Search for/in (certain) files

filesearch.php

<?php

class Filesearch {
  // Config
  public $string = '';
  public $extensions = array();
  public $dir = '';
  public $exclude = array();
  public $in_files = true;
  public $matcher = 'ci';
  public $padding = 2;

  // Process
  public $match_extensions = array();
  public $matcher_callback = null;
  public $regex_string = '';
  public $file_contents = array();
  public $_start = 0;
  public $_end = 0;

  // Result
  public $num_scanned_files = 0;
  public $num_matched_files = 0;
  public $num_matched_lines = 0;
  public $matched_files = array();
  public $matched_lines = array();

  function __construct( $dir, $string, $options = array() ) {
    foreach ( $options as $name => $value ) {
      if ( isset($this->$name) ) {
        $this->$name = $value;
      }
    }

    if ( is_string($this->extensions) ) {
      $this->extensions = array_filter(explode(',', $this->extensions));
    }

    $this->exclude = array_filter(array_map('realpath', $this->exclude));

    $this->dir = str_replace(array('\\', '/'), DIRECTORY_SEPARATOR, $dir);
    $this->string = $string;
  }

  function find() {
    $this->_start = microtime(1);

    switch ($this->matcher) {
      case 're':
      case 'reci':
        $ci = $this->matcher == 'reci' ? 'i' : '';
        $this->matcher_callback = function($string, $pattern) use ($ci) {
          return preg_match('/' . $pattern . "/$ci", $string) > 0;
        };
        $this->regex_string = '/(' . $this->string . ")/$ci";
        break;

      case 'cs':
        $this->matcher_callback = function($haystack, $needle) {
          return strpos($haystack, $needle) !== FALSE;
        };
        $this->regex_string = '/(' . preg_quote($this->string, '/') . ')/';
        break;

      case 'ci':
      default:
        $this->matcher_callback = function($haystack, $needle) {
          return stripos($haystack, $needle) !== FALSE;
        };
        $this->regex_string = '/(' . preg_quote($this->string, '/') . ')/i';
        break;
    }

    $this->match_extensions = in_array('*', $this->extensions) ?: array_flip(array_map('strtolower', $this->extensions));

    set_time_limit(0);
    $this->read($this->dir);

    $this->_end = microtime(1);
    $this->printStatsSummary();
  }

  function dirContents( $dir ) {
    $files = array();

    $map = opendir($dir);
    while ( $file = readdir($map) ) {
      $filepath = $dir . DIRECTORY_SEPARATOR . $file;
      if ( is_readable($filepath) && $file[0] != '.' ) {
        $files[] = $filepath;
      }
    }

    natcasesort($files);
    return $files;
  }

  function read( $dir ) {
    $dir = rtrim($dir, '\\/');

    foreach ( $this->dirContents($dir) as $filepath ) {
      if ( is_file($filepath) ) {
        if ( $this->potentialFile($filepath) ) {
          $this->num_scanned_files++;
          if ( $this->match($filepath) ) {
            // match
            $this->num_matched_files++;
            $this->matched_files[] = $filepath;
            $this->matched_lines[$filepath] = array();

            $this->printFileMatch($filepath);
          }
        }
      }
      else if ( is_dir($filepath) ) {
        if ( $this->potentialDir($filepath) ) {
          // recurse
          $this->read($filepath);
        }
      }
    }
  }

  function getStatsTitle() {
    $_end = $this->_end ?: microtime(1);
    $time = number_format($_end - $this->_start, 2);
    $delim = in_array($this->matcher, ['re', 'reci']) ? '/' : '"';
    return $this->num_matched_lines . ' x <code>' . $delim . html($this->string) . $delim . '</code> in ' . $this->num_matched_files . ' / ' . $this->num_scanned_files . ' files (' . $time . ' sec)';
  }

  function printStatsSummary() {
    echo '<script>';
    echo 'document.getElementById("top-results-summary").innerHTML = "' . javascript($this->getStatsTitle()) . '";';
    echo 'document.title = "' . javascript(strip_tags($this->getStatsTitle())) . '";';
    echo '</script>';
  }

  function htmlClass( $string ) {
    $string = strtolower($string);
    $string = preg_replace('#[^a-z0-9\-]+#', '-', $string);
    $string = trim($string, '-');
    $string = preg_replace('#\-+#', '-', $string);
    return $string;
  }

  function printFileMatch( $filepath ) {
    $linesBefore = $this->num_matched_lines;
    $lineMatches = '';
    if ( $this->in_files ) {
      $lineMatches .= '<div class="match-groups">';
      $lineMatches .= $this->printLineMatches($filepath);
      $lineMatches .= '</div>';
    }
    $linesAfter = $this->num_matched_lines;
    $matches = $linesAfter - $linesBefore;

