From 5a90e7b727bd224bea05bbdafe7031c3f4356dc0 Mon Sep 17 00:00:00 2001
From: Philip Sargent <philip.sargent@gmail.com>
Date: Sat, 18 Mar 2023 03:03:06 +0000
Subject: fix more twiddly variations of QMs

---
 parsers/QMs.py | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)

(limited to 'parsers/QMs.py')

diff --git a/parsers/QMs.py b/parsers/QMs.py
index 5a0be79..56c97e4 100644
--- a/parsers/QMs.py
+++ b/parsers/QMs.py
@@ -147,20 +147,32 @@ def parse_KH_QMs(kh, inputFile, ticked):
             # <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>\s*(?P<grade>[ABCDX?V])<dd>(?P<location_description>.*)(\[(?P<station_name>.*)\])?",
+                r"name=\"[CB](?P<year>\d*)-(?P<cave>\d*)-(?P<number>\d*).*</a>\s*(?P<grade>[ABCDX?V])<dd>(?P<location_description>[^[]*)(\[\s*(?P<completion>[^]]*)\s*\])?\s*(\[\s*(?P<station_name>[^]]*)\s*\])?",
                 dataline,
             )
             if res:
                 res = res.groupdict()
                 year = int(res["year"])
                 
+                completion = res["completion"]
+                station_name = res["station_name"]
+                if not completion and station_name:
+                    if station_name.startswith("<a href"):
+                        completion = station_name
+                        station_name = ""
+                if completion and not station_name:
+                    if not ticked:
+                        station_name = completion
+                        
+                if completion:
+                    completion = completion.replace("<a href=\"","<a href=\"/1623/161/")
                 nearest_station_name = ""
                 resolution_station_name = ""
-                if res["station_name"]:
+                if station_name:
                     if ticked:
-                        resolution_station_name = res["station_name"].replace("<a href=\"","<a href=\"/1623/161/")
+                        resolution_station_name = station_name.replace("<a href=\"","<a href=\"/1623/161/")
                     else:
-                        nearest_station_name =  res["station_name"]
+                        nearest_station_name =  station_name.replace("<a href=\"","<a href=\"/1623/161/")
                 lookupAttribs = {
                     #'found_by':placeholder,
                     "blockname": "",
@@ -172,7 +184,7 @@ def parse_KH_QMs(kh, inputFile, ticked):
                 nonLookupAttribs = {
                     "ticked": ticked,
                     "page_ref": "",
-                    "completion_description": "",
+                    "completion_description": completion,
                     "nearest_station_name": nearest_station_name,
                     "resolution_station_name": resolution_station_name,
                     "location_description": res["location_description"].replace("<a href=\"","<a href=\"/1623/161/"),
-- 
cgit v1.2.3