server/src/app/Libraries/Sparql/SparqlQueryAnalyser.php
author ymh <ymh.work@gmail.com>
Mon, 31 Oct 2016 14:24:23 +0100
changeset 386 c731ab9b934d
child 387 7fba86fa8604
permissions -rw-r--r--
implement first version of sparql client interface
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
386
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
<?php
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
namespace CorpusParole\Libraries\Sparql;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
use CorpusParole\Libraries\Utils;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
class SparqlQueryAnalyser {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
    const SPARQL_PREFIX_BASE_REGEXP = '(((?:prefix\s+([\p{L}-\d]+)\s*\:)|base)\s*\<((?:(?:https?|ftp)://)(?:\S+(?::\S*)?@)?(?:(?!10(?:\.\d{1,3}){3})(?!127(?:\.\d{1,3}){3})(?!169\.254(?:\.\d{1,3}){2})(?!192\.168(?:\.\d{1,3}){2})(?!172\.(?:1[6-9]|2\d|3[0-1])(?:\.\d{1,3}){2})(?:[1-9]\d?|1\d\d|2[01]\d|22[0-3])(?:\.(?:1?\d{1,2}|2[0-4]\d|25[0-5])){2}(?:\.(?:[1-9]\d?|1\d\d|2[0-4]\d|25[0-4]))|(?:(?:[a-z\x{00a1}-\x{ffff}0-9]+-?)*[a-z\x{00a1}-\x{ffff}0-9]+)(?:\.(?:[a-z\x{00a1}-\x{ffff}0-9]+-?)*[a-z\x{00a1}-\x{ffff}0-9]+)*(?:\.(?:[a-z\x{00a1}-\x{ffff}]{2,})))(?::\d{2,5})?(?:/[^\s]*)?)\>)\s*';
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
    const SPARQL_SELECT_QUERY_REGEXP = '^(?:\s*(?:'.self::SPARQL_PREFIX_BASE_REGEXP.')*select)';
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
    const SPARQL_ASK_QUERY_REGEXP = '^(?:\s*(?:'.self::SPARQL_PREFIX_BASE_REGEXP.')*ask)';
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
    const SPARQL_GRAPH_QUERY_REGEXP = '^(?:\s*(?:'.self::SPARQL_PREFIX_BASE_REGEXP.')*(?:(?:construct)|(?:describe)))';
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
    const SPARQL_LIMIT_OFFSET_QUERY_REGEXP = '(?:(?:(limit\s+(\d+))|(offset\s+(\d+)))\s*)+\s*$';
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
    const UNKNOWN_QUERY = 0;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
    const SELECT_QUERY = 1;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    18
    const GRAPH_QUERY = 2;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
    const ASK_QUERY = 2;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
    private $query;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
    private $queryType = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
    private $rawPrefixes = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
    private $prefixes = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    25
    private $limit = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
    private $offset = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    27
    private $rawQuery = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    28
    private $countVar = false;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    29
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    30
    public function __construct($query) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    31
        $this->query = $query;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    32
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    33
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    34
    public function getQueryType() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    35
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    36
        if($this->queryType === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    37
            if(preg_match("%".self::SPARQL_SELECT_QUERY_REGEXP."%iu", $this->query) === 1) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    38
                $this->queryType = self::SELECT_QUERY;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    39
            } elseif(preg_match("%".self::SPARQL_GRAPH_QUERY_REGEXP."%iu", $this->query) === 1) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    40
                $this->queryType = self::GRAPH_QUERY;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    41
            } elseif(preg_match("%".self::SPARQL_ASK_QUERY_REGEXP."%iu", $this->query) === 1) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    42
                $this->queryType = self::ASK_QUERY;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    43
            } else {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    44
                $this->queryType = self::UNKNOWN_QUERY;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
            }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    46
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    47
        return $this->queryType;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    48
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    50
    private function extractPrefix() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
        $prefixes = [];
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
        $rawPrefixes = [];
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
        $res = preg_replace_callback("%".self::SPARQL_PREFIX_BASE_REGEXP."%iu", function($m) use (&$prefixes, &$rawPrefixes) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
            $rawPrefixes[] = trim($m[0]);
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
            $prefixes[$m[3]?$m[3]:""] = $m[4];
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
            return "";
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
        }, $this->query);
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
        return [$rawPrefixes, $prefixes, trim($res)];
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
    public function getRawPrefixes() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    63
        if($this->rawPrefixes === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    64
            list($this->rawPrefixes, $this->prefixes, $this->rawQuery) = $this->extractPrefix();
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    65
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    66
        return $this->rawPrefixes;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    67
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    68
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
    public function getPrefixes() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
        if($this->prefixes === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    71
            list($this->rawPrefixes, $this->prefixes, $this->rawQuery) = $this->extractPrefix();
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    72
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    73
        return $this->prefixes;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    74
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    75
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    76
    public function getRawQuery() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    77
        if($this->rawQuery === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    78
            list($this->rawPrefixes, $this->prefixes, $this->rawQuery) = $this->extractPrefix();
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    79
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    80
        return $this->rawQuery;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    81
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    82
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    83
    public function getCountVar() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    84
        if($this->countVar === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    85
            $this->countVar = "?count_cp_".hash('md5', $this->query);
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    86
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    87
        return $this->countVar;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    88
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    89
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    90
    public function getCountQuery() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    91
        return implode(" ", $this->getRawPrefixes())." select (count(*) as ".$this->getCountVar().") { ".$this->getRawQuery()." }";
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    92
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    93
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    94
    private function setLimitOffset() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    95
        if(preg_match("%".self::SPARQL_LIMIT_OFFSET_QUERY_REGEXP."%iu", $this->query, $m) === 1) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    96
            for($i=0;$i<(count($m)-1)/2;$i++) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    97
                if(Utils::startsWith(strtolower($m[2*$i+1]), "limit")) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    98
                    $this->limit = intval($m[$i*2+2]);
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
    99
                } elseif (Utils::startsWith(strtolower($m[2*$i+1]), "offset")) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   100
                    $this->offset = intval($m[$i*2+2]);
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   101
                }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   102
            }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   103
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   104
        if($this->limit === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   105
            $this->limit = null;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   106
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   107
        if($this->offset === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   108
            $this->offset = null;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   109
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   110
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   111
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   112
    public function getLimit() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   113
        if($this->limit === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   114
            $this->setLimitOffset();
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   115
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   116
        return $this->limit;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   117
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   118
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   119
    public function getOffset() {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   120
        if($this->offset === false) {
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   121
            $this->setLimitOffset();
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   122
        }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   123
        return $this->offset;
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   124
    }
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   125
c731ab9b934d implement first version of sparql client interface
ymh <ymh.work@gmail.com>
parents:
diff changeset
   126
}