xref: /webtrees/app/Report/ReportParserBase.php (revision f315390b2ef3b402a974292b894bb8d92201eb00)
1ef0d468bSGreg Roach<?php
23976b470SGreg Roach
3ef0d468bSGreg Roach/**
4ef0d468bSGreg Roach * webtrees: online genealogy
5d11be702SGreg Roach * Copyright (C) 2023 webtrees development team
6ef0d468bSGreg Roach * This program is free software: you can redistribute it and/or modify
7ef0d468bSGreg Roach * it under the terms of the GNU General Public License as published by
8ef0d468bSGreg Roach * the Free Software Foundation, either version 3 of the License, or
9ef0d468bSGreg Roach * (at your option) any later version.
10ef0d468bSGreg Roach * This program is distributed in the hope that it will be useful,
11ef0d468bSGreg Roach * but WITHOUT ANY WARRANTY; without even the implied warranty of
12ef0d468bSGreg Roach * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13ef0d468bSGreg Roach * GNU General Public License for more details.
14ef0d468bSGreg Roach * You should have received a copy of the GNU General Public License
1589f7189bSGreg Roach * along with this program. If not, see <https://www.gnu.org/licenses/>.
16ef0d468bSGreg Roach */
17fcfa147eSGreg Roach
18e7f56f2aSGreg Roachdeclare(strict_types=1);
19e7f56f2aSGreg Roach
2076692c8bSGreg Roachnamespace Fisharebest\Webtrees\Report;
21ef0d468bSGreg Roach
22208e9f76SGreg Roachuse DomainException;
237774ce41SGreg Roachuse Exception;
248792b649SGreg Roachuse XMLParser;
25299d100dSGreg Roach
26b6f35a76SGreg Roachuse function fclose;
27b6f35a76SGreg Roachuse function feof;
28b6f35a76SGreg Roachuse function fread;
29b19e047dSGreg Roachuse function method_exists;
30b6f35a76SGreg Roachuse function sprintf;
31b6f35a76SGreg Roachuse function xml_error_string;
32b6f35a76SGreg Roachuse function xml_get_current_line_number;
33b6f35a76SGreg Roachuse function xml_get_error_code;
34b6f35a76SGreg Roachuse function xml_parse;
35b6f35a76SGreg Roachuse function xml_parser_create;
36b6f35a76SGreg Roachuse function xml_parser_free;
37b6f35a76SGreg Roachuse function xml_parser_set_option;
38b6f35a76SGreg Roachuse function xml_set_character_data_handler;
39b6f35a76SGreg Roachuse function xml_set_element_handler;
40b6f35a76SGreg Roach
41b6f35a76SGreg Roachuse const XML_OPTION_CASE_FOLDING;
42b19e047dSGreg Roach
43ef0d468bSGreg Roach/**
44ef0d468bSGreg Roach * Class ReportParserBase
45ef0d468bSGreg Roach */
46c1010edaSGreg Roachclass ReportParserBase
47c1010edaSGreg Roach{
48b112142eSGreg Roach    // The XML parser
49b112142eSGreg Roach    protected XMLParser $xml_parser;
50ef0d468bSGreg Roach
51ef0d468bSGreg Roach    /** @var string Text contents of tags */
5277bab461SGreg Roach    protected string $text = '';
53ef0d468bSGreg Roach
54ef0d468bSGreg Roach    /**
55ef0d468bSGreg Roach     * Create a parser for a report
56ef0d468bSGreg Roach     *
57ef0d468bSGreg Roach     * @param string $report The XML filename
587774ce41SGreg Roach     *
597774ce41SGreg Roach     * @throws Exception
60ef0d468bSGreg Roach     */
6176f666f4SGreg Roach    public function __construct(string $report)
62c1010edaSGreg Roach    {
63ef0d468bSGreg Roach        $this->xml_parser = xml_parser_create();
648a4ee39cSGreg Roach
65*f315390bSGreg Roach        xml_parser_set_option($this->xml_parser, XML_OPTION_CASE_FOLDING, 0);
668a4ee39cSGreg Roach
678a4ee39cSGreg Roach        xml_set_element_handler(
688a4ee39cSGreg Roach            $this->xml_parser,
69af14d238SGreg Roach            function ($parser, string $name, array $attrs): void {
708a4ee39cSGreg Roach                $this->startElement($parser, $name, $attrs);
718a4ee39cSGreg Roach            },
72af14d238SGreg Roach            function ($parser, string $name): void {
738a4ee39cSGreg Roach                $this->endElement($parser, $name);
748a4ee39cSGreg Roach            }
758a4ee39cSGreg Roach        );
768a4ee39cSGreg Roach
778a4ee39cSGreg Roach        xml_set_character_data_handler(
788a4ee39cSGreg Roach            $this->xml_parser,
799d454b6bSGreg Roach            function ($parser, string $data): void {
808a4ee39cSGreg Roach                $this->characterData($parser, $data);
818a4ee39cSGreg Roach            }
828a4ee39cSGreg Roach        );
83ef0d468bSGreg Roach
846c84de1dSGreg Roach        $fp = fopen($report, 'rb');
857774ce41SGreg Roach
86e364afe4SGreg Roach        while ($data = fread($fp, 4096)) {
87ef0d468bSGreg Roach            if (!xml_parse($this->xml_parser, $data, feof($fp))) {
88208e9f76SGreg Roach                throw new DomainException(sprintf(
89ef0d468bSGreg Roach                    'XML error: %s at line %d',
90ef0d468bSGreg Roach                    xml_error_string(xml_get_error_code($this->xml_parser)),
91ef0d468bSGreg Roach                    xml_get_current_line_number($this->xml_parser)
92ef0d468bSGreg Roach                ));
93ef0d468bSGreg Roach            }
94ef0d468bSGreg Roach        }
95ef0d468bSGreg Roach
967774ce41SGreg Roach        fclose($fp);
977774ce41SGreg Roach
98ef0d468bSGreg Roach        xml_parser_free($this->xml_parser);
99ef0d468bSGreg Roach    }
100ef0d468bSGreg Roach
101ef0d468bSGreg Roach    /**
102ef0d468bSGreg Roach     * XML handler for an opening (or self-closing) tag.
103ef0d468bSGreg Roach     *
104ef0d468bSGreg Roach     * @param resource      $parser The resource handler for the xml parser
105ef0d468bSGreg Roach     * @param string        $name   The name of the xml element parsed
10609482a55SGreg Roach     * @param array<string> $attrs  An array of key value pairs for the attributes
10718d7a90dSGreg Roach     *
10818d7a90dSGreg Roach     * @return void
109ef0d468bSGreg Roach     */
110af14d238SGreg Roach    protected function startElement($parser, string $name, array $attrs): void
111c1010edaSGreg Roach    {
112a6f13a4aSGreg Roach        $method = $name . 'StartHandler';
113208e9f76SGreg Roach
114ef0d468bSGreg Roach        if (method_exists($this, $method)) {
11579a908a7SGreg Roach            $this->$method($attrs);
116ef0d468bSGreg Roach        }
117ef0d468bSGreg Roach    }
118ef0d468bSGreg Roach
119ef0d468bSGreg Roach    /**
120ef0d468bSGreg Roach     * XML handler for a closing tag.
121ef0d468bSGreg Roach     *
122ef0d468bSGreg Roach     * @param resource $parser the resource handler for the xml parser
123ef0d468bSGreg Roach     * @param string   $name   the name of the xml element parsed
12418d7a90dSGreg Roach     *
12518d7a90dSGreg Roach     * @return void
126ef0d468bSGreg Roach     */
127af14d238SGreg Roach    protected function endElement($parser, string $name): void
128c1010edaSGreg Roach    {
129a6f13a4aSGreg Roach        $method = $name . 'EndHandler';
130208e9f76SGreg Roach
131ef0d468bSGreg Roach        if (method_exists($this, $method)) {
13279a908a7SGreg Roach            $this->$method();
133ef0d468bSGreg Roach        }
134ef0d468bSGreg Roach    }
135ef0d468bSGreg Roach
136ef0d468bSGreg Roach    /**
137ef0d468bSGreg Roach     * XML handler for character data.
138ef0d468bSGreg Roach     *
139ef0d468bSGreg Roach     * @param resource $parser The resource handler for the xml parser
140ef0d468bSGreg Roach     * @param string   $data   The name of the xml element parsed
14118d7a90dSGreg Roach     *
14218d7a90dSGreg Roach     * @return void
143ef0d468bSGreg Roach     */
14424f2a3afSGreg Roach    protected function characterData($parser, string $data): void
145c1010edaSGreg Roach    {
146ef0d468bSGreg Roach        $this->text .= $data;
147ef0d468bSGreg Roach    }
148ef0d468bSGreg Roach}
149