recursiveloader: Support finding file & dist entries

2 files changed, 120 insertions, 7 deletions

diff --git a/gemato/recursiveloader.py b/gemato/recursiveloader.py
index 5198ba5..9b72c0c 100644
--- a/gemato/recursiveloader.py
+++ b/gemato/recursiveloader.py
@@ -72,3 +72,56 @@ class ManifestRecursiveLoader(object):
                 break
             for mpath, e in to_load:
                 self.load_manifest(mpath, e)
+
+    def find_timestamp(self):
+        """
+        Find a timestamp entry and return it. Returns None if there
+        is no timestamp.
+        """
+
+        self.load_manifests_for_path('')
+        for p, m in self._iter_manifests_for_path(''):
+            for e in m.entries:
+                if isinstance(e, gemato.manifest.ManifestEntryTIMESTAMP):
+                    return e
+        return None
+
+    def find_path_entry(self, path):
+        """
+        Find a matching entry for path @path and return it. Returns
+        None when no path matches. DIST entries are not included.
+        """
+
+        self.load_manifests_for_path(path)
+        for relpath, m in self._iter_manifests_for_path(path):
+            for e in m.entries:
+                if isinstance(e, gemato.manifest.ManifestEntryIGNORE):
+                    # ignore matches recursively, so we process it separately
+                    # py<3.5 does not have os.path.commonpath()
+                    fullpath = os.path.join(relpath, e.path)
+                    if (path + '/').startswith(fullpath + '/'):
+                        return e
+                elif isinstance(e, gemato.manifest.ManifestEntryDIST):
+                    # distfiles are not local files, so skip them
+                    pass
+                elif isinstance(e, gemato.manifest.ManifestPathEntry):
+                    fullpath = os.path.join(relpath, e.path)
+                    if fullpath == path:
+                        return e
+        return None
+
+    def find_dist_entry(self, filename, relpath=''):
+        """
+        Find a matching entry for distfile @filename and return it.
+        If @relpath is provided, loads all Manifests up to @relpath
+        (which can be e.g. a relevant package directory).
+        Returns None when no DIST entry matches.
+        """
+
+        self.load_manifests_for_path(relpath+'/')
+        for p, m in self._iter_manifests_for_path(relpath+'/'):
+            for e in m.entries:
+                if isinstance(e, gemato.manifest.ManifestEntryDIST):
+                    if e.path == filename:
+                        return e
+        return None
diff --git a/tests/test_recursiveloader.py b/tests/test_recursiveloader.py
index 6bbc7bd..b60c44c 100644
--- a/tests/test_recursiveloader.py
+++ b/tests/test_recursiveloader.py
@@ -3,6 +3,7 @@
 # (c) 2017 Michał Górny
 # Licensed under the terms of 2-clause BSD license
 
+import datetime
 import io
 import os
 import tempfile
@@ -15,12 +16,17 @@ class BasicNestingTest(unittest.TestCase):
     DIRS = ['sub', 'sub/deeper']
     FILES = {
         'Manifest': u'''
-MANIFEST sub/Manifest 65 MD5 6af76e314820a44aba2b4bd3e6280c20
+TIMESTAMP 2017-01-01T01:01:01Z
+MANIFEST sub/Manifest 128 MD5 30fd28b98a23031c72793908dd35c530
+DIST topdistfile-1.txt 0 MD5 d41d8cd98f00b204e9800998ecf8427e
 ''',
         'sub/Manifest': u'''
-MANIFEST deeper/Manifest 0 MD5 d41d8cd98f00b204e9800998ecf8427e
+MANIFEST deeper/Manifest 50 MD5 0f7cd9ed779a4844f98d28315dd9176a
+DIST subdistfile-1.txt 0 MD5 d41d8cd98f00b204e9800998ecf8427e
+''',
+        'sub/deeper/Manifest': u'''
+DATA test 0 MD5 d41d8cd98f00b204e9800998ecf8427e
 ''',
-        'sub/deeper/Manifest': u'',
     }
 
     def setUp(self):
@@ -62,16 +68,55 @@ MANIFEST deeper/Manifest 0 MD5 d41d8cd98f00b204e9800998ecf8427e
         self.assertIn('sub/Manifest', m.loaded_manifests)
         self.assertIn('sub/deeper/Manifest', m.loaded_manifests)
 
