diff options
Diffstat (limited to 'parsers/logbooks.py')
-rw-r--r-- | parsers/logbooks.py | 39 |
1 files changed, 3 insertions, 36 deletions
diff --git a/parsers/logbooks.py b/parsers/logbooks.py index 4554b08..22674c6 100644 --- a/parsers/logbooks.py +++ b/parsers/logbooks.py @@ -282,7 +282,7 @@ def LoadLogbookForExpedition(expedition): """ Parses all logbook entries for one expedition """ expowebbase = os.path.join(settings.EXPOWEB, "years") - year = str(expedition.year) + #year = str(expedition.year) yearlinks = settings.LOGBOOK_PARSER_SETTINGS logbook_parseable = False @@ -311,50 +311,17 @@ def LoadLogbookForExpedition(expedition): parser(expedition.year, expedition, txt) SetDatesFromLogbookEntries(expedition) - return "TOLOAD: " + year + " " + str(expedition.personexpedition_set.all()[1].logbookentry_set.count()) + " " + str(models.PersonTrip.objects.filter(personexpedition__expedition=expedition).count()) + #return "TOLOAD: " + year + " " + str(expedition.personexpedition_set.all()[1].logbookentry_set.count()) + " " + str(models.PersonTrip.objects.filter(personexpedition__expedition=expedition).count()) def LoadLogbooks(): """ This is the master function for parsing all logbooks into the Troggle database. Requires yearlinks, which is a list of tuples for each expedition with expedition year, logbook path, and parsing function. """ - - #Deletion has been moved to a seperate function to enable the non-destructive importing - #models.LogbookEntry.objects.all().delete() - expowebbase = os.path.join(settings.EXPOWEB, "years") - #yearlinks = [ ("2001", "2001/log.htm", Parseloghtml01), ] #overwrite - #yearlinks = [ ("1996", "1996/log.htm", Parseloghtml01),] # overwrite - - yearlinks = settings.LOGBOOK_PARSER_SETTINGS expos = models.Expedition.objects.all() for expo in expos: print("\nLoading Logbook for: " + expo.year) - logbook_parseable = False - - if expo.year in yearlinks: - #print(yearlinks[expo.year]) - year_settings = yearlinks[expo.year] - file_in = open(os.path.join(expowebbase, year_settings[0])) - txt = file_in.read().decode("latin1") - file_in.close() - parsefunc = year_settings[1] - logbook_parseable = True - else: - try: - file_in = open(os.path.join(expowebbase, expo.year, settings.DEFAULT_LOGBOOK_FILE)) - txt = file_in.read().decode("latin1") - file_in.close() - logbook_parseable = True - print("No set parser found using default") - parsefunc = settings.DEFAULT_LOGBOOK_PARSER - except (IOError): - logbook_parseable = False - print("Couldn't open default logbook file and nothing in settings for expo " + expo.year) - - if logbook_parseable: - parser = globals()[parsefunc] - parser(expo.year, expo, txt) - SetDatesFromLogbookEntries(expo) + LoadLogbookForExpedition(expo) dateRegex = re.compile(r'<span\s+class="date">(\d\d\d\d)-(\d\d)-(\d\d)</span>', re.S) |