Correct successors/ancestors for fusion+split, fix #43

CHANGELOG.md

@@ -1,6 +1,10 @@
 # CHANGELOG
 
-## 8.1.1 — 2017-05-12
+## 8.1.3 — 2017-05-18
+
+* Correct successors/ancestors in case of fusions then splits.
+
+## 8.1.2 — 2017-05-12
 
 * Add 2-letters ISO codes for overseas collectivities.
 

README.md

@@ -43,9 +43,3 @@ Note that the duration of the whole test suite run is about 5 minutes.
 That's why you would probably prefer to run a particular test:
 
     $ python -m pytest tests/test_actions.py::test_change_name
-
-
-
-## TODO
-
-* Collect feedback from reusers

exports/communes/communes_1962-03-07.csv

@@ -21422,7 +21422,7 @@ COM55150@1942-01-01,55150,1942-01-01 00:00:00,9999-12-31 23:59:59,Demange-aux-Ea
 COM55151@1942-01-01,55151,1942-01-01 00:00:00,1972-12-31 23:59:59,Deuxnouds-aux-Bois,COM55274@1942-01-01,,DEP55@1860-07-01,NULL,330
 COM55152@1942-01-01,55152,1942-01-01 00:00:00,1972-12-31 23:59:59,Deuxnouds-devant-Beauzée,COM55040@1973-01-01,,DEP55@1860-07-01,NULL,330
 COM55153@1942-01-01,55153,1942-01-01 00:00:00,9999-12-31 23:59:59,Dieppe-sous-Douaumont,,,DEP55@1860-07-01,183,0
-COM55154@1942-01-01,55154,1942-01-01 00:00:00,1984-12-31 23:59:59,Dieue,COM55154@1985-01-01,COM55204@1942-01-01,DEP55@1860-07-01,262,120
+COM55154@1942-01-01,55154,1942-01-01 00:00:00,1984-12-31 23:59:59,Dieue,COM55154@1985-01-01;COM55204@1985-01-01,COM55204@1942-01-01,DEP55@1860-07-01,262,120
 COM55155@1942-01-01,55155,1942-01-01 00:00:00,9999-12-31 23:59:59,Dombasle-en-Argonne,,,DEP55@1860-07-01,432,0
 COM55156@1942-01-01,55156,1942-01-01 00:00:00,9999-12-31 23:59:59,Dombras,,,DEP55@1860-07-01,132,0
 COM55157@1942-01-01,55157,1942-01-01 00:00:00,9999-12-31 23:59:59,Dommartin-la-Montagne,,,DEP55@1860-07-01,60,0

geohisto/actions.py

@@ -112,7 +112,7 @@ def creation_delegated(towns, record):
         towns.delete(current_town)
 
 
-@in_case_of(CHANGE_NAME_REINSTATEMENT, REINSTATEMENT)
+@in_case_of(REINSTATEMENT)
 def reinstatement(towns, record):
     current_town = towns.get_current(record.depcom, record.eff)
 
@@ -134,13 +134,43 @@ def reinstatement(towns, record):
     towns.upsert(new_town)
 
     old_town = current_town.generate(
-        nccenr=record.nccanc or record.nccoff,
+        nccenr=record.nccoff,
         end_datetime=min(current_town.end_datetime, record.eff - DELTA),
         modification=record.mod
     )
     if new_town.valid_at(old_town.end_datetime + DELTA):
         old_town = old_town.add_successor(new_town.id)
     towns.upsert(old_town)
+    towns.replace_successor(old_town, new_town,
+                            valid_datetime=new_town.start_datetime - DELTA)
+
+
+@in_case_of(CHANGE_NAME_REINSTATEMENT)
+def change_name_reinstatement(towns, record):
+    current_town = towns.get_current(record.depcom, record.eff)
+    new_town = current_town.generate(
+        id=compute_id(current_town.depcom, record.effdate),
+        start_datetime=record.eff,
+        end_datetime=END_DATETIME,
+        nccenr=record.nccoff,
+        successors='',
+        modification=0
+    )
+    towns.upsert(new_town)
+
+    old_town = current_town.generate(
+        nccenr=record.nccanc or record.nccoff,
+        end_datetime=min(current_town.end_datetime, record.eff - DELTA),
+        modification=record.mod
+    )
+    old_town = old_town.add_successor(new_town.id)
+    for ancestor in old_town.get_ancestors(towns):
+        for guessed_successor in towns.valid_at(
+                old_town.end_datetime + DELTA, depcom=ancestor.depcom):
+            if (guessed_successor and guessed_successor.id != old_town.id and
+                    guessed_successor.id != new_town.id):
+                old_town = old_town.add_successor(guessed_successor.id)
+    towns.upsert(old_town)
 
 
 @in_case_of(SPLITING)
@@ -153,16 +183,34 @@ def spliting(towns, record):
     towns.upsert(current_town)
 
