*team parsing much improved. Copes with everything now.

2 files changed, 41 insertions, 4 deletions

diff --git a/parsers/people.py b/parsers/people.py
index 6a4456b..0f3a66e 100644
--- a/parsers/people.py
+++ b/parsers/people.py
@@ -133,15 +133,32 @@ def who_is_this(year,possibleid):
         return personexpedition.person
     else:
         return None
+        
+def known_foreigner(id):
+    '''If this someone from ARGE or a known Austrian? Name has to be exact, no soft matching
+    '''
+    friends = ["P. Jeutter", "K. Jäger", "S. Steinberger", "R. Seebacher", 
+        "Dominik Jauch", "Fritz Mammel", "Marcus Scheuerman", 
+        "Uli Schütz", "Wieland Scheuerle",
+        "Kai Schwekend", "Regina Kaiser", "Thilo Müller","Wieland Scheuerle",
+        "Florian Gruner", "Helmut Stopka-Ebeler", "Aiko", "Mark Morgan"]
+
+    if id in friends:
+        return True
+    else:
+        return False
+
     
 # Refactor. The dict GetPersonExpeditionNameLookup(expo) indexes by name and has values of personexpedition
-# This is convoluted, the whole personexpedition concept is unnecessary.
+# This is convoluted, the whole personexpedition concept is unnecessary?
 
 Gpersonexpeditionnamelookup = { }
 def GetPersonExpeditionNameLookup(expedition):
     global Gpersonexpeditionnamelookup
     
     def apply_variations(f, l):
+        '''Be generous in guessing possible matches. Any duplicates will be ruled as invalid.
+        '''
         f = f.lower()
         l = l.lower()
         variations = []
@@ -151,8 +168,9 @@ def GetPersonExpeditionNameLookup(expedition):
         variations.append(f + " " + l)
         variations.append(f + " " + l[0])
         variations.append(f + l[0])
-        variations.append(f + l[0] + '.')
+        variations.append(f + " " +l[0] + '.')
         variations.append(f[0] + " " + l)
+        variations.append(f[0] + ". " + l)
         variations.append(f[0] + l)
         variations.append(f[0] + l[0]) # initials e.g. gb or bl
         return variations
@@ -188,25 +206,36 @@ def GetPersonExpeditionNameLookup(expedition):
                 
             if f == "Robert".lower():
                 possnames += apply_variations("Bob", l)
+            if f == "Rob".lower():
+                possnames += apply_variations("Robert", l)
+                
             if f == "Andrew".lower():
                 possnames += apply_variations("Andy", l)
             if f == "Andy".lower():
                 possnames += apply_variations("Andrew", l)
             if f == "Michael".lower():
                 possnames += apply_variations("Mike", l)
+                
             if f == "David".lower():
                 possnames += apply_variations("Dave", l)
             if f == "Dave".lower():
                 possnames += apply_variations("David", l)
+                
             if f == "Peter".lower():
                 possnames += apply_variations("Pete", l)
             if f == "Pete".lower():
                 possnames += apply_variations("Peter", l)
+                
             if f == "Olly".lower():
                 possnames += apply_variations("Oliver", l)
             if f == "Oliver".lower():
                 possnames += apply_variations("Olly", l)
                
+            if f == "Ollie".lower():
+                possnames += apply_variations("Oliver", l)
+            if f == "Oliver".lower():
+                possnames += apply_variations("Ollie", l)
+
             if f == "Becka".lower():
                 possnames += apply_variations("Rebecca", l)
          
diff --git a/parsers/survex.py b/parsers/survex.py
index 714ba67..8d6cb33 100644
--- a/parsers/survex.py
+++ b/parsers/survex.py
@@ -14,7 +14,7 @@ from django.utils.timezone import make_aware
 import troggle.settings as settings
 from troggle.core.models.caves import Cave, Entrance, QM, LogbookEntry
 from troggle.core.utils import get_process_memory, chaosmonkey
-from troggle.parsers.people import GetPersonExpeditionNameLookup
+from troggle.parsers.people import GetPersonExpeditionNameLookup, known_foreigner
 from troggle.parsers.logbooks import GetCaveLookup
 from troggle.core.models.troggle import DataIssue, Expedition
 from troggle.core.models.survex import SurvexPersonRole, Wallet, SurvexDirectory, SurvexFile, SurvexBlock, SurvexStation
@@ -225,7 +225,7 @@ class LoadingSurvex():
         (NB PersonTrip is a logbook thing)
         """
         def record_team_member(tm, survexblock):
-            tm = tm.strip('\"\'')
+            tm = tm.strip('\"\'').strip()
             # Refactor. The dict GetPersonExpeditionNameLookup(expo) indexes by name and has values of personexpedition
             # This is convoluted, the whole personexpedition concept is unnecessary.
             
@@ -251,6 +251,10 @@ class LoadingSurvex():
                     personrole.expeditionday = survexblock.expeditionday 
                     self.currentpersonexped.append(personexpedition) # used in push/pop block code
                     personrole.save()
+                elif known_foreigner(tm): # note, not using .lower()
+                    message = "- *team {} '{}' known foreigner on *team {} ({})  in '{}'".format(expo.year, tm, survexblock.survexfile.path, survexblock, line)
+                    print(self.insp+message)
+                    # DataIssue.objects.create(parser='survex', message=message, url=get_offending_filename(survexblock.survexfile.path))
                 else:
                     # we know the date and expo, but can't find the person
                     message = "! *team {} '{}' FAIL personexpedition lookup on *team {} ({})  in '{}'".format(expo.year, tm, survexblock.survexfile.path, survexblock, line)
@@ -377,6 +381,10 @@ class LoadingSurvex():
                             pr.person = pr.personexpedition.person
                             pr.save()
                             self.currentpersonexped.append(pe) # used in push/pop block code
+                        elif known_foreigner(pr.personname): # note, not using .lower()
+                            message = "- *team {} '{}' known foreigner on *date {} ({})  in '{}'".format(expo.year, pr.personname, survexblock.survexfile.path, survexblock, line)
+                            print(self.insp+message)
+                            # DataIssue.objects.create(parser='survex', message=message, url=get_offending_filename(survexblock.survexfile.path))
                         else:
                             message = "! *team {} '{}' FAIL personexpedition lookup on *date {} ({})  '{}'".format(year, pr.personname, survexblock.survexfile.path, survexblock, pr.personname)
                             print(self.insp+message)