finding the right survex file for a cave

1 files changed, 39 insertions, 12 deletions

diff --git a/parsers/caves.py b/parsers/caves.py
index e7eebb4..391501e 100644
--- a/parsers/caves.py
+++ b/parsers/caves.py
@@ -1,7 +1,7 @@
 import os
 import re
 import string
-
+import sys
 
 from pathlib import Path
 
@@ -103,15 +103,37 @@ def set_dummy_entrance(id, slug, cave, msg="DUMMY"):
         # print(message)
 
     
-def create_new_cave(svxpath, msg=None):
+def create_new_cave(svxpath, svxid=None, msg=None):
     """This is called only when a new survex file is edited online which has a path on the
     :loser: repo which is not recognised as a known cave.
     ALSO called by survex parser when it finds a cave it doesn't recognise
     """
-    # e.g. svxpath = "caves-1623/666/beast" .svx
+    # e.g. svxpath = "caves-1623/666/beast" .svx - from the *inlcude tree
+    # e.g. svxid = "caves-1623/666/beast"
     print(f"Create new cave at {svxpath} - {msg}")
     #
-    survex_file = svxpath.replace("caves-","") + ".svx"
+    survex_file = ""
+    if svxid:
+        sv = Path(settings.SURVEX_DATA, svxid + ".svx")
+        if sv.is_file:
+            survex_file = svxid + ".svx"
+    else:
+        sv = Path(settings.SURVEX_DATA, svxpath + ".svx")
+        if sv.is_file:
+            survex_file = svxpath + ".svx"
+        
+    if survex_file: 
+        # message = f"Found a survex file {survex_file=} {svxpath=} {svxid=} "   
+        # DataIssue.objects.create(parser="caves", message=message)
+        # print(message, file=sys.stderr) 
+        # print(message)
+        pass
+    else:
+        message = f"NOT found a survex file {svxpath=} {svxid=}"   
+        DataIssue.objects.create(parser="caves", message=message)
+        print(message, file=sys.stderr) 
+        print(message)
+      
     parts = svxpath.split("/")
     a = parts[0][-4:]
     caveid = parts[1]
@@ -142,9 +164,8 @@ def create_new_cave(svxpath, msg=None):
         print(message)
         raise    
     
-    # we know what the survex file is, we don't need to use the guess.
-    # But this sets the survex file on  the cave from the first one we find, not necessarily the best survex file for this cave
-    cave.survex_file=survex_file
+    # But this sets the survex file on  the Cave from the first one we find, not necessarily the best survex file for this cave
+    cave.survex_file=survex_file # primary survex file for Cave
     cave.areacode=areacode
     cave.save()
     return cave
@@ -230,6 +251,9 @@ def do_pending_cave(slug, caveid, url, areacode, msg=None):
         """Guesses at and finds a survex file for this pending cave.
         Convoluted. Needs rewriting.
         Pointless if this cave is being created because we found a survex file...
+        
+        On problem is that the Cave name may have different capitalisation from the survex filename,
+        e.g. 2018-NTU-02 has a survex file 2018-ntu-02.svx
         """
         if k[0:3] == "162":
             id = Path(k[5:])
@@ -244,6 +268,7 @@ def do_pending_cave(slug, caveid, url, areacode, msg=None):
             if Path(settings.SURVEX_DATA, survex_file).is_file():
                 return survex_file
 
+        # This should find the file even if the capitalisation is different, or if the directory name is totaly different
         survex_file = ""
         d = Path(settings.SURVEX_DATA, f"caves-{areacode}/{id}")
         if d.is_dir():
@@ -257,10 +282,12 @@ def do_pending_cave(slug, caveid, url, areacode, msg=None):
                         prime_suspect = survex_file
             if prime_suspect:
                 survex_file = prime_suspect
-            # message = f" ! {k:14} Found a survex file which might be the right one: {survex_file} - {msg}"
-            # DataIssue.objects.create(parser='caves', message=message, url=url)
-            # print(message)
-        return survex_file
+                # message = f" ! {k:14} Found a survex file which might be the right one: {survex_file} - {msg}"
+                # DataIssue.objects.create(parser='caves', message=message, url=url)
+                # print(message)
+                if Path(settings.SURVEX_DATA, survex_file).is_file():
+                    return survex_file
+        return ""
 
     g = GetCaveLookup()
     with transaction.atomic():
@@ -324,7 +351,7 @@ def do_pending_cave(slug, caveid, url, areacode, msg=None):
         )
         if cave:
             cave.save()  # must save to have id before foreign keys work. This is also a ManyToMany key.
-            message = f" ! {slug:18} Pending cave write-up url: {url} - {msg}"
+            message = f" ! {slug:18} Pending cave write-up url: {url} - {survex_file=} - {msg}"
             DataIssue.objects.create(parser="caves", message=message, url=url)
             print(message)