[Assorted-commits] SF.net SVN: assorted: [191] movie-lookup

[<yan...@us...>]

Revision: 191
          http://assorted.svn.sourceforge.net/assorted/?rev=191&view=rev
Author:   yangzhang
Date:     2007-12-23 17:39:35 -0800 (Sun, 23 Dec 2007)

Log Message:
-----------
added movie-lookup

Added Paths:
-----------
    movie-lookup/
    movie-lookup/trunk/
    movie-lookup/trunk/README
    movie-lookup/trunk/src/
    movie-lookup/trunk/src/MovieLookup.hs
    movie-lookup/trunk/src/Setup.hs
    movie-lookup/trunk/src/hbo.bash
    movie-lookup/trunk/src/lookup.bash
    movie-lookup/trunk/src/movie-lookup.cabal

Added: movie-lookup/trunk/README
===================================================================
--- movie-lookup/trunk/README	                        (rev 0)
+++ movie-lookup/trunk/README	2007-12-24 01:39:35 UTC (rev 191)
@@ -0,0 +1,7 @@
+- Build the `movie-lookup` binary.
+
+- Go to HBO's online schedule and select whatever options yield the largest
+  amount of information in a printable view. Copy the plain-text body of this
+  page into `hbo.txt`.
+
+- Run `hbo.bash`, which uses `lookup.bash`, which uses `movie-lookup`.

Added: movie-lookup/trunk/src/MovieLookup.hs
===================================================================
--- movie-lookup/trunk/src/MovieLookup.hs	                        (rev 0)
+++ movie-lookup/trunk/src/MovieLookup.hs	2007-12-24 01:39:35 UTC (rev 191)
@@ -0,0 +1,66 @@
+module Main where
+
+import Control.Arrow
+import Control.Monad
+import Data.ByteString.Char8 (ByteString)
+import qualified Data.ByteString.Char8 as BS
+import Data.Char
+import Data.List
+import Data.String
+import Debug.Trace
+import System.Environment
+import System.FilePath
+import System.Path.Glob
+import Text.Regex
+
+t x = trace (show x) x
+
+-- TODO Commons
+
+fork :: [Either a b] -> ([a],[b])
+fork = foldr f ([],[])
+  where f e (as,bs) = case e of
+                        Left a  -> (a:as,   bs)
+                        Right b -> (  as, b:bs)
+
+simplify = BS.map toLower >>>
+           BS.words >>> BS.unwords >>>
+           BS.unpack >>> strip >>> BS.pack
+
+simplifyTitle s =
+  let simplified = BS.unpack $ simplify s
+      rgx = mkRegex "^the |, the$"
+      dearticled = subRegex rgx simplified ""
+  in  dearticled
+
+subRegexBS regex target replace = BS.pack $ subRegex regex (BS.unpack target) replace
+
+main = do
+  -- load the indexes; their union is the catalog
+  (dir:_) <- getArgs
+  indexes <- glob $ dir </> "*"
+  catalog <- fmap (BS.lines . BS.concat) $ mapM BS.readFile indexes
+
+  -- process queries
+  queries <- fmap BS.lines BS.getContents
+  let (hits, misses) = query catalog queries
+
+  -- output
+  putStrLn "hits"
+  putStrLn "----"
+  forM_ hits $ \(title, url) -> do
+    putStrLn $ baseUrl ++ BS.unpack url
+  putStrLn ""
+  putStrLn "misses"
+  putStrLn "------"
+  mapM_ BS.putStrLn misses
+
+baseUrl = "http://www.rottentomatoes.com"
+
+query :: [ByteString] -> [ByteString] -> ([(ByteString, ByteString)],[ByteString])
+query catalog = map queryOne >>> fork
+  where queryOne q = case find (cmp q) catalog of
+                       Just entry -> Left (q, subRegexBS rgx entry "\\1")
+                       Nothing    -> Right q
+        cmp q entry = simplifyTitle q == simplifyTitle (subRegexBS rgx entry "\\2")
+        rgx = mkRegex ".*href=\"([^\"]*)\">([^<]*)<.*"

