update boarddurationrender

* add selector frontend
* basic celery backend
* client library

Dockerfile

@@ -0,0 +1,8 @@
+FROM alpine:edge
+
+ADD ["requirements.txt", "/"]
+RUN echo "http://dl-cdn.alpinelinux.org/alpine/edge/testing/" >> /etc/apk/repositories && \
+    apk add --update --no-cache libpng libpng-dev freetype freetype-dev g++ python3 python3-dev libstdc++ openblas-dev && \
+    pip3 --no-cache-dir install -r requirements.txt && \
+    apk del libpng-dev freetype-dev g++ python3-dev openblas-dev && \
+    rm requirements.txt

analysis/analyzers/__init__.py

@@ -12,7 +12,7 @@ from .render import Render
 from .render.biogames import SimulationRoundsRender, BoardDurationHistRender, BoardDurationBoxRender, \
 	ActivityMapperRender, StoreRender, SimulationOrderRender, SimulationGroupRender
 from .render.default import PrintRender, JSONRender, TrackRender, HeatMapRender, LogEntryCountAnalyzerPlot, \
-	LogEntryCountCSV
+	LogEntryCountCSV, KMLRender
 from .render.locomotion import LocomotionActionRelativeRender, LocomotionActionAbsoluteRender, \
 	LocomotionActionRatioRender
 
@@ -41,6 +41,7 @@ __MAPPING__ = {
 	LocationAnalyzer: [
 		TrackRender,
 		HeatMapRender,
+		KMLRender,
 	],
 	ActivityMapper: [
 		ActivityMapperRender

analysis/analyzers/analyzer/__init__.py

@@ -2,7 +2,7 @@ import logging
 from collections import KeysView
 from typing import Type, Sized, Collection
 
-from analyzers.settings import LogSettings
+from analysis.analyzers.settings import LogSettings
 
 log: logging.Logger = logging.getLogger(__name__)
 

analysis/analyzers/analyzer/biogames.py

@@ -3,8 +3,8 @@ from collections import defaultdict, namedtuple, OrderedDict
 from types import SimpleNamespace
 from typing import List, NamedTuple
 
-from util import json_path, combinate
-from util.download import download_board, get_board_data
+from analysis.util import json_path, combinate
+from analysis.util.download import download_board, get_board_data
 from . import Result, LogSettings, Analyzer, ResultStore
 from .default import CategorizerStub, Store
 

analysis/analyzers/analyzer/default.py

@@ -1,7 +1,7 @@
 import logging
 from collections import defaultdict, OrderedDict
 
-from util import json_path
+from analysis.util import json_path
 from . import Result, LogSettings, Analyzer, ResultStore
 
 
@@ -16,9 +16,9 @@ class LocationAnalyzer(Analyzer):
 		super().__init__(settings)
 		self.entries = []
 
-	def result(self, store: ResultStore) -> None:
+	def result(self, store: ResultStore, **kwargs) -> None:
 		self.log.debug(len(self.entries))
-		store.add(Result(type(self), list(self.entries)))
+		store.add(Result(type(self), list(self.entries), name=kwargs['name']))
 
 	def process(self, entry: dict) -> bool:
 		if entry[self.settings.type_field] in self.settings.spatials:

analysis/analyzers/analyzer/locomotion.py

@@ -1,4 +1,4 @@
-import util
+from analysis import util
 from . import Analyzer, LogSettings, Result, ResultStore
 
 

analysis/analyzers/render/__init__.py

@@ -5,7 +5,7 @@ from .. import Result
 class Render:
 	result_types = []
 
-	def render(self, results: List[Result], name=None):
+	def render(self, results: List[Result], name=None) -> [str]:
 		raise NotImplementedError()
 
 	def filter(self, results: List[Result]):

analysis/analyzers/render/biogames.py

@@ -9,8 +9,8 @@ from scipy.interpolate import interp1d
 import networkx as nx
 import itertools
 
-from analyzers import Store, BiogamesStore, SimulationOrderAnalyzer
-from util.meta_temp import CONFIG_NAMES
+from analysis.analyzers import Store, BiogamesStore, SimulationOrderAnalyzer
+from analysis.util.meta_temp import CONFIG_NAMES
 from . import Render
 from .. import Result, SimulationRoundsAnalyzer, BoardDurationAnalyzer, ActivityMapper
 
@@ -135,16 +135,16 @@ class BoardDurationHistRender(Render):
 class BoardDurationBoxRender(Render):
 	result_types = [BoardDurationAnalyzer]
 
-	def render(self, results: List[Result], name=None):
+	def render(self, results: List[Result], name=None) -> [str]:
 		data = defaultdict(list)
 		for result in self.filter(results):
-			get = result.get()
-			for board in get:
+			for board in result.get():
 				duration = board['active'] if 'active' in board else 0
 				data[board['id']].append(duration)
 		data_tuples = [(key, data[key]) for key in sorted(data)]
 		data_tuples = sorted(data_tuples, key=lambda x: sum(x[1]))
-		plot(data_tuples)
+		plot(data_tuples, name=name)
+		return [name]
 
 
 class ActivityMapperRender(Render):
@@ -152,12 +152,14 @@ class ActivityMapperRender(Render):
 
 	def render(self, results: List[Result], name=None):
 		print(os.getcwd())
+		files = []
 		for result in self.filter(results):
 			data = result.get()
-			with open(os.path.join("static", "progress", "data", data['instance'] + "_" + str(name) + ".json"),
-			          "w") as out:
+			path = os.path.join("/tmp", data['instance'] + "_" + str(name) + ".json")
+			with open(path, "w") as out:
 				json.dump(data, out, indent=1)
-		return "ok"
+			files.append(path)
+		return files
 
 
 class StoreRender(Render):

analysis/analyzers/render/default.py

@@ -2,11 +2,13 @@ import json
 import logging
 from typing import List
 
+import datetime
 import matplotlib.pyplot as plt
 
-from analyzers import LogEntryCountAnalyzer
+from analysis.analyzers import LogEntryCountAnalyzer
+from analysis.util.meta_temp import KML_PATTERN
 from . import Render, Result
-from .. import LocationAnalyzer
+from analysis.analyzers import LocationAnalyzer
 
 log = logging.getLogger(__name__)
 
@@ -39,6 +41,34 @@ class TrackRender(Render):
 		return dumps
 
 
+def format_time(ts):
+	return datetime.datetime.fromtimestamp(ts/1000).strftime("%Y-%m-%dT%H:%M:%S.%f")
+
+
+class KMLRender(Render):
+	result_types = [LocationAnalyzer]
+
+	def render(self, results: List[Result], name=None):
+		files = []
+		for result in self.filter(results):
+			times = ["<when>{time}</when>".format(time=format_time(entry["timestamp"])) for entry in result.get()]
+			coords = [
+				"<gx:coord>{long} {lat} 0.0</gx:coord>"
+					.format(
+					lat=entry['location']['coordinates'][1],
+					long=entry['location']['coordinates'][0])
+				for entry in result.get()
+			]
+			filename = str(result.name)+".kml"
+			print(filename)
+			with open(filename, "w") as out:
+				out.write(KML_PATTERN.format(name=str(result.name), coordinates="\n".join(coords), when="\n".join(times)))
+			files.append(filename)
+		return files
+
+
+
+
 class HeatMapRender(TrackRender):
 	weight = 0.01
 

analysis/analyzers/render/wip.py

@@ -2,8 +2,8 @@ import json
 
 import numpy as np
 
-import analyzers
-from util.geo import calc_distance
+import analysis.analyzers
+from analysis.util.geo import calc_distance
 
 
 def time_distribution(store):
@@ -97,7 +97,7 @@ def time_distribution(store):
 
 from collections import defaultdict
 import matplotlib.pyplot as plt
-from util.meta_temp import CONFIG_NAMES
+from analysis.util.meta_temp import CONFIG_NAMES
 
 keys = [
 	"simu",

analysis/analyzers/settings.py

@@ -1,12 +1,12 @@
 import json
 import sys
-from sources import SOURCES
+from clients.webclients import CLIENTS
 
 
 def load_source(config):
-	if config["type"] in SOURCES:
-		source = SOURCES[config["type"]]()
-		source.connect(**config)
+	if config["type"] in CLIENTS:
+		source = CLIENTS[config["type"]](**config)
+		source.login()
 		return source
 
