Implement ProcessingState class and use for images

The idea is to have a class that has the knowledge of the
currently being processed media and also has tools for
that.

The long term idea is to make reprocessing easier by for
example hiding the way the original comes into the
processing code.

mediagoblin/media_types/image/processing.py

@@ -19,7 +19,6 @@ import os
 import logging
 
 from mediagoblin import mg_globals as mgg
-from mediagoblin.decorators import get_workbench
 from mediagoblin.processing import BadMediaFail, \
     create_pub_filepath, FilenameBuilder
 from mediagoblin.tools.exif import exif_fix_image_orientation, \
@@ -95,21 +94,21 @@ def sniff_handler(media_file, **kw):
     return False
 
 
-@get_workbench
+def process_image(entry):
-def process_image(entry, workbench=None):
     """Code to process an image. Will be run by celery.
 
     A Workbench() represents a local tempory dir. It is automatically
     cleaned up when this function exits.
     """
+    proc_state = entry.proc_state
+    workbench = proc_state.workbench
+
     # Conversions subdirectory to avoid collisions
     conversions_subdir = os.path.join(
         workbench.dir, 'conversions')
     os.mkdir(conversions_subdir)
-    queued_filepath = entry.queued_media_file
+
-    queued_filename = workbench.localized_file(
+    queued_filename = proc_state.get_queued_filename()
-        mgg.queue_store, queued_filepath,
-        'source')
     name_builder = FilenameBuilder(queued_filename)
 
     # EXIF extraction
@@ -147,8 +146,7 @@ def process_image(entry, workbench=None):
     mgg.public_store.copy_local_to_storage(queued_filename, original_filepath)
 
     # Remove queued media file from storage and database
-    mgg.queue_store.delete_file(queued_filepath)
+    proc_state.delete_queue_file()
-    entry.queued_media_file = []
 
     # Insert media file information into database
     media_files_dict = entry.setdefault('media_files', {})

mediagoblin/processing/__init__.py

@@ -74,6 +74,37 @@ class FilenameBuilder(object):
                              ext=self.ext)
 
 
+class ProcessingState(object):
+    def __init__(self, entry):
+        self.entry = entry
+        self.workbench = None
+        self.queued_filename = None
+
+        # Monkey patch us onto the entry
+        entry.proc_state = self
+
+    def set_workbench(self, wb):
+        self.workbench = wb
+
+    def get_queued_filename(self):
+        """
+        Get the a filename for the original, on local storage
+        """
+        if self.queued_filename is not None:
+            return self.queued_filename
+        queued_filepath = self.entry.queued_media_file
+        queued_filename = self.workbench.localized_file(
+            mgg.queue_store, queued_filepath,
+            'source')
+        self.queued_filename = queued_filename
+        return queued_filename
+
+    def delete_queue_file(self):
+        queued_filepath = self.entry.queued_media_file
+        mgg.queue_store.delete_file(queued_filepath)
+        self.entry.queued_media_file = []
+
+
 def mark_entry_failed(entry_id, exc):
     """
     Mark a media entry as having failed in its conversion.

mediagoblin/processing/task.py

@@ -22,7 +22,7 @@ from celery import registry, task
 
 from mediagoblin import mg_globals as mgg
 from mediagoblin.db.models import MediaEntry
-from mediagoblin.processing import mark_entry_failed, BaseProcessingFail
+from . import mark_entry_failed, BaseProcessingFail, ProcessingState
 from mediagoblin.tools.processing import json_processing_callback
 
 _log = logging.getLogger(__name__)
@@ -85,8 +85,11 @@ class ProcessMedia(task.Task):
 
             _log.debug('Processing {0}'.format(entry))
 
-            # run the processing code
+            proc_state = ProcessingState(entry)
-            entry.media_manager['processor'](entry)
+            with mgg.workbench_manager.create() as workbench:
+                proc_state.set_workbench(workbench)
+                # run the processing code
+                entry.media_manager['processor'](entry)
 
             # We set the state to processed and save the entry here so there's
             # no need to save at the end of the processing stage, probably ;)