resolvekit 0.1.11__tar.gz → 0.1.12__tar.gz

{resolvekit-0.1.11 → resolvekit-0.1.12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: resolvekit
-Version: 0.1.11
+Version: 0.1.12
 Summary: Entity and place resolution system that maps messy place/entity strings and codes to canonical entities
 Keywords: entity-resolution,geocoding,place-names,data-commons,iso-codes,offline,disambiguation,normalization
 Author: Jorge Rivera

{resolvekit-0.1.11 → resolvekit-0.1.12}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "resolvekit"
-version = "0.1.11"
+version = "0.1.12"
 description = "Entity and place resolution system that maps messy place/entity strings and codes to canonical entities"
 requires-python = ">=3.12"
 

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/_data/geo/continental_unions/metadata.json

@@ -10,7 +10,7 @@
     "fts": "fts5",
     "symspell": "symspell.dict"
   },
-  "build_timestamp": "2026-06-14T09:07:23.981230+00:00",
+  "build_timestamp": "2026-06-15T10:41:55.048834+00:00",
   "source_datasets": [
     "datacommons"
   ],
@@ -19,7 +19,7 @@
   },
   "description": null,
   "checksums": {
-    "sqlite": "8bc1a7a67d2c118aeb0348e8aa3fb12f51c9ade8c62a45afe764b84f85624288",
+    "sqlite": "e346f2503428e44cc0b7cf688356eae9522d640e0b3c6ae78f1ba4636399af39",
     "symspell": "4af3c92b484012bc0cbcd147409f8377ffe7bd6b02e501c3ea4d8475250da6dd"
   },
   "distribution": "bundled",
@@ -35,7 +35,7 @@
   "quality_metrics": {
     "entity_count": 33,
     "names_count": 98,
-    "codes_count": 3,
+    "codes_count": 4,
     "relations_count": 0,
     "names_coverage": 1.0,
     "codes_coverage": 0.030303030303030304,

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/_data/geo/countries/metadata.json

@@ -13,7 +13,7 @@
     "fts": "fts5",
     "symspell": "symspell.dict"
   },
-  "build_timestamp": "2026-06-14T21:33:50.501680+00:00",
+  "build_timestamp": "2026-06-15T11:23:37.041970+00:00",
   "source_datasets": [
     "datacommons"
   ],
@@ -23,8 +23,8 @@
   },
   "description": null,
   "checksums": {
-    "sqlite": "c46ad9f54dc83a09e338ec546edc72ed753756b8fd50613c04fe247e65c032f1",
-    "symspell": "a8254c4a5a18e45adf3e30d4b8d239e6c02a9876baff50c05c0d8c8153e1e815",
+    "sqlite": "e2a930d8eb4f2a31d35492da727da24871b79d10ed6704161f91f538ea9db25d",
+    "symspell": "a74c9af96955d8d6353acff8349deff435628bd26d03356a5b8ac9d6cecbed59",
     "calibrator": "900c09348742718985f3236f270d0cdc2d4941cc38fade9b7310185d6cf845a7"
   },
   "distribution": "bundled",
@@ -39,8 +39,8 @@
   "allow_new_entities": false,
   "quality_metrics": {
     "entity_count": 241,
-    "names_count": 4355,
-    "codes_count": 9129,
+    "names_count": 4361,
+    "codes_count": 9371,
     "relations_count": 1858,
     "names_coverage": 1.0,
     "codes_coverage": 1.0,

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/_data/geo/countries/symspell.dict

