From d64948749e2f963f1938016764ce8256efd796dd Mon Sep 17 00:00:00 2001
From: Philip Sargent <philip.sargent@gmail.com>
Date: Sat, 18 Mar 2023 00:57:40 +0000
Subject: more QMs parsed correctly

---
 parsers/QMs.py | 31 ++++++++++++++++++++++++-------
 1 file changed, 24 insertions(+), 7 deletions(-)

(limited to 'parsers/QMs.py')

diff --git a/parsers/QMs.py b/parsers/QMs.py
index f3044ff..5a0be79 100644
--- a/parsers/QMs.py
+++ b/parsers/QMs.py
@@ -141,16 +141,26 @@ def parse_KH_QMs(kh, inputFile, ticked):
     with open(os.path.join(settings.EXPOWEB, inputFile), "r") as khQMfile:
         khQMs = khQMfile.readlines()
         nqms = 0
-        for line in khQMs:
+        line = 0
+        fails = 0
+        for dataline in khQMs:
             # <dt><a href="sibria.htm#qC1997-161-27" name="C1997-161-27">C1997-161-27</a> A<dd>Sib: pitch at end of Fuzzy Logic [Paradox Rift - continues] [sep.fuzzy.13]
+            line += 1
             res = re.search(
-                r"name=\"[CB](?P<year>\d*)-(?P<cave>\d*)-(?P<number>\d*).*</a> (?P<grade>[ABCDX])<dd>(?P<location_description>.*)\[(?P<nearest_station_name>.*)\]",
-                line,
+                r"name=\"[CB](?P<year>\d*)-(?P<cave>\d*)-(?P<number>\d*).*</a>\s*(?P<grade>[ABCDX?V])<dd>(?P<location_description>.*)(\[(?P<station_name>.*)\])?",
+                dataline,
             )
             if res:
                 res = res.groupdict()
                 year = int(res["year"])
-               
+                
+                nearest_station_name = ""
+                resolution_station_name = ""
+                if res["station_name"]:
+                    if ticked:
+                        resolution_station_name = res["station_name"].replace("<a href=\"","<a href=\"/1623/161/")
+                    else:
+                        nearest_station_name =  res["station_name"]
                 lookupAttribs = {
                     #'found_by':placeholder,
                     "blockname": "",
@@ -163,13 +173,20 @@ def parse_KH_QMs(kh, inputFile, ticked):
                     "ticked": ticked,
                     "page_ref": "",
                     "completion_description": "",
-                    "nearest_station_name": res["nearest_station_name"],
-                    "location_description": res["location_description"],
+                    "nearest_station_name": nearest_station_name,
+                    "resolution_station_name": resolution_station_name,
+                    "location_description": res["location_description"].replace("<a href=\"","<a href=\"/1623/161/"),
                 }
                 # Create new. We know it doesn't exist as we deleted evrything when we started.
                 instance = QM.objects.create(**nonLookupAttribs, **lookupAttribs)
-                
                 nqms += 1
+            else:
+                if dataline.startswith("<dt><a href"):
+                    fails += 1
+                    message = f" ! - {inputFile} line {line} Parse error \n{str(dataline)} "
+                    print(message)
+                    DataIssue.objects.create(parser="QMs", message=message)
+    print(f" - {fails:2g} parsing errors in {inputFile}")
     return nqms
 
 
-- 
cgit v1.2.3