more cave parsing data fixes

1 files changed, 13 insertions, 10 deletions

diff --git a/parsers/caves.py b/parsers/caves.py
index cce680d..ba8f60e 100644
--- a/parsers/caves.py
+++ b/parsers/caves.py
@@ -45,7 +45,7 @@ def readcaves():
         
         # For those caves which do not have cave_data/1623-xxx.html XML files even though they exist and have surveys
         # also needs to be done *before* entrances so that the entrance-cave links work properly.
-        pending = ["2007-04", "2007-05", "2007-06", "2007-07", "2007-12", "2009-01", "2009-02", 
+        pending = ["2007-05", "2007-06", "2007-12", "2009-01", "2009-02", 
                 "2010-06", "2010-07", "2012-ns-01", "2012-ns-02", "2010-04", "2012-ns-05", "2012-ns-06", 
                 "2012-ns-07", "2012-ns-08", "2012-ns-12", "2012-ns-14", "2012-ns-15", "2014-bl888", 
                 "2018-pf-01", "2018-pf-02", "haldenloch"]
@@ -58,13 +58,16 @@ def readcaves():
                         underground_description = "Pending cave write-up - creating as empty object. No XML file available yet.",
                         survex_file = "caves-1623/" + k + "/" + k +".svx",
                         url = url,
-                        notes="_Survex file found in loser repo but no description in expoweb")
+                        notes="_Survex file found in loser repo but no description in expoweb <br>\n"+
+                            "INSTRUCTIONS: First open 'This survex file' (link above the CaveView panel) to find the date and info. Then <br>\n" +
+                            "search in the Expo for that year e.g. <a href='/expedition/2007'>2007</a> to find a relevant logbook entry, then <br>\n" +
+                            "click on 'New Entrance' at the bottom of this page as we need to create the entrance *first*.")
                 if cave:
                     cave.save() # must save to have id before foreign keys work. This is also a ManyToMany key.
                     #print(f' ! - READ CAVES:  cave {k} {cave}')
                     cave.area.add(area_1623[0])
                     cave.save()
-                    message = " ! {:11s} {}".format(cave.unofficial_number, cave.underground_description)
+                    message = f" ! {k:12} {cave.underground_description}"
                     DataIssue.objects.create(parser='caves', message=message, url=url)
                     print(message)
                     
@@ -108,7 +111,7 @@ def readentrance(filename):
     # Note: these are HTML files in the EXPOWEB repo, not from the loser repo.
     with open(os.path.join(settings.ENTRANCEDESCRIPTIONS, filename)) as f:
         contents = f.read()
-    context = "in file %s" % filename
+    context = filename
     #print("Reading file ENTRANCE {} / {}".format(settings.ENTRANCEDESCRIPTIONS, filename))
     entrancecontentslist = getXML(contents, "entrance", maxItems = 1, context = context)
     if len(entrancecontentslist) != 1:
@@ -199,7 +202,7 @@ def readcave(filename):
     # Note: these are HTML files in the EXPOWEB repo, not from the loser repo.
     with open(os.path.join(settings.CAVEDESCRIPTIONS, filename)) as f:
         contents = f.read()
-    context = " in file %s" % filename
+    context = filename
     cavecontentslist = getXML(contents, "cave", maxItems = 1, context = context)
     if len(cavecontentslist) != 1:
         message = f'! BAD CAVE at "{filename}"'
@@ -310,13 +313,13 @@ def readcave(filename):
                     
             if survex_file[0]:
                 if not (Path(SURVEX_DATA) / survex_file[0]).is_file():
-                    message = f' ! survex filename does not exist :LOSER:"{survex_file[0]}" in "{filename}"'
+                    message = f' ! {slug:12} survex filename does not exist :LOSER:"{survex_file[0]}" in "{filename}"'
                     DataIssue.objects.create(parser='caves', message=message, url=f'/cave/{slug}/edit/')
                     print(message)
                     
             if description_file[0]:
                 if not (Path(EXPOWEB) / description_file[0]).is_file():
-                    message = f' ! description filename does not exist :{EXPOWEB}:"{description_file[0]}" in "{filename}"'
+                    message = f' ! {slug:12} description filename does not exist :{EXPOWEB}:"{description_file[0]}" in "{filename}"'
                     DataIssue.objects.create(parser='caves', message=message, url=f'/cave/{slug}/edit/')
                     print(message)
                     #c.description_file="" # done only once, to clear out cruft.
@@ -327,14 +330,14 @@ def getXML(text, itemname, minItems = 1, maxItems = None, printwarnings = True,
     if len(items) < minItems and printwarnings:
         message = " ! %(count)i %(itemname)s found, at least %(min)i expected" % {"count": len(items),
                                                                            "itemname": itemname,
-                                                                           "min": minItems} + context
-        DataIssue.objects.create(parser='caves', message=message)
+                                                                           "min": minItems} + " in file " + context
+        DataIssue.objects.create(parser='caves', message=message, url=""+context)
         print(message)
         
     if maxItems is not None and len(items) > maxItems and printwarnings:
         message = " ! %(count)i %(itemname)s found, no more than %(max)i expected in this XML unit " % {"count": len(items),
                                                                                "itemname": itemname,
-                                                                               "max": maxItems} + context
+                                                                               "max": maxItems} + " in file " + context
         DataIssue.objects.create(parser='caves', message=message)
         print(message)
     return items
\ No newline at end of file