delinquent_files/test_regex.php

Thu, 22 Feb 2024 20:21:22 +0100

author
Peter Gervai <grin@grin.hu>
date
Thu, 22 Feb 2024 20:21:22 +0100
changeset 8
38415be9f910
parent 4
aa3e6379dc3d
permissions
-rwxr-xr-x

Skipping {{nobots}} pages

#!/usr/bin/php
<?php
## By Peter 'grin' Gervai, 2022
## GPLv3+ and CC-By-Sa-4.0 
##
## $Id$
##
## This file is the testing "environment" for the regex matching.
## Tries to test unlink (removal) and replacement rules.
##
## The input files are in _t_*.in, and the unlink and replace
## expected results are in _t_*.unlink and _t_*.replace.
## The tests are listed below in the array, with the original
## image and its expected good matching pattern, which is also
## verified. The replacement image also set there.
##

$DEBUG=0;

require_once( './matcher.inc' );
require_once( './debug.inc' );

$d = new Debug;
#$d->set_level(9); # trace

$matcher = new Matcher($d);


# files are in
#  input:            _t_{id}.in
#  test data:        _t_{id}.data
#  good replacement: _t_{id}.replace
#  good unlink:      _t_{id}.unlink

$dir_tests = './test';
$testnames = collect_test_names($dir_tests);

$data_keys = array( 'img', 'pattern', 'repl_img' );

foreach ($testnames as $t) {
    $tests[$t] = get_test_data( $dir_tests, $t, $data_keys );
}

$test_res = array( 'ok' => 0, 'bad' => 0 );

# run
foreach ($tests as $key => $t) {
    #print( "Running test '$key'..." );
    #print("IN=" . $t["in"] . "\n");
    #print("Im=" . $t["img"] . "\n");
    #print("RE=" . $t["unlink"] . "\n");

    $img = $t["img"];
    $pattern = test_prepare_pattern( $img );
    if( $DEBUG ) {print(" Pat=$pattern\n");}
    $res = test_result( $pattern, $t["pattern"], "pattern", $img, $key );
    if( $res == 1 ) {
        $test_res['bad']++;
    } else {
        $test_res['ok']++;
    }

    $text = read_text_from_file( "${dir_tests}/_t_${key}.in" );
    $t_unlink = read_text_from_file( "${dir_tests}/_t_${key}.unlink" );
    $t_replace = read_text_from_file( "${dir_tests}/_t_${key}.replace" );

    $res = test_regex_unlink( $text, $pattern );
    $res = test_result( $res, $t_unlink, "unlink", $img, $key );
    if( $res == 1 ) {
        $test_res['bad']++;
    } else {
        $test_res['ok']++;
    }

    $res = test_regex_replace( $text, $pattern, $t["repl_img"] );
    $res = test_result( $res, $t_replace, "replace", $img, $key );
    if( $res == 1 ) {
        $test_res['bad']++;
    } else {
        $test_res['ok']++;
    }
}

print( "\nResults:\n" );
print( "OK : " . $test_res['ok'] . "\nBAD: " . $test_res['bad'] ."\n" );


## collect the name of the tests (verbose code :))
function collect_test_names($dir) {
    $dlist = scandir($dir);

    foreach ($dlist as $key => $val) {
        if( !is_dir( $dir . '/' . $val  ) &&  preg_match( '/^_t_(.+)\.data$/', $val, $matches ) ) {
            $tests[] = $matches[1];
        }
    }
    return $tests;
}


function get_test_data($dir, $t, $keys) {
    $f = fopen( "${dir}/_t_${t}.data", 'r');
    if( $f ) {
        while( ($line = fgets($f)) !== false ) {
            // process a line
            if( preg_match( '/^(\S+)\s*:\s*(.+)$/', $line, $matches ) ) {
                # print( "Test $t ${matches[1]} => ${matches[2]}\n" );
                if( !in_array( $matches[1], $keys ) ) {
                    trigger_error( "Unknown key '$matches[1]' in ${t}.data", E_USER_ERROR );
                }
                $test[$matches[1]] = $matches[2];
            }
        }
    } else {
        trigger_error( "test $t data file is missing", E_USER_ERROR );
    }

    foreach ($keys as $k) {
        if( !array_key_exists( $k, $test ) ) {
            trigger_error( "test $t data is missing $k key", E_USER_ERROR );
        }
    }

    return $test;
}


function test_result( $result, $expected, $name, $img, $id ) {
    global $dir_tests;
    if( $result <> $expected ) {
        #print( "${test} BAD $name:$id ($img)! result=\n$result\n\nexpect=\n$expected\n\n" );
        if( $name == 'pattern' ) {
            print( "${id} BAD $name ($img)! expected '$expected', result '$result', fix in ${dir_tests}/_t_${id}.data, expected is in _bad_${id}.${name}\n" );
        } else {
            print( "${id} BAD $name ($img)! diff -u ${dir_tests}/_t_${id}.${name} _bad_${id}.${name}\n" );
        }
        write_text_to_file( "_bad_${id}.${name}", $result );
        return 1;
    } else {
        print( "${id} OK $name ($img)!\n" );
    }
    return 0;
}


function read_text_from_file( $fname ) {
    $f = fopen( $fname, 'r') or die( "Cannot read file $fname" );
    $text = fread( $f, 1e8 );
    fclose($f);
    return $text;
}


function write_text_to_file($fname, $text) {
    #print("Recording into $fname text '$text'");
    $f = fopen( $fname, 'w') or die( "Cannot create file $fname" );
    $res = fwrite( $f, $text );
    fclose($f);
}


function test_prepare_pattern( $file ) {
    global $matcher;
    return $matcher->matcher_prepare_pattern( $file );
}


function test_regex_unlink( $text, $pattern ) {
    global $matcher;
    return $matcher->matcher_do_unlink( $text, $pattern );
}


function test_regex_replace( $text, $pattern, $newimg ) {
    global $matcher;
    return $matcher->matcher_do_replacement( $text, $pattern, $newimg );
}

mercurial