@@ -118,7 +118,7 @@ américa	1
 américaines	1
 ancienne	1
 ancienne république yougoslave de macédoine	1
-and	71
+and	72
 andorra	6
 andorre	1
 angola	7
@@ -415,7 +415,7 @@ bósnia	1
 bósnia e herzegovina	1
 cabo	7
 cabo verde	5
-caicos	7
+caicos	8
 caicosinseln	1
 caimán	1
 caledonia	5
@@ -494,7 +494,7 @@ china, región administrativa especial de hong kong	1
 china, región administrativa especial de macao	1
 china, republic of	1
 china, taiwan province of	1
-chine	3
+chine	4
 chine (république populaire de)	1
 chine,	2
 chine, région administrative spéciale de hong kong	1
@@ -1117,7 +1117,7 @@ holy see (vatican city state)	1
 honduran	1
 honduran republic	1
 honduras	7
-hong	11
+hong	12
 hong kong	1
 hong kong sar	1
 hong kong sar, china	1
@@ -1125,6 +1125,7 @@ hong kong special administrative region	1
 hong kong special administrative region of china	1
 hong kong special administrative region of the people's republic of china	1
 hong kong, china	1
+hong kong, chine	1
 hongkong	1
 hongrie	1
 hu	1
@@ -1227,7 +1228,7 @@ island territory of curaçao	1
 islanda	1
 islande	1
 islandia	1
-islands	52
+islands	53
 islands,	1
 islas	9
 islas caimán	1
@@ -1381,7 +1382,7 @@ kolumbien	1
 komoren	1
 komori	2
 kong	10
-kong,	1
+kong,	2
 kongo	4
 kongo (republik kongo)	1
 kongo)	1
@@ -1515,7 +1516,7 @@ líbano	2
 líbia	1
 lúcia	1
 ma	1
-maarten	5
+maarten	7
 macao	14
 macao s.a.r	1
 macao sar	1
@@ -1614,7 +1615,7 @@ mauritânia	1
 maurício	1
 maw	1
 mayen	9
-mayotte	1
+mayotte	3
 mazedonien	1
 mda	1
 mejicana	1
@@ -2505,8 +2506,8 @@ singapore city	1
 singapour	1
 singapur	2
 singapura	1
-sint	9
-sint maarten	1
+sint	11
+sint maarten	3
 sint maarten (dutch part)	1
 sint maarten (dutch part), kingdom of the netherlands	1
 sint martin	1
@@ -3201,11 +3202,11 @@ turkmenia	1
 turkmenistan	5
 turkmenistán	1
 turkménistan	1
-turks	6
+turks	7
 turks & caicos	1
 turks & caicos islands	1
 turks and caicos	1
-turks and caicos islands	1
+turks and caicos islands	2
 turks-	1
 turks- und caicosinseln	1
 turks-et-caïcos	1

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/_data/geo/regions/metadata.json

@@ -10,7 +10,7 @@
     "fts": "fts5",
     "symspell": "symspell.dict"
   },
-  "build_timestamp": "2026-06-14T09:07:23.453133+00:00",
+  "build_timestamp": "2026-06-15T10:41:54.481467+00:00",
   "source_datasets": [
     "datacommons"
   ],
@@ -19,7 +19,7 @@
   },
   "description": null,
   "checksums": {
-    "sqlite": "16a109da6c7c772097e0b0d54ac8b06b98af642fe53e5143e049df77de3449b6",
+    "sqlite": "f75d22afa159aa281115bd296e4df73c5ac6fc664cc14bffb262a6c8d3401ec6",
     "symspell": "0ceb0ae2f947178b5d6bbcf25995c4cdb2429d695ac13799f54bb8558b0b94c4"
   },
   "distribution": "bundled",
@@ -35,7 +35,7 @@
   "quality_metrics": {
     "entity_count": 829,
     "names_count": 935,
-    "codes_count": 1621,
+    "codes_count": 1668,
     "relations_count": 804,
     "names_coverage": 1.0,
     "codes_coverage": 1.0,

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/_data/manifest.json

@@ -8,9 +8,9 @@
       "entity_types": [
         "geo.country"
       ],
-      "size_mb": 2.61,
+      "size_mb": 2.62,
       "checksums": {
-        "sqlite": "c46ad9f54dc83a09e338ec546edc72ed753756b8fd50613c04fe247e65c032f1"
+        "sqlite": "e2a930d8eb4f2a31d35492da727da24871b79d10ed6704161f91f538ea9db25d"
       }
     },
     {
@@ -159,9 +159,9 @@
         "geo.region",
         "geo.subregion"
       ],
-      "size_mb": 1.08,
+      "size_mb": 1.09,
       "checksums": {
-        "sqlite": "16a109da6c7c772097e0b0d54ac8b06b98af642fe53e5143e049df77de3449b6"
+        "sqlite": "f75d22afa159aa281115bd296e4df73c5ac6fc664cc14bffb262a6c8d3401ec6"
       }
     },
     {
@@ -174,7 +174,7 @@
       ],
       "size_mb": 0.1,
       "checksums": {
-        "sqlite": "8bc1a7a67d2c118aeb0348e8aa3fb12f51c9ade8c62a45afe764b84f85624288"
+        "sqlite": "e346f2503428e44cc0b7cf688356eae9522d640e0b3c6ae78f1ba4636399af39"
       }
     },
     {
@@ -196,9 +196,9 @@
       "entity_types": [
         "org.development_finance_provider"
       ],
-      "size_mb": 0.31,
+      "size_mb": 0.32,
       "checksums": {
-        "sqlite": "8ffa163bb1874b98b9f2121f8c23aee24be4a17366b6af282cf5b71836bfb35f"
+        "sqlite": "5ca8f8a55d4791be539601c9e1154e6f0685f75a9d39df4670f7ee1f3822fc72"
       }
     },
     {

