From bce766ee7f08ca4d992470fd674b220060d52d46 Mon Sep 17 00:00:00 2001
From: Sean Kelly <kelly@seankelly.biz>
Date: Thu, 23 Apr 2020 17:35:14 -0500
Subject: [PATCH] Resolve #44 - `manifest_url` for SIP labels

-   Test case to expose the issue
-   New constant for SIP manifest locations
-   Slight PEP-8 cosmetic fix (spaces around `+`)
---
 src/pds/aipgen/constants.py             |  3 +++
 src/pds/aipgen/sip.py                   |  6 +++---
 src/pds/aipgen/tests/test_functional.py | 23 +++++++++++++++++++++--
 3 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/src/pds/aipgen/constants.py b/src/pds/aipgen/constants.py
index df33a14..34fcfbc 100644
--- a/src/pds/aipgen/constants.py
+++ b/src/pds/aipgen/constants.py
@@ -62,3 +62,6 @@
     'SHA-1':   'sha1',
     'SHA-256': 'sha256',
 }
+
+# The "well-defined" location for SIP manifests
+SIP_MANIFEST_URL = 'https://pds-gamma.jpl.nasa.gov/data/pds4/manifests/'
diff --git a/src/pds/aipgen/sip.py b/src/pds/aipgen/sip.py
index 0daa027..f7ae4b2 100644
--- a/src/pds/aipgen/sip.py
+++ b/src/pds/aipgen/sip.py
@@ -33,7 +33,7 @@
 
 from .constants import (
     INFORMATION_MODEL_VERSION, PDS_NS_URI, XML_SCHEMA_INSTANCE_NS_URI, XML_MODEL_PI,
-    PDS_SCHEMA_URL, AIP_PRODUCT_URI_PREFIX, PDS_LABEL_FILENAME_EXTENSION, HASH_ALGORITHMS
+    PDS_SCHEMA_URL, AIP_PRODUCT_URI_PREFIX, PDS_LABEL_FILENAME_EXTENSION, HASH_ALGORITHMS, SIP_MANIFEST_URL
 )
 from .utils import (
     getPrimariesAndOtherInfo, getMD5, getLogicalIdentifierAndFileInventory, parseXML, getDigest, addLoggingArguments
@@ -353,8 +353,8 @@ def _writeLabel(logicalID, versionID, title, digest, size, numEntries, hashName,
     deep.append(etree.Comment('MD5 digest checksum for the manifest file'))
     etree.SubElement(deep, prefix + 'manifest_checksum').text = digest
     etree.SubElement(deep, prefix + 'checksum_type').text = 'MD5'
-    etree.SubElement(deep, prefix + 'manifest_url').text = 'file:' + os.path.abspath(manifestFile)
-    etree.SubElement(deep, prefix + 'aip_lidvid').text = AIP_PRODUCT_URI_PREFIX + logicalID.split(':')[-1]+ '_v' + versionID + '::1.0'
+    etree.SubElement(deep, prefix + 'manifest_url').text = SIP_MANIFEST_URL
+    etree.SubElement(deep, prefix + 'aip_lidvid').text = AIP_PRODUCT_URI_PREFIX + logicalID.split(':')[-1] + '_v' + versionID + '::1.0'
 
     aipMD5 = getMD5(aipFile) if aipFile else '00000000000000000000000000000000'
     etree.SubElement(deep, prefix + 'aip_label_checksum').text = aipMD5
diff --git a/src/pds/aipgen/tests/test_functional.py b/src/pds/aipgen/tests/test_functional.py
index bdb8585..eae8944 100644
--- a/src/pds/aipgen/tests/test_functional.py
+++ b/src/pds/aipgen/tests/test_functional.py
@@ -32,8 +32,10 @@
 '''PDS AIP-GEN functional tests'''
 
 
-import unittest, tempfile, shutil, os, pkg_resources, filecmp
+from lxml import etree
 from pds.aipgen.sip import produce
+from pds.aipgen.constants import PDS_NS_URI
+import unittest, tempfile, shutil, os, pkg_resources, filecmp
 
 
 class SIPFunctionalTestCase(unittest.TestCase):
@@ -41,6 +43,7 @@ class SIPFunctionalTestCase(unittest.TestCase):
 
     TODO: factor this out so we can generically do AIP and other file-based functional tests too.
     '''
+    _urlXPath = f'./{{{PDS_NS_URI}}}Information_Package_Component_Deep_Archive/{{{PDS_NS_URI}}}manifest_url'
     def setUp(self):
         super(SIPFunctionalTestCase, self).setUp()
         self.input = pkg_resources.resource_stream(__name__, 'data/ladee_test/mission_bundle/LADEE_Bundle_1101.xml')
@@ -49,7 +52,7 @@ def setUp(self):
         os.chdir(self.testdir)
     def test_sip_of_a_ladee(self):
         '''Test if the SIP manifest of LADEE bundle works as expected'''
-        manifest, label = produce(
+        manifest, ignoredLabel = produce(
             bundle=self.input,
             hashName='md5',
             registryServiceURL=None,
@@ -61,6 +64,22 @@ def test_sip_of_a_ladee(self):
             aipFile=None
         )
         self.assertTrue(filecmp.cmp(manifest, self.valid), "SIP manifest doesn't match the valid version")
+    def test_label_url(self):
+        '''Test if the label of a SIP manifest has the right ``manifest_url``'''
+        ignoredManifest, label = produce(
+            bundle=self.input,
+            hashName='md5',
+            registryServiceURL=None,
+            insecureConnectionFlag=True,
+            site='PDS_ATM',
+            offline=True,
+            baseURL='https://atmos.nmsu.edu/PDS/data/PDS4/LADEE/',
+            allCollections=True,
+            aipFile=None
+        )
+        matches = etree.parse(label).getroot().findall(self._urlXPath)
+        self.assertEqual(1, len(matches))
+        self.assertEqual('https://pds-gamma.jpl.nasa.gov/data/pds4/manifests/', matches[0].text)
     def tearDown(self):
         self.input.close()
         os.chdir(self.cwd)