From d041afa83bcea5b439907861765ffd87ced1a9fb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fr=C3=A9d=C3=A9ric=20P=C3=A9ters?= <fpeters@entrouvert.com>
Date: Thu, 20 Sep 2012 16:31:34 +0200
Subject: [PATCH] add possibility to add files

---
 server/index.html       |  9 +++++--
 server/search_server.py | 60 ++++++++++++++++++++++++++++++++++-------
 server/upload.html      | 38 ++++++++++++++++++++++++++
 3 files changed, 95 insertions(+), 12 deletions(-)
 create mode 100644 server/upload.html

diff --git a/server/index.html b/server/index.html
index 09c377b..df7eb64 100644
--- a/server/index.html
+++ b/server/index.html
@@ -13,7 +13,12 @@ function display_results(data) {
   var items = [];
   $('#debug').html(data['qtime']);
   $.each(data['results'], function(key, val) {
-     items.push('<li><span class="score">' + Math.min(5, parseInt(val['score']*5)) + '</span> <a href="pdfs/' + val['id'] + '">' + val['id'] + '</a></li>');
+     if (val['id'].indexOf('N-') == 0) {
+       url = 'pdfs-N/' + val['id'];
+     } else {
+       url = 'pdfs/' + val['id'];
+     }
+     items.push('<li><span class="score">' + Math.min(5, parseInt(val['score']*5)) + '</span> <a href="' + url + '">' + val['id'] + '</a></li>');
   });
   $('#results').html($('<ul/>', {html: items.join('')}));
   var orig_offset = $('#offset').val()*1;
@@ -29,7 +34,6 @@ function display_results(data) {
   }
   $('#offset').val(orig_offset);
   if ($('#offset').val()*1 + $('#count').val()*1 < data['hits']*1) {
-    console.log('there');
     var next_start = $('#offset').val()*1 + NUM_RESULTS;
     $('#offset').val(next_start);
     link = '<a class="nav next" href="?' + $('form').serialize() + '">Résultats suivants</a>';
@@ -91,6 +95,7 @@ $(function() {
 
     <div id="appbar">
     <h2>Prototype de moteur de recherche - GED PFWB</h2>
+    <a href="upload.html">Ajouter un fichier</a>
     </div>
 
   <form>
diff --git a/server/search_server.py b/server/search_server.py
index 33a11af..7853dea 100755
--- a/server/search_server.py
+++ b/server/search_server.py
@@ -32,6 +32,7 @@ import json
 import cairo
 import poppler
 import hashlib
+import subprocess
 
 class SearchHandler(SCGIHandler):
     debug = False
@@ -57,14 +58,24 @@ class SearchHandler(SCGIHandler):
             print 'Request received at', time.strftime('[%Y-%m-%d %H:%M]')
             print ' - body size:', bodysize
 
-        if bodysize != 0:
-            return self.error_page(output, 'this server only supports GET')
-
         uri = env.get('REQUEST_URI')[len('/search/'):]
         if self.debug:
             print ' - uri:', uri
 
-        if uri.startswith('pdfs/'):
+        if uri == 'upload':
+            if bodysize == 0:
+                return self.redirect_home(env, output)
+            hdr = {'content-type': env.get('CONTENT_TYPE')}
+            field = cgi.FieldStorage(input, headers=hdr, environ=env).value[0]
+            filename = 'pdfs-N/N-%s' % os.path.basename(field.filename)
+            fd = file(filename, 'w')
+            fd.write(field.value)
+            fd.close()
+            if not self.index(filename=filename, id=os.path.basename(filename)):
+                 return self.error_page(output, 'failed to index')
+            return self.redirect_home(env, output)
+
+        if uri.startswith('pdfs/') or uri.startswith('pdfs-N/'):
             preview_filename = '/tmp/preview-%s.png' % hashlib.sha1(uri).hexdigest()
             if uri.endswith('.png'):
                 uri = uri[:-4]
@@ -91,17 +102,13 @@ class SearchHandler(SCGIHandler):
             return
 
         if not env.get('QUERY_STRING'):
-            # homepage
-            print >> output, 'Content-type: text/html'
-            print >> output, ''
-            output.write(file('search.html').read())
-            return
+             return self.redirect_home(env, output)
 
         qs = cgi.parse_qs(env.get('QUERY_STRING'))
         try:
             query = unicode(qs['q'][0], 'utf-8')
         except KeyError:
-            return self.error_page('missing q parameter')
+            return self.error_page(output, 'missing q parameter')
         if 'count' in qs:
             count = int(qs['count'][0])
         else:
@@ -123,6 +130,39 @@ class SearchHandler(SCGIHandler):
         print >> output, ''
         json.dump(response, output)
 
+    def redirect_home(self, env, output):
+        # go to homepage
+        root_url = 'http://%s' % env.get('SERVER_NAME')
+        if env.get('SERVER_PORT') != '80':
+            root_url = root_url + ':%s' % env.get('SERVER_PORT')
+
+        print >> output, 'Status: 302 Redirect'
+        print >> output, 'Content-type: text/plain'
+        print >> output, 'Location: ' + root_url
+        print >> output, ''
+        print >> output, 'Redirection'
+
+    def index(self, filename, id):
+        cmd = ['pdftotext', filename, '-']
+        try:
+            p = subprocess.Popen(cmd,
+                             close_fds=True,
+                             stdin=subprocess.PIPE,
+                             stdout=subprocess.PIPE,
+                             stderr=subprocess.PIPE)
+        except OSError, e:
+            if self.debug:
+                print 'E: error running pdftotext'
+            return False
+        stdout, stderr = p.communicate()
+        if p.returncode != 0:
+           if self.debug:
+               print 'E: error running pdftotext (rc:%d)' % p.returncode
+           return False
+        self.solr.add([{'id': id, 'text': unicode(stdout, 'utf-8')}])
+        self.solr.commit()
+        return True
+
     def error_page(self, output, message):
         print >> output, 'Content-type: text/plain'
         print >> output, ''
diff --git a/server/upload.html b/server/upload.html
new file mode 100644
index 0000000..1f74944
--- /dev/null
+++ b/server/upload.html
@@ -0,0 +1,38 @@
+<!DOCTYPE html>
+<html lang="fr">
+ <head>
+  <meta charset="utf-8"/>
+  <link rel="stylesheet" type="text/css" media="all" href="style.css"/>
+  <script src="js/jquery-1.7.2.min.js"></script>
+  <title>Recherche</title>
+ </head>
+ <body>
+
+  <div id="wrap">
+    <div id="header">
+      <h1>PFWB</h1>
+      <span>Prototype</span>
+    </div>
+    <div id="splash">
+    </div>
+    <div id="content">
+
+    <div id="appbar">
+    <h2>Prototype de moteur de recherche - GED PFWB</h2>
+    <a href="./">Retourner à la recherche</a>
+    </div>
+
+  <form method="post" action="search/upload" enctype="multipart/form-data">
+   Fichier à ajouter : <input name="file" type="file"/>
+   <input type="submit" value="Uploader"/>
+  </form>
+
+    <div id="footer">
+        Entr'ouvert — 2012
+    </div>
+  </div>
+
+
+ </body>
+</html>
+