initial refactoring

1 files changed, 59 insertions, 49 deletions

diff --git a/parsers/caves.py b/parsers/caves.py
index de7c8e8..2489d6e 100644
--- a/parsers/caves.py
+++ b/parsers/caves.py
@@ -274,8 +274,11 @@ def do_pending_cave(k, url, area):
 
 def readentrance(filename):
     """Reads an entrance description from the .html file
-    Convoluted. Sorry.This is as I inherited it and I haven't fiddled with it. Needs rewriting
     """
+    def getXMLmax1(field):
+        return getXML(entrancecontents, field, maxItems=1, context=context)
+        # return getXML(entrancecontents, field, maxItems=1, context=context)[0]
+        
     global entrances_xslug
     global caves_xslug
     global areas_xslug
@@ -287,36 +290,44 @@ def readentrance(filename):
     # print("Reading file ENTRANCE {} / {}".format(ENTRANCEDESCRIPTIONS, filename))
     entrancecontentslist = getXML(contents, "entrance", maxItems=1, context=context)
     if len(entrancecontentslist) != 1:
-        message = f'! BAD ENTRANCE at "{filename}"'
+        message = f'! BAD ENTRANCE at "{filename}". Loading aborted. '
         DataIssue.objects.create(parser="entrances", message=message)
         print(message)
     else:
         entrancecontents = entrancecontentslist[0]
-        non_public = getXML(entrancecontents, "non_public", maxItems=1, context=context)
-        name = getXML(entrancecontents, "name", maxItems=1, context=context)
-        slugs = getXML(entrancecontents, "slug", context=context)
-        entrance_description = getXML(entrancecontents, "entrance_description", maxItems=1, context=context)
-        explorers = getXML(entrancecontents, "explorers", maxItems=1, context=context)
-        map_description = getXML(entrancecontents, "map_description", maxItems=1, context=context)
-        location_description = getXML(entrancecontents, "location_description", maxItems=1, context=context)
         lastvisit = getXML(entrancecontents, "last visit date", maxItems=1, minItems=0, context=context)
-        approach = getXML(entrancecontents, "approach", maxItems=1, context=context)
-        underground_description = getXML(entrancecontents, "underground_description", maxItems=1, context=context)
-        photo = getXML(entrancecontents, "photo", maxItems=1, context=context)
-        marking = getXML(entrancecontents, "marking", maxItems=1, context=context)
-        marking_comment = getXML(entrancecontents, "marking_comment", maxItems=1, context=context)
-        findability = getXML(entrancecontents, "findability", maxItems=1, context=context)
-        findability_description = getXML(entrancecontents, "findability_description", maxItems=1, context=context)
-        alt = getXML(entrancecontents, "alt", maxItems=1, context=context)
-        northing = getXML(entrancecontents, "northing", maxItems=1, context=context)
-        easting = getXML(entrancecontents, "easting", maxItems=1, context=context)
-        tag_station = getXML(entrancecontents, "tag_station", maxItems=1, context=context)
-        exact_station = getXML(entrancecontents, "exact_station", maxItems=1, context=context)
-        other_station = getXML(entrancecontents, "other_station", maxItems=1, context=context)
-        other_description = getXML(entrancecontents, "other_description", maxItems=1, context=context)
-        bearings = getXML(entrancecontents, "bearings", maxItems=1, context=context)
-        url = getXML(entrancecontents, "url", maxItems=1, context=context)
+        slugs = getXML(entrancecontents, "slug", context=context)
         
+        alt = getXMLmax1("alt")
+        approach = getXMLmax1("approach")
+        bearings = getXMLmax1("bearings")
+        easting = getXMLmax1("easting")
+        entrance_description = getXMLmax1("entrance_description")
+        exact_station = getXMLmax1("exact_station")
+        explorers = getXMLmax1("explorers")
+        findability = getXMLmax1("findability")
+        findability_description = getXMLmax1("findability_description")
+        location_description = getXMLmax1("location_description")
+        map_description = getXMLmax1("map_description")
+        marking = getXMLmax1("marking")
+        marking_comment = getXMLmax1("marking_comment")
+        name = getXMLmax1("name")
+        non_public = getXMLmax1("non_public")
+        northing = getXMLmax1("northing")
+        other_description = getXMLmax1("other_description")
+        other_station = getXMLmax1("other_station")
+        photo = getXMLmax1("photo")
+        tag_station = getXMLmax1("tag_station")
+        underground_description = getXMLmax1("underground_description")
+        url = getXMLmax1("url")
+
+        if len(slugs) >1:
+            # Only ever one of these per entrance in the expo dataset
+            message = f" ! - More than one slug for an entrance: {entrance}, slugs: {slugs}. Aborting."
+            DataIssue.objects.create(parser="entrances", message=message, url=f"/cave/{slug}/edit/")
+            print(message)
+            return
+            
         e, state = Entrance.objects.update_or_create(
             name=name[0],
             non_public={
@@ -325,35 +336,31 @@ def readentrance(filename):
                 "true": True,
                 "false": False,
             }[non_public[0]],
+            alt=alt[0],
+            approach=approach[0],
+            bearings=bearings[0],
+            easting=easting[0],
             entrance_description=entrance_description[0],
+            exact_station=exact_station[0],
             explorers=explorers[0],
-            map_description=map_description[0],
-            location_description=location_description[0],
+            filename=filename,
+            findability=findability[0],
+            findability_description=findability_description[0],
             lastvisit=lastvisit[0],
-            approach=approach[0],
-            underground_description=underground_description[0],
-            photo=photo[0],
+            location_description=location_description[0],
+            map_description=map_description[0],
             marking=marking[0],
             marking_comment=marking_comment[0],
-            findability=findability[0],
-            findability_description=findability_description[0],
-            alt=alt[0],
             northing=northing[0],
-            easting=easting[0],
-            tag_station=tag_station[0],
-            exact_station=exact_station[0],
-            other_station=other_station[0],
             other_description=other_description[0],
-            bearings=bearings[0],
-            url=url[0],
-            filename=filename,
+            other_station=other_station[0],
+            photo=photo[0],
             slug=slugs[0],
+            tag_station=tag_station[0],
+            underground_description=underground_description[0],
+            url=url[0],
         )
-        if len(slugs) >1:
-            # Only ever one of these in the expo dataset
-            message = f" ! - More than one slug for an entrance: {entrance}, slugs: {slugs}. Aborting."
-            DataIssue.objects.create(parser="entrances", message=message, url=f"/cave/{slug}/edit/")
-            print(message)
+
 
 def readcave(filename):
     """Reads an enrance description from the .html file
@@ -538,10 +545,13 @@ def readcave(filename):
             print(message)
 
 
-def getXML(text, itemname, minItems=1, maxItems=None, printwarnings=True, context=""):
-    """Reads a single XML tag"""
+def getXML(text, itemname, minItems=1, maxItems=None, context=""):
+    """Reads a single XML tag
+    Should throw exception rather than producing error message here,
+    then handle exception in calling routine where it has the context.
+    """
     items = re.findall("<%(itemname)s>(.*?)</%(itemname)s>" % {"itemname": itemname}, text, re.S)
-    if len(items) < minItems and printwarnings:
+    if len(items) < minItems:
         message = (
             " ! %(count)i x %(itemname)s found, at least %(min)i expected. Load ABORT. "
             % {"count": len(items), "itemname": itemname, "min": minItems}
@@ -551,7 +561,7 @@ def getXML(text, itemname, minItems=1, maxItems=None, printwarnings=True, contex
         DataIssue.objects.create(parser="caves", message=message, url="" + context)
         print(message)
 
-    if maxItems is not None and len(items) > maxItems and printwarnings:
+    if maxItems is not None and len(items) > maxItems:
         message = (
             " ! %(count)i x %(itemname)s found, no more than %(max)i expected in this XML unit. Load ABORT. "
             % {"count": len(items), "itemname": itemname, "max": maxItems}