 
@@ -28,13 +28,15 @@ class LogSettings:
 		self.boards = json_dict['boards']
 		for mod in json_dict['analyzers']:
 			for name in json_dict['analyzers'][mod]:
-				print(mod, name)
+				print(mod, name, getattr(sys.modules[mod], name))
 				self.analyzers.append(getattr(sys.modules[mod], name))
 		self.sequences = json_dict['sequences']
 		if 'custom' in json_dict:
 			self.custom = json_dict['custom']
 		if "source" in json_dict:
 			self.source = load_source(json_dict['source'])
+		if "render" in json_dict:
+			self.render = json_dict['render']
 
 	def __repr__(self):
 		return str({
@@ -51,3 +53,7 @@ class LogSettings:
 
 def load_settings(file: str) -> LogSettings:
 	return LogSettings(json.load(open(file)))
+
+
+def parse_settings(config: str) -> LogSettings:
+	return LogSettings(json.loads(config))

analysis/log_analyzer.py

@@ -2,16 +2,16 @@ import json
 import logging
 from typing import List
 
-import analyzers
-from analyzers import get_renderer, render
-from analyzers.analyzer import ResultStore
-from analyzers.analyzer.default import write_logentry_count_csv, write_simulation_flag_csv
-from analyzers.render import wip
-from analyzers.render.default import LogEntryCountCSV
-from analyzers.render.wip import time_distribution, plot_data
-from analyzers.settings import LogSettings, load_settings
-from loaders import LOADERS
-from util.processing import grep, run_analysis, src_file
+from analysis import analyzers
+from analysis.analyzers import get_renderer, render
+from analysis.analyzers.analyzer import ResultStore
+from analysis.analyzers.analyzer.default import write_logentry_count_csv, write_simulation_flag_csv
+from analysis.analyzers.render import wip
+from analysis.analyzers.render.default import LogEntryCountCSV, KMLRender
+from analysis.analyzers.render.wip import time_distribution, plot_data
+from analysis.analyzers.settings import LogSettings, load_settings, parse_settings
+from analysis.loaders import LOADERS
+from analysis.util.processing import grep, run_analysis, src_file
 
 logging.basicConfig(format='%(levelname)s %(name)s:%(message)s', level=logging.DEBUG)
 log: logging.Logger = logging.getLogger(__name__)
@@ -26,35 +26,38 @@ def urach_logs(log_ids, settings):
 
