import csv import os import re # import pickle # import shelve import time from random import randint from datetime import datetime, date from pathlib import Path from django.conf import settings from django.template.defaultfilters import slugify from django.utils.timezone import get_current_timezone, make_aware from troggle.core.models.troggle import DataIssue, Expedition from troggle.core.utils import TROG, save_carefully from troggle.core.models.caves import Cave, LogbookEntry, PersonTrip, GetCaveLookup from parsers.people import GetPersonExpeditionNameLookup ''' Parses and imports logbooks in all their wonderful confusion # When we edit logbook entries, allow a "?" after any piece of data to say we've frigged it and # it can be checked up later from the hard-copy if necessary; or it's not possible to determin (name, trip place, etc) ''' todo=''' - refactor everything with some urgency, esp. LoadLogbookForExpedition() - profile the code to find bad repetitive things, of which there are many. - far too many uses of Django field dereferencing to get values, which is SLOW - Logbooks 1987, 1988, 1989 all crash on MySql - but not sqlite - with db constraint fail. Edit logbook to fix. - import/parse/re-export-as-html the 'artisanal-format' old logbooks so that we keep only a modern HTML05 format. Then we can retire the old parsers and reduce the volume of code here substantially. - rewrite to use generators rather than storing everything intermediate in lists - to reduce memory impact. - We should ensure logbook.html is utf-8 and stop this crap: file_in = open(logbookfile,'rb') txt = file_in.read().decode("latin1") - this is a slow and uncertain function: cave = getCaveByReference(caveRef) - use Fixtures https://docs.djangoproject.com/en/4.1/ref/django-admin/#django-admin-loaddata to cache data for old logbooks. New design needed, with a mechanism for flagging fixtures as outdated after edits. ''' MAX_LOGBOOK_ENTRY_TITLE_LENGTH = 200 DEFAULT_LOGBOOK_PARSER = "parser_html" DEFAULT_LOGBOOK_FILE = "logbook.html" # All years since 2010 use the default value for Logbook parser # but several don't work, and are skipped by the parsing code, e.g. 1983 LOGBOOK_PARSER_SETTINGS = { "2010": ("logbook.html", "parser_html"), "2009": ("2009logbook.txt", "parser_wiki"), "2008": ("2008logbook.txt", "parser_wiki"), "2007": ("logbook.html", "parser_html"), "2006": ("logbook.html", "parser_html"), # "2006": ("logbook/logbook_06.txt", "parser_wiki"), "2006": ("logbook.html", "parser_html"), "2005": ("logbook.html", "parser_html"), "2004": ("logbook.html", "parser_html"), "2003": ("logbook.html", "parser_html_03"), "2002": ("logbook.html", "parser_html"), "2001": ("log.htm", "parser_html_01"), "2000": ("log.htm", "parser_html_01"), "1999": ("log.htm", "parser_html_01"), "1998": ("log.htm", "parser_html_01"), "1997": ("log.htm", "parser_html_01"), "1996": ("log.htm", "parser_html_01"), "1995": ("log.htm", "parser_html_01"), "1994": ("log.htm", "parser_html_01"), "1993": ("log.htm", "parser_html_01"), "1992": ("log.htm", "parser_html_01"), "1991": ("log.htm", "parser_html_01"), "1990": ("log.htm", "parser_html_01"), "1989": ("log.htm", "parser_html_01"), #crashes MySQL "1988": ("log.htm", "parser_html_01"), #crashes MySQL "1987": ("log.htm", "parser_html_01"), #crashes MySQL "1985": ("log.htm", "parser_html_01"), "1984": ("log.htm", "parser_html_01"), "1983": ("log.htm", "parser_html_01"), "1982": ("log.htm", "parser_html_01"), } entries = { "2022": 64, "2019": 56, "2018": 74, "2017": 60, "2016": 81, "2015": 79, "2014": 