Encode download url, fix tests and style updates.

Author: cslzchen

mfr/core/exceptions.py

@@ -147,9 +147,8 @@ def __init__(self, message, *args, metadata_url: str='', response: str='', **kwa
 
 
 class QueryParameterError(ProviderError):
-    """The MFR related errors raised from a :class:`mfr.core.provider` and relating to query parameters
-    should inherit from MetadataError
-    This error is thrown when a query parameter is used missused
+    """The MFR related errors raised from a :class:`mfr.core.provider`and relating to query
+    parameters. This error is thrown when the query has an invalid value.
     """
 
     __TYPE = 'query_parameter'

mfr/core/provider.py

@@ -49,6 +49,14 @@ def download(self):
 class ProviderMetadata:
 
     def __init__(self, name, ext, content_type, unique_key, download_url, is_public=False):
+        """
+        :param name: the file name
+        :param ext: the file extension
+        :param content_type: the file type
+        :param unique_key:
+        :param download_url:
+        :param is_public:
+        """
         self.name = name
         self.ext = ext
         self.content_type = content_type

mfr/core/utils.py

@@ -78,25 +78,27 @@ def make_renderer(name, metadata, file_path, url, assets_url, export_url):
     normalized_name = (name and name.lower()) or 'none'
     if metadata.is_public:
         try:
+            # Use the public renderer if exist
             return driver.DriverManager(
                 namespace='mfr.public_renderers',
                 name=normalized_name,
                 invoke_on_load=True,
                 invoke_args=(metadata, file_path, url, assets_url, export_url),
             ).driver
         except:
-            # Check for a public renderer, if one doesn't exist, use a regular one
-            # Real exceptions handled by main driver.DriverManager
+            # If public render does not exist, use default renderer by MFR
+            # If public render exists but exceptions occurs, delay the exception handling
             pass
 
     try:
+        # Use the default MFR handler
         return driver.DriverManager(
             namespace='mfr.renderers',
             name=normalized_name,
             invoke_on_load=True,
             invoke_args=(metadata, file_path, url, assets_url, export_url),
         ).driver
-    except RuntimeError:
+    except:
         raise exceptions.MakeRendererError(
             namespace='mfr.renderers',
             name=normalized_name,
@@ -110,6 +112,7 @@ def make_renderer(name, metadata, file_path, url, assets_url, export_url):
             }
         )
 
+
 def sizeof_fmt(num, suffix='B'):
     if abs(num) < 1000:
         return '%3.0f%s' % (num, suffix)

mfr/extensions/office365/render.py

@@ -1,19 +1,25 @@
 import os
+from urllib import parse
+
 import furl
+from mako.lookup import TemplateLookup
 
 from mfr.core import extension
-from mako.lookup import TemplateLookup
 from mfr.extensions.office365 import settings
 
+from logging import getLogger
+logger = getLogger(__name__)
+
 
 class Office365Renderer(extension.BaseRenderer):
-    """A renderer for use with public .docx files.
+    """A renderer for .docx files that are publicly available.
+
+    Office online can render `.docx` files to `.pdf` for us. This renderer will only be made
+    if a query param with `public_file=true` is present. It then generates and embeds an
+    office online URL into an `iframe` and returns the template. The file it is trying to
+    render MUST be public.
 
-    Office online can render .docx files to pdf for us.
-    This renderer will only ever be made if a query param with `public_file=1` is sent.
-    It then generates and embeds an office online url into an
-    iframe and returns the template. The file it is trying to render MUST
-    be available publically online. This renderer will not work if testing locally.
+    Note: this renderer DOES NOT work locally.
 
     """
 
@@ -23,9 +29,12 @@ class Office365Renderer(extension.BaseRenderer):
         ]).get_template('viewer.mako')
 
     def render(self):
-        download_url = furl.furl(self.metadata.download_url).set(query='')
-        url = settings.OFFICE_BASE_URL + download_url.url
-        return self.TEMPLATE.render(base=self.assets_url, url=url)
+        download_url = furl.furl(self.metadata.download_url).set(query='').url
+        logger.info('download_url = ' + download_url)
+        encoded_download_url = parse.quote(download_url)
+        office_render_url = settings.OFFICE_BASE_URL + encoded_download_url
+        logger.info('office_render_url = ' + office_render_url)
+        return self.TEMPLATE.render(base=self.assets_url, url=office_render_url)
 
     @property
     def file_required(self):

mfr/providers/osf/provider.py

@@ -1,22 +1,21 @@
-import os
 import json
 import hashlib
 import logging
-from urllib.parse import urlparse
 import mimetypes
+import os
+from urllib.parse import urlparse
 
 import furl
 import aiohttp
 from aiohttp.errors import ContentEncodingError
 
 from waterbutler.core import streams
 
-from mfr.core import exceptions
-from mfr.core import provider
+from mfr.core import exceptions, provider
+from mfr.core.exceptions import TooBigToRenderError
 from mfr.core.utils import sizeof_fmt
 from mfr.providers.osf import settings
 from mfr.settings import MAX_FILE_SIZE_TO_RENDER
-from mfr.core.exceptions import TooBigToRenderError
 
 logger = logging.getLogger(__name__)
 
@@ -121,18 +120,16 @@ async def metadata(self):
         unique_key = hashlib.sha256((metadata['data']['etag'] + cleaned_url.url).encode('utf-8')).hexdigest()
 
         is_public = False
-
-        if 'public_file' in cleaned_url.args:
-            if cleaned_url.args['public_file'] not in ['0', '1']:
+        public_file = cleaned_url.args.get('public_file', None)
+        if public_file:
+            if public_file not in ['true', 'false']:
                 raise exceptions.QueryParameterError(
-                    'The `public_file` query paramter should either `0`, `1`, or unused. Instead '
-                    'got  {}'.format(cleaned_url.args['public_file']),
+                    'Invalid value for query parameter `public_file`: {}'.format(cleaned_url.args['public_file']),
                     url=download_url,
                     provider=self.NAME,
                     code=400,
                 )
-
-            is_public = cleaned_url.args['public_file'] == '1'
+            is_public = public_file == 'true'
 
         return provider.ProviderMetadata(name, ext, content_type,
                     unique_key, download_url, is_public=is_public)

tests/extensions/office365/test_renderer.py

@@ -1,3 +1,5 @@
+from urllib import parse
+
 import furl
 import pytest
 
@@ -20,12 +22,7 @@ def file_path():
 
 @pytest.fixture
 def url():
-    return 'http://osf.io/file/test.pdf'
-
-
-@pytest.fixture
-def assets_url():
-    return 'http://mfr.osf.io/assets'
+    return parse.quote('http://osf.io/file/test.pdf')
 
 
 @pytest.fixture
@@ -40,7 +37,7 @@ def renderer(metadata, file_path, url, assets_url, export_url):
 
 class TestOffice365Renderer:
 
-    def test_render_pdf(self, renderer, metadata, assets_url):
+    def test_render_pdf(self, renderer, metadata):
         download_url = furl.furl(metadata.download_url).set(query='')
         body_url = settings.OFFICE_BASE_URL + download_url.url
         body = renderer.render()