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/_data/org/providers/metadata.json

@@ -10,7 +10,7 @@
     "fts": "fts5",
     "symspell": "symspell.dict"
   },
-  "build_timestamp": "2026-05-31T15:20:14.507864+00:00",
+  "build_timestamp": "2026-06-15T10:42:07.671754+00:00",
   "source_datasets": [
     "datacommons"
   ],
@@ -19,7 +19,7 @@
   },
   "description": null,
   "checksums": {
-    "sqlite": "8ffa163bb1874b98b9f2121f8c23aee24be4a17366b6af282cf5b71836bfb35f",
+    "sqlite": "5ca8f8a55d4791be539601c9e1154e6f0685f75a9d39df4670f7ee1f3822fc72",
     "symspell": "73dc2fdf8b1292bfececc56052ffe59e27bd503112f270d40f2b4bc4843d9785"
   },
   "distribution": "bundled",
@@ -35,7 +35,7 @@
   "quality_metrics": {
     "entity_count": 254,
     "names_count": 376,
-    "codes_count": 254,
+    "codes_count": 358,
     "relations_count": 0,
     "names_coverage": 1.0,
     "codes_coverage": 1.0,

{resolvekit-0.1.11 → resolvekit-0.1.12}/src/resolvekit/builder/oecd_dac.py

@@ -254,6 +254,18 @@ def _attach_names_and_code(
     _insert_code(contrib, entity_id=entity_id, system=code_system, value=code_value)
 
 
+def _attach_dotstat_code(
+    contrib: GraphContribution,
+    *,
+    entity_id: str,
+    dotstatcode: str | None,
+) -> None:
+    if dotstatcode:
+        _insert_code(
+            contrib, entity_id=entity_id, system="oecd:dotstat", value=dotstatcode
+        )
+
+
 # ──────────────────────────────────────────────────────────────────────────────
 # Per-section upsert helpers
 # ──────────────────────────────────────────────────────────────────────────────
@@ -297,6 +309,9 @@ def _upsert_recipients(
             code_system="oecd:recipient",
             code_value=code,
         )
+        _attach_dotstat_code(
+            geo_contrib, entity_id=entity_id, dotstatcode=row.get("dotstatcode")
+        )
 
 
 def _upsert_country_providers(
@@ -326,6 +341,9 @@ def _upsert_country_providers(
             code_system="oecd:provider",
             code_value=code,
         )
+        _attach_dotstat_code(
+            geo_contrib, entity_id=entity_id, dotstatcode=row.get("dotstatcode")
+        )
 
 
 def _upsert_multilateral_providers(
@@ -375,6 +393,13 @@ def _upsert_multilateral_providers(
                                     "value_norm": _normalizer.normalize(label),
                                 }
                             )
+                # dotstat code is crosswalk coverage — attach even when the donor
+                # name is suppressed (EU Institutions case).
+                _attach_dotstat_code(
+                    geo_contrib,
+                    entity_id=mapped_entity_id,
+                    dotstatcode=row.get("dotstatcode"),
+                )
             elif db_label == "org" and org_contrib is not None:
                 _attach_names_and_code(
                     org_contrib,
@@ -384,6 +409,11 @@ def _upsert_multilateral_providers(
                     code_system="oecd:provider",
                     code_value=code,
                 )
+                _attach_dotstat_code(
+                    org_contrib,
+                    entity_id=mapped_entity_id,
+                    dotstatcode=row.get("dotstatcode"),
+                )
             # if target DB is unavailable, skip gracefully
         elif org_contrib is not None:
             entity_id = f"org/oecd:provider:{code}"
@@ -411,6 +441,9 @@ def _upsert_multilateral_providers(
                 code_system="oecd:provider",
                 code_value=code,
             )
+            _attach_dotstat_code(
+                org_contrib, entity_id=entity_id, dotstatcode=row.get("dotstatcode")
+            )
 
 
 def _resolve_channel_entity_id(