SFR-1877: rewrite URLs as fulfillment URLs

CHANGELOG.md

@@ -1,10 +1,9 @@
 # CHANGELOG
 ## unreleased version -- v0.12.4
 ## Added
-New /fulfill endpoint with ability to check for NYPL login in Bearer authorization header
-Fulfill endpoint returns pre-signed URLs for objects in private buckets when user is logged in
-Change default development port to 5050 due to macOS Monterey and higher occupying port 5000 by default
-
+- New /fulfill endpoint with ability to check for NYPL login in Bearer authorization header
+- Fulfill endpoint returns pre-signed URLs for objects in private buckets when user is logged in
+- Change default development port to 5050 due to macOS Monterey and higher occupying port 5000 by default
 
 ## unreleased version -- v0.12.4
 ## Added

api/blueprints/drbEdition.py

@@ -27,13 +27,13 @@ def editionFetch(editionID):
 
     filteredFormats = APIUtils.formatFilters(terms)
 
-    edition = dbClient.fetchSingleEdition(editionID) 
+    edition = dbClient.fetchSingleEdition(editionID)
     if edition:
         statusCode = 200
         records = dbClient.fetchRecordsByUUID(edition.dcdw_uuids)
 
         responseBody = APIUtils.formatEditionOutput(
-            edition, records=records, dbClient=dbClient, showAll=showAll, formats=filteredFormats, reader=readerVersion
+            edition, request=request, records=records, dbClient=dbClient, showAll=showAll, formats=filteredFormats, reader=readerVersion
         )
     else:
         statusCode = 404

api/blueprints/drbLink.py

@@ -1,4 +1,4 @@
-from flask import Blueprint, current_app
+from flask import Blueprint, current_app, request
 from ..db import DBClient
 from ..utils import APIUtils
 from logger import createLog
@@ -18,7 +18,7 @@ def linkFetch(linkID):
 
     if link:
         statusCode = 200
-        responseObject = APIUtils.formatLinkOutput(link)
+        responseObject = APIUtils.formatLinkOutput(link, request=request)
     else:
         statusCode = 404
         responseObject = {'message': 'Unable to locate link #{}'.format(linkID)}

api/blueprints/drbSearch.py

@@ -77,7 +77,7 @@ def standardQuery():
     dataBlock = {
         'totalWorks': searchResult.hits.total.value,
         'works': APIUtils.formatWorkOutput(
-            works, results, dbClient=dbClient, formats=filteredFormats, reader=readerVersion
+            works, results, request=request, dbClient=dbClient, formats=filteredFormats, reader=readerVersion
         ),
         'paging': paging,
         'facets': facets

api/blueprints/drbWork.py

@@ -20,7 +20,7 @@ def workFetch(uuid):
     terms = {}
     for param in ['filter']:
         terms[param] = APIUtils.extractParamPairs(param, searchParams)
-    
+
     showAll = searchParams.get('showAll', ['true'])[0].lower() != 'false'
     readerVersion = searchParams.get('readerVersion', [None])[0]\
         or current_app.config['READER_VERSION']
@@ -30,8 +30,10 @@ def workFetch(uuid):
     work = dbClient.fetchSingleWork(uuid)
     if work:
         statusCode = 200
-        responseBody = APIUtils.formatWorkOutput(work, None, showAll=showAll, dbClient=dbClient, formats=filteredFormats, reader=readerVersion)
-
+        responseBody = APIUtils.formatWorkOutput(work, None, showAll=showAll,
+            request=request, dbClient=dbClient, formats=filteredFormats, reader=readerVersion
+        )
+
     else:
         statusCode = 404
         responseBody = {

api/utils.py

@@ -2,6 +2,7 @@
 from datetime import datetime
 from hashlib import scrypt
 from flask import jsonify
+from itertools import repeat
 from math import ceil
 from model import Collection, Edition
 import re
@@ -140,7 +141,7 @@ def formatPagingOptions(page, pageSize, totalHits):
 
     @classmethod
     def formatWorkOutput(
-        cls, works, identifiers, dbClient, showAll=True, formats=None, reader=None
+        cls, works, identifiers, dbClient, request, showAll=True, formats=None, reader=None,
     ):
         #Multiple formatted works with formats specified
         if isinstance(works, list):
@@ -233,16 +234,25 @@ def addWorkMeta(cls, work, **kwargs):
 
     @classmethod
     def formatEditionOutput(
-        cls, edition, records=None, dbClient=None, showAll=False, formats=None, reader=None
+        cls, edition, request, records=None, dbClient=None, showAll=False, formats=None, reader=None
     ):
         editionWorkTitle = edition.work.title
         editionWorkAuthors = edition.work.authors
         editionInCollection = cls.checkEditionInCollection(None, edition, dbClient)
 
-        return cls.formatEdition(
+        formattedEdition = cls.formatEdition(
             edition, editionWorkTitle, editionWorkAuthors, editionInCollection, records, formats, showAll=showAll, reader=reader
         )
 
+        if formattedEdition.get("instances"):
+            for instance in formattedEdition['instances']:
+                for item in instance['items']:
+                    # Map over item links and patch with pre-signed URL where necessary
+                    item['links']= list(map(APIUtils.replacePrivateLinkUrl, item['links'], repeat(request)))
+
+        return formattedEdition
+
+
     @classmethod
     def checkEditionInCollection(cls, work, edition, dbClient):
 
