pyannotators-entityfishing 1.6.136__tar.gz → 1.6.140__tar.gz

{pyannotators_entityfishing-1.6.136 → pyannotators_entityfishing-1.6.140}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: pyannotators-entityfishing
-Version: 1.6.136
+Version: 1.6.140
 Summary: Annotator based on entity-fishing
 Project-URL: Homepage, https://github.com/oterrier/pyannotators_entityfishing/
 Author-email: Olivier Terrier <olivier.terrier@kairntech.com>

{pyannotators_entityfishing-1.6.136 → pyannotators_entityfishing-1.6.140}/src/pyannotators_entityfishing/__init__.py

@@ -1,3 +1,3 @@
 """Annotator based on entity-fishing"""
 
-__version__ = "1.6.136"
+__version__ = "1.6.140"

{pyannotators_entityfishing-1.6.136 → pyannotators_entityfishing-1.6.140}/tools/build_poi.py

@@ -7,6 +7,7 @@
 A POI is a building, monument, venue, transport hub, museum, park, square or cemetery.
 The mapping is a list of Wikidata classes, gathered from the subclass trees of hub classes,
 minus the natural places (continents, bodies of water, landforms: the AFP does not want them)
+but the mountains, lakes, forests, waterfalls, canyons and caves, which it wants
 and two kinds of concepts:
 
 - a settlement: a population (P1082) marks a city, a town or an island, which stay Location;
@@ -84,9 +85,24 @@ PRECISE_HUBS = {
     "Q39614": ("cemetery", 2),
     "Q200141": ("necropolis", 2),
 }
-# Natural places, which the AFP does not want as POI and which stay Location: every subclass of a
-# continent, a body of water or a landform (the Everest, the Lake Geneva), whatever hub brings it in.
+# Natural places, which stay Location: every subclass of a continent, a body of water or a landform
+# (the Mediterranean, the Alps, the Strait of Hormuz), whatever hub brings it in...
 NATURAL_ROOTS = {"Q5107": "continent", "Q15324": "body of water", "Q271669": "landform"}
+# ...but those of these hubs, which the AFP wants as POI (answer of 2026-10-14): a place a country
+# or a city holds, the Mont Ventoux, the Lake Geneva, Sherwood Forest. Walked like the broad hubs.
+NATURAL_HUBS = {
+    "Q8502": "mountain",
+    "Q8072": "volcano",
+    "Q207326": "summit",
+    "Q54050": "hill",
+    "Q23397": "lake",
+    "Q131681": "reservoir",
+    "Q4421": "forest",
+    "Q34038": "waterfall",
+    "Q150784": "canyon",
+    "Q35509": "cave",
+    "Q11269813": "cave with prehistoric art",  # not under cave: Lascaux, Chauvet
+}
 # Subclasses the hubs bring in that are not points of interest, found by
 # `python tools/wdfilter.py audit tests/data_v2/POI.json`.
 EXCLUDE = {
@@ -115,6 +131,25 @@ EXCLUDE = {
     "Q666294": "list of Natural Reserves in Russia",
     "Q2501058": "Off-Off-Broadway",
     "Q7307215": "ship refit",
+    "Q6617796": "list of ecoregions in Nigeria",
+    "Q30889174": "Chongar bridges",
+    "Q12363245": "no English label",
+    # brought in by the natural hubs: man-made or managed, not places one names
+    "Q131596": "farm",
+    "Q819511": "spoil tip",
+    "Q1413117": "retention basin",
+    "Q1434085": "underground hangar",
+    "Q1911141": "molehill",
+    "Q2039061": "windbreak",
+    "Q2137499": "infiltration basin",
+    "Q2527200": "compost heap",
+    "Q2586008": "commercial forest",
+    "Q2710737": "stockpile",
+    "Q2843044": "tree farm",
+    "Q3632569": "settling basin",
+    "Q4086881": "bioswale",
+    "Q4410701": "cultivated forest",
+    "Q4994812": "fire-fighting pond",
 }
 
 # Gold sets, sampled on structural properties rather than on the P31 classes the
@@ -129,9 +164,9 @@ GOLD = {
     "POI_stations": ("?x wdt:P296 ?code . ?x wdt:P625 ?c .", "a station code (P296)"),
     "POI_museums": ("?x wdt:P31/wdt:P279* wd:Q33506 .", "instance of museum or a subclass"),
     "POI_parks": ("?x wdt:P31/wdt:P279* wd:Q22698 . ?x wdt:P625 ?c .", "instance of park or a subclass"),
-    "LOC_mountains": ("?x wdt:P2660 ?prominence .", "a topographic prominence (P2660): stays Location"),
-    "LOC_lakes": ("?x wdt:P4511 ?depth . ?x wdt:P625 ?c .", "a vertical depth (P4511) and coordinates: stays Location"),
-    "LOC_waterfalls": ("?x wdt:P31/wdt:P279* wd:Q34038 .", "instance of waterfall or a subclass: stays Location"),
+    "POI_mountains": ("?x wdt:P2660 ?prominence .", "a topographic prominence (P2660)"),
+    "POI_lakes": ("?x wdt:P4511 ?depth . ?x wdt:P625 ?c .", "a vertical depth (P4511) and coordinates"),
+    "POI_waterfalls": ("?x wdt:P31/wdt:P279* wd:Q34038 .", "instance of waterfall or a subclass"),
     "POI_squares": ("?x wdt:P31/wdt:P279* wd:Q174782 .", "instance of square or a subclass"),
     "LOC_rivers": ("?x wdt:P403 ?mouth . ?x wdt:P2043 ?length .", "a watercourse with a mouth (P403): stays Location"),
 }
@@ -142,10 +177,14 @@ def poi_classes() -> list[str]:
     kept: set[str] = set()
     for hub in BROAD_HUBS:
         kept.update(n.qid for n in taxonomy(hub, max_depth=2, counts=False) if n.sites >= MIN_SITES or n.depth == 0)
+    wanted: set[str] = set()
+    for hub in NATURAL_HUBS:
+        wanted.update(n.qid for n in taxonomy(hub, max_depth=2, counts=False) if n.sites >= MIN_SITES or n.depth == 0)
+    kept |= wanted
     for hub, (_, depth) in PRECISE_HUBS.items():
         kept.update(n.qid for n in taxonomy(hub, max_depth=depth, counts=False) if n.depth <= 1 or n.sites >= MIN_SITES)
     natural = {r["c"].rsplit("/", 1)[1] for root in NATURAL_ROOTS for r in _sparql(f"SELECT ?c WHERE {{ ?c wdt:P279* wd:{root} }}", timeout=300)}
-    return sorted(kept - EXCLUDE.keys() - natural, key=lambda q: int(q[1:]))
+    return sorted(kept - EXCLUDE.keys() - (natural - wanted), key=lambda q: int(q[1:]))
 
 
 def build_mapping(qids: list[str]) -> dict: