restructured the app based on the Flask tutorial, cleaned up unused code

oduwsdl · May 28, 2018 · 443e2634c1c4d2277032009e337f173b209022bb · 443e263
1 parent 7592549
commit 443e2634c1c4d2277032009e337f173b209022bb
diff --git a/Dockerfile b/Dockerfile
@@ -1,5 +1,9 @@
 FROM python:3.6.4-stretch

+# the flask documentation recommends WSGI server waitress for use
+RUN pip install waitress
+
+# TODO: publish archiveit_utilities so that we don't need to do this
 RUN git clone https://github.com/shawnmjones/archiveit_utilities.git

 RUN cd archiveit_utilities && pip install .
@@ -16,4 +20,5 @@ ENV FLASK_APP=mementoembed/mementoembed.py

 CMD ["flask", "run", "--host", "0.0.0.0", "--port", "5550"]

-# CMD ["/bin/bash"]
+# TODO: actually use waitress
+# CMD ["waitress-serve", "--call", "mementoembed:create_app"]
diff --git a/bin/fetch_surrogate_data b/bin/fetch_surrogate_data
@@ -4,7 +4,7 @@ import requests
 import errno
 import json

-from mementoembed import Surrogate, identify_archive, identify_collection, get_archive_favicon
+from mementoembed import MementoSurrogate
 from archivenow import archivenow

 if __name__ == '__main__':
@@ -53,18 +53,22 @@ if __name__ == '__main__':

    logger.info("creating surrogate information")

-    s = Surrogate(urim, resp.text, resp.headers)
+    s = MementoSurrogate(urim)

    output = {
        "URI-M": urim,
+        "Memento-Datetime": s.memento_datetime,
        "Text Snippet": s.text_snippet,
        "Striking Image": s.striking_image,
        "Title": s.title,
-        "Member of Archive": identify_archive(urim),
-        "Archive Collection": identify_collection(urim),
-        "Archive Favicon": get_archive_favicon(urim),
-        "Memento-Datetime": mdt
-        # "Site Favicon":
+        "Member of Archive": s.archive_name,
+        "Archive URI": s.archive_uri,
+        "Archive Collection": s.collection_name,
+        "Archive Collection URI": s.archive_collection_uri,
+        "Archive Favicon": s.archive_favicon,
+        "Original Favicon": s.original_favicon,
+        "Original Link Status": s.original_link_status,
+        "Generated Datetime": s.surrogate_creation_time
    }

    print(json.dumps(output))
diff --git a/mementoembed/__init__.py b/mementoembed/__init__.py
@@ -1,13 +1,125 @@
+import os
+import json
+
+import htmlmin
+import dicttoxml
+import requests
+
+from flask import Flask, request, render_template, make_response
+from cachecontrol import CacheControl
+from cachecontrol.caches.file_cache import FileCache
+
 from .mementosurrogate import MementoSurrogate, NotMementoException, \
    MementoConnectionTimeout, MementoImageConnectionError, \
    MementoImageConnectionTimeout
-from .archiveinfo import identify_archive, identify_collection, \
-    get_archive_favicon, archive_names, get_collection_uri, \
-    get_archive_uri

 __all__ = [
-    "MementoSurrogate", "identify_archive", 
-    "identify_collection", "get_archive_favicon",
-    "archive_names", "get_collection_uri",
-    "get_archive_uri"
-    ]
+    "MementoSurrogate", "NotMementoException", "MementoConnectionTimeout",
+    "MementoImageConnectionError", "MementoImageConnectionTimeout"
+    ]
+
+user_agent_string = "ODU WS-DL Researcher Shawn M. Jones <sjone@cs.odu.edu>"
+
+def create_app():
+
+    app = Flask(__name__, instance_relative_config=True)
+
+    # pylint: disable=no-member
+    app.logger.info("loading Flask app for {}".format(app.name))
+
+    #pylint: disable=unused-variable
+    @app.route('/', methods=['GET', 'HEAD'])
+    def front_page():
+        return render_template('index.html')
+
+    @app.route('/services/oembed', methods=['GET', 'HEAD'])
+    def oembed_endpoint():
+
+        urim = request.args.get("url")
+        responseformat = request.args.get("format")
+
+        # JSON is the default
+        if responseformat == None:
+            responseformat = "json"
+
+        if responseformat != "json":
+            if responseformat != "xml":
+                return "The provider cannot return a response in the requested format.", 501
+
+        app.logger.debug("received url {}".format(urim))
+        app.logger.debug("format: {}".format(responseformat))
+
+        if os.environ.get('FLASK_ENV'):
+            if 'development' in os.environ.get('FLASK_ENV').lower():
+                session = CacheControl(requests.session(), cache=FileCache('.web_cache', forever=True))
+            else:
+                session = CacheControl(requests.session(), cache=FileCache('.web_cache'))    
+        else:
+            session = CacheControl(requests.session(), cache=FileCache('.web_cache'))
+
+        try:
+            
+            s = MementoSurrogate(
+                urim=urim,
+                session=session,
+                logger=app.logger
+            )
+
+            output = {}
+
+            output["type"] = "rich"
+            output["version"] = "1.0"
+
+            output["url"] = urim
+            output["provider_name"] = s.archive_name
+            output["provider_uri"] = s.archive_uri
+
+            urlroot = request.url_root
+            urlroot = urlroot if urlroot[-1] != '/' else urlroot[0:-1]
+
+            app.logger.info("generating oEmbed output for {}".format(urim))
+            output["html"] = htmlmin.minify( render_template(
+                "social_card.html",
+                urim = urim,
+                urir = s.original_uri,
+                image = s.striking_image,
+                archive_uri = s.archive_uri,
+                archive_favicon = s.archive_favicon,
+                archive_collection_id = s.collection_id,
+                archive_collection_uri = s.collection_uri,
+                archive_collection_name = s.collection_name,
+                archive_name = s.archive_name,
+                original_favicon = s.original_favicon,
+                original_domain = s.original_domain,
+                original_link_status = s.original_link_status,
+                surrogate_creation_time = s.creation_time,
+                memento_datetime = s.memento_datetime,
+                me_title = s.title,
+                me_snippet = s.text_snippet,
+                server_domain = urlroot
+            ), remove_empty_space=True, 
+            remove_optional_attribute_quotes=False )
+
+            output["width"] = 500
+
+            #TODO: fix this to the correct height!
+            output["height"] = 200
+
+            if responseformat == "json":
+                response = make_response(json.dumps(output, indent=4))
+                response.headers['Content-Type'] = 'application/json'
+            else:
+                response = make_response( dicttoxml.dicttoxml(output, custom_root='oembed') )
+                response.headers['Content-Type'] = 'text/xml'
+
+            app.logger.info("returning output as application/json...")
+
+        except NotMementoException:
+            return render_template(
+                'make_your_own_memento.html',
+                urim = urim
+                ), 400
+
+        return response
+
+    return app
diff --git a/mementoembed/archiveinfo.py b/mementoembed/archiveinfo.py