xref: /dokuwiki/inc/Search/Index/TupleOps.php (revision 7fcedc39d164da3013a88e9ad6660009940e851b) 
1<?php
2
3namespace dokuwiki\Search\Index;
4
5/**
6 * Provides operations on tuple records used in our indexes
7 *
8 * Tuples consist of a key (typically a RID from another Index) and a number (usually a count).
9 * Used to store page <-> word counts for example
10 */
11class TupleOps
12{
13    /**
14     * Insert or replace a tuple in a line
15     *
16     * @param string $record This is the current row value to be modified
17     * @param int|string $key The foreign rid or identifier
18     * @param int $count The count to store
19     * @return string A new row value
20     * @author Tom N Harris <tnharris@whoopdedo.org>
21     *
22     */
23    public static function updateTuple($record, $key, $count)
24    {
25        if ($record != '') {
26            // remove any current version of the tuple
27            $record = preg_replace('/(^|:)' . preg_quote($key, '/') . '\*\d*/', '', $record);
28        }
29        $record = trim($record, ':');
30        if ($count) {
31            if ($record) {
32                return "{$key}*{$count}:" . $record;
33            } else {
34                return "{$key}*{$count}";
35            }
36        }
37        return $record;
38    }
39
40    /**
41     * Sum the counts in a list of tuples
42     *
43     * @param string $record The row value to parse
44     * @return int sum of all counts
45     * @author Tom N Harris <tnharris@whoopdedo.org>
46     */
47    public static function aggregateTupleCounts($record)
48    {
49        $freq = 0;
50        $parts = explode(':', $record);
51        foreach ($parts as $tuple) {
52            if ($tuple === '') continue;
53            list(/* $key */, $cnt) = explode('*', $tuple);
54            $freq += (int)$cnt;
55        }
56        return $freq;
57    }
58
59    /**
60     * Split a line into an array of tuples
61     *
62     * The given key of the given $filtermap defines which tuples to extract, the value
63     * gives the name in the output array. This basically allows to map RIDs to their
64     * respective real values. The result will contain the counts associated with the
65     * mapped keys.
66     *
67     * @param string $record The row value to parse
68     * @param array $filtermap Associative array of ($key => $mapping)
69     * @return array mapped counts
70     * @author Andreas Gohr <andi@splitbrain.org>
71     *
72     * @author Tom N Harris <tnharris@whoopdedo.org>
73     */
74    public static function parseTuples($record, $filtermap)
75    {
76        $result = array();
77        if ($record == '') return $result;
78        $parts = explode(':', $record);
79        foreach ($parts as $tuple) {
80            if ($tuple === '') continue;
81            list($key, $cnt) = explode('*', $tuple);
82            if (!$cnt) continue;
83            if (empty($filtermap[$key])) continue;
84            $mapped = $filtermap[$key];
85            $result[$mapped] = $cnt;
86        }
87        return $result;
88    }
89}
90