65, "2013": 51, "2012": 75, "2011": 68, "2010": 22, "2009": 52, "2008": 49, "2007": 111, "2006": 60, "2005": 55, "2004": 76, "2003": 40, "2002": 31, "2001": 48, "2000": 54, "1999": 79, "1998": 43, "1997": 53, "1996": 94, "1995": 41, "1994": 32, "1993": 41, "1992": 61, "1991": 38, "1990": 87, "1989": 1,"1988": 1,"1987": 1, "1985": 24,"1984": 32,"1983": 52,"1982": 42,} # Logbooks log.htm exist for 1983, 84, 85, 87, 88, 89 but have no full-working parser, or need hand-editing. logentries = [] # the entire logbook for one year is a single object: a list of entries noncaveplaces = [ "QMplaceholder", "Journey", "Loser Plateau", "UNKNOWN", 'plateau', 'base camp', 'basecamp', 'top camp', 'topcamp' ] logdataissues = TROG['issues']['logdataissues'] trips ={} # # the logbook loading section # def set_trip_id(year, seq): tid= f"{year}_s{seq:02d}" return tid rx_tripperson = re.compile(r'(?i)(.*?)$') rx_round_bracket = re.compile(r"[\(\[].*?[\)\]]") def GetTripPersons(trippeople, expedition, logtime_underground, tid=None): res = [ ] author = None #print(f'# {tid}') for tripperson in re.split(r",|\+|&|&(?!\w+;)| and ", trippeople): tripperson = tripperson.strip() # mul = re.match(r"(?i)(.*?)$", tripperson) mul = rx_tripperson.match(tripperson) if mul: tripperson = mul.group(1).strip() if tripperson and tripperson[0] != '*': tripperson = re.sub(rx_round_bracket, "", tripperson).strip() if tripperson =="Wiggy": tripperson = "Phil Wigglesworth" if tripperson =="Animal": tripperson = "Mike Richardson" if tripperson =="MikeTA": tripperson = "Mike Richardson" personyear = GetPersonExpeditionNameLookup(expedition).get(tripperson.lower()) if not personyear: message = f" ! - {expedition.year} No name match for: '{tripperson}' in entry {tid=} for this expedition year." print(message) DataIssue.objects.create(parser='logbooks', message=message) logdataissues[tid]=message res.append((personyear, logtime_underground)) if mul: author = personyear if not author: if not res: return None, None author = res[-1][0] return res, author def EnterLogIntoDbase(date, place, title, text, trippeople, expedition, logtime_underground, tid=None): """ saves a logbook entry and related persontrips Does NOT save the expeditionday_id - all NULLs. why? Because we are deprecating expeditionday ! troggle.log shows that we are creating lots of duplicates, which is no no problem with SQL as they just overwrite but we are saving the same thing too many times.. """ try: trippersons, author = GetTripPersons(trippeople, expedition, logtime_underground, tid=tid) except: message = f" ! - {expedition.year} Skipping logentry: {title} - GetTripPersons FAIL" DataIssue.objects.create(parser='logbooks', message=message) logdataissues["title"]=message print(message) raise return if not author: message = f" ! - {expedition.year} Warning: logentry: {title} - no author for entry '{tid}'" DataIssue.objects.create(parser='logbooks', message=message) logdataissues["title"]=message print(message) #return # This needs attention. The slug field is derived from 'title' # both GetCaveLookup() and GetTripCave() need to work together better. None of this data is *used* though? #tripCave = GetTripCave(place): lplace = place.lower() cave=None if lplace not in noncaveplaces: cave = GetCaveLookup().get(lplace) y = str(date)[:4] text = text.replace('src="', f'src="/years/{y}/' ) text = text.replace("src='", f"src='/years/{y}/" ) #Check for an existing copy of the current entry, and save expeditionday = expedition.get_expedition_day(date) lookupAttribs={'date':date, 