Feature/search hyphenate addresses

requirements.in

@@ -7,7 +7,7 @@ django-modeltranslation
 flake8
 requests
 requests_cache
-git+https://github.com/City-of-Helsinki/django-munigeo@v0.2.76#egg=django-munigeo
+git+https://github.com/City-of-Helsinki/django-munigeo@v0.2.83#egg=django-munigeo
 pytz
 django-cors-headers
 django-extensions

requirements.txt

@@ -87,7 +87,7 @@ django-mptt==0.13.4
     # via
     #   -r requirements.in
     #   django-munigeo
-django-munigeo @ git+https://github.com/City-of-Helsinki/django-munigeo@v0.2.76
+django-munigeo @ git+https://github.com/City-of-Helsinki/django-munigeo@v0.2.83
     # via -r requirements.in
 django-polymorphic==3.1.0
     # via -r requirements.in

services/fixtures/exclusion_words.json

@@ -0,0 +1,18 @@
+[
+    {
+        "model": "services.exclusionword",
+        "pk": 1,
+        "fields": {
+            "word": "katu",
+            "language_short": "fi"
+        }
+    },
+    {
+        "model": "services.exclusionword",
+        "pk": 2,
+        "fields": {
+            "word": "tie",
+            "language_short": "fi"
+        }
+    }
+    ]

services/management/commands/index_search_columns.py

@@ -1,11 +1,14 @@
 import logging
+from datetime import datetime, timedelta
 
 from django.contrib.postgres.search import SearchVector
 from django.core.management.base import BaseCommand
+from django.utils import timezone
 from munigeo.models import Address, AdministrativeDivision
 
 from services.models import Service, ServiceNode, Unit
-from services.search.utils import hyphenate
+from services.search.constants import HYPHENATE_ADDRESSES_MODIFIED_WITHIN_DAYS
+from services.search.utils import get_foreign_key_attr, hyphenate
 
 logger = logging.getLogger("search")
 
@@ -27,28 +30,42 @@ def get_search_column(model, lang):
     return search_column
 
 
-def generate_syllables(model):
+def generate_syllables(
+    model, hyphenate_all_addresses=False, hyphenate_addresses_from=None
+):
     """
     Generates syllables for the given model.
     """
     # Disable sending of signals
     model._meta.auto_created = True
+    save_kwargs = {}
     num_populated = 0
-    for row in model.objects.all():
+    if model.__name__ == "Address" and not hyphenate_all_addresses:
+        save_kwargs["skip_modified_at"] = True
+        if not hyphenate_addresses_from:
+            hyphenate_addresses_from = Address.objects.latest(
+                "modified_at"
+            ).modified_at - timedelta(days=HYPHENATE_ADDRESSES_MODIFIED_WITHIN_DAYS)
+        qs = model.objects.filter(modified_at__gte=hyphenate_addresses_from)
+    else:
+        qs = model.objects.all()
+    for row in qs:
         row.syllables_fi = []
         for column in model.get_syllable_fi_columns():
-            row_content = getattr(row, column, None)
+            row_content = get_foreign_key_attr(row, column)
             if row_content:
                 # Rows migth be of type str or Array, if str
                 # cast to array by splitting.
                 if isinstance(row_content, str):
                     row_content = row_content.split()
                 for word in row_content:
                     syllables = hyphenate(word)
-                    for s in syllables:
-                        row.syllables_fi.append(s)
-            row.save()
+                    if len(syllables) > 1:
+                        for s in syllables:
+                            row.syllables_fi.append(s)
+            row.save(**save_kwargs)
             num_populated += 1
+
     # Enable sending of signals
     model._meta.auto_created = False
     return num_populated
@@ -85,13 +102,44 @@ def index_servicenodes(lang):
 