 
 if __name__ == '__main__':
-	settings: LogSettings = load_settings("biogames2.json")
-	log_ids_urach: List[str] = urach_logs([
-		# "34fecf49dbaca3401d745fb467",
-		#	"44ea194de594cd8d63ac0314be",
-		#	"57c444470dbf88605433ca935c",
-		#	"78e0c545b594e82edfad55bd7f",
-		#	"91abfd4b31a5562b1c66be37d9",
+	settings = {}
+	log_ids_gf = []
+	# settings: LogSettings = load_settings("biogames2.json")
+	# log_ids_urach: List[str] = urach_logs([
+	# 	# "34fecf49dbaca3401d745fb467",
+	# 	#	"44ea194de594cd8d63ac0314be",
+	# 	#	"57c444470dbf88605433ca935c",
+	# 	#	"78e0c545b594e82edfad55bd7f",
+	# 	#	"91abfd4b31a5562b1c66be37d9",
+	# 	# "597b704fe9ace475316c345903",
+	# 	# "e01a684aa29dff9ddd9705edf8",
 	# 	"597b704fe9ace475316c345903",
 	# 	"e01a684aa29dff9ddd9705edf8",
-		"597b704fe9ace475316c345903",
-		"e01a684aa29dff9ddd9705edf8",
-		"fbf9d64ae0bdad0de7efa3eec6",
 	# 	"fbf9d64ae0bdad0de7efa3eec6",
-		"fe1331481f85560681f86827ec",  # urach
-		# "fe1331481f85560681f86827ec"]
-		"fec57041458e6cef98652df625",
-	]
-		, settings)
-	log_ids_gf = grep(["9d11b749c78a57e786bf5c8d28",  # filderstadt
-					   "a192ff420b8bdd899fd28573e2",  # eichstätt
-					   "3a3d994c04b1b1d87168422309",  # stadtökologie
-					   "fe1331481f85560681f86827ec",  # urach
-					   "96f6d9cc556b42f3b2fec0a2cb7ed36e"  # oberelsbach
-					   ],
-					  "/home/clemens/git/ma/test/src",
-					  settings)
-	log_ids = src_file("/home/clemens/git/ma/test/filtered_5_actions")
+	# 	# "fbf9d64ae0bdad0de7efa3eec6",
+	# 	"fe1331481f85560681f86827ec",  # urach
+	# 	# "fe1331481f85560681f86827ec"]
+	# 	"fec57041458e6cef98652df625",
+	# ]
+	# 	, settings)
+	# log_ids_gf = grep(["9d11b749c78a57e786bf5c8d28",  # filderstadt
+	# 				   "a192ff420b8bdd899fd28573e2",  # eichstätt
+	# 				   "3a3d994c04b1b1d87168422309",  # stadtökologie
+	# 				   "fe1331481f85560681f86827ec",  # urach
+	# 				   "96f6d9cc556b42f3b2fec0a2cb7ed36e"  # oberelsbach
+	# 				   ],
+	# 				  "/home/clemens/git/ma/test/src",
+	# 				  settings)
+	# log_ids = src_file("/home/clemens/git/ma/test/filtered_5_actions")
 
-	#store: ResultStore = run_analysis(log_ids_gf, settings, LOADERS)
+	if False:
+		store: ResultStore = run_analysis(log_ids_gf, settings, LOADERS)
 	# store: ResultStore = run_analysis(log_ids, settings, LOADERS)
 
 	if False:
@@ -69,7 +72,7 @@ if __name__ == '__main__':
 		# render(analyzers.ProgressAnalyzer, store.get_all())
 
 	if False:
-		from analyzers.postprocessing import graph
+		from analysis.analyzers.postprocessing import graph
 
 		g = graph.Cache(settings)
 		g.run(store)
@@ -85,7 +88,7 @@ if __name__ == '__main__':
 	if False:
 		time_distribution(store)
 
-	if True:
+	if False:
 		# spatial_data = get_data_distance(store,relative_values=False)
 		# temporal_data = get_data(store,relative_values=False)
 		# spatial_data_rel = get_data_distance(store,relative_values=True)
@@ -104,6 +107,18 @@ if __name__ == '__main__':
 		# plot_time_space_rel(combined, keys)
 		plot_data(combined, wip.keys)
 
+	if True:
+		settings: LogSettings = load_settings("../oeb_kml.json")
+		log_ids = src_file("/home/clemens/git/ma/test/oeb_2016_path")
+		log_ids = log_ids[0:2]
+		print(log_ids)
+		store: ResultStore = run_analysis(log_ids, settings, LOADERS)
+		print("render")
+		kml = KMLRender()
+		kml.render(store.get_all())
+		print("done")
+		#for cat in store.get_categories():
+		#	render(analyzers.ActivityMapper, store.get_category(cat), name=cat)
 