'title':title} # 'cave' is converted to a string doing this, which renders as the cave slug. # but it is a db query which we should try to avoid - rewrite this #NEW slug for a logbook entry here! Unique id + slugified title fragment if tid is not None: slug = tid + "_" + slugify(title)[:10].replace('-','_') else: slug = str(randint(1000,10000)) + "_" + slugify(title)[:10].replace('-','_') nonLookupAttribs={'place':place, 'text':text, 'expedition':expedition, 'cave_slug':str(cave), 'slug': slug} # This creates the lbo instance of LogbookEntry lbo, created=save_carefully(LogbookEntry, lookupAttribs, nonLookupAttribs) for tripperson, time_underground in trippersons: lookupAttribs={'personexpedition':tripperson, 'logbook_entry':lbo} nonLookupAttribs={'time_underground':time_underground, 'is_logbook_entry_author':(tripperson == author)} # this creates the PersonTrip instance. save_carefully(PersonTrip, lookupAttribs, nonLookupAttribs) def ParseDate(tripdate, year): """ Interprets dates in the expo logbooks and returns a correct datetime.date object """ dummydate = date(1970, 1, 1) month = 1 day = 1 # message = f" ! - Trying to parse date in logbook: {tripdate} - {year}" # print(message) try: mdatestandard = re.match(r"(\d\d\d\d)-(\d\d)-(\d\d)", tripdate) mdategoof = re.match(r"(\d\d?)/0?(\d)/(20|19)?(\d\d)", tripdate) if mdatestandard: if not (mdatestandard.group(1) == year): message = f" ! - Bad date (year) in logbook: {tripdate} - {year}" DataIssue.objects.create(parser='logbooks', message=message) logdataissues["tripdate"]=message return dummydate else: year, month, day = int(mdatestandard.group(1)), int(mdatestandard.group(2)), int(mdatestandard.group(3)) elif mdategoof: if not (not mdategoof.group(3) or mdategoof.group(3) == year[:2]): message = " ! - Bad date mdategoof.group(3) in logbook: " + tripdate + " - " + mdategoof.group(3) DataIssue.objects.create(parser='logbooks', message=message) logdataissues["tripdate"]=message return dummydate else: yadd = int(year[:2]) * 100 day, month, year = int(mdategoof.group(1)), int(mdategoof.group(2)), int(mdategoof.group(4)) + yadd else: year = 1970 message = f" ! - Bad date in logbook: {tripdate} - {year}" DataIssue.objects.create(parser='logbooks', message=message) logdataissues["tripdate"]=message return date(year, month, day) except: message = f" ! - Failed to parse date in logbook: {tripdate} - {year}" DataIssue.objects.create(parser='logbooks', message=message) logdataissues["tripdate"]=message return datetime.date(1970, 1, 1) # (2006 - not any more), 2008 - 2009 def parser_wiki(year, expedition, txt): global logentries global logdataissues logbook_entry_count = 0 trippara = re.findall(r"===(.*?)===([\s\S]*?)(?====)", txt) for triphead, triptext in trippara: logbook_entry_count += 1 tid = set_trip_id(year,logbook_entry_count) tripheadp = triphead.split("|") if not (len(tripheadp) == 3): message = " ! - Bad no of items in tripdate in logbook: " + tripdate + " - " + tripheadp DataIssue.objects.create(parser='logbooks', message=message) logdataissues["tripdate"]=message tripdate, tripplace, trippeople = tripheadp tripsplace = tripplace.split(" - ") tripcave = tripsplace[0].strip() if len(tripsplace) == 1: tripsplace = tripsplace[0] else: tripsplace = tripsplace[1] #print(f"! LOGBOOK {year} {logbook_entry_count:2} {len(triptext):4} '{tripsplace}'") tul = re.findall(r"T/?U:?\s*(\d+(?:\.\d*)?|unknown)\s*(hrs|hours)?", triptext) if tul: tu = tul[0][0] else: tu = "" ldate = ParseDate(tripdate.strip(), year) tripid ="" entrytuple = (ldate, tripcave, tripsplace, triptext, trippeople, expedition, tu, tripid) logentries.append(entrytuple) # 2002, 2004, 2005, 2007, 2010 - now # 2006 wiki text is incomplete, but the html all there. So using this parser now. def parser_html(year, expedition, txt): global logentries global logdataissues tripparas = re.findall(r"
