xref: /PHP-8.4/ext/fileinfo/create_data_file.php (revision bcd3eec4)
1#!/usr/bin/env php
2/* This is a generated file, do not modify */
3/* Usage: php create_data_file.php /path/to/magic.mgc > data_file.c */
4<?php
5    /*--- Initialization of our translation table ---*/
6
7    // By default, everything gets mapped to its \o notation
8    // (not \x, because by C's norm, \x eats as many chars as possible, while \o stops at exactly 3;
9    // thus \x0ABACK_2_MALICE is interpreted as hex \x0ABAC (which overflows) followed by string K_2_MALICE,
10    // while \o0120 is unambiguously a CR followed by digit 0).
11    for ($i = 0; $i < 0x100; ++$i) {
12        $map[chr($i)] = sprintf('\%03o', $i);
13    }
14    // \0 is a shortcut for \x00; as the majority of the input file is \0's,
15    // we divide the generated file's size by nearly 2 (30 MB -> 16 MB).
16    $map[chr(0)] = '\0';
17    $map["\n"] = '\n';
18    // Displayable ASCII can be output as is: strings for file types will appear readable.
19    for ($i = ord(' '); $i < 0x7F; ++$i) {
20        $map[chr($i)] = chr($i);
21    }
22    // … Except digits following a \0: \012 will be interpreted as octal 012, and not \0 followed by 12.
23    // Then we have to express \0 in a full unambiguous 3-chars octal code.
24    for ($i = ord('0'); $i <= ord('9'); ++$i) {
25        $map[chr(0).chr($i)] = '\000'.chr($i);
26    }
27    // … Except " and \ because we enclose the result into quotes and escape with \.
28    $map['"'] = '\"';
29    $map['\\'] = '\\\\';
30
31    /*--- File generation ---*/
32
33    // https://github.com/php/php-src/pull/10422
34    // Some compilers (GCC, clang) do not like long lists; some (MSVC) do not like long strings.
35    // CHUNK_SIZE splitting our ~10 MB binary source should give a good compromise between both.
36    const CHUNK_SIZE = 1024;
37
38    $dta = file_get_contents( $argv[1] );
39    $chunks = str_split($dta, CHUNK_SIZE);
40    $chunks[count($chunks) - 1] = str_pad($chunks[count($chunks) - 1], CHUNK_SIZE, chr(0));
41
42    echo 'const unsigned char php_magic_database[' . count($chunks) . '][' . CHUNK_SIZE . "] = {\n";
43    foreach ($chunks as $chunk) {
44        echo '"' . strtr($chunk, $map) . '",' . "\n";
45    }
46    echo "};\n";
47?>
48