From 7dfd08c0866a2e5cf1b765e0ec1546447380fb4e Mon Sep 17 00:00:00 2001
From: Jared Smith <jared.smith.jrod@gmail.com>
Date: Wed, 17 Aug 2016 15:04:01 -0700
Subject: [PATCH] Storing the size and mimetypes from uploads

size is stored as bytes

there's a bit of redundancy with getting the mimetypes of files, by
default, mimetype is determined by the url, fallback is done by using
`python-magic` module
---
 ckan/lib/uploader.py        | 24 ++++++++++++++++++++++++
 ckan/logic/action/create.py | 12 ++++++++----
 ckan/logic/action/update.py |  6 +++++-
 requirements.in             |  1 +
 4 files changed, 38 insertions(+), 5 deletions(-)

diff --git a/ckan/lib/uploader.py b/ckan/lib/uploader.py
index d65d15a3888..7465b280f49 100644
--- a/ckan/lib/uploader.py
+++ b/ckan/lib/uploader.py
@@ -4,6 +4,7 @@
 import cgi
 import datetime
 import logging
+import magic
 
 import ckan.lib.munge as munge
 import ckan.logic as logic
@@ -96,6 +97,8 @@ def __init__(self, object_type, old_filename=None):
         self.storage_path = None
         self.filename = None
         self.filepath = None
+        self.filesize = 0 # bytes
+        self.mimetype = None
         path = get_storage_path()
         if not path:
             return
@@ -166,10 +169,18 @@ def upload(self, max_size=2):
                     raise logic.ValidationError(
                         {self.file_field: ['File upload too large']}
                     )
+            output_file.seek(0, os.SEEK_END)
+            self.filesize = output_file.tell()
             output_file.close()
             os.rename(self.tmp_filepath, self.filepath)
             self.clear = True
 
+            try:
+                self.mimetype = magic.from_file(filepath, mime=True)
+            except:
+                # Not that important if call above fails
+                self.mimetype = None
+
         if (self.clear and self.old_filename
                 and not self.old_filename.startswith('http')):
             try:
@@ -192,6 +203,8 @@ def __init__(self, resource):
             if e.errno != 17:
                 raise
         self.filename = None
+        self.filesize = 0 # bytes
+        self.mimetype = None
 
         url = resource.get('url')
         upload_field_storage = resource.pop('upload', None)
@@ -254,6 +267,7 @@ def upload(self, id, max_size=10):
                 current_size = current_size + 1
                 # MB chunks
                 data = self.upload_file.read(2 ** 20)
+
                 if not data:
                     break
                 output_file.write(data)
@@ -262,8 +276,18 @@ def upload(self, id, max_size=10):
                     raise logic.ValidationError(
                         {'upload': ['File upload too large']}
                     )
+
+            output_file.seek(0, os.SEEK_END)
+            self.filesize = output_file.tell()
             output_file.close()
             os.rename(tmp_filepath, filepath)
+
+            try:
+                self.mimetype = magic.from_file(filepath, mime=True)
+            except:
+                # Not that important if call above fails
+                self.mimetype = None
+
             return
 
         # The resource form only sets self.clear (via the input clear_upload)
diff --git a/ckan/logic/action/create.py b/ckan/logic/action/create.py
index 799f1064430..23ac48a59e9 100644
--- a/ckan/logic/action/create.py
+++ b/ckan/logic/action/create.py
@@ -3,6 +3,7 @@
 '''API functions for adding data to CKAN.'''
 
 import logging
+import mimetypes
 import random
 import re
 from socket import error as socket_error
@@ -295,6 +296,13 @@ def resource_create(context, data_dict):
         pkg_dict['resources'] = []
 
     upload = uploader.get_resource_uploader(data_dict)
+    ## Get out resource_id resource from model as it will not appear in
+    ## package_show until after commit
+    upload.upload(context['package'].resources[-1].id,
+                  uploader.get_max_resource_size())
+
+    data_dict['mimetype'] = mimetypes.guess_type(data_dict['url'])[0] or upload.mimetype
+    data_dict['size'] = upload.filesize
 
     pkg_dict['resources'].append(data_dict)
 
@@ -307,10 +315,6 @@ def resource_create(context, data_dict):
         errors = e.error_dict['resources'][-1]
         raise ValidationError(errors)
 
-    ## Get out resource_id resource from model as it will not appear in
-    ## package_show until after commit
-    upload.upload(context['package'].resources[-1].id,
-                  uploader.get_max_resource_size())
     model.repo.commit()
 
     ##  Run package show again to get out actual last_resource
diff --git a/ckan/logic/action/update.py b/ckan/logic/action/update.py
index 667021cb14e..c8193740c9f 100644
--- a/ckan/logic/action/update.py
+++ b/ckan/logic/action/update.py
@@ -6,6 +6,7 @@
 import datetime
 import time
 import json
+import mimetypes
 
 from ckan.common import config
 import paste.deploy.converters as converters
@@ -93,6 +94,10 @@ def resource_update(context, data_dict):
         plugin.before_update(context, pkg_dict['resources'][n], data_dict)
 
     upload = uploader.get_resource_uploader(data_dict)
+    upload.upload(id, uploader.get_max_resource_size())
+
+    data_dict['mimetype'] = mimetypes.guess_type(data_dict['url'])[0] or upload.mimetype
+    data_dict['size'] = upload.filesize
 
     pkg_dict['resources'][n] = data_dict
 
@@ -105,7 +110,6 @@ def resource_update(context, data_dict):
         errors = e.error_dict['resources'][n]
         raise ValidationError(errors)
 
-    upload.upload(id, uploader.get_max_resource_size())
     model.repo.commit()
 
     resource = _get_action('resource_show')(context, {'id': id})
diff --git a/requirements.in b/requirements.in
index fed7a94fda3..c97b96ffaed 100644
--- a/requirements.in
+++ b/requirements.in
@@ -13,6 +13,7 @@ Pairtree==0.7.1-T
 passlib==1.6.5
 paste==1.7.5.1
 psycopg2==2.4.5
+python-magic>=0.4.6
 pysolr==3.5.0
 Pylons==0.9.7
 python-dateutil>=1.5.0,<2.0.0