fix tests after updating testdata/

contrib/docker/config.ini

@@ -7,7 +7,7 @@
 [DEFAULT]
 db_uri:      	 postgresql://docker:docker@db_1:5432/debsources
 root_dir:    	 /opt/debsources
-cache_dir:   	 %(root_dir)s/cache
+cache_dir:   	 %(root_dir)s/testdata/cache
 local_dir:   	 %(root_dir)s/local
 sources_dir: 	 %(root_dir)s/testdata/sources
 python_dir:  	 %(root_dir)s/python

doc/testing.txt

@@ -72,7 +72,7 @@ updated to avoid test failures. Here is the recommended procedures to do that:
 1. start with *clean slate*: clean your DB (e.g., `dropdb debsources`) and your
    local sources directory (e.g., `rm -rf /srv/debsources/sources`). Then
    re-inizialize an empty DB (e.g.,
-   `bin/debsources-admin --createdb postgres:///debsources`)
+   `createdb debsources; bin/debsources-dbadmin --createdb postgres:///debsources`)
 
 2. do a *full update run* using a version of the code base that is trusted
    enough to create the new reference version of the DB to be used for tests

lib/debsources/license_helper.py

@@ -101,11 +101,6 @@ def license_url(package, version):
 
 
 def get_license(package, version, path, c):
-    # if not license_path:
-    #     # retrieve license from DB
-    #     return qry.get_license_w_path(session, package, version, path)
-    # import pdb; pdb.set_trace()
-
     # pathlib.Path uses a str to internally represent a path. In case of
     # invalid bytes for utf8 encoding, surrogate escape sequences are
     # used. debian.copyright seems to only work with utf8 when reading the

lib/debsources/plugins/hook_copyright.py

@@ -38,7 +38,7 @@ def parse_license_file(path):
         for line in licenses:
             fields = line.rstrip().split(b'\t')
             license = fields[0].decode('utf8')
-            filename = fields[1]
+            filename = Path(fields[1].decode('utf8', 'surrogateescape'))
             license_list.append((license, filename))
     return license_list
 

lib/debsources/tests/test_fs_storage.py

@@ -31,7 +31,7 @@ class FsStorageTests(unittest.TestCase):
     @istest
     def assertWalkLength(self):
         self.assertEqual(len([f for f in walk(make_path(''))]),
-                         288)
+                         261)
 
     @istest
     def assertWalkTestChecksums(self):

lib/debsources/tests/test_stats.py

@@ -49,9 +49,9 @@ def diskUsagesMatchReferenceDb(self):
             'wheezy': 39688,
             'jessie': 50528,
             'sid': 54456,
-            'experimental': 12968,
+            'experimental': 12964,
         }
-        total_size = 180732
+        total_size = 180728
         self.assertSuiteCountsEqual(sizes, statistics.disk_usage)
         self.assertEqual(total_size, statistics.disk_usage(self.session))
 
@@ -120,13 +120,13 @@ def slocCountsMatchReferenceDb(self):
     @istest
     def ctagsCountsMatchReferenceDb(self):
         ctags = {
-            'squeeze': 30644,
-            'wheezy': 20150,
-            'jessie': 23444,
-            'sid': 28352,
+            'squeeze': 31015,
+            'wheezy': 20521,
+            'jessie': 23815,
+            'sid': 28723,
             'experimental': 17284,
         }
-        total_ctags = 116089
+        total_ctags = 116832
         self.assertSuiteCountsEqual(ctags, statistics.ctags)
         self.assertEqual(total_ctags, statistics.ctags(self.session))
 
@@ -135,7 +135,7 @@ def slocPerPkgMatchReferenceDb(self):
         LARGEST = ('cvsnt', '2.5.03.2382-3', 293583)
         SMALLEST = ('susv3', '6.1', 10)
         LARGEST_exp = ('beignet', '1.0.0-1', 81413)
-        SMALLEST_exp = ('ledger', '3.0.0~20130313+b608ed2-1', 45848)
+        SMALLEST_exp = ('ledger', '3.0.0~20130313+b608ed2-1', 46060)
 
         slocs_all = statistics.sloc_per_package(self.session)
         self.assertEqual(slocs_all[0], LARGEST)
@@ -149,9 +149,9 @@ def slocPerPkgMatchReferenceDb(self):
     @istest
     def areaFiltersMatchReferenceDb(self):
         self.assertEqual(statistics.disk_usage(self.session),
-                         180732)
+                         180728)
         self.assertEqual(statistics.disk_usage(self.session, areas=['main']),
-                         155176)
+                         155172)
         self.assertEqual(statistics.disk_usage(self.session,
                                                suite='wheezy', areas=['main']),
                          35824)
@@ -172,7 +172,7 @@ def areaFiltersMatchReferenceDb(self):
                              statistics.sloccount_lang(self.session, 'ansic'))
 
         area_count = statistics.ctags(self.session, areas=['main'])
-        self.assertEqual(area_count, 87507)
+        self.assertEqual(area_count, 88250)
         self.assertLessEqual(area_count, statistics.ctags(self.session))
 
     @istest
@@ -194,7 +194,7 @@ def test_group_by_stats(self):
         self.assertEqual(stats['etch'], 32736)
 
         stats = dict(statistics.stats_grouped_by(self.session, 'ctags'))
-        self.assertEqual(stats['wheezy'], 20150)
+        self.assertEqual(stats['wheezy'], 20521)
 
         stats = dict(statistics.stats_grouped_by(self.session,
                                                  'source_packages'))

lib/debsources/tests/test_updater.py

@@ -281,16 +281,16 @@ def tearDown(self):
 
     @istest
     def sizeMatchesReferenceDb(self):
-        EXPECTED_SIZE = 180732
+        EXPECTED_SIZE = 180728
         self.assertEqual(EXPECTED_SIZE, self.stats['total.disk_usage'])
 
     @istest
     def statsMatchReferenceDb(self):
         expected_stats = {  # just a few samples
-            'total.ctags': 116089,
-            'debian_sid.ctags': 28352,
-            'debian_squeeze.ctags': 30644,
-            'debian_experimental.disk_usage': 12968,
+            'total.ctags': 116832,
+            'debian_sid.ctags': 28723,
+            'debian_squeeze.ctags': 31015,
+            'debian_experimental.disk_usage': 12964,
             'total.source_files': 9333,
             'debian_experimental.source_files': 1396,
             'debian_jessie.source_files': 2038,
@@ -306,7 +306,8 @@ def statsMatchReferenceDb(self):
             'debian_wheezy.sloccount.python': 2798,
             'debian_squeeze.sloccount.ruby': 193,
             'debian_wheezy.sloccount.ruby': 193,
-            'total.sloccount': 1250618,
+            'total.sloccount': 1250830,
+            'total.sloccount.javascript': 212,
             'debian_squeeze.sloccount': 315750,
         }
         self.assertDictContainsSubset(expected_stats, self.stats)

lib/debsources/tests/test_webapp.py

@@ -516,8 +516,8 @@ def test_api_checksum_search_within_package(self):
 
     def test_api_search_ctag(self):
         rv = json.loads(self.app.get('/api/ctag/?ctag=name').data)
-        self.assertEqual(rv["count"], 193)
-        self.assertEqual(len(rv["results"]), 193)
+        self.assertEqual(rv["count"], 195)
+        self.assertEqual(len(rv["results"]), 195)
 
     def test_api_search_ctag_within_package(self):
         rv = json.loads(self.app.get(
@@ -555,8 +555,8 @@ def test_info_version(self):
     def test_api_stats_suite(self):
         rv = json.loads(self.app.get('/api/stats/jessie/').data)
         self.assertEqual(rv["suite"], "jessie")
-        # self.assertEqual(rv["results"]["debian_jessie.ctags"], 21767)
-        # self.assertEqual(rv["results"]["debian_jessie.disk_usage"], 43032)
+        self.assertEqual(rv["results"]["debian_jessie.ctags"], 23815)
+        self.assertEqual(rv["results"]["debian_jessie.disk_usage"], 50528)
         self.assertEqual(rv["results"]["debian_jessie.source_files"], 2038)
         self.assertEqual(rv["results"]["debian_jessie.sloccount.python"], 2916)