parse_doc.php

<?php
$base_dir = $argv[1];
$rs_dir = opendir($base_dir);
while ($filename = readdir($rs_dir)) {
    if (false === strpos($filename, '.doc')) {
        continue;
    }
    $filepath = $base_dir . '/' . $filename;
    $arr_line = file($filepath);
    $boundary = '';
    $arr_part = [];
    $part = '';
    $part_is_text = 0;
    foreach ($arr_line as $line) {
        $line = trim($line);
        if (preg_match('/^Content-Type:multipart.*boundary="([a-z0-9A-Z_-]+)"$/', $line, $m)) {
            $boundary = $m[1];
        }
        if ($boundary) {
            if (false !== strpos($line, $boundary)) {
                if ($part) {
                    $arr_part[] = $part;
                }
                $part = '';
                $part_is_text = 0;
            } elseif (preg_match('/^Content-Type:text.*/', $line)) {
                $part_is_text = 1;
            } elseif (preg_match('/:/', $line)) {
                //is header
            } elseif ($line) {
                if ($part_is_text) {
                    $part .= trim($line);
                }
            }
        }
    }
    $arr_mobile = [];
    foreach ($arr_part as $part) {
        $utf8_content = iconv('GBK', 'UTF8', base64_decode($part));
        if (preg_match('/1[0-9]{10}/', $utf8_content, $m)) {
            $arr_mobile[] = $m[0];
        }
    }
    $mobile = join(',', $arr_mobile);
    if (preg_match('/.*_(.+)-(.+)\.doc/', $filename, $m)) {
        $name = $m[1];
        $city = $m[2];
        printf("%s\t%s\t%s\n", $name, $city, $mobile);
    } else {
        echo "==={$filename}====不合法\n";
    }


}
closedir($rs_dir);