)? # second date \s*(?:\s*)? \s*
)? \s*
)? # second date \s*(?:\s*)? \s*
)? \s*
", "", ltriptext).strip() entrytuple = (ldate, tripcave, triptitle, ltriptext, trippeople, expedition, tu, tripid1) logentries.append(entrytuple) # main parser for 1991 - 2001. simpler because the data has been hacked so much to fit it # trying it out for years 1982 - 1990 too. Some logbook editing required by hand.. place def parser_html_01(year, expedition, txt): global logentries global logdataissues errorcount = 0 tripparas = re.findall(r"
]*>(T/?U.*)', triptext) if mtu: tu = mtu.group(1) triptext = triptext[:mtu.start(0)] + triptext[mtu.end():] else: tu = "" triptitles = triptitle.split(" - ") tripcave = triptitles[0].strip() ltriptext = triptext mtail = re.search(r'(?:[^<]*|\s|/|-|&|?p>|\((?:same day|\d+)\))*$', ltriptext) if mtail: ltriptext = ltriptext[:mtail.start(0)] ltriptext = re.sub(r"
", "", ltriptext) ltriptext = re.sub(r"\s*?\n\s*", " ", ltriptext) ltriptext = re.sub(r"|
", "\n\n", ltriptext).strip()
ltriptext = re.sub(r"?u>", "_", ltriptext)
ltriptext = re.sub(r"?i>", "''", ltriptext)
ltriptext = re.sub(r"?b>", "'''", ltriptext)
if ltriptext == "":
message = " ! - Zero content for logbook entry!: " + tid
DataIssue.objects.create(parser='logbooks', message=message)
logdataissues[tid]=message
print(message)
entrytuple = (ldate, tripcave, triptitle, ltriptext,
trippeople, expedition, tu, tid)
logentries.append(entrytuple)
except:
message = f" ! - Skipping logentry {year} due to exception in: {tid}"
DataIssue.objects.create(parser='logbooks', message=message)
logdataissues[tid]=message
print(message)
errorcount += 1
raise
if errorcount >5 :
message = f" !!- TOO MANY ERRORS - aborting at '{tid}' logbook: {year}"
DataIssue.objects.create(parser='logbooks', message=message)
logdataissues[tid]=message
print(message)
return
# parser for 2003
def parser_html_03(year, expedition, txt):
global logentries
global logdataissues
tripparas = re.findall(r"
", "\n\n", ltriptext).strip() ltriptext = re.sub(r"[^\s0-9a-zA-Z\-.,:;'!&()\[\]<>?=+*%]", "_NONASCII_", ltriptext) entrytuple = (ldate, tripcave, triptitle, ltriptext, trippeople, expedition, tu, tid) logentries.append(entrytuple) def LoadLogbookForExpedition(expedition): """ Parses all logbook entries for one expedition """ # absolutely horrid. REFACTOR THIS (all my fault..) global logentries global logdataissues global entries logbook_parseable = False yearlinks = LOGBOOK_PARSER_SETTINGS expologbase = os.path.join(settings.EXPOWEB, "years") logentries=[] year = expedition.year expect = entries[year] # print(" - Logbook for: " + year) def cleanerrors(year): global logdataissues dataissues = DataIssue.objects.filter(parser='logbooks') for di in dataissues: ph = year if re.search(ph, di.message) is not None: #print(f' - CLEANING dataissue {di.message}') di.delete() #print(f' - CLEAN {year} {len(logdataissues)} {type(logdataissues)} data issues for this year') dellist = [] for key, value in logdataissues.items(): #print(f' - CLEANING logdataissues [{key}]: {value}') if key.startswith(year): #print(f' - CLEANING logdataissues [{key:12}]: {value} ') dellist.append(key) for i in dellist: del logdataissues[i] cleanerrors(year) if year in yearlinks: logbookpath = Path(expologbase) / year / yearlinks[year][0] expedition.logbookfile = yearlinks[year][0] parsefunc = yearlinks[year][1] else: logbookpath = Path(expologbase) / year / DEFAULT_LOGBOOK_FILE expedition.logbookfile = DEFAULT_LOGBOOK_FILE parsefunc = DEFAULT_LOGBOOK_PARSER expedition.save() try: file_in = open(logbookpath,'rb') txt = file_in.read().decode("utf-8") file_in.close() logbook_parseable = True except (IOError): logbook_parseable = False print(" ! Couldn't open logbook as UTF-8 " + logbookpath) except: logbook_parseable = False print(" ! Very Bad Error opening " + logbookpath) if logbook_parseable: parser = globals()[parsefunc] print(f' - {year} parsing with {parsefunc}') parser(year, expedition, txt) # this launches the right parser for this year i=0 for entrytuple in logentries: # date, tripcave, triptitle, text, trippeople, expedition, logtime_underground, tripid1 = entrytuple try: date, tripcave, triptitle, text, trippeople, expedition, logtime_underground, tripid1 = entrytuple except ValueError: # cope with removal of entry_type but still in cache files. Remove in Dec. 2022. date, tripcave, triptitle, text, trippeople, expedition, logtime_underground, entry_type, tripid1 = entrytuple print(f' - Exception entry_type "{entry_type}" {tripid1}') EnterLogIntoDbase(date, tripcave, triptitle, text, trippeople, expedition, 0, tripid1) i +=1 if len(logentries) == expect: # print(f"OK {year} {len(logentries):5d} is {expect}\n") pass else: print(f"Mismatch {year} {len(logentries):5d} is not {expect}\n") return len(logentries) def LoadLogbooks(): """ This is the master function for parsing all logbooks into the Troggle database. This should be rewritten to use coroutines to load all logbooks from disc in parallel, but must be serialised to write to database as sqlite is single-user. """ global logdataissues global entries logdataissues = {} DataIssue.objects.filter(parser='logbooks').delete() expos = Expedition.objects.all() if len(expos) <= 1: message = f" ! - No expeditions found. Load 'people' first" DataIssue.objects.create(parser='logbooks', message=message) logdataissues[f"sqlfail 0000"]=message print(message) noexpo = ["1986", "2020", "2021",] #no expo lostlogbook = ["1976", "1977", "1978", "1979", "1980", "1981"] sqlfail = ["1987", "1988", "1989"] # breaks mysql with db constraint fail - debug locally first] nologbook = noexpo + lostlogbook + sqlfail nlbe={} expd ={} actuals = [] for expo in expos: # pointless as we explicitly know the years in this code. year = expo.year TROG['pagecache']['expedition'][year] = None # clear cache if year in sqlfail: print(" - Logbook for: " + year + " NO parsing attempted - known sql failures") message = f" ! - Not even attempting to parse logbook for {year} until code fixed" DataIssue.objects.create(parser='logbooks', message=message) logdataissues[f"sqlfail {year}"]=message print(message) if year not in nologbook: if year in entries: actuals.append(expo) else: print(" - No Logbook yet for: " + year) # catch case when preparing for next expo for ex in actuals: nlbe[ex] = LoadLogbookForExpedition(ex) # this actually loads the logbook for one expo # tried to use map with concurrent threads - but sqlite database is not concurrent, so failed with database lock # yt = 0 # for r in map(LoadLogbookForExpedition, actuals): # yt = r yt = 0 for e in nlbe: yt += nlbe[e] print(f"total {yt:,} log entries parsed in all expeditions") # dateRegex = re.compile(r'(\d\d\d\d)-(\d\d)-(\d\d)', re.S) # expeditionYearRegex = re.compile(r'(.*?)', re.S) # titleRegex = re.compile(r'