fixed circular ref on setup & in-memory db

This commit is contained in:
Philip Sargent 2020-06-08 00:11:09 +01:00
parent 9237a6262e
commit 538a3b6ca8
2 changed files with 146 additions and 121 deletions

View File

@ -151,6 +151,10 @@ def get_name(pe):
return pe.person.first_name
class UploadFileForm(forms.Form):
# Because this has EXECUTABLE statements in its signature (the fields) they get
# executed when this module is LOADED. Which barfs horribly.
# so all replaced by an __init__ method instead.
def __init__(self):
title = forms.CharField(max_length=50)
file = forms.FileField()
html = forms.CharField(widget=TinyMCE(attrs={'cols': 80, 'rows': 30}))
@ -159,8 +163,9 @@ class UploadFileForm(forms.Form):
slug = forms.CharField(max_length=50)
date = forms.DateField(required=False)
#This next line is the one that causes django.setup() to BARF LOUDLY
caves = [cave.slug for cave in Cave.objects.all()]
#caves.sort()
#caves.sort() # sort needs rewriting for python3
caves = ["-----"] + caves
cave = forms.ChoiceField([(c, c) for c in caves], required=False)

View File

@ -9,11 +9,13 @@ os.environ['PYTHONPATH'] = settings.PYTHON_PATH
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'settings')
from django.core import management
from django.db import connection, close_old_connections
from django.db import connection, close_old_connections, connections
from django.contrib.auth.models import User
from django.http import HttpResponse
from django.core.urlresolvers import reverse
print(" 1 settings on loading databaseReset.py")
from troggle.core.models_caves import Cave, Entrance
import troggle.parsers.caves
#import troggle.settings
@ -36,67 +38,117 @@ if os.geteuid() == 0:
print("This script should be run as expo not root - quitting")
exit()
dbengine = ""
dbname = ""
dbdefault =""
expouser=settings.EXPOUSER
expouserpass=settings.EXPOUSERPASS
expouseremail=settings.EXPOUSER_EMAIL
print(" - settings: {} ({:.5}...) <{}> on module loading".format(expouser, expouserpass, expouseremail))
def reinit_db():
"""Rebuild database from scratch. Deletes the file first if sqlite is used,
otherwise it drops the database and creates it.
Note - initial loading of troggle.sqlite will already have populated the models
in memory (django python models, not the database), so there is already a full load
of stuff known. Deleting the db file does not clear memory.
"""
django.db.close_old_connections() # wipes an in-memory sqlite db
currentdbname = settings.DATABASES['default']['NAME']
if settings.DATABASES['default']['NAME'] == ':memory:':
pass
elif settings.DATABASES['default']['ENGINE'] == 'django.db.backends.sqlite3':
if currentdbname == ':memory:':
# closing connections should wipe the in-memory database
django.db.close_old_connections()
for conn in django.db.connections.all():
print(" ! Closing another connection to db...")
conn.close()
elif django.db.connections.databases['default']['ENGINE'] == 'django.db.backends.sqlite3':
try:
os.remove(currentdbname)
except OSError:
print(" ! OSError on removing: " + currentdbname + " (Is the file open in another app?\n")
raise
else:
cursor = connection.cursor()
cursor = django.db.connection.cursor()
cursor.execute("DROP DATABASE %s" % currentdbname)
cursor.execute("CREATE DATABASE %s" % currentdbname)
cursor.execute("ALTER DATABASE %s CHARACTER SET=utf8" % currentdbname)
cursor.execute("USE %s" % currentdbname)
print(" - SETTINGS: {} ({:.5}...) <{}> before calling syncuser()".format(expouser, expouserpass, expouseremail))
syncuser()
def syncuser():
"""Sync user - needed after reload
"""
print(" - Synchronizing user on: " + settings.DATABASES['default']['NAME'])
#Sync user - needed after reload
print(" - Migrating: " + settings.DATABASES['default']['NAME'])
print(django.db.connections.databases['default']['NAME'])
management.call_command('migrate', interactive=False)
print(" - done migration on: " + settings.DATABASES['default']['NAME'])
try:
print(" - Setting up admin user on: " + settings.DATABASES['default']['NAME'])
print(django.db.connections.databases['default']['NAME'])
print(" - user: {} ({:.5}...) <{}> ".format(expouser, expouserpass, expouseremail))
user = User.objects.create_user(expouser, expouseremail, expouserpass)
user.is_staff = True
user.is_superuser = True
user.save()
except:
print(" ! INTEGRITY ERROR user on: " + settings.DATABASES['default']['NAME'])
print(django.db.connections.databases['default']['NAME'])
print(" ! You probably have not got a clean db when you thought you had.\n")
raise
print(" ! Also you are probably NOT running an in-memory db now.\n")
memdumpsql(fn='integrityfail.sql')
django.db.connections.databases['default']['NAME'] = ':memory:'
#raise
def memdumpsql(fn):
djconn = django.db.connection
from dump import _iterdump
with open(fn, 'w') as f:
for line in _iterdump(djconn):
f.write('%s\n' % line.encode("utf8"))
return True
def store_dbsettings():
global dbengine
global dbname
global dbdefault
dbengine = settings.DATABASES['default']['ENGINE']
dbname = settings.DATABASES['default']['NAME']
dbdefault = settings.DATABASES['default']
def restore_dbsettings():
settings.DATABASES['default'] = dbdefault
settings.DATABASES['default']['ENGINE'] = dbengine
settings.DATABASES['default']['NAME'] = dbname
django.db.connections.databases['default'] = dbdefault
django.db.connections.databases['default']['ENGINE'] = dbengine
django.db.connections.databases['default']['NAME'] = dbname
def set_in_memory_dbsettings():
django.db.close_old_connections() # needed if MySQL running?
settings.DATABASES['default'] = {'ENGINE': 'django.db.backends.sqlite3',
'AUTOCOMMIT': True,
'ATOMIC_REQUESTS': False,
'NAME': ':memory:',
'CONN_MAX_AGE': 0,
'TIME_ZONE': 'UTC',
'OPTIONS': {},
'HOST': '',
'USER': '',
'TEST': {'COLLATION': None, 'CHARSET': None, 'NAME': None, 'MIRROR': None},
'PASSWORD': '',
'PORT': ''}
settings.DATABASES['default']['ENGINE'] = 'django.db.backends.sqlite3'
settings.DATABASES['default']['NAME'] = ':memory:'
django.db.connections.databases['default']['ENGINE'] = 'django.db.backends.sqlite3'
django.db.connections.databases['default']['NAME'] = ':memory:'
def dirsredirect():
"""Make directories that troggle requires and sets up page redirects
"""
#should also deal with permissions here.
#if not os.path.isdir(settings.PHOTOS_ROOT):
#os.mkdir(settings.PHOTOS_ROOT)
for oldURL, newURL in [("indxal.htm", reverse("caveindex"))]:
f = troggle.flatpages.models.Redirect(originalURL = oldURL, newURL = newURL)
f.save()
def import_caves():
print("Importing Caves")
print("Importing Caves to ",end="")
print(django.db.connections.databases['default']['NAME'])
troggle.parsers.caves.readcaves()
def import_people():
print("Importing People (folk.csv)")
print("Importing People (folk.csv) to ",end="")
print(django.db.connections.databases['default']['NAME'])
troggle.parsers.people.LoadPersonsExpos()
def import_surveyscans():
@ -152,9 +204,6 @@ class JobQueue():
"""A list of import operations to run. Always reports profile times
in the same order.
"""
dbengine = ""
dbname = ""
dbdefault =""
def __init__(self,run):
self.runlabel = run
@ -163,7 +212,7 @@ class JobQueue():
self.results_order=[
"date","runlabel","reinit", "caves", "people",
"logbooks", "QMs", "scans", "survexblks", "survexpos",
"tunnel", "surveyimgs", "test", "dirsredirect", "syncuser" ]
"tunnel", "surveyimgs", "test" ]
for k in self.results_order:
self.results[k]=[]
self.tfile = "import_profile.json"
@ -202,19 +251,13 @@ class JobQueue():
json.dump(self.results, f)
return True
def memdumpsql(self):
djconn = django.db.connection
from dump import _iterdump
with open('memdump.sql', 'w') as f:
for line in _iterdump(djconn):
f.write('%s\n' % line.encode("utf8"))
return True
def runqonce(self):
"""Run all the jobs in the queue provided - once
"""
print("** Running job ", self.runlabel)
print("** Running job ", self.runlabel,end=" to ")
print(django.db.connections.databases['default']['NAME'])
jobstart = time.time()
self.results["date"].pop()
self.results["date"].append(jobstart)
@ -235,43 +278,18 @@ class JobQueue():
print("** Ended job %s - %.1f seconds total." % (self.runlabel,jobduration))
return True
def store_dbsettings(self):
self.dbengine = settings.DATABASES['default']['ENGINE']
self.dbname = settings.DATABASES['default']['NAME']
self.dbdefault = settings.DATABASES['default']
def restore_dbsettings(self):
settings.DATABASES['default'] = self.dbdefault
settings.DATABASES['default']['ENGINE'] = self.dbengine
settings.DATABASES['default']['NAME'] = self.dbname
def set_in_memory_dbsettings(self):
django.db.close_old_connections() # needed if MySQL running?
settings.DATABASES['default'] = {'ENGINE': 'django.db.backends.sqlite3',
'AUTOCOMMIT': True,
'ATOMIC_REQUESTS': False,
'NAME': ':memory:',
'CONN_MAX_AGE': 0,
'TIME_ZONE': 'UTC',
'OPTIONS': {},
'HOST': '',
'USER': '',
'TEST': {'COLLATION': None, 'CHARSET': None, 'NAME': None, 'MIRROR': None},
'PASSWORD': '',
'PORT': ''}
settings.DATABASES['default']['ENGINE'] = 'django.db.backends.sqlite3'
settings.DATABASES['default']['NAME'] = ":memory:"
def append_placeholders(self):
for j in self.results_order:
self.results[j].append(None) # append a placeholder
def run_now_django_tests(self,n):
self.store_dbsettings()
store_dbsettings()
# this leaves the db set to :memory: whatever it was initially
management.call_command('test', verbosity=n)
django.db.close_old_connections()
self.restore_dbsettings()
restore_dbsettings()
# and whatever I do, it stays that way !
def skip_memory_phase(self):
if not self.runlabel:
@ -289,36 +307,33 @@ class JobQueue():
relinquish some kind of db connection (not fixed yet)
"""
self.loadprofiles()
self.store_dbsettings()
store_dbsettings()
print("-- phase 0 ", settings.DATABASES['default']['ENGINE'], settings.DATABASES['default']['NAME'])
#print "-- DATABASES.default", settings.DATABASES['default']
print("-- start ", settings.DATABASES['default']['ENGINE'], settings.DATABASES['default']['NAME'])
print(django.db.connections.databases['default']['NAME'])
if self.dbname ==":memory:":
if dbname ==":memory:":
# just run, and save the sql file
self.runqonce()
self.memdumpsql() # saved contents of scratch db, could be imported later..
memdumpsql('memdump.sql') # saved contents of scratch db, could be imported later..
self.saveprofiles()
elif self.skip_memory_phase():
self.runqonce()
self.saveprofiles()
else:
# run all the imports through :memory: first
self.set_in_memory_dbsettings()
set_in_memory_dbsettings()
print("-- phase 1 ", settings.DATABASES['default']['ENGINE'], settings.DATABASES['default']['NAME'])
#print("-- DATABASES.default", settings.DATABASES['default'])
# but because the user may be expecting to add this to a db with lots of tables already there,
# the jobqueue may not start from scratch so we need to initialise the db properly first
# because we are using an empty :memory: database
# But initiating twice crashes it; so be sure to do it once only.
# Damn. syncdb() is still calling MySQL somehow **conn_params not sqlite3. So crashes on expo server.
# Damn. migrate() is still calling MySQL somehow **conn_params not sqlite3. So crashes on expo server.
if ("reinit",reinit_db) not in self.queue:
reinit_db()
if ("dirsredirect",dirsredirect) not in self.queue:
dirsredirect()
if ("caves",import_caves) not in self.queue:
import_caves() # sometime extract the initialising code from this and put in reinit...
if ("people",import_people) not in self.queue:
@ -327,21 +342,22 @@ class JobQueue():
django.db.close_old_connections() # maybe not needed here
self.runqonce()
self.memdumpsql()
memdumpsql('memdump2.sql')
self.showprofile()
# restore the original db and import again
# if we wanted to, we could re-import the SQL generated in the first pass to be
# blazing fast. But for the present just re-import the lot.
self.restore_dbsettings()
restore_dbsettings()
print("-- phase 2 ", settings.DATABASES['default']['ENGINE'], settings.DATABASES['default']['NAME'])
print(django.db.connections.databases['default']['NAME'])
django.db.close_old_connections() # maybe not needed here
for j in self.results_order:
self.results[j].pop() # throw away results from :memory: run
self.append_placeholders()
django.db.close_old_connections() # magic rune. works. found by looking in django.db__init__.py
django.db.close_old_connections()
#django.setup() # should this be needed?
self.runqonce()
@ -353,13 +369,7 @@ class JobQueue():
"""Prints out the time it took to run the jobqueue
"""
for k in self.results_order:
if k =="dirsredirect":
break
if k =="surveyimgs":
break
elif k =="syncuser":
break
elif k =="test":
if k =="test":
break
elif k =="date":
print(" days ago ", end=' ')
@ -384,6 +394,8 @@ class JobQueue():
# prints one place to the left of where you expect
if r[len(r)-1]:
s = r[i]-r[len(r)-1]
elif r[len(r)-2]:
s = r[i]-r[len(r)-2]
else:
s = 0
days = (s)/(24*60*60)
@ -408,34 +420,35 @@ def usage():
profile - print the profile from previous runs. Import nothing.
reset - normal usage: clear database and reread everything from files - time-consuming
caves - read in the caves (must run first after reset)
people - read in the people from folk.csv (must run before logbooks)
caves - read in the caves (must run first after initialisation)
people - read in the people from folk.csv (must run after 'caves')
logbooks - read in the logbooks
QMs - read in the QM csv files (older caves only)
scans - the survey scans in all the wallets (must run before survex)
survex - read in the survex files - all the survex blocks but not the x/y/z positions
survexpos - just the x/y/z Pos out of the survex files
survexpos - just the x/y/z Pos out of the survex files (not needed)
tunnel - read in the Tunnel files - which scans the survey scans too
reinit - clear database (delete everything) and make empty tables. Import nothing.
syncuser - needed after reloading database from SQL backup
autologbooks - Not used. read in autologbooks (what are these?)
dumplogbooks - Not used. write out autologbooks (not working?)
surveyimgs - Not used. read in scans by-expo, must run after "people".
and [runlabel] is an optional string identifying this run of the script
in the stored profiling data 'import-profile.json'
if [runlabel] is absent or begins with "F-" then it will skip the :memory: pass
caves and logbooks must be run on an empty db before the others as they
set up db tables used by the others.
the in-memory phase is on an empty db, so always runs reinit, caves & people for this phase
the commands are first run on an in-memory empty database before being run on
the actual persistent database. This is very fast and checks for import errors.
the initial in-memory phase is on an empty db, so always runs caves & people for this phase
""")
if __name__ == "__main__":
django.setup()
if os.geteuid() == 0:
print("Do not run as root or using sudo - file permissions for cache files and logs will break")
@ -446,8 +459,17 @@ if __name__ == "__main__":
else:
runlabel=None
store_dbsettings()
set_in_memory_dbsettings()
print(" - django.setup - next")
try:
django.setup()
except:
print(" ! COMPLICATED FAILURE. Does not occur with a valid 'troggle.sqlite' database in place.")
raise
print(" - django.setup - done")
jq = JobQueue(runlabel)
#jq.run_now_django_tests(1)
if len(sys.argv)==1:
usage()
@ -455,7 +477,7 @@ if __name__ == "__main__":
elif "test" in sys.argv:
jq.enq("caves",import_caves)
jq.enq("people",import_people)
jq.run_now_django_tests(2)
#jq.run_now_django_tests(2)
elif "caves" in sys.argv:
jq.enq("caves",import_caves)
elif "logbooks" in sys.argv:
@ -466,7 +488,6 @@ if __name__ == "__main__":
jq.enq("QMs",import_QMs)
elif "reset" in sys.argv:
jq.enq("reinit",reinit_db)
jq.enq("dirsredirect",dirsredirect)
jq.enq("caves",import_caves)
jq.enq("people",import_people)
jq.enq("scans",import_surveyscans)
@ -493,9 +514,6 @@ if __name__ == "__main__":
# writeCaves()
elif "profile" in sys.argv:
jq.loadprofiles()
# need to increment everything runq does
print("!! - days before appears as 0.00 - to be fixed")
jq.append_placeholders()
jq.showprofile()
exit()
elif "help" in sys.argv:
@ -506,5 +524,7 @@ if __name__ == "__main__":
print("%s not recognised as a command." % sys.argv[1])
exit()
#jq.run_now_django_tests(1)
jq.run()
jq.showprofile()