Include ftm version in index version prefix

yente/search/indexer.py

@@ -20,8 +20,8 @@
     NAME_PHONETIC_FIELD,
 )
 from yente.provider import SearchProvider, with_provider
-from yente.search.util import parse_index_name
-from yente.search.util import construct_index_name
+from yente.search.versions import parse_index_name
+from yente.search.versions import construct_index_name
 from yente.data.util import expand_dates, phonetic_names
 from yente.data.util import index_name_parts, index_name_keys
 

yente/search/status.py

@@ -1,7 +1,7 @@
 from yente import settings
 from yente.logs import get_logger
 from yente.provider import SearchProvider
-from yente.search.util import parse_index_name
+from yente.search.versions import parse_index_name
 from yente.data.manifest import Catalog
 
 log = get_logger(__name__)

yente/search/util.py → yente/search/versions.py

@@ -1,9 +1,19 @@
+from functools import cache
 from typing import Tuple
 from normality import slugify
+import followthemoney
 
 from yente import settings
 
 
+@cache
+def system_version() -> str:
+    """Get the current version of the system."""
+    parts = [v.rjust(2, "0") for v in followthemoney.__version__.split(".")]
+    ftm_version = "".join(parts)[:6]
+    return f"{settings.INDEX_VERSION}{ftm_version}-"
+
+
 def parse_index_name(index: str) -> Tuple[str, str]:
     """
     Parse a given index name.
@@ -19,31 +29,33 @@ def parse_index_name(index: str) -> Tuple[str, str]:
     if "-" not in index_end:
         raise ValueError("Index name does not contain a version.")
     dataset, index_version = index_end.split("-", 1)
-    if not index_version.startswith(settings.INDEX_VERSION):
+    sys_version = system_version()
+    if not index_version.startswith(sys_version):
         raise ValueError("Index version does not start with the correct prefix.")
-    dataset_version = index_version[len(settings.INDEX_VERSION) :]
+    dataset_version = index_version[len(sys_version) :]
     if len(dataset_version) < 1:
         raise ValueError("Index version must be at least one character long.")
     return (dataset, dataset_version)
 
 
-def construct_index_name(ds_name: str, ds_version: str | None = None) -> str:
+def construct_index_name(dataset: str, version: str | None = None) -> str:
     """
     Given a dataset and optionally a version construct a properly versioned index name.
     """
-    if len(str(ds_name)) < 1:
+    if len(str(dataset)) < 1:
         raise ValueError("Dataset name must be at least one character long.")
-    base = f"{settings.ENTITY_INDEX}-{ds_name}"
-    if ds_version is None:
+    base = f"{settings.ENTITY_INDEX}-{dataset}"
+    if version is None:
         return base
-    return f"{base}-{construct_index_version(ds_version)}"
+    return f"{base}-{construct_index_version(version)}"
 
 
 def construct_index_version(version: str) -> str:
     """Given a version ID, return a version string with the version prefix."""
     if len(version) < 1:
         raise ValueError("Version must be at least one character long.")
-    combined = slugify(f"{settings.INDEX_VERSION}{version}", "-")
-    if combined is None or len(combined) < len(settings.INDEX_VERSION) + 1:
-        raise ValueError("Invalid version: %s%s." % (settings.INDEX_VERSION, version))
+    sys_version = system_version()
+    combined = slugify(f"{sys_version}{version}", "-")
+    if combined is None or len(combined) < len(sys_version) + 1:
+        raise ValueError("Invalid version: %s%s." % (sys_version, version))
     return combined

yente/settings.py

@@ -181,7 +181,7 @@ def random_cron() -> str:
 INDEX_SHARDS = int(env_legacy("YENTE_INDEX_SHARDS", "YENTE_ELASTICSEARCH_SHARDS", "1"))
 INDEX_NAME = env_legacy("YENTE_INDEX_NAME", "YENTE_ELASTICSEARCH_INDEX", "yente")
 ENTITY_INDEX = f"{INDEX_NAME}-entities"
-INDEX_VERSION = env_str("YENTE_INDEX_VERSION", "009")
+INDEX_VERSION = env_str("YENTE_INDEX_VERSION", "011")
 assert len(INDEX_VERSION) == 3, "Index version must be 3 characters long."
 
 # ElasticSearch-only options: