Made shas unique when combined with version, added additional logging (boostorg#1736)

Author: daveoconnor

libraries/tasks.py

@@ -48,6 +48,7 @@ def get_and_store_library_version_documentation_urls_for_version(version_pk):
     try:
         version = Version.objects.get(pk=version_pk)
     except Version.DoesNotExist:
+        logger.error(f"Version does not exist for {version_pk=}")
         raise
 
     if version_missing_docs(version):

versions/management/commands/import_archives_release_data.py

@@ -36,6 +36,7 @@ def command(release: str, new: bool):
     it will import the release data for all releases that are greater than or equal to
     the minimum release defined in settings.MIN_ARCHIVES_RELEASE.
     """
+    logger.info(f"import_archive_release_data {release=} {new=}")
     last_release = settings.MIN_ARCHIVES_RELEASE
 
     if release:
@@ -44,21 +45,23 @@ def command(release: str, new: bool):
         versions = [Version.objects.most_recent()]
     else:
         versions = Version.objects.filter(name__gte=last_release)
+    logger.info(f"import_archive_release_data {versions=}")
 
     for v in versions:
-        logger.info(f"Processing release {v.name}")
+        logger.info(f"Processing release {v.name=}")
         version_num = v.name.replace("boost-", "")
         try:
             archives_urls = get_archives_download_uris_for_release(version_num)
             binaries_urls = get_binary_download_uris_for_release(version_num)
             file_urls = archives_urls + binaries_urls
         except requests.exceptions.HTTPError:
-            print(f"Skipping {version_num}, error retrieving release data")
+            logger.info(f"Skipping {version_num}, error retrieving release data")
             continue
 
         download_data = []
         checksums = dict()
         for url in file_urls:
+            logger.info(f"Processing {v.name=} {url=}")
             try:
                 if "/binaries/" in url:
                     if not checksums:
@@ -70,6 +73,6 @@ def command(release: str, new: bool):
             except (requests.exceptions.HTTPError, ValueError):
                 print(f"Skipping {url}; error retrieving download data")
                 continue
-
+            logger.info(f"Data for {v.name=} at {url=}: {download_data=}")
             store_release_downloads_for_version(v, download_data)
-            print(f"Stored download data for {v.name}")
+            logger.info(f"Stored download data from {url=} for {v.name}")

versions/migrations/0024_alter_versionfile_checksum_and_more.py

@@ -0,0 +1,22 @@
+# Generated by Django 4.2.16 on 2025-08-22 22:43
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("versions", "0023_version_fully_imported"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="versionfile",
+            name="checksum",
+            field=models.CharField(default=None, max_length=64),
+        ),
+        migrations.AlterUniqueTogether(
+            name="versionfile",
+            unique_together={("version", "checksum")},
+        ),
+    ]

versions/models.py

@@ -221,12 +221,15 @@ class VersionFile(models.Model):
     operating_system = models.CharField(
         choices=OPERATING_SYSTEM_CHOICES, max_length=15, default=Unix
     )
-    checksum = models.CharField(max_length=64, unique=True, default=None)
+    checksum = models.CharField(max_length=64, default=None)
     url = models.URLField()
     display_name = models.CharField(max_length=256, blank=True, null=True)
 
     objects = VersionFileManager()
 
+    class Meta:
+        unique_together = [["version", "checksum"]]
+
 
 # TODO: should this go in a new `reviews` app?
 class Review(models.Model):

versions/releases.py

@@ -380,6 +380,7 @@ def store_release_downloads_for_version(version, release_data):
             - display_name (str): The name of the release file.
     """
     for data in release_data:
+        logger.info(f"Storing download data for {version.name}: {data=}")
         VersionFile.objects.update_or_create(
             version=version,
             checksum=data["checksum"],

versions/tasks.py

@@ -143,18 +143,7 @@ def import_version(
         },
     )
 
-    if created:
-        logger.info(
-            "import_versions_created_version",
-            version_name=name,
-            version_id=version.pk,
-        )
-    else:
-        logger.info(
-            "import_versions_updated_version",
-            version_name=name,
-            version_id=version.pk,
-        )
+    logger.info(f"import_versions_version {created=} {name=} {version.pk} ")
 
     # Get the release date for the version
     if get_release_date and not version.release_date:
@@ -325,18 +314,12 @@ def import_library_versions(version_name, token=None, version_type="tag"):
             library_version = save_library_version_by_library_key(
                 library_name, version, gitmodule
             )
-            if library_version:
-                logger.info(
-                    "import_library_versions_by_library_key",
-                    version_name=version_name,
-                    library_name=library_name,
-                )
-            else:
-                logger.info(
-                    "import_library_versions_skipped_library",
-                    version_name=version_name,
-                    library_name=library_name,
-                )
+
+            logger.info(
+                f"import_library_versions_by_library {version_name=} "
+                f"{library_name=} {library_version=} "
+            )
+
             continue
 
         if not libraries_json:
@@ -345,11 +328,11 @@ def import_library_versions(version_name, token=None, version_type="tag"):
             library_version = save_library_version_by_library_key(
                 library_name, version, gitmodule
             )
+
             if not library_version:
                 logger.info(
-                    "import_library_versions_skipped_library",
-                    version_name=version_name,
-                    library_name=library_name,
+                    f"import_library_versions_skipped_library "
+                    f"{version_name=} {library_name=}"
                 )
             continue
 
@@ -401,7 +384,7 @@ def import_library_versions(version_name, token=None, version_type="tag"):
     # For any libraries no longer in gitmodules we want to remove master and develop
     #  references from the library_versions list.
     if version_name in ["master", "develop"]:
-        logger.info("Triggering  removed submodules garbage collection")
+        logger.info("Triggering removed submodules garbage collection")
         gc_removed_submodules.delay(library_keys, version_name)
 
     # Retrieve and store the docs url for each library-version in this release
@@ -413,16 +396,17 @@ def import_library_versions(version_name, token=None, version_type="tag"):
 
 @app.task
 def import_release_downloads(version_pk):
+    logger.info(f"import_release_downloads w/ {version_pk=}")
     version = Version.objects.get(pk=version_pk)
     version_num = version.name.replace("boost-", "")
     if version_num < "1.63.0":
         # Downloads are in Sourceforge for older versions, and that has
         # not been implemented yet
-        logger.info("import_release_downloads_skipped", version_name=version.name)
+        logger.info(f"import_release_downloads_skipped {version.name=}")
         return
-
+    logger.info(f"import_release_downloads starting {version.name=}")
     call_command("import_archives_release_data", release=version_num)
-    logger.info("import_release_downloads_complete", version_name=version.name)
+    logger.info(f"import_release_downloads_complete {version.name=}")
 
 
 @app.task