ERP5OOo: py3

bt5/erp5_base/DocumentTemplateItem/portal_components/document.erp5.OOoDocument.py

@@ -30,6 +30,7 @@
 import contextlib
 import re, zipfile
 from io import BytesIO
+import six
 from warnings import warn
 from AccessControl import ClassSecurityInfo
 from OFS.Image import Pdata
@@ -39,8 +40,8 @@ from Products.ERP5Type import Permissions, PropertySheet
 from Products.ERP5Type.Cache import CachingMethod
 from erp5.component.document.File import File
 from erp5.component.document.Document import Document, \
-       VALID_IMAGE_FORMAT_LIST, ConversionError, NotConvertedError
-from Products.ERP5Type.Utils import bytes2str, fill_args_from_request, str2bytes
+       VALID_IMAGE_FORMAT_LIST, VALID_TEXT_FORMAT_LIST, ConversionError, NotConvertedError
+from Products.ERP5Type.Utils import guessEncodingFromText, bytes2str, fill_args_from_request, str2bytes
 
 # Mixin Import
 from erp5.component.mixin.BaseConvertableFileMixin import BaseConvertableFileMixin
@@ -336,7 +337,14 @@ class OOoDocument(OOoDocumentExtensibleTraversableMixin, BaseConvertableFileMixi
         # store conversion
         self.setConversion(data, mime, format=original_format, **kw)
 
-    return self.getConversion(format=original_format, **kw)
+    mime, data = self.getConversion(format=original_format, **kw)
+    if format in VALID_TEXT_FORMAT_LIST:
+      # Libreoffice conversions on cloudooo usually have a BOM, we are using guessEncodingFromText
+      # here mostly as a convenient way to decode with the encoding from BOM
+      data = data.decode(guessEncodingFromText(data) or 'ascii')
+      if six.PY2:
+        data = data.encode('utf-8')
+    return mime, data
 
   security.declareProtected(Permissions.ModifyPortalContent,
                             '_populateConversionCacheWithHTML')

product/ERP5OOo/transforms/odt_to_doc.py

@@ -27,7 +27,7 @@ class OdtToDoc:
     raise AttributeError(attr)
 
   def convert(self, orig, data, cache=None, filename=None, context=None, **kwargs):
-    data = str(orig)
+    data = bytes(orig)
     doc = OOOdCommandTransform(context, filename, data, self.inputs[0])
     msword = doc.convertTo('doc')
     if cache is not None:

product/ERP5OOo/transforms/odt_to_pdf.py

@@ -27,7 +27,7 @@ class OdtToPdf:
     raise AttributeError(attr)
 
   def convert(self, orig, data, cache=None, filename=None, context=None, **kwargs):
-    data = str(orig)
+    data = bytes(orig)
     doc = OOOdCommandTransform(context, filename, data, self.inputs[0])
     pdf = doc.convertTo('pdf')
     if cache is not None:

product/ERP5OOo/transforms/odt_to_xml.py

@@ -28,7 +28,7 @@ class OdtToXml:
     raise AttributeError(attr)
 
   def convert(self, orig, data, cache=None, filename=None, context=None, **kwargs):
-    data = str(orig)
+    data = bytes(orig)
     doc = OOOdCommandTransform(context, filename, data, self.inputs[0])
     builder = OOoBuilder(doc)
     content = builder.extract('content.xml')

product/ERP5OOo/transforms/oood_commandtransform.py

@@ -143,7 +143,7 @@ class OOOdCommandTransform(commandtransform):
             image = OFSImage(image.getId(), image.getTitle(), image_data)
 
           # image should be OFSImage
-          data = str(image.data)
+          data = bytes(image.data)
           width = image.width
           height = image.height
           if height: