troggle-unchained/parsers/scans.py

143 lines
4.9 KiB
Python
Raw Normal View History

2021-05-03 20:36:29 +01:00
import sys
import os
import types
import stat
import csv
import re
import datetime
from functools import reduce
import settings
from troggle.core.models.survex import SingleScan, Wallet, DrawingFile
from troggle.core.models.troggle import DataIssue
2021-05-04 20:57:16 +01:00
from troggle.core.utils import save_carefully, GetListDir
2021-05-03 20:36:29 +01:00
'''Searches through all the survey scans directories (wallets) in expofiles, looking for images to be referenced.
'''
contentsjson = "contents.json"
indexhtml = "walletindex.html"
wallet_blank_json = {
"cave": "",
"date": "",
"description url": "/caves",
"description written": False,
"electronic survey": False,
"elev drawn": False,
"elev not required": False,
"name": "",
"people": [
"Unknown"
],
"plan drawn": False,
"plan not required": False,
"qms written": False,
"survex file": [],
"survex not required": False,
"website updated": False}
wallet_blank_html = '''<html><body><H1>Wallet WALLET</H1>
<p>List of trips: <a href="http://expo.survex.com/expedition/YEAR">expedition/YEAR</a>
- troggle-processed .svx files and logbook entries on server</p>
<p>Date: </p><p>People: Unknown,</p>
<p>Cave <a href='http://expo.survex.com/caves/'>Guidebook description</a>
- A description is indicated as being needed, so may need adding into this cave page.
<p>Survex file: not identified yet
<H2>Issues</H2>
<p>The description needs writing</p>
<p>The QMs needs writing</p><p>The website is marked as needing updating (using the guidebook description)</p>
<p>Tunnel / Therion drawing files need drawing</p>
<H2>Files</H2>
<UL>
</UL>
</body></html>
'''
2021-05-03 20:36:29 +01:00
def get_or_create_placeholder(year):
""" All surveys must be related to a logbookentry. We don't have a way to
automatically figure out which survey went with which logbookentry,
so we create a survey placeholder logbook entry for each year. This
function always returns such a placeholder, and creates it if it doesn't
exist yet.
"""
lookupAttribs={'date__year':int(year), 'title':"placeholder for surveys",}
nonLookupAttribs={'text':"surveys temporarily attached to this should be re-attached to their actual trips", 'date':datetime.date(int(year),1,1)}
placeholder_logbook_entry, newly_created = save_carefully(LogbookEntry, lookupAttribs, nonLookupAttribs)
return placeholder_logbook_entry
# def listdir(*directories):
# '''WHAT is this ?! Some python2 fossil. The only use of settings.SURVEYS
# local import statement to be avoided, please.
# '''
# try:
# return os.listdir(os.path.join(settings.SURVEYS, *directories))
# except:
# import urllib.request, urllib.parse, urllib.error
# url = settings.SURVEYS + reduce(lambda x, y: x + "/" + y, ["listdir"] + list(directories))
# folders = urllib.request.urlopen(url.replace("#", "%23")).readlines()
# return [folder.rstrip(r"/") for folder in folders]
2021-05-03 20:36:29 +01:00
def LoadListScansFile(wallet):
gld = [ ]
# flatten out any directories in these wallet folders - should not be any
for (fyf, ffyf, fisdiryf) in GetListDir(wallet.fpath):
if fisdiryf:
gld.extend(GetListDir(ffyf))
else:
gld.append((fyf, ffyf, fisdiryf))
c=0
for (fyf, ffyf, fisdiryf) in gld:
if re.search(r"\.(?:png|jpg|jpeg|pdf|svg|gif)(?i)$", fyf):
singlescan = SingleScan(ffile=ffyf, name=fyf, wallet=wallet)
singlescan.save()
c+=1
if c>=10:
print(".", end='')
c = 0
# this iterates through the scans directories (either here or on the remote server)
# and builds up the models we can access later
def load_all_scans():
print(' - Loading Survey Scans')
SingleScan.objects.all().delete()
Wallet.objects.all().delete()
print(' - deleting all scansFolder and scansSingle objects')
2022-03-15 20:53:55 +00:00
DataIssue.objects.filter(parser='scans').delete()
2021-05-03 20:36:29 +01:00
# first do the smkhs (large kh survey scans) directory
2022-04-18 20:48:49 +01:00
manywallets_smkhs = Wallet(fpath=os.path.join(settings.SCANS_ROOT, "../surveys/smkhs"), walletname="smkhs")
2021-05-03 20:36:29 +01:00
print("smkhs", end=' ')
if os.path.isdir(manywallets_smkhs.fpath):
manywallets_smkhs.save()
LoadListScansFile(manywallets_smkhs)
# iterate into the surveyscans directory
print(' - ', end=' ')
2022-04-18 20:48:49 +01:00
for f, ff, fisdir in GetListDir(settings.SCANS_ROOT):
2021-05-03 20:36:29 +01:00
if not fisdir:
continue
# do the year folders
if re.match(r"\d\d\d\d$", f):
print("%s" % f, end=' ')
for fy, ffy, fisdiry in GetListDir(ff):
if fisdiry:
wallet = Wallet(fpath=ffy, walletname=fy)
wallet.save()
LoadListScansFile(wallet)
# do the
elif f != "thumbs":
wallet = Wallet(fpath=ff, walletname=f)
wallet.save()
LoadListScansFile(wallet)
print("", flush=True)