 
 class Command(BaseCommand):
-    def handle(self, *args, **kwargs):
+
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "--hyphenate_addresses_from",
+            nargs="?",
+            type=str,
+            help="Hyphenate addresses whose modified_at timestamp starts at given timestamp YYYY-MM-DDTHH:MM:SS",
+        )
+
+        parser.add_argument(
+            "--hyphenate_all_addresses",
+            action="store_true",
+            help="Hyphenate all addresses",
+        )
+
+    def handle(self, *args, **options):
+        hyphenate_all_addresses = options.get("hyphenate_all_addresses", None)
+        hyphenate_addresses_from = options.get("hyphenate_addresses_from", None)
+
+        if hyphenate_addresses_from:
+            try:
+                hyphenate_addresses_from = timezone.make_aware(
+                    datetime.strptime(hyphenate_addresses_from, "%Y-%m-%dT%H:%M:%S")
+                )
+            except ValueError as err:
+                raise ValueError(err)
         for lang in ["fi", "sv", "en"]:
             key = "search_column_%s" % lang
             # Only generate syllables for the finnish language
             if lang == "fi":
                 logger.info(f"Generating syllables for language: {lang}.")
                 logger.info(f"Syllables generated for {generate_syllables(Unit)} Units")
+                num_populated = generate_syllables(
+                    Address,
+                    hyphenate_all_addresses=hyphenate_all_addresses,
+                    hyphenate_addresses_from=hyphenate_addresses_from,
+                )
+                logger.info(f"Syllables generated for {num_populated} Addresses")
                 logger.info(
                     f"Syllables generated for {generate_syllables(Service)} Services"
                 )

services/migrations/0101_exclusionword.py

@@ -0,0 +1,37 @@
+# Generated by Django 4.1.13 on 2024-05-13 10:52
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("services", "0100_alter_unitconnection_section_type"),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name="ExclusionWord",
+            fields=[
+                (
+                    "id",
+                    models.AutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                ("word", models.CharField(max_length=100, verbose_name="Word")),
+                (
+                    "language_short",
+                    models.CharField(max_length=2, verbose_name="Language short"),
+                ),
+            ],
+            options={
+                "verbose_name": "Exclusion word",
+                "verbose_name_plural": "Exclusion words",
+                "ordering": ["-id"],
+            },
+        ),
+    ]

services/models/__init__.py

@@ -2,7 +2,7 @@
 from .department import Department
 from .keyword import Keyword
 from .notification import Announcement, ErrorMessage
-from .search_rule import ExclusionRule
+from .search_rule import ExclusionRule, ExclusionWord
 from .service import Service, UnitServiceDetails
 from .service_mapping import ServiceMapping
 from .service_node import ServiceNode

services/models/search_rule.py

@@ -13,3 +13,16 @@ class Meta:
 
     def __str__(self):
         return "%s : %s" % (self.word, self.exclusion)
+
+
+class ExclusionWord(models.Model):
+    word = models.CharField(max_length=100, verbose_name=_("Word"))
+    language_short = models.CharField(max_length=2, verbose_name=_("Language short"))
+
+    class Meta:
+        ordering = ["-id"]
+        verbose_name = _("Exclusion word")
+        verbose_name_plural = _("Exclusion words")
+
+    def __str__(self):
+        return self.word

services/search/api.py

@@ -27,9 +27,10 @@
 from drf_spectacular.utils import extend_schema, OpenApiParameter
 from munigeo import api as munigeo_api
 from munigeo.models import Address, AdministrativeDivision
-from rest_framework import serializers
+from rest_framework import serializers, status
 from rest_framework.exceptions import ParseError
 from rest_framework.generics import GenericAPIView
+from rest_framework.response import Response
 
 from services.api import (
     TranslatedModelSerializer,
@@ -60,6 +61,7 @@
     get_search_exclusions,
     get_service_node_results,
     get_trigram_results,
+    has_exclusion_word_in_query,
     set_address_fields,
     set_service_node_unit_count,
     set_service_unit_count,
@@ -331,7 +333,6 @@ def get(self, request):
             raise ParseError("Supply search terms with 'q=' ' or input=' '")
 
         if not re.match(r"^[\w\såäö.'+&|-]+$", q_val):
-
             raise ParseError(
                 "Invalid search terms, only letters, numbers, spaces and .'+-&| allowed."
             )
@@ -447,6 +448,13 @@ def get(self, request):
                     search_query_str += f"& {q}:*"
             else:
                 search_query_str = f"{q}:*"
+
+        if has_exclusion_word_in_query(q_vals, language_short):
+            return Response(
+                f"Search query {q_vals} would return too many results",
+                status=status.HTTP_400_BAD_REQUEST,
+            )
+
         search_fn = "to_tsquery"
         if use_websearch:
             exclusions = get_search_exclusions(q)

services/search/constants.py

@@ -11,9 +11,11 @@
     "Address",
 )
 QUERY_PARAM_TYPE_NAMES = [m.lower() for m in SEARCHABLE_MODEL_TYPE_NAMES]
-# None will slice to the end of list, e.g. no limit.
+# None will slice to the end of list, i.e., no limit.
 DEFAULT_MODEL_LIMIT_VALUE = None
 # The limit value for the search query that search the search_view. "NULL" = no limit
 DEFAULT_SEARCH_SQL_LIMIT_VALUE = "NULL"
 DEFAULT_TRIGRAM_THRESHOLD = 0.15
 DEFAULT_RANK_THRESHOLD = 1
+
+HYPHENATE_ADDRESSES_MODIFIED_WITHIN_DAYS = 7

services/search/tests/conftest.py

@@ -24,6 +24,7 @@
 from services.models import (
     Department,
     ExclusionRule,
+    ExclusionWord,
     Service,
     ServiceNode,
     Unit,
@@ -277,6 +278,15 @@ def addresses(streets, municipality):
         number=1,
         full_name="Tarkk'ampujankatu 1",
     )
+    Address.objects.create(
+        municipality_id=municipality.id,
+        location=Point(60.44879002342721, 22.283629416961055),
+        id=7,
+        street_id=46,
+        number=1,
+        full_name="Kellonsoittajankatu 1",
+    )
+    generate_syllables(Address)
     Address.objects.update(search_column_fi=get_search_column(Address, "fi"))
     return Address.objects.all()
 
@@ -314,10 +324,17 @@ def streets():
     Street.objects.create(id=43, name="Markulantie", municipality_id="turku")
     Street.objects.create(id=44, name="Yliopistonkatu", municipality_id="turku")
     Street.objects.create(id=45, name="Tarkk'ampujankatu", municipality_id="turku")
+    Street.objects.create(id=46, name="Kellonsoittajankatu", municipality_id="turku")
     return Street.objects.all()
 
 
 @pytest.fixture
 def exclusion_rules():
     ExclusionRule.objects.create(id=1, word="tekojää", exclusion="-nurmi")
     return ExclusionRule.objects.all()
+
+
+@pytest.fixture
+def exclusion_words():
+    ExclusionWord.objects.create(id=1, word="katu", language_short="fi")
+    return ExclusionWord.objects.all()

services/search/tests/test_api.py

@@ -14,6 +14,7 @@ def test_search(
     accessibility_shortcoming,
     municipality,
     exclusion_rules,
+    exclusion_words,
 ):
     # Search for "museo" in entities: units,services and servicenods
     url = reverse("search") + "?q=museo&type=unit,service,servicenode"
@@ -121,6 +122,22 @@ def test_search(
     assert kurrapolku["location"]["type"] == "Point"
     assert kurrapolku["location"]["coordinates"][0] == 60.479032
     assert kurrapolku["location"]["coordinates"][1] == 22.25417
+    # Test search with excluded word
+    url = reverse("search") + "?q=katu"
+    response = api_client.get(url)
+    assert response.status_code == 400
+    url = reverse("search") + "?q=Katu"
+    response = api_client.get(url)
+    assert response.status_code == 400
+    url = reverse("search") + "?q=koti katu"
+    response = api_client.get(url)
+    assert response.status_code == 400
+    # Test search with 'kello'
+    url = reverse("search") + "?q=kello&type=address"
+    response = api_client.get(url)
+    results = response.json()["results"]
+    assert len(results) == 1
+    assert results[0]["name"]["fi"] == "Kellonsoittajankatu 1"
     # Test address search with apostrophe in query
     url = reverse("search") + "?q=tarkk'ampujankatu&type=address"
     response = api_client.get(url)