allow .conda packages to use cached .tar.bz2 entries for otherwise si…

…milar packages
alexhall · May 13, 2019 · f87945e · f87945e
1 parent daa9cac
commit f87945e
Show file tree

Hide file tree

Showing 4 changed files with 36 additions and 23 deletions.
diff --git a/conda/core/package_cache_data.py b/conda/core/package_cache_data.py
@@ -474,15 +474,19 @@ def make_actions_for_record(pref_or_spec):
         sha256 = pref_or_spec.get("sha256")
         size = pref_or_spec.get("size")
         md5 = pref_or_spec.get("md5")
+        legacy_bz2_size = pref_or_spec.get("legacy_bz2_size")
+        legacy_bz2_md5 = pref_or_spec.get("legacy_bz2_md5")
 
         def pcrec_matches(pcrec):
             matches = True
-            if sha256 is not None and pcrec.sha256 is not None:
-                matches = sha256 == pcrec.sha256
+            # sha256 is overkill for things that are already in the package cache.
+            #     It's just a quick match.
+            # if sha256 is not None and pcrec.sha256 is not None:
+            #     matches = sha256 == pcrec.sha256
             if size is not None and pcrec.size is not None:
-                matches = size == pcrec.size
+                matches = pcrec.size in (size, legacy_bz2_size)
             if matches and md5 is not None and pcrec is not None:
-                matches = md5 == pcrec.md5
+                matches = pcrec.md5 in (md5, legacy_bz2_md5)
             return matches
 
         if md5:

diff --git a/conda/core/subdir_data.py b/conda/core/subdir_data.py
@@ -361,25 +361,29 @@ def _process_raw_repodata_str(self, raw_repodata_str):
             k[:-6] + _tar_bz2 for k in iterkeys(conda_packages)
         )
 
-        for fn, info in concatv(
-            iteritems(conda_packages),
-            ((k, legacy_packages[k]) for k in use_these_legacy_keys),
-        ):
-            info['fn'] = fn
-            info['url'] = join_url(channel_url, fn)
-            if add_pip and info['name'] == 'python' and info['version'].startswith(('2.', '3.')):
-                info['depends'].append('pip')
-            info.update(meta_in_common)
-            if info.get('record_version', 0) > 1:
-                log.debug("Ignoring record_version %d from %s",
-                          info["record_version"], info['url'])
-                continue
-            package_record = PackageRecord(**info)
-
-            _package_records.append(package_record)
-            _names_index[package_record.name].append(package_record)
-            for ftr_name in package_record.track_features:
-                _track_features_index[ftr_name].append(package_record)
+        for group, copy_legacy_md5 in ((iteritems(conda_packages), True),
+                                       (((k, legacy_packages[k]) for k in use_these_legacy_keys), False)):
+            for fn, info in group:
+                info['fn'] = fn
+                info['url'] = join_url(channel_url, fn)
+                if copy_legacy_md5:
+                    counterpart = fn.replace('.conda', '.tar.bz2')
+                    if counterpart in legacy_packages:
+                        info['legacy_bz2_md5'] = legacy_packages[counterpart].get('md5')
+                        info['legacy_bz2_size'] = legacy_packages[counterpart].get('size')
+                if add_pip and info['name'] == 'python' and info['version'].startswith(('2.', '3.')):
+                    info['depends'].append('pip')
+                info.update(meta_in_common)
+                if info.get('record_version', 0) > 1:
+                    log.debug("Ignoring record_version %d from %s",
+                              info["record_version"], info['url'])
+                    continue
+                package_record = PackageRecord(**info)
+
+                _package_records.append(package_record)
+                _names_index[package_record.name].append(package_record)
+                for ftr_name in package_record.track_features:
+                    _track_features_index[ftr_name].append(package_record)
 
         self._internal_state = _internal_state
         return _internal_state

diff --git a/conda/models/records.py b/conda/models/records.py
@@ -244,6 +244,9 @@ class PackageRecord(DictSafeMixin, Entity):
     fn = FilenameField(aliases=('filename',))
 
     md5 = StringField(default=None, required=False, nullable=True, default_in_dump=False)
+    legacy_bz2_md5 = StringField(default=None, required=False, nullable=True,
+                                 default_in_dump=False)
+    legacy_bz2_size = IntegerField(required=False, nullable=True, default_in_dump=False)
     url = StringField(default=None, required=False, nullable=True, default_in_dump=False)
     sha256 = StringField(default=None, required=False, nullable=True, default_in_dump=False)
 

diff --git a/tests/core/test_package_cache_data.py b/tests/core/test_package_cache_data.py
@@ -50,6 +50,8 @@
       "depends": [
         "vc >=14.1,<15.0a0"
       ],
+      "legacy_bz2_md5": "a46cf10ba0eece37dffcec2d45a1f4ec",
+      "legacy_bz2_size": 131285,
       "license": "zlib",
       "license_family": "Other",
       "md5": "edad165fc3d25636d4f0a61c42873fbc",