| 1 | Index: scripts/imdb.pl |
|---|
| 2 | =================================================================== |
|---|
| 3 | --- scripts/imdb.pl (revision 13165) |
|---|
| 4 | +++ scripts/imdb.pl (working copy) |
|---|
| 5 | @@ -94,6 +94,9 @@ |
|---|
| 6 | my $ldata = lc($data); |
|---|
| 7 | my $start = index($ldata, lc($beg)) + length($beg); |
|---|
| 8 | my $finish = index($ldata, lc($end), $start); |
|---|
| 9 | + |
|---|
| 10 | + # if (defined $opt_d) { printf("ParseBetween -> $beg ($start) $end ($finish)\n"); } |
|---|
| 11 | + |
|---|
| 12 | if ($start != (length($beg) -1) && $finish != -1) { |
|---|
| 13 | my $result = substr($data, $start, $finish - $start); |
|---|
| 14 | # return w/ decoded numeric character references |
|---|
| 15 | @@ -130,13 +133,12 @@ |
|---|
| 16 | $title = $1; |
|---|
| 17 | } |
|---|
| 18 | |
|---|
| 19 | - # parse director |
|---|
| 20 | - my $director = parseBetween($response, ">Directed by</h5>", "/a><br/>"); |
|---|
| 21 | - $director = parseBetween($director, "/\">", "<"); |
|---|
| 22 | + # parse director(s) |
|---|
| 23 | + my $director = parseBetween($response, ">Director", ">Writer"); |
|---|
| 24 | + $director = join(",", ($director =~ m/$name_link_pat/g)); |
|---|
| 25 | |
|---|
| 26 | - # parse writer |
|---|
| 27 | - # (Note: this takes the 'first' writer, may want to include others) |
|---|
| 28 | - my $data = parseBetween($response, ">Writing credits <a href=\"/wga\">(WGA)</a></h5>", "\n<br/>"); |
|---|
| 29 | + # parse writer(s) |
|---|
| 30 | + my $data = parseBetween($response, ">Writer", ">Release"); |
|---|
| 31 | my $writer = join(",", ($data =~ m/$name_link_pat/g)); |
|---|
| 32 | |
|---|
| 33 | # parse plot |
|---|
| 34 | @@ -417,16 +419,23 @@ |
|---|
| 35 | # possible matches are grouped in several catagories: |
|---|
| 36 | # exact, partial, and approximate |
|---|
| 37 | my $popular_results = parseBetween($response, "<b>Popular Titles</b>", |
|---|
| 38 | - "</ol>"); |
|---|
| 39 | + "<p>"); |
|---|
| 40 | my $exact_matches = parseBetween($response, "<b>Titles (Exact Matches)</b>", |
|---|
| 41 | - "</ol>"); |
|---|
| 42 | + "<p>"); |
|---|
| 43 | my $partial_matches = parseBetween($response, "<b>Titles (Partial Matches)</b>", |
|---|
| 44 | - "</ol>"); |
|---|
| 45 | + "<p>"); |
|---|
| 46 | # my $approx_matches = parseBetween($response, "<b>Approximate Matches</b>", |
|---|
| 47 | # "</ol>"); |
|---|
| 48 | + |
|---|
| 49 | + |
|---|
| 50 | + |
|---|
| 51 | +#print "--------------------------\n"; |
|---|
| 52 | +#print $popular_results, "\n"; |
|---|
| 53 | +#print "EXACT $exact_matches\n"; |
|---|
| 54 | + |
|---|
| 55 | # parse movie list from matches |
|---|
| 56 | - my $beg = "<li>"; |
|---|
| 57 | - my $end = "</li>"; |
|---|
| 58 | + my $beg = "<td"; |
|---|
| 59 | + my $end = "</td>"; |
|---|
| 60 | my $count = 0; |
|---|
| 61 | my @movies; |
|---|
| 62 | |
|---|
| 63 | @@ -445,7 +454,14 @@ |
|---|
| 64 | my $year; |
|---|
| 65 | my $type; |
|---|
| 66 | my $title; |
|---|
| 67 | + |
|---|
| 68 | + if (defined $opt_d) {print "data = $data\n";}; |
|---|
| 69 | + if (defined $opt_d) {print "Item start tag = $beg\n";}; |
|---|
| 70 | + if (defined $opt_d) {print "start = $start finish = $finish\n";}; |
|---|
| 71 | + |
|---|
| 72 | while ($start != -1 && $start < length($data)) { |
|---|
| 73 | + |
|---|
| 74 | + if (defined $opt_d) { print "Item start = $start ";}; |
|---|
| 75 | $start += length($beg); |
|---|
| 76 | my $entry = substr($data, $start, $finish - $start); |
|---|
| 77 | $start = index($data, $beg, $finish + 1); |
|---|
| 78 | @@ -465,6 +481,7 @@ |
|---|
| 79 | if ($lhs =~ m/<a href="\/title\/tt(\d+)\/.*\">(.+)<\/a>/i) { |
|---|
| 80 | $movienum = $1; |
|---|
| 81 | $title = $2; |
|---|
| 82 | + if (defined $opt_d) { print "Movie $movienum, Title $title";}; |
|---|
| 83 | } else { |
|---|
| 84 | if (defined $opt_d) { |
|---|
| 85 | print("Unrecognized entry format\n"); |
|---|
