diff options
author | Martin Green <martin.speleo@gmail.com> | 2023-07-10 17:00:56 +0200 |
---|---|---|
committer | Martin Green <martin.speleo@gmail.com> | 2023-07-10 17:00:56 +0200 |
commit | 67361fa66c1aa2885ba90b68b9d9181f7a5b981b (patch) | |
tree | 256d4bcc22760cc396a09782df3760e345dc00da /parsers/logbooks.py | |
parent | b27852c1f3fd3280e36ca0973b34a64c361b536a (diff) | |
parent | 8ff438942dd0981d2dd442dae3e61fbd5a84280e (diff) | |
download | troggle-67361fa66c1aa2885ba90b68b9d9181f7a5b981b.tar.gz troggle-67361fa66c1aa2885ba90b68b9d9181f7a5b981b.tar.bz2 troggle-67361fa66c1aa2885ba90b68b9d9181f7a5b981b.zip |
Merge branch 'master' of ssh://expo.survex.com/home/expo/troggle
Diffstat (limited to 'parsers/logbooks.py')
-rw-r--r-- | parsers/logbooks.py | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/parsers/logbooks.py b/parsers/logbooks.py index e7c7363..5b9bb6e 100644 --- a/parsers/logbooks.py +++ b/parsers/logbooks.py @@ -57,7 +57,7 @@ LOGBOOK_PARSER_SETTINGS = { LOGBOOKS_DIR = "years" # subfolder of settings.EXPOWEB ENTRIES = { - "2023": 1, + "2023": 11, "2022": 90, "2019": 55, "2018": 95, @@ -295,7 +295,7 @@ def parser_html(year, expedition, txt, seq=""): headmatch = re.match(r"(?i)(?s).*<body[^>]*>(.*?)<hr.*", txt) headpara = headmatch.groups()[0].strip() - # print(f" - headpara:\n'{headpara}'") + #print(f" - headpara:\n'{headpara}'") if len(headpara) > 0: frontpath = Path(settings.EXPOWEB, LOGBOOKS_DIR, year, "frontmatter.html") with open(frontpath, "w") as front: @@ -305,7 +305,7 @@ def parser_html(year, expedition, txt, seq=""): endmatch = re.match(r"(?i)(?s).*<hr\s*/>([\s\S]*?)(?=</body)", txt) endpara = endmatch.groups()[0].strip() - # print(f" - endpara:\n'{endpara}'") + #print(f" - endpara:\n'{endpara}'") if len(endpara) > 0: endpath = Path(settings.EXPOWEB, LOGBOOKS_DIR, year, "endmatter.html") with open(endpath, "w") as end: @@ -333,7 +333,7 @@ def parser_html(year, expedition, txt, seq=""): if s: tripid, tripid1, tripdate, trippeople, triptitle, triptext, tu = s.groups() else: # allow title and people to be swapped in order - msg = f" !- {year} Can't parse:{logbook_entry_count} '{trippara[:50]}'..." + msg = f" !- {year} Can't parse:{logbook_entry_count} '{trippara[:55]}'...'{trippara}'" print(msg) DataIssue.objects.create(parser="logbooks", message=msg) @@ -353,7 +353,7 @@ def parser_html(year, expedition, txt, seq=""): tripid, tripid1, tripdate, triptitle, trippeople, triptext, tu = s2.groups() else: # if not re.search(r"Rigging Guide", trippara): - msg = f" !- Logbook. Can't parse entry on 2nd pass:{logbook_entry_count} '{trippara[:50]}'..." + msg = f" !- Logbook. Can't parse entry on 2nd pass:{logbook_entry_count} '{trippara[:55]}'...'{trippara}'" print(msg) DataIssue.objects.create(parser="logbooks", message=msg) continue |