    $numMatches = $this->in_files ? ' (' . $matches . ')' : '';

    echo "\n" . '<div class="match-file">';
    echo '<h3 id="' . $this->htmlClass($filepath) . '" class="match-file-title">' . $filepath . $numMatches . '</h3>' . "\n";
    echo $lineMatches;
    echo $this->printStatsSummary();
    echo '</div>' . "\n\n";

    flush();
    // sleep(1);
  }

  function printLineMatches( $filepath ) {
    $lines = $this->getFileLines($filepath);
    $matchedGroups = $this->getLineMatches($filepath);

    $html = '';
    foreach ( $matchedGroups as $group ) {
      $matchedLine = key($group);
      $linesPre = array_slice($lines, 0, $matchedLine);
      $codePre = implode("\n", $linesPre);

      $lastFunction = '<no function or class>';
      if ( preg_match_all('#(function\s+[a-zA-Z0-9_]+[\sa-zA-Z0-9,_&:\(\)$\[\]=\'"-\.]*)\s*{#', $codePre, $functionMatches) ) {
        $lastFunction = trim(end($functionMatches[1]));
      }
      else if ( preg_match_all('#(class\s+[a-zA-Z0-9_]+[\sa-zA-Z0-9,_&:\(\)$\[\]=\'"-]*)\s*{#', $codePre, $functionMatches) ) {
        $lastFunction = trim(end($functionMatches[1]));
      }

      $html .= '<div class="match-function-context">' . html($lastFunction) . '</div>';
      $html .='<ul class="match-group">';
      foreach ( $group as $line => $match ) {
        $classes = $match ? 'match-line-match' : '';

        $printLine = $lines[$line];
        if ( $printLine === '' ) {
          $printLine = '&nbsp;';
        }
        else if ( $match ) {
          $i = preg_match_all($this->regex_string, $printLine, $rematches, PREG_OFFSET_CAPTURE);
          $rematches = $rematches[0];

          $parts = array();
          $lastEnd = 0;
          foreach ($rematches as $rematch) {
            $parts[] = html(substr($printLine, $lastEnd, $rematch[1] - $lastEnd));
            $parts[] = '<span class="match">' . html($rematch[0]) . '</span>';

            $lastEnd = $rematch[1] + strlen($rematch[0]);
          }
          $parts[] = html(substr($printLine, $lastEnd));

          $printLine = implode($parts);
        }
        else {
          $printLine = html($printLine);
        }

        $lineNumber = str_pad($line + 1, 4, ' ', STR_PAD_LEFT);
        $lineNumber = '<span class="ln">' . $lineNumber . ' </span>';

        $html .='<li class="match-line ' . $classes . '">' . $lineNumber . $printLine . '</li>';
      }
      $html .='</ul>';
    }

    return $html;
  }

  function getLineMatches( $filepath, $context = null ) {
    $context === null && $context = $this->padding;
    $matcher = $this->matcher_callback;

    $lines = $this->getFileLines($filepath);
    $maxLine = count($lines) - 1;

    $matchedLines = array();
    foreach ( $lines as $i => $line ) {
      if ( $matcher($line, $this->string) ) {
        $this->num_matched_lines++;
        $this->matched_lines[$filepath][] = $i;

        if ( $context ) {
          for ( $j=max(0, $i-$context), $m=min($maxLine, $i+$context); $j<=$m; $j++ ) {
            $matchedLines[$j] = !empty($matchedLines[$j]) || $i == $j;
          }
        }
        else {
          $matchedLines[$i] = true;
        }
      }
    }

    $matchedGroups = array();
    $lastLine = -9;
    foreach ( $matchedLines as $line => $match ) {
      if ( $line != $lastLine+1 ) {
        $matchedGroups[] = array();
        $group = &$matchedGroups[count($matchedGroups)-1];
      }

