First Code 2 - Merrill Alegado

<?php 
require_once("./simple_html_dom.php");   
 
//getAuthorURLs();
//getPoemURLs();
getPoemTexts();
 
function getAuthorURLs() {
  //Load html from url
  $url = "http://www.poetiv.com";
  $ch = curl_init($url);
  $fp = fopen("./poetiv.html", "w");
  curl_setopt($ch, CURLOPT_FILE, $fp);
  curl_setopt($ch, CURLOPT_HEADER, 0);
  curl_exec($ch);
  fclose($fp);
 
  //output csv file
  $csv_out = fopen("AuthorURLs.csv","w");
 
  //Create Dom from html
  $dom = new simple_html_dom();
  $dom->load_file( "poetiv.html" );
 
  $div_left_list = $dom->find("#fleft");
 
  //line to debug size of list
  //echo count($div_left_list) . "\n";
 
  foreach ( $div_left_list as &$div ) {
    $a_list = $div->find("a");
    foreach ( $a_list as &$anchor ) {
      //put the data into an array and then write array out to a csv file.
      $csv_array = array( $anchor->plaintext, $anchor->getAttribute("href") );
      fputcsv( $csv_out, $csv_array );
    }
  }
 
  $div_right_list = $dom->find("#fright");
 
  foreach ( $div_right_list as &$div ) {
    $a_list = $div->find("a");
    foreach ( $a_list as &$anchor ) {
      //put the data into an array and then write array out to a csv file.
      $csv_array = array( $anchor->plaintext, $anchor->getAttribute("href") );
      fputcsv( $csv_out, $csv_array );
    }
  }
 
  //close the csv file output.
  fclose( $csv_out );
}
 
function getPoemURLs() {
  $input_csv = fopen( "AuthorURLs.csv", "r" );
  $author_names = array();
  $author_urls = array();
 
  while (($columns = fgetcsv($input_csv, 1000, ",")) !== FALSE) {
    array_push( $author_names, $columns[0] );
    array_push( $author_urls,  $columns[1] );
  }
  fclose( $input_csv );
 
  $output_csv = fopen( "PoemURLs.csv", "w" );
 
  for ( $i=0; $i<count($author_names); $i++ ) {
    echo "getting page(s) for " . $author_names[$i] . " " . trim($author_urls[$i]) . "\n";
    findPoemURLs( trim($author_urls[$i]), $output_csv, trim($author_names[$i]) );
 
    for ( $j=2; $j<16; $j++ ) {
      $author_url =  str_replace( ".html", "", $author_urls[$i] ) . $j . ".html";
      echo "getting page(s) for " . $author_names[$i] . " " . trim($author_url) . "\n";
      findPoemURLs( trim($author_url), $output_csv, trim($author_names[$i]) );
    }
  }
  fclose( $output_csv );
}
 
function findPoemURLs( $url, $out, $author_name ) {
   $ch = curl_init($url);
   $fp = fopen("./author.html", "w");
   curl_setopt($ch, CURLOPT_FILE, $fp);
   curl_setopt($ch, CURLOPT_HEADER, 0);
   curl_exec($ch);
   fclose($fp);
 
   if ( filesize( "./author.html" ) > 1000 ) {
    $html = new simple_html_dom();
    $html->load_file( "./author.html" );
 
    $fleft = $html->find("#fleft");
    foreach ( $fleft as &$myFleft ) {
     $aList = $myFleft->find("a");
     foreach ( $aList as &$anchor ) {
       $poem_name = trim($anchor->plaintext);
       $poem_url = trim($anchor->getAttribute('href'));
       $poem_row = array( $author_name, $poem_name, $poem_url );
       fputcsv( $out, $poem_row );
     }
    }
 
    $fright = $html->find("#fright");
    foreach ( $fright as &$myFright ) {
     $aList = $myFright->find("a");
     foreach ( $aList as &$anchor ) {
       $poem_name = trim($anchor->plaintext);
       $poem_url = trim($anchor->getAttribute('href'));
       $poem_row = array( $author_name, $poem_name, $poem_url );
       fputcsv( $out, $poem_row );
     }
    }
 
    $html->clear();
   }
   unlink( "./author.html" );
}
 
function getPoemTexts() {
  $input_csv = fopen( "PoemURLs.csv", "r" );
  $author_names = array();
  $poem_names = array();
  $poem_urls = array();
 
  while (($columns = fgetcsv($input_csv, 1000, ",")) !== FALSE) {
    array_push( $author_names, $columns[0] );
    array_push( $poem_names,  $columns[1] );
    array_push( $poem_urls,  $columns[2] );
  }
  fclose( $input_csv );
 
  $dbh = mysql_connect( 'localhost', 'poetiv', 'alegado' );
  mysql_select_db( 'poetiv', $dbh );
 
  $output_csv = fopen( "Poems.csv", "w" );
 
  $simple = new simple_html_dom();
 
  for ( $i=0; $i<count($poem_urls); $i++ ) {
    //$simple->load_file( trim($poem_urls[$i]) );
    $ch = curl_init( trim($poem_urls[$i]) );
    $fp = fopen("./poem.html", "w");
    curl_setopt($ch, CURLOPT_FILE, $fp);
    curl_setopt($ch, CURLOPT_HEADER, 0);
    curl_exec($ch);
    fclose($fp);
 
    if ( filesize("./poem.html") > 800 ) {
     $simple->load_file( "./poem.html" );
     $pre_list = $simple->find('pre');
     $poem_text = '';
     foreach ( $pre_list as &$pre ) {
       $poem_text = trim($pre->plaintext);
       $poem_text = str_replace('"','',$poem_text);
     }
 
     echo $author_names[$i] . ", " . $poem_names[$i] . ", " . $poem_urls[$i] . ", " . $poem_text . "\n";
     $poem_data = array( $author_names[$i], $poem_names[$i], $poem_text, $poem_urls[$i] );
     fputcsv( $output_csv, $poem_data );   
     mysql_query( 'INSERT INTO poetiv_poems ( author_name, poem_name, poem_text, poem_url ) VALUES ( "' . $author_names[$i] . '", "' . $poem_names[$i] . '", "' . $poem_text . '", "' . $poem_urls[$i] . '" )' );
 
     usleep( 200000 );
    } 
  }
 
  fclose( $output_csv );
  mysql_close( $dbh );
}
 
?>