 
-@in_case_of(DELETION_PARTITION, DELETION_FUSION,
-            FUSION_ASSOCIATION_ASSOCIATED, CREATION_DELEGATED)
+@in_case_of(DELETION_PARTITION, DELETION_FUSION, CREATION_DELEGATED)
 def deletion(towns, record):
     current_town = towns.get_current(record.depcom, record.eff)
+    old_town = current_town.generate(
+        nccenr=record.nccoff,
+        end_datetime=record.eff - DELTA,
+        modification=record.mod
+    )
+    successor = towns.get_current(record.comech, record.eff)
+    old_town = old_town.add_successor(successor.id)
+    towns.upsert(old_town)
+
+
+@in_case_of(FUSION_ASSOCIATION_ASSOCIATED)
+def fusion_association_associated(towns, record):
+    current_town = towns.get_current(record.depcom, record.eff)
 
     end_datetime = record.eff - DELTA
 
-    # It happens with `Lamarche-en-Woëvre` for instance
+    # TODO: move to specials.
+    # It happens only with `Lamarche-en-Woëvre` for instance
     # because the reinstatement is at the same date of the (re)fusion,
     # so we set the end_date just after the start_date exceptionnally.
+    # Record(depcom='55273', mod=330, eff=datetime.datetime(1983, 1, 1, 0, 0),
+    # nccoff='Lamarche-en-Woëvre', nccanc='', comech='55386', dep='55',
+    # com='273', depanc='', last=None, effdate=datetime.date(1983, 1, 1))
+    # Current town: <Town (COM55273@1983-01-01): Lamarche-en-Woëvre
+    # from 1983-01-01 to 9999-12-31 with successors  and mod 0>
     has_temporary_existence = current_town.start_datetime == record.eff
     if has_temporary_existence:
         end_datetime = record.eff + DELTA
@@ -174,6 +222,13 @@ def deletion(towns, record):
     )
     successor = towns.get_current(record.comech, record.eff)
     old_town = old_town.add_successor(successor.id)
+    if successor.modification == CHANGE_NAME_REINSTATEMENT:
+        # Deal with fusions then splits declared in the wrong order.
+        if old_town.depcom not in successor.successors:
+            new_town = towns.get_current(
+                old_town.depcom, successor.end_datetime + DELTA)
+            successor = successor.add_successor(new_town.id)
+            towns.upsert(successor)
     towns.upsert(old_town)
 
 

geohisto/models.py

@@ -181,6 +181,13 @@ def add_ancestor(self, ancestor):
             ancestor = self.ancestors + ';' + ancestor
         return self._replace(**{'ancestors': ancestor})
 
+    def get_ancestors(self, towns):
+        """Iterator across ancestors of the current town."""
+        for town in towns.with_successors():
+            for successor_id in town.successors.split(';'):
+                if successor_id == self.id:
+                    yield town
+
     def add_successor(self, successor):
         """Append the given successor to the current list if any."""
         if self.successors:
@@ -192,6 +199,13 @@ def replace_successor(self, old_successor, new_successor):
         successors = self.successors.replace(old_successor, new_successor)
         return self._replace(**{'successors': successors.strip(';')})
 
+    def remove_successor(self, successor):
+        """Remove the given successor from the current list."""
+        successors = ';'.join([
+            succ for succ in self.successors.split(';')
+            if succ != successor])
+        return self._replace(**{'successors': successors})
+
     def clear_successors(self):
         """Remove all successors."""
         return self._replace(**{'successors': ''})

geohisto/utils.py

@@ -88,7 +88,13 @@ def compute_ancestors(towns):
         for successor_id in town.successors.split(';'):
             successor = towns.retrieve(successor_id)
             if successor:
-                successor = successor.add_ancestor(town.id)
-                towns.upsert(successor)
+                # Avoid weird parenthood relations.
+                if (town.id not in town.successors and
+                        town.end_datetime <= successor.end_datetime):
+                    successor = successor.add_ancestor(town.id)
+                    towns.upsert(successor)
+                else:
+                    town = town.remove_successor(successor.id)
+                    towns.upsert(town)
             else:
                 print('Successor not found for', town.repr_insee)

tests/test_actions.py

@@ -156,14 +156,24 @@ def test_change_name_reinstatement_after_fusion():
         dep='28', com='159', mod=SPLITING,
         effdate=date(1987, 1, 1),
         nccoff='Framboisière', comech='28368')
+    fusion_association_associated = record_factory(
+        dep='28', com='368', mod=FUSION_ASSOCIATION_ASSOCIATED,
+        effdate=date(1972, 12, 22), comech='28159',
+        nccoff='Saucelle')
+    reinstatement_record = record_factory(
+        dep='28', com='368', mod=REINSTATEMENT,
+        effdate=date(1987, 1, 1), comech='28159',
+        nccoff='Saucelle', nccanc='Framboisière-la-Saucelle')
     history = [
         change_name_fusion_record, change_name_reinstatement_record,
-        spliting_record
+        spliting_record, fusion_association_associated, reinstatement_record
     ]
     compute(towns, history)
     framb, framb_saucelle, framb2 = towns.filter(depcom='28159')
+    saucelle, saucelle2 = towns.filter(depcom='28368')
+    assert saucelle.successors == framb_saucelle.id
     assert framb_saucelle.id == 'COM28159@1972-12-22'
-    assert framb_saucelle.successors == framb2.id
+    assert framb_saucelle.successors == framb2.id + ';' + saucelle2.id
     assert framb_saucelle.modification == CHANGE_NAME_REINSTATEMENT
     assert framb.id == 'COM28159@1942-01-01'
     assert framb.successors == framb_saucelle.id