Ticket #3111: imdb_fix.patch

mythvideo/scripts/imdb.pl

@@ -115 +115 @@
    my $year = $2;
 
    # parse director
-   my $director = parseBetween($response, ">Directed by</b>", "/a><br>");
+   my $director = parseBetween($response, ">Directed by</h5>", "/a><br>");
    $director = parseBetween($director, "/\">", "<");
 
    # parse writer
    # (Note: this takes the 'first' writer, may want to include others)
-   my $writer = parseBetween($response, ">Writing credits</b>", "</table>");
+   my $writer = parseBetween($response, ">Writing credits <a href=\"/wga\">(WGA)</a></h5>", "</table>");
    $writer = parseBetween($writer, "/\">", "</");
 
    # parse plot
-   my $plot = parseBetween($response, ">Plot Outline:</b> ", "<br>");
+   my $plot = parseBetween($response, ">Plot Outline:</h5> \n", "<br>");
    if (!$plot) {
-      $plot = parseBetween($response, ">Plot Summary:</b> ", "<br>");
+      $plot = parseBetween($response, ">Plot Summary:</h5> \n", "<br>");
    }
 
    if ($plot) {
@@ -139 +139 @@
       $plot =~ s/$title_link_pat/$1/g;
 
       # plot ends at first remaining link
-      my $plot_end = index($plot, "<a href=\"");
+      my $plot_end = index($plot, "<a class=\"tn15more inline\" href");
       if ($plot_end != -1) {
          $plot = substr($plot, 0, $plot_end);
       }
@@ -151 +151 @@
 
    # parse MPAA rating
    my $ratingcountry = "USA";
-   my $movierating = parseBetween($response, ">MPAA</a>:</b> ", "<br>");
+   my $movierating = parseBetween($response, ">MPAA</a>:</h5> \n", "</div" );
    if (!$movierating) {
-       $movierating = parseBetween($response, ">Certification:</b>", "<br>");
+       $movierating = parseBetween($response, ">Certification:</h5>", "</div>");
        $movierating = parseBetween($movierating, "certificates=$ratingcountry",
                                    "/a>");
        $movierating = parseBetween($movierating, ">", "<");
    }
 
    # parse movie length
-   my $runtime = parseBetween($response, ">Runtime:</b>\n", " min");
+   my $runtime = parseBetween($response, ">Runtime:</h5>\n", " min");
 
    # parse cast
    #  Note: full cast would be from url:
@@ -168 +168 @@
    my @actors;
    my $cast = "";
    my $count = 0;
-   my $data = parseBetween($response, "Cast overview, first billed only:",
+   my $data = parseBetween($response, "\(Cast overview, first billed only\)",
                                "/table>");
    if ($data) {
       my $beg = "/\">";
@@ -178 +178 @@
       my $actor;
       while ($start != -1) {
          $start += length($beg);
-         $actor = substr($data, $start, $finish - $start);
+         if(substr($data, $start, 1) ne '<') {
+               $actor = substr($data, $start, $finish - $start);
          # add to array
          $actors[$count++] = $actor;
-
+        }
          # advance data to next movie
          $data = substr($data, - (length($data) - $finish));
          $start = index($data, $beg);
@@ -194 +195 @@
    # parse genres
    my $lgenres = "";
    $count = 0;
-   $data = parseBetween($response, "<b class=\"ch\">Genre:</b>","<b class=\"ch\">User Comments:</b>");
+   $data = parseBetween($response, "<h5>Genre:</h5>","<h5>User Comments:</h5>");
    if ($data) {
       my $genre_pat = qr'/Sections/Genres/(?:[a-z ]+/)*">([^<]+)<'im;
       $lgenres = join(',', ($data =~ /$genre_pat/g));
@@ -204 +205 @@
    my @countries;
    my $lcountries = "";
    $count = 0;
-   $data = parseBetween($response, "<b class=\"ch\">Country:</b>","<br>");
+   $data = parseBetween($response, "<h5>Country:</h5>","</div>");
    if ($data) {
       my $beg = "/\">";
       my $end = "</a>";
@@ -428 +429 @@
    # check to see if we got a results page or a movie page
    #    looking for 'add=<movieid>" target=' which only exists
    #    in a movie description page
-   my $movienum = parseBetween($response, "add=", "\" target=");
+   my $movienum = parseBetween($response, "add=", "\"");
    if ($movienum) {
        if (defined $opt_d) { printf("# redirected to movie page\n"); }
        my $movietitle = parseBetween($response, "<title>", "</title>");