+    def test_find_timestamp(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertEqual(m.find_timestamp().ts,
+                datetime.datetime(2017, 1, 1, 1, 1, 1))
+
+    def test_find_path_entry(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertIsNone(m.find_path_entry('test'))
+        self.assertIsNone(m.find_path_entry('sub/test'))
+        self.assertEqual(m.find_path_entry('sub/deeper/test').path, 'test')
+
+    def test_find_top_dist_entry(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertEqual(m.find_dist_entry('topdistfile-1.txt').path, 'topdistfile-1.txt')
+        self.assertIsNone(m.find_dist_entry('subdistfile-1.txt'))
+
+    def test_find_sub_dist_entry(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertEqual(m.find_dist_entry('topdistfile-1.txt', 'sub').path, 'topdistfile-1.txt')
+        self.assertEqual(m.find_dist_entry('subdistfile-1.txt', 'sub').path, 'subdistfile-1.txt')
+
+    def test_find_sub_dist_entry_with_slash_path(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertEqual(m.find_dist_entry('topdistfile-1.txt', 'sub/').path, 'topdistfile-1.txt')
+        self.assertEqual(m.find_dist_entry('subdistfile-1.txt', 'sub/').path, 'subdistfile-1.txt')
+
+    def test_find_sub_dist_entry_with_file_path(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertEqual(m.find_dist_entry('topdistfile-1.txt', 'sub/file').path, 'topdistfile-1.txt')
+        self.assertEqual(m.find_dist_entry('subdistfile-1.txt', 'sub/file').path, 'subdistfile-1.txt')
+
 
 class MultipleManifestTest(unittest.TestCase):
     DIRS = ['sub']
     FILES = {
         'Manifest': u'''
 MANIFEST sub/Manifest.a 0 MD5 d41d8cd98f00b204e9800998ecf8427e
-MANIFEST sub/Manifest.b 0 MD5 d41d8cd98f00b204e9800998ecf8427e
+MANIFEST sub/Manifest.b 32 MD5 95737355786df5760d6369a80935cf8a
 ''',
         'sub/Manifest.a': u'',
-        'sub/Manifest.b': u'',
+        'sub/Manifest.b': u'''
+TIMESTAMP 2017-01-01T01:01:01Z
+''',
     }
 
     def setUp(self):
@@ -98,15 +143,24 @@ MANIFEST sub/Manifest.b 0 MD5 d41d8cd98f00b204e9800998ecf8427e
         self.assertIn('sub/Manifest.a', m.loaded_manifests)
         self.assertIn('sub/Manifest.b', m.loaded_manifests)
 
+    def test_find_timestamp(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        # here it is expected to fail since TIMESTAMP is supposed
+        # to be top-level
+        self.assertIsNone(m.find_timestamp())
+
 
 class MultipleTopLevelManifestTest(unittest.TestCase):
     FILES = {
         'Manifest': u'''
 MANIFEST Manifest.a 0 MD5 d41d8cd98f00b204e9800998ecf8427e
-MANIFEST Manifest.b 0 MD5 d41d8cd98f00b204e9800998ecf8427e
+MANIFEST Manifest.b 32 MD5 95737355786df5760d6369a80935cf8a
 ''',
         'Manifest.a': u'',
-        'Manifest.b': u'',
+        'Manifest.b': u'''
+TIMESTAMP 2017-01-01T01:01:01Z
+''',
     }
 
     def setUp(self):
@@ -126,3 +180,9 @@ MANIFEST Manifest.b 0 MD5 d41d8cd98f00b204e9800998ecf8427e
         m.load_manifests_for_path('')
         self.assertIn('Manifest.a', m.loaded_manifests)
         self.assertIn('Manifest.b', m.loaded_manifests)
+
+    def test_find_timestamp(self):
+        m = gemato.recursiveloader.ManifestRecursiveLoader(
+            os.path.join(self.dir, 'Manifest'))
+        self.assertEqual(m.find_timestamp().ts,
+                datetime.datetime(2017, 1, 1, 1, 1, 1))