 # for analyzers in analyzers:
 #	if analyzers.name() in ["LogEntryCount", "ActionSequenceAnalyzer"]:

analysis/util/download.py

@@ -1,31 +1,29 @@
 import logging
 import os
 
-from util import json_path
+from analysis.util import json_path
 
 logger = logging.getLogger(__name__)
 
 
-def download_board(board_id, instance_config_id, sequence_id, source):
-	local_file = "static/progress/images/{config_id}/{sequence_id}/{board_id}".format(
-		config_id=instance_config_id,
-		sequence_id=sequence_id,
-		board_id=board_id)
-	if os.path.exists(local_file):
+def download_board(board_id, instance_config_id, sequence_id, source, path="/data/results/"):
+	local_file = os.path.join("static", instance_config_id, sequence_id, board_id)
+	abs_path = os.path.join(path, local_file)
+	if os.path.exists(abs_path):
 		return local_file
 	url = "/game2/editor/config/{config_id}/sequence/{sequence_id}/board/{board_id}/".format(
 		config_id=instance_config_id,
 		sequence_id=sequence_id,
 		board_id=board_id
 	)
-	board = source._get(url)
+	board = source.get(url)
 	if not board.ok:
-		raise ConnectionError()
+		raise ConnectionError(url, board, board.status_code)
 	data = board.json()
 	preview_url = json_path(data, "preview_url.medium")
 	logger.debug(preview_url)
-	os.makedirs(local_file[:-len(board_id)], exist_ok=True)
-	source.download_file(preview_url, local_file)
+	os.makedirs(abs_path[:-len(board_id)], exist_ok=True)
+	source.download_file(preview_url, abs_path)
 	return local_file
 
 
@@ -68,10 +66,11 @@ def get_json(source, url):
 	if url in cache:
 		return cache[url]
 	try:
-		data = source.get_json(url)
+		data = source.get(url).json()
 
 	except Exception as e:
-		print("exception", e, e.args)  # TODO: logging
+		print("exception", e, e.args)
+		logger.exception(e)
 		data = None
 	cache[url] = data
 	return data

analysis/util/meta_temp.py

@@ -97,3 +97,20 @@ CONFIG_NAMES = {
 	'fe43a0f0-3dea-11e6-a065-00199963ac6e': u'Vorlagen',
 	'ff8f1e8f-6cf5-4a7b-835b-5e2226c1e771': u'Bad Urach'
 }
+
+
+KML_PATTERN="""<?xml version="1.0" encoding="UTF-8"?>
+<kml xmlns="http://www.opengis.net/kml/2.2" xmlns:gx="http://www.google.com/kml/ext/2.2">
+  <Document>
+    <Placemark>
+    	<gx:MultiTrack>
+    		<gx:Track>
+    			{when}
+    			{coordinates}
+    		</gx:Track>
+    	</gx:MultiTrack>
+      {coordinates}
+    </Placemark>
+  </Document>
+</kml>
+"""

analysis/util/processing.py

@@ -1,8 +1,8 @@
 import logging
 from typing import List
 
-from analyzers.analyzer import ResultStore, Analyzer
-from analyzers.settings import LogSettings
+from analysis.analyzers.analyzer import ResultStore, Analyzer
+from analysis.analyzers.settings import LogSettings
 
 log: logging.Logger = logging.getLogger(__name__)
 
@@ -31,6 +31,7 @@ def process_log(logfile: str, settings: LogSettings, loaders) -> List[Analyzer]:
 def run_analysis(log_ids: list, settings, loaders):
 	store: ResultStore = ResultStore()
 	for log_id in log_ids:
+		log.info("LOG_ID: "+ str(log_id))
 		for analysis in process_log(log_id, settings, loaders):
 			log.info("* Result for " + analysis.name())
 			analysis.result(store, name=log_id)

