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: [PATCH] more QMs parsed correctly

---
 parsers/QMs.py          | 31 ++++++++++++++++++++++++-------
 parsers/survex.py       | 11 ++---------
 templates/cave_qms.html | 12 ++++++------
 templates/qm.html       |  7 ++++---
 4 files changed, 36 insertions(+), 25 deletions(-)

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
 
 
diff --git a/parsers/survex.py b/parsers/survex.py
index 79e90d1..1e961c8 100644
--- a/parsers/survex.py
+++ b/parsers/survex.py
@@ -1313,7 +1313,7 @@ class LoadingSurvex:
             blockname = survexblock.name
         else:
             blockname = survexblock.name[:6] + survexblock.name[-1:]
-        # logslug = f'D{int(qmyear)}_{blockname}_{int(qm_no):03d}'
+            # logslug = f'D{int(qmyear)}_{blockname}_{int(qm_no):03d}'
         qm_ticked = False # default
         
         qm_no = qmline.group(1)  # this is NOT unique across multiple survex files
@@ -1340,14 +1340,7 @@ class LoadingSurvex:
         else:
             qm_ticked = True
             # print(f"{survexblock.survexfile.cave} {survexblock}:{qm_no}{qm_grade} {qmline.group(4)}", file=sys.stderr)
-        if resolution_station_name:
-            qm_ticked = True
-        # if qmline.group(6) and qmline.group(6) != "-":
-            # resolution_station_name = qmline.group(6)
-            # if qmline.group(7):
-                # resolution_station_name = resolution_station_name + "." + qmline.group(7)
-        # else:
-            # resolution_station_name = ""
+            
         qm_notes = qmline.group(5)
         # qm_notes = qmline.group(8)
 
diff --git a/templates/cave_qms.html b/templates/cave_qms.html
index f55ce9b..1839004 100644
--- a/templates/cave_qms.html
+++ b/templates/cave_qms.html
@@ -27,7 +27,7 @@
     <h3>Open leads</h3>
     <ul>{% for QM in cave.get_open_QMs %}
         <li><a href="{{QM.get_absolute_url}}">{{QM}}</a>
-        {% if QM.nearest_station_name %}&sect;<em>{{QM.nearest_station_name}}</em>&sect;{% endif %}
+        {% if QM.nearest_station_name %}&sect;<em>{{QM.nearest_station_name|safe}}</em>&sect;{% endif %}
 {{QM.location_description}} <b>{{QM.grade}}</b> 
         {% if QM.block %} <a href="/survexfile/{{QM.block.survexfile.path}}.svx">{{QM.block}}.svx</a> {{QM.block.date}} {% endif %}</li>
     {% empty %}
@@ -36,13 +36,13 @@
     </ul>
 
 &sect; QM.nearest_station_name<br />
-&star; QM.resolution_station_name<br />
+&star; QM.resolution_station_name - when ticked.<br />
     <h3>Ticked off</h3>
     <ul>{% for QM in cave.get_ticked_QMs %}
         <li><a href="{{QM.get_absolute_url}}">{{QM}}</a>
-        {% if QM.nearest_station_name %}&sect;<em>{{QM.nearest_station_name}}</em>&sect;{% endif %}
-        {% if QM.resolution_station_name %}&star;<em>{{QM.resolution_station_name}}</em>&star;{% endif %}
-        {{QM.location_description}} <b>{{QM.grade}}</b> 
+        {% if QM.nearest_station_name %}&sect;<em>{{QM.nearest_station_name|safe}}</em>&sect;{% endif %}
+        {% if QM.resolution_station_name %}<b>&star;<em>{{QM.resolution_station_name|safe}}</em>&star;</b>{% endif %}
+        {{QM.location_description|safe}} <b>{{QM.grade}}</b> 
         {% if QM.block %} <a href="/survexfile/{{QM.block.survexfile.path}}.svx">{{QM.block}}.svx</a> {{QM.block.date}} {% endif %}
         
             {% if QM.completion_description %}
@@ -57,5 +57,5 @@
     </ul>
 </p>
 &sect; QM.nearest_station_name<br />
-&star; QM.resolution_station_name<br />
+&star; QM.resolution_station_name - when ticked.<br />
 {% endblock %}
\ No newline at end of file
diff --git a/templates/qm.html b/templates/qm.html
index 1afaa15..d10e742 100644
--- a/templates/qm.html
+++ b/templates/qm.html
@@ -31,17 +31,18 @@ Tick off completion description: '{{qm.completion_description}}' </a> <br />
 Tick off url: <a href="/{{qm.get_completion_url}}">{{qm.get_completion_url}}</a> WRONG <br />
 Tick off date: '{{qm.completion_date}}' [For survex file ticks. not implemented yet]</a> <br />
 {% endif %}
+QM page_ref: '{{qm.page_ref}}'<br>
 
 {% if qm.cave %}
 <h3>Cave: {{qm.cave|safe }}</h3>
-<a href="{% url 'caveQMs' qm.cave|safe %}">All QMs on this cave </a>{{ qm.cave|safe }} <br>
-<a href="/{{ qm.cave.url }}"> Cave description </a> {{ qm.cave|safe }}
+{{ qm.cave|safe }}<a href="{% url 'caveQMs' qm.cave|safe %}">All QMs </a> <br>
+{{ qm.cave|safe }}<a href="/{{ qm.cave.url }}"> description </a> <br />
+{{ qm.cave }} <a href="/survexfile/{{ qm.cave.kataster_number }}"> survexfiles </a> <br />
 {% else %}
 <h3>This QM is OPEN</h3>
 Comment: '{{qm.comment}}'</a> <br />
 {% endif %}
 
-QM page_ref: '{{qm.page_ref}}'<br>
 
 <h3>QM Number: {{qm.number}} Grade: {{qm.grade}} </h3>