@@ -430,7 +440,7 @@ def formatRecord(cls, record, itemsByLink):
         return outRecord
 
     @classmethod
-    def formatLinkOutput(cls, link):
+    def formatLinkOutput(cls, link, request):
         linkItem = dict(link.items[0])
         linkItem['item_id'] = link.items[0].id
 
@@ -447,6 +457,9 @@ def formatLinkOutput(cls, link):
         linkDict['work']['editions'] = [linkEdition]
         linkDict['work']['editions'][0]['items'] = [linkItem]
 
+        # Amend link to include /fulfill link if appropriate
+        linkDict = APIUtils.replacePrivateLinkUrl(linkDict, request)
+
         return linkDict
 
     @classmethod
@@ -545,7 +558,8 @@ def generate_presigned_url(s3_client, client_method, method_parameters, expires_
     def getPresignedUrlFromObjectUrl(s3Client, url):
         """
         Given the URL of an S3 resource, generate a presigned Amazon S3 URL
-        that can be used to access that resource.
+        that can be used to access that resource. This function assumes S3
+        URLs in the "virtual hosted bucket" style, not deprecated path style.
 
         :param s3_client: A Boto3 Amazon S3 client
         :param url: The URL of the desired resource
@@ -571,3 +585,15 @@ def getPresignedUrlFromObjectUrl(s3Client, url):
             {'Bucket': bucketName,'Key': objectKey},
             timeValid
         )
+
+    @staticmethod
+    def replacePrivateLinkUrl(link, request):
+        """
+        Given a link object, return a link object with the url replaced if
+        the link has flags indicating it should be fulfilled via /fulfill
+        """
+        if link['flags'].get("edd") or not link['flags'].get("nypl_login"):
+            return link
+        else:
+            link['url'] = request.host + "/fulfill/" + str(link['link_id'])
+        return link

tests/helper.py

@@ -1,6 +1,5 @@
 import os
 
-
 class TestHelpers:
     ENV_VARS = {
         'POSTGRES_HOST': 'test_psql_host',

tests/unit/test_api_edition_blueprint.py

@@ -1,10 +1,9 @@
-from flask import Flask
+from flask import Flask, request
 import pytest
 
 from api.blueprints.drbEdition import editionFetch
 from api.utils import APIUtils
 
-
 class TestEditionBlueprint:
     @pytest.fixture
     def mockUtils(self, mocker):
@@ -15,7 +14,7 @@ def mockUtils(self, mocker):
             formatResponseObject=mocker.DEFAULT,
             formatEditionOutput=mocker.DEFAULT
         )
-    
+
     @pytest.fixture
     def testApp(self):
         flaskApp = Flask('test')
@@ -48,7 +47,8 @@ def test_editionFetch_success_noFormat(self, mockUtils, testApp, mocker):
 
             mockUtils['normalizeQueryParams'].assert_called_once()
             mockUtils['formatEditionOutput'].assert_called_once_with(
-                mockEdition, records='testRecords', dbClient=mockDB, showAll=True, formats=[], reader='test'
+                mockEdition, records='testRecords', dbClient=mockDB, showAll=True, formats=[],
+                reader='test', request=request
             )
             mockUtils['formatResponseObject'].assert_called_once_with(
                 200, 'singleEdition', 'testEdition'
@@ -86,8 +86,9 @@ def test_editionFetch_success_format(self, mockUtils, testApp, mocker):
                 mocker.call('filter', queryParams)
             ])
             mockUtils['formatEditionOutput'].assert_called_once_with(
-                mockEdition, records='testRecords', showAll=True, dbClient=mockDB, 
-                                    formats=['application/html+edd', 'application/x.html+edd'], reader='test'
+                mockEdition, records='testRecords', showAll=True,
+                dbClient=mockDB,request=request,
+                formats=['application/html+edd', 'application/x.html+edd'], reader='test'
             )
             mockUtils['formatResponseObject'].assert_called_once_with(
                 200, 'singleEdition', 'testEdition'

tests/unit/test_api_link_blueprint.py

@@ -1,4 +1,4 @@
-from flask import Flask
+from flask import Flask, request
 import pytest
 
 from api.blueprints.drbLink import linkFetch
@@ -36,7 +36,7 @@ def test_linkFetch_success(self, mockUtils, testApp, mocker):
             assert testAPIResponse == 'singleLinkResponse'
             mockDBClient.assert_called_once_with('testDBClient')
 
-            mockUtils['formatLinkOutput'].assert_called_once_with('dbLinkRecord')
+            mockUtils['formatLinkOutput'].assert_called_once_with('dbLinkRecord', request=request)
             mockUtils['formatResponseObject'].assert_called_once_with(
                 200, 'singleLink', 'testLink'
             )

tests/unit/test_api_search_blueprint.py

@@ -1,11 +1,10 @@
-from flask import Flask
+from flask import Flask, request
 import pytest
 
 from api.blueprints.drbSearch import standardQuery
 from api.utils import APIUtils
 from api.elastic import ElasticClientError
 
-
 class TestSearchBlueprint:
     @pytest.fixture
     def mockUtils(self, mocker):
@@ -139,7 +138,8 @@ def test_standardQuery(self, mockUtils, mockHits, testApp, mocker):
                 testResultIds,
                 dbClient=mockDB,
                 formats=['text/html'],
-                reader='test'
+                reader='test',
+                request=request
             )
 
             mockUtils['formatResponseObject'].assert_called_once_with(