
630007225e5Sgerardnico
631007225e5Sgerardnico        $this->plainTextId++;
632007225e5Sgerardnico
633007225e5Sgerardnico        /**
634007225e5Sgerardnico         * Length
635007225e5Sgerardnico         */
636007225e5Sgerardnico        $len = strlen($text);
637007225e5Sgerardnico        $this->stats[self::PLAINTEXT][$this->plainTextId]['len'] = $len;
638007225e5Sgerardnico
639007225e5Sgerardnico
640007225e5Sgerardnico        /**
641007225e5Sgerardnico         * Multi-formatting
642007225e5Sgerardnico         */
643007225e5Sgerardnico        if ($this->formattingBracket > 1) {
644007225e5Sgerardnico            $numberOfFormats = 1 * ($this->formattingBracket - 1);
645007225e5Sgerardnico            $this->stats[self::PLAINTEXT][$this->plainTextId]['multiformat'] += $numberOfFormats;
646007225e5Sgerardnico        }
647007225e5Sgerardnico
648007225e5Sgerardnico        /**
649007225e5Sgerardnico         * Total
650007225e5Sgerardnico         */
651007225e5Sgerardnico        $this->stats[self::PLAINTEXT][0] += $len;
652007225e5Sgerardnico    }
653007225e5Sgerardnico
654007225e5Sgerardnico    public function internalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
655007225e5Sgerardnico    {
656007225e5Sgerardnico        $this->stats[Analytics::INTERNAL_MEDIAS_COUNT]++;
657007225e5Sgerardnico    }
658007225e5Sgerardnico
659007225e5Sgerardnico    public function externalmedia($src, $title = null, $align = null, $width = null, $height = null, $cache = null, $linking = null)
660007225e5Sgerardnico    {
661007225e5Sgerardnico        $this->stats[Analytics::EXTERNAL_MEDIAS]++;
662007225e5Sgerardnico    }
663007225e5Sgerardnico
664007225e5Sgerardnico    public function reset()
665007225e5Sgerardnico    {
666007225e5Sgerardnico        $this->stats = array();
667007225e5Sgerardnico        $this->metadata = array();
668007225e5Sgerardnico        $this->headerId = 0;
669007225e5Sgerardnico    }
670007225e5Sgerardnico
671007225e5Sgerardnico    public function setMeta($key, $value)
672007225e5Sgerardnico    {
673007225e5Sgerardnico        $this->metadata[$key] = $value;
674007225e5Sgerardnico    }
675007225e5Sgerardnico
676007225e5Sgerardnico
677007225e5Sgerardnico}
678007225e5Sgerardnico
679