clients/webclients.py

@@ -0,0 +1,121 @@
+import json
+import logging
+
+import os
+import shutil
+import tempfile
+import typing
+
+import requests
+
+log: logging.Logger = logging.getLogger(__name__)
+
+
+class Client:
+	host: str = ""
+	cookies: typing.Dict[str, str] = {}
+	headers: typing.Dict[str, str] = {}
+
+	def url(self, path):
+		if self.host:
+			return self.host + path
+		return path
+
+	def get(self, url, **kwargs) -> requests.models.Response:
+		log.info("GET " + str(url))
+		return requests.get(self.url(url), cookies=self.cookies, headers=self.headers, **kwargs)
+
+	def post(self, url, data, **kwargs) -> requests.models.Response:
+		log.info("POST " + str(url))
+		return requests.post(self.url(url), data, cookies=self.cookies, headers=self.headers, **kwargs)
+
+	def download_file(self, url, target, **kwargs) -> bool:
+		with open(target, "wb") as out:
+			try:
+				download = self.get(url, stream=True, **kwargs)
+				shutil.copyfileobj(download.raw, out)
+			except Exception as e:
+				log.exception(e)
+				os.remove(target)
+				return False
+		return True
+
+	def download_files(self, urls, **kwargs) -> tempfile.TemporaryDirectory:
+		target = tempfile.TemporaryDirectory()
+		for path in urls:
+			filename = os.path.join(target.name, path.split("/")[-1])
+			self.download_file(path, filename, **kwargs)
+		return target
+
+	def login(self):
+		pass
+
+	def list(self):
+		pass
+
+
+class BiogamesClient(Client):
+	config_fields: typing.Dict[str, typing.List[str]] = {
+		'login': ('username', 'password', 'host'),
+		'session': ('sessionid', 'csrftoken', 'host'),
+	}
+	login_url: str = "/game2/auth/json-login"
+	list_url: str = "/game2/instance/log/list/"
+	headers: typing.Dict[str, str] = {'Accept': 'application/json'}
+
+	def __init__(self, **kwargs):
+		match = {j: all([i in kwargs for i in self.config_fields[j]]) for j in self.config_fields}
+		valid = filter(lambda x: match[x], match)
+		if not valid:
+			raise ValueError("missing parameter (" + str(self.config_fields) + ")")
+		self.config = kwargs
+		self.cookies = {}
+		self.host = self.config['host']
+		if 'session' in valid:
+			self.cookies = kwargs
+
+	def login(self) -> bool:
+		csrf_request = self.get(self.list_url)
+		if not csrf_request.ok:
+			log.exception(ConnectionError("Unable to obtain CSRF token (" + str(csrf_request) + ")"))
+			return False
+		if not 'csrftoken' in self.cookies:
+			self.cookies['csrftoken'] = csrf_request.cookies['csrftoken']
+		login_payload = {
+			'username': self.config['username'],
+			'password': self.config['password'],
+			'next': '',
+			'csrfmiddlewaretoken': 'csrftoken',
+		}
+		login = self.post(self.login_url, json.dumps(login_payload))
+		if not login.ok:
+			log.exception(ConnectionError("Unable to authenticate", login, login.text))
+			return False
+		self.cookies['sessionid'] = login.cookies['sessionid']
+		print(self.cookies)
+		return True
+
+	def list(self) -> dict:
+		print(self.cookies)
+		logs = self.get(self.list_url)
+		if not logs.ok:
+			raise ConnectionError("HTTP fail", logs, logs.text)
+		return logs.json()
+
+	def load_all_logs(self) -> tempfile.TemporaryDirectory:
+		return self.download_files([i["file_url"] for i in self.list()])
+
+
+CLIENTS: typing.Dict[str, typing.Type[Client]] = {
+	"Biogames": BiogamesClient,
+}
+
+if __name__ == '__main__':
+	# c = BiogamesClient(host="http://biodiv", username="ba", password="853451")
+	# print(c.login())
+	# print(json.dumps(c.list(), indent=1))
+	# print(type(c.load_all_logs()))
+	# print(type(c.get("/")))
+	c = BiogamesClient(host="http://biodiv", **{'csrftoken': 'IgbwP83iEibW6RE7IADIFELYdbx0dvqQ',
+