From dcd2d2bf4a9b3dd8a053c943f221e2d7e9e8f53e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fr=C3=A9d=C3=A9ric=20P=C3=A9ters?= <fpeters@entrouvert.com>
Date: Thu, 1 Dec 2011 11:01:34 +0100
Subject: [PATCH] index document contents

---
 tabellio/documents/configure.zcml             |  8 +++
 tabellio/documents/indexer.py                 | 51 +++++++++++++++++--
 .../types/tabellio.documents.document.xml     |  2 +-
 3 files changed, 57 insertions(+), 4 deletions(-)

diff --git a/tabellio/documents/configure.zcml b/tabellio/documents/configure.zcml
index c5bf900..3315f43 100644
--- a/tabellio/documents/configure.zcml
+++ b/tabellio/documents/configure.zcml
@@ -12,6 +12,14 @@
 
   <i18n:registerTranslations directory="locales" />
 
+
+  <plone:behavior
+      title="Document SearchableText indexer behavior"
+      description="Enables the dynamic SearchableText indexer for Documents"
+      provides="tabellio.documents.indexer.IDocumentIndexer"
+      for="plone.dexterity.interfaces.IDexterityContent"
+      />
+
   <plone:behavior
       title="Authors SearchableText indexer behavior"
       description="Enables the dynamic SearchableText indexer for Authors fields"
diff --git a/tabellio/documents/indexer.py b/tabellio/documents/indexer.py
index 6ccc9ca..5f12b27 100644
--- a/tabellio/documents/indexer.py
+++ b/tabellio/documents/indexer.py
@@ -1,5 +1,7 @@
 from zope.interface import Interface
+from ZODB.POSException import ConflictError
 from five import grok
+from Products.CMFCore.utils import getToolByName
 from plone.dexterity.utils import iterSchemata
 from plone.indexer import indexer
 from plone.z3cform import z2
@@ -84,6 +86,11 @@ grok.global_adapter(dossierInterveningPersonsIndexer, name="interveningPersonsDo
 grok.global_adapter(questionInterveningPersonsIndexer, name="interveningPersonsDoc")
 
 
+class IDocumentIndexer(Interface):
+    """Dexterity behavior interface for enabling the dynamic SearchableText
+       indexer on Document objecgs."""
+
+
 class IAuthorsIndexer(Interface):
     """Dexterity behavior interface for enabling the dynamic SearchableText
        indexer on Authors fields."""
@@ -98,8 +105,7 @@ class FakeView(object):
         self.request = request
 
 
-@indexer(IAuthorsIndexer)
-def dynamic_searchable_text_indexer(obj):
+def raw_author_dynamic_searchable_text_indexer(obj):
     """Dynamic searchable text indexer.
     """
 
@@ -136,10 +142,49 @@ def dynamic_searchable_text_indexer(obj):
     return ' '.join(indexed)
 
 
-grok.global_adapter(dynamic_searchable_text_indexer,
+@indexer(IAuthorsIndexer)
+def author_dynamic_searchable_text_indexer(obj):
+    return raw_author_dynamic_searchable_text_indexer(obj)
+
+
+grok.global_adapter(author_dynamic_searchable_text_indexer,
                     name='SearchableText')
 
 
+@indexer(IDocumentIndexer)
+def document_dynamic_searchable_text_indexer(obj):
+    """Dynamic searchable text indexer.
+    """
+
+    title_and_authors = raw_author_dynamic_searchable_text_indexer(obj)
+
+    data = obj.file
+    if not data or data.getSize() == 0:
+        return title_and_authors
+
+    # if there is no path to text/plain, do nothing
+    transforms = getToolByName(obj, 'portal_transforms')
+    if not transforms._findPath(data.contentType, 'text/plain'):
+        return title_and_authors
+
+    # convert it to text/plain
+    print 'indexing pdf'
+    try:
+        datastream = transforms.convertTo(
+            'text/plain', data.data, mimetype=data.contentType,
+            filename=data.filename)
+        return title_and_authors + ' ' + datastream.getData()
+    except (ConflictError, KeyboardInterrupt):
+        raise
+
+    return title_and_authors
+
+
+grok.global_adapter(document_dynamic_searchable_text_indexer,
+                    name='SearchableText')
+
+
+
 def get_field_widget(obj, field):
     """Returns the field widget of a field in display mode without
     touching any form.
diff --git a/tabellio/documents/profiles/default/types/tabellio.documents.document.xml b/tabellio/documents/profiles/default/types/tabellio.documents.document.xml
index 73cf705..dcd0239 100644
--- a/tabellio/documents/profiles/default/types/tabellio.documents.document.xml
+++ b/tabellio/documents/profiles/default/types/tabellio.documents.document.xml
@@ -23,7 +23,7 @@
  <!-- enabled behaviors -->
  <property name="behaviors">
    <element value="plone.app.content.interfaces.INameFromTitle" />
-   <element value="tabellio.documents.indexer.IAuthorsIndexer" />
+   <element value="tabellio.documents.indexer.IDocumentIndexer" />
  </property>
 
  <!-- View information -->