[RP] add school_type (#153)

* [RP] add school_type * using walrus to make code more concise Co-authored-by: Knut Hühne <[email protected]> --------- Co-authored-by: Knut Hühne <[email protected]>
Datenschule · Mar 5, 2025 · 9d7acf2 · 9d7acf2
1 parent bbcf299
commit 9d7acf2
Showing 1 changed file with 35 additions and 0 deletions.
diff --git a/jedeschule/spiders/rheinland_pfalz.py b/jedeschule/spiders/rheinland_pfalz.py
@@ -7,6 +7,29 @@
 from jedeschule.items import School
 from jedeschule.spiders.school_spider import SchoolSpider
 
+school_types = {
+    'BEA': 'BEA',  # I could not find the meaning of this abbreviation
+    'BBS': 'Berufsbildende Schule',
+    'FWS': 'Freie Waldorfschule',
+    'GHS': 'Grund- und Hauptschule (org. verbunden)',
+    'GRS+': 'Grund- und Realschule plus (org. verbunden)',
+    'GS': 'Grundschule',
+    'GY': 'Gymnasium',
+    'HS': 'Hauptschule',
+    'IGS': 'Integrierte Gesamtschule',
+    'Koll': 'Kolleg',
+    'Koll/AGY': 'Kolleg und Abendgymnasium (org.verbunden)',
+    'RS': 'Realschule',
+    'RS+': 'Realschule plus',
+    'RS+FOS': 'Realschule plus mit Fachoberschule',
+    'StudSem': 'Studienseminar'
+
+    # Förderschulen (special education schools) come in a variety of abbreviations
+    # The following are some examples from the dataset
+    # SFGLS, SFG, SFGM, SFE, SFL, SFLG, SFBLS, SFMG, SFLS
+    # so we will treat them a bit differently, see below in the normalize step
+}
+
 
 class RheinlandPfalzSpider(CrawlSpider, SchoolSpider):
     name = "rheinland-pfalz"
@@ -43,6 +66,17 @@ def parse_school(self, response):
     def normalize(self, item: Item) -> School:
         zip, city = item.get("Anschrift")[-1].split(" ", 1)
         email = item.get("E-Mail", "").replace("(at)", "@")
+
+        if kurzbezeichnung := item.get('Kurzbezeichnung'):
+            first_part = kurzbezeichnung.split(" ")[0]
+            # special handling for special education schools
+            if first_part.startswith('SF'):
+                school_type = 'Förderschule'
+            else:
+                school_type = school_types.get(first_part, None)
+        else:
+            school_type = None
+
         return School(
             name=item.get("name"),
             id="RP-{}".format(item.get("id")),
@@ -56,4 +90,5 @@ def normalize(self, item: Item) -> School:
             provider=item.get("Träger"),
             fax=item.get("Telefax"),
             phone=item.get("Telefon"),
+            school_type=school_type
         )