Add ability to publish package indices using AptByHash format

closes pulp#795

CHANGES/795.feature

@@ -0,0 +1,2 @@
+Added an option to publish `by-hash/` files to mitigate the Hash Sum Mismatch error in debian repos as specified here: https://wiki.debian.org/DebianRepository/Format#indices_acquisition_via_hashsums_.28by-hash.29.
+Use the APT_BY_HASH setting to enable this feature.

docs/feature_overview.rst

@@ -144,6 +144,22 @@ That way, the repository version created in the target repository, can be meanin
 We are also planning to expand the advanced copy feature with a :ref:`dependency solving <dependency_solving>` mechanism in the future.
 
 
+.. _apt_by_hash:
+
+ApyByHash
+--------------------------------------------------------------------------------
+
+AptByHash is a feature that mitigates commonplace 'Hash Sum Mismatch' errors during an 'apt-get-update'.
+It adds the checksum of the package metadata to the the packages' names.
+These files are then stored within a 'by-hash' directory within each release architecture in the specified debian repository.
+The client will then use the filename to identify the expected checksum and download a file whose name matches the checksum.
+
+Please note that this feature is disabled by default and should be enabled prior to use.
+To do this, set `APT_BY_HASH = True` in `/pulp_deb/app/settings.py`.
+
+In addition, you are responsible for setting up a reverse proxy with cache in order to cache the by-hash files.
+
+
 Roadmap and Experimental
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 

pulp_deb/app/settings.py

@@ -5,5 +5,6 @@
     http://docs.pulpproject.org/en/3.0/nightly/plugins/plugin-writer/index.html
 """
 
+APT_BY_HASH = False
 FORBIDDEN_CHECKSUM_WARNINGS = True
 FORCE_IGNORE_MISSING_PACKAGE_INDICES = False

pulp_deb/app/tasks/publishing.py

@@ -2,6 +2,7 @@
 import os
 import shutil
 from contextlib import suppress
+from pathlib import Path
 
 from datetime import datetime, timezone
 from debian import deb822
@@ -44,6 +45,7 @@
     CHECKSUM_TYPE_MAP,
 )
 
+from pulp_deb.app.settings import APT_BY_HASH
 
 import logging
 from gettext import gettext as _
@@ -352,6 +354,23 @@ def finish(self):
                 publication=self.parent.publication, file=File(open(gz_package_index_path, "rb"))
             )
             gz_package_index.save()
+
+            # Generating metadata files using checksum
+            if APT_BY_HASH:
+                for path, index in (
+                    (package_index_path, package_index),
+                    (gz_package_index_path, gz_package_index),
+                ):
+                    for checksum in settings.ALLOWED_CONTENT_CHECKSUMS:
+                        if checksum in CHECKSUM_TYPE_MAP:
+                            hashed_index_path = _fetch_file_checksum(path, index, checksum)
+                            hashed_index = PublishedMetadata.create_from_file(
+                                publication=self.parent.publication,
+                                file=File(open(path, "rb")),
+                                relative_path=hashed_index_path,
+                            )
+                            hashed_index.save()
+
             self.parent.add_metadata(package_index)
             self.parent.add_metadata(gz_package_index)
         # Publish Sources Indices file
@@ -406,6 +425,7 @@ def __init__(
         self.release["Components"] = ""  # Will be set later
         if release.description != NULL_VALUE:
             self.release["Description"] = release.description
+        self.release["Acquire-By-Hash"] = "yes" if APT_BY_HASH else "no"
 
         for checksum_type, deb_field in CHECKSUM_TYPE_MAP.items():
             if checksum_type in settings.ALLOWED_CONTENT_CHECKSUMS:
@@ -480,3 +500,10 @@ def _zip_file(file_path):
         with GzipFile(gz_file_path, "wb") as f_out:
             shutil.copyfileobj(f_in, f_out)
     return gz_file_path
+
+
+def _fetch_file_checksum(file_path, index, checksum):
+    digest = getattr(index.contentartifact_set.first().artifact, checksum)
+    checksum_type = CHECKSUM_TYPE_MAP[checksum]
+    hashed_path = Path(file_path).parents[0] / "by-hash" / checksum_type / digest
+    return hashed_path