From 88833941b5fec8e2dcfb129301455044a33e850b Mon Sep 17 00:00:00 2001
From: Philip Sargent <philip.sargent@gmail.com>
Date: Sat, 22 Nov 2025 14:32:16 +0200
Subject: [PATCH] nearly parsed JSON into correct form for database

---
 parsers/logbooks.py | 31 ++++++++++++++++++++++++++-----
 1 file changed, 26 insertions(+), 5 deletions(-)

diff --git a/parsers/logbooks.py b/parsers/logbooks.py
index e8239a4..bc2fc56 100644
--- a/parsers/logbooks.py
+++ b/parsers/logbooks.py
@@ -523,6 +523,8 @@ def parser_html(year, expedition, txt, seq=""):
    
         entry = LogbookEntryData(ldate, place, tripcave, triptitle, tripcontent, trippersons, author, guests, expedition, tu, tid)
         logentries.append(entry)
+        if str(ldate) == "2025-07-08":
+            print(f"PARSED from html\n",entry,"\n")
     return logentries
 
 
@@ -656,10 +658,9 @@ def parse_logbook_for_expedition(expedition, blog=False):
     """
     def load_from_json():
         entries = []
+        logentries = []
         for jsonfile in json_entries_dir.rglob("*.json"):
             with open(jsonfile, 'r', encoding='utf-8') as json_f:
-                print(f"OPENING {jsonfile}")
-                
                 message = ""
                 try:
                     entrydict = json.load(json_f)
@@ -674,10 +675,30 @@ def parse_logbook_for_expedition(expedition, blog=False):
                     DataIssue.objects.update_or_create(parser="logbooks", message=message, url=jsonurl)
                     return None  
                 entries.append(entrydict)
+                logentries.append(parse_from_dict(entrydict))
         
         check_number_of_entries(entries) # temp check on pre-parsed list
-        return entries
+        check_number_of_entries(logentries) 
+        return logentries
+
+    def parse_from_dict(entrydict):
+        ldate = entrydict["date"]
+        place = entrydict["place"]
+        tripcave = entrydict["cave"]
+        triptitle = entrydict["title"]
+        tripcontent = entrydict["text"]
+        trippersons = None
+        author = entrydict["author"]["slug"]
+        guests = entrydict["other_people"]
+        expedition = entrydict["expedition"]
+        tu = entrydict["time_underground"]
+        tid = entrydict["slug"] # Is this right ? Or is it the end txt?
+
+        logentry = LogbookEntryData(ldate, place, tripcave, triptitle, tripcontent, trippersons, author, guests, expedition, tu, tid)
+        if ldate == "2025-07-08":
+            print(f"PARSED from JSON\n",logentry,"\n")
         
+        return logentry
         
     def check_number_of_entries(logentries):
         if logentries:
@@ -700,11 +721,11 @@ def parse_logbook_for_expedition(expedition, blog=False):
     json_entries_dir = settings.EXPOWEB / "years" / year / "log_entries"
 
     if json_entries_dir.is_dir():
-        print(f" # WARNING year {year} has JSON-encoded logbook entries. Should use these instead of the archive .html file")
+        print(f" # WARNING year {year} has JSON-encoded logbook entries. Using these instead of the archive .html file.")
         logentries = load_from_json()
         
         logentries = []
-        # check_number_of_entries()
+        # check_number_of_entries(logentries)
         # return logentries