content_md5 needs to compute its value against original value.

As TextDocument can store this original value in data property (Ingestion Input)
or text_content property (UI Input).
content_md5 is computed against getData, then TextDocument instances must fallback
on getTextContent if data is empty.
getData usage is unified among all types of documents.

Additional Comments:
TextDocument._convert() still rely on text_content value as raw_data.
because if data exists, text_content also, as a base conversion of data.



git-svn-id: https://svn.erp5.org/repos/public/erp5/trunk@41849 20353a03-c40f-0410-a6d1-a30d3c3de9de

product/ERP5/Document/TextDocument.py

@@ -390,3 +390,19 @@ class TextDocument(CachedConvertableMixin, BaseConvertableFileMixin,
       LOG('TextDocument', WARNING,
                 'Usage of text_format is deprecated, use text_content instead')
       return self._setContentType(value)
+
+    def getData(self, default=_MARKER):
+      """getData must returns original content but TextDocument accepts
+      data or text_content to store original content.
+      Fallback on text_content property if data is not defined
+      """
+      if not self.hasData():
+        if default is _MARKER:
+          return self.getTextContent()
+        else:
+          return self.getTextContent(default)
+      else:
+        if default is _MARKER:
+          return File.getData(self)
+        else:
+          return File.getData(self, default)

product/ERP5/mixin/cached_convertable.py

@@ -247,7 +247,7 @@ class CachedConvertableMixin:
   def updateContentMd5(self):
     """Update md5 checksum from the original file
     """
-    mime, data = self.convert(None)
+    data = self.getData()
     if data is not None:
       if isinstance(data, Pdata):
         self._setContentMd5(hashPdataObject(aq_base(data)))