Added: movie-lookup/trunk/src/Setup.hs
===================================================================
--- movie-lookup/trunk/src/Setup.hs	                        (rev 0)
+++ movie-lookup/trunk/src/Setup.hs	2007-12-24 01:39:35 UTC (rev 191)
@@ -0,0 +1,4 @@
+#!/usr/bin/env runhaskell
+
+import Distribution.Simple
+main = defaultMain


Property changes on: movie-lookup/trunk/src/Setup.hs
___________________________________________________________________
Name: svn:executable
   + *

Added: movie-lookup/trunk/src/hbo.bash
===================================================================
--- movie-lookup/trunk/src/hbo.bash	                        (rev 0)
+++ movie-lookup/trunk/src/hbo.bash	2007-12-24 01:39:35 UTC (rev 191)
@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+
+set -o errexit -o nounset
+
+cat hbo.txt |
+sed 's/[^\.]*\.\.\.\.// ; s/\.\..*//' |
+sort -u |
+./lookup.bash


Property changes on: movie-lookup/trunk/src/hbo.bash
___________________________________________________________________
Name: svn:executable
   + *

Added: movie-lookup/trunk/src/lookup.bash
===================================================================
--- movie-lookup/trunk/src/lookup.bash	                        (rev 0)
+++ movie-lookup/trunk/src/lookup.bash	2007-12-24 01:39:35 UTC (rev 191)
@@ -0,0 +1,35 @@
+#!/usr/bin/env bash
+
+set -o errexit -o nounset
+
+# get (cache) all the index pages (300-400 of these!)
+
+tmp="${TEMP:-/tmp}"
+movietmp="$tmp/movies.txt"
+if [ ! -d "$tmp/www.rottentomatoes.com" ]
+then wget -P "$tmp" -r -l 1 http://www.rottentomatoes.com/features/stats/index.php
+fi
+
+# search for the movies (read from stdin)
+
+if [ ! -f "$movietmp" ]
+then dist/build/movie-lookup/movie-lookup "$tmp/www.rottentomatoes.com/features/stats/index" > "$movietmp"
+fi
+
+# download the movies
+
+grep ^http: "$movietmp" |
+wget -q -i- -x -P "$tmp" -nc || true
+
+# extract scores
+
+grep 'critics_tomatometer_score_txt_percentage' "$tmp"/www.rottentomatoes.com/m/*/index.html |
+grep -v '<div' |
+sed 's/<span.*//'
+
+# tomatometer looks like:
+# <div id="critics_tomatometer_score_txt">
+# 13<span id="critics_tomatometer_score_txt_percentage">%</span>
+# </div>
+
+# vim:et:sw=2:ts=2


Property changes on: movie-lookup/trunk/src/lookup.bash
___________________________________________________________________
Name: svn:executable
   + *

Added: movie-lookup/trunk/src/movie-lookup.cabal
===================================================================
--- movie-lookup/trunk/src/movie-lookup.cabal	                        (rev 0)
+++ movie-lookup/trunk/src/movie-lookup.cabal	2007-12-24 01:39:35 UTC (rev 191)
@@ -0,0 +1,18 @@
+Name:                movie-lookup
+Version:             0.0
+Description:         Retrieve ratings from RottenTomatoes.com.
+License:             GPL
+License-file:        LICENSE
+Author:              Yang Zhang
+Maintainer:          gmail:yaaang
+Build-Depends:       base,
+                     FilePath >= 0.11,
+                     MissingH >= 0.18,
+                     regex-base >= 0.71,
+                     regex-compat >= 0.71,
+                     regex-posix >= 0.71,
+                     unix >= 1.0
+
+Executable:          movie-lookup
+Main-is:             MovieLookup.hs
+ghc-options:         -O


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.