      $group[$line] = $match;
      $lastLine = $line;
    }

    return $matchedGroups;
  }

  function getFileContents( $filepath ) {
    if ( !isset($this->file_contents[$filepath]) ) {
      $this->file_contents[$filepath] = file_get_contents($filepath);
    }

    return $this->file_contents[$filepath];
  }

  function getFileLines( $filepath ) {
    $contents = $this->getFileContents($filepath);

    return preg_split('/(\r\n|\r|\n)/', $contents);
  }

  function potentialFile( $filepath ) {
    $name = basename($filepath);
    return $name[0] != '.' && $this->matchExtension($name) && !in_array(realpath($filepath), $this->exclude);
  }

  function matchExtension( $name ) {
    if ($this->match_extensions === true) {
      return true;
    }

    $ext = strtolower(substr(strrchr($name, '.'), 1));
    return isset($this->match_extensions[$ext]);
  }

  function potentialDir( $filepath ) {
    $name = basename($filepath);
    return $name[0] != '.' && !in_array(realpath($filepath), $this->exclude);
  }

  function match( $filepath ) {
    return $this->matchFileName($filepath) || ( $this->in_files && $this->matchFileContents($filepath) );
  }

  function matchFileName( $filepath ) {
    $matcher = $this->matcher_callback;
    return $matcher($filepath, $this->string);
  }

  function matchFileContents( $filepath ) {
    $matcher = $this->matcher_callback;
    $contents = file_get_contents($filepath);
    $match = $matcher($contents, $this->string);
    if ( $match ) {
      $this->file_contents[$filepath] = $contents;
    }
    return $match;
  }

}

header('Content-type: text/html; charset=utf-8');

if ( isset($_GET['extensions']) && $_GET['extensions'] != @$_COOKIE['fs_extensions'] ) {
  $_COOKIE['fs_extensions'] = $_GET['extensions'];
  setcookie('fs_extensions', $_COOKIE['fs_extensions'], time()+86400*90);
}

if ( isset($_GET['path']) && $_GET['path'] != @$_COOKIE['fs_path'] ) {
  $_COOKIE['fs_path'] = $_GET['path'];
  setcookie('fs_path', $_COOKIE['fs_path'], time()+86400*90);
}

if ( isset($_GET['exclude']) ) {
  $excludes = explode(',', (string) @$_COOKIE['fs_excludes']);
  $excludes = array_filter(array_unique(array_merge($excludes, (array) $_GET['exclude'])));
  sort($excludes);
  $excludes = implode(',', $excludes);
  if ( $excludes != @$_COOKIE['fs_excludes'] ) {
    setcookie('fs_excludes', $_COOKIE['fs_path'] = $excludes, time()+86400*90);
  }
}

$matchers = array('ci' => 'Case insensitive', 'cs' => 'Case sensitive', 're' => 'RegExp', 'reci' => 'RegExp CI');
$paddings = array(0 => 'none', 1 => '1 line', 2 => '2 lines', 3 => '3 lines', 5 => '5 lines', 10 => '10 lines');

$excludes = array_filter(explode(',', (string) @$_COOKIE['fs_excludes']));
$excludeDirs = array_filter((array)@$_GET['exclude']) ?: array('');

?>
<!doctype html>
<html>

<head>
<meta charset="utf-8" />
<link rel="shortcut icon" href="/favicon.png" />
<style>
html, body { margin: 0; padding: 0; width: -webkit-fit-content; min-width: 100%; }
body { padding: 6px 0; }
p { margin: 2px 8px; }
label { display: inline-block; min-width: 7em; }
input:not([type=checkbox]):not([type=submit]):not([type=reset]) { width: 40em; }
.exclude a { display: inline-block; margin-left: .5em; text-decoration: none; -webkit-transform: scale(1.8); font-weight: bold; }
h3, ul, li { margin: 0; padding: 0; display: block; list-style: none; }
.match-file { margin-bottom: 8px; }
.match-file-title { font-family: monospace; font-size: 1.2rem; font-weight: normal; background-color: #444; color: #fff; padding: 3px 6px; }
.match-groups { font-family: monospace; background-color: #888; }
.match-function-context { color: white; }
.match-function-context:first-child { padding-top: 4px; }
.match-group { padding: 4px 0; line-height: 1.4; }
.match-line { white-space: pre; tab-size: 4; background-color: #ccc; border-bottom: solid 1px rgba(0, 0, 0, 0.06); }
.match-line-match { background-color: #ddd; position: relative; }
span.match { background-color: #f88; color: #000; padding: 2px; }
span.ln { color: #d63900; }
.result-options { text-align: right; padding: 4px 8px; }
#top-results-summary { float: left; }

body:not(.ln) .ln { display: none; }
body.fn .match-file > :not(h3) { display: none; }
</style>
</head>

<body class="ln">
<form id="search_form" method="get" action>
  <p><label>String: </label><input id="string" name="string" value="<?= html(@$_GET['string']) ?>" /></p>
  <p><label>Extensions: </label><input name="extensions" value="<?= html(@$_COOKIE['fs_extensions']) ?>" /></p>
  <p><label>Path: </label><input name="path" value="<?= html(@$_COOKIE['fs_path']) ?>" /></p>
  <?php foreach ( $excludeDirs as $exclude ) { ?>
    <p class="exclude"><label>Ignore: </label><input autocomplete="off" list="excludes" name="exclude[]" value="<?= html($exclude) ?>" /> <a class="more-exclude" href="#">+</a></p>
  <?php } ?>
  <p class="cb"><label><input type="checkbox" name="in_files" value="1" checked /> Within files</label></p>
  <p><label>Find method: </label><select name="matcher"><?= options($matchers, isset($_GET['matcher']) ? $_GET['matcher'] : 'cs') ?></select></p>
  <p><label>Show context: </label><select name="padding"><?= options($paddings, isset($_GET['padding']) ? $_GET['padding'] : 10) ?></select></p>

  <datalist id="excludes"><option value="<?= implode('"><option value="', $excludes) ?>"></datalist>

  <p>
    <input type="submit" />
    <input type="reset" />
  </p>
</form>

<script>
function $(sel) { return document.querySelector(sel); }
function fillExcludes(path) {
  <? if (!empty($_GET['excludes'])): ?>return console.log('no excludes');<? endif ?>

  var oldEl;
  while (oldEl = $('p.exclude + p.exclude')) {
    oldEl.remove();
  }

  var added = false;
  [].forEach.call($('#excludes').options, el => {
    if (el.value.startsWith(path)) {
      var newEl = addExclude();
      newEl.querySelector('input').value = el.value;
      added = true;
    }
  });
  if (added) {
    $('p.exclude').remove();
  }
}
function addExclude() {
  var els = document.querySelectorAll('p.exclude');
  var lastEl = els[els.length-1];
  var newEl = lastEl.cloneNode(true);
  newEl.querySelector('input').value = '';
  lastEl.parentNode.insertBefore(newEl, lastEl.nextElementSibling);
  return newEl;
}
$('#search_form input[name="path"]').onchange = function(e) {
  fillExcludes(this.value);
};
$('#search_form').onclick = function(e) {
  if (e.target.classList.contains('more-exclude')) {
    e.preventDefault();

    var newEl = addExclude();
    setTimeout(function() {
      newEl.querySelector('input').focus();
    });
  }
};
var $path = $('input[name="path"]');
if ($path.value) {
  fillExcludes($path.value);
}
</script>

<?php

if ( isset($_GET['string'], $_GET['path']) ) {
  ?>
  <div class="result-options" style="max-width: calc(100vw - 20px); box-sizing: border-box">
    <div id="top-results-summary"></div>
    <label><input type="checkbox" onclick="document.body.classList.toggle('fn', this.checked)" /> Show ONLY file names</label>
    <label><input type="checkbox" onclick="document.body.classList.toggle('ln', this.checked)" checked /> Show line numbers</label>
  </div>
  <?php
  $_GET['in_files'] = !empty($_GET['in_files']);

  $search = new Filesearch($_GET['path'], $_GET['string'], $_GET);
  // print_r($search);

  $search->find();

  $time = number_format($search->_end - $search->_start, 4);
  echo '<p>' . $search->getStatsTitle() . '</p>';
  // print_r($search);
}

?>
</body>

</html>
<?php

function javascript( $str ) {
  return addslashes($str);
}

function html( $str ) {
  return htmlspecialchars($str, ENT_COMPAT, 'UTF-8');
}

function options( $options, $selected ) {
  $html = '';
  foreach ( $options AS $v => $label ) {
    $html .= '<option value="' . html($v) . '"';
    if ( $selected == $v ) {
      $html .= ' selected';
    }
    $html .= '>' . html($label) . '</option>';
  }
  return $html;
}