catch nonUTF8 survex files, DataIssues url editor

1 files changed, 13 insertions, 2 deletions

diff --git a/parsers/survex.py b/parsers/survex.py
index 70ff0d3..8baa9bf 100644
--- a/parsers/survex.py
+++ b/parsers/survex.py
@@ -1140,7 +1140,18 @@ class LoadingSurvex():
                 return
         self.svxfileslist.append(path)
 
-        svxlines = fin.read().splitlines()
+        try:
+            svxlines = fin.read().splitlines()
+        except UnicodeDecodeError:
+            # some bugger put an umlaut in a non-UTF survex file ?!
+            message = f"    ! ERROR *include file '{path}' in '{survexblock}' has UnicodeDecodeError"
+            print(message)
+            print(message,file=sys.stderr)
+            offendingfile = "/survexfile/" + path + ".svx"
+            DataIssue.objects.create(parser='survex', message=message,  url=offendingfile)
+            return # skip this survex file and all things *included in it
+       
+        
         for svxline in svxlines:
             self.lineno += 1
             thissvxline += 1
@@ -1153,7 +1164,7 @@ class LoadingSurvex():
                 print(message)
                 print(message,file=sys.stderr)
                 DataIssue.objects.create(parser='survex', message=message)
-                return # skip this survex file
+                return # skip this survex file and all things *included in it
 
             includestmt =self.rx_include.match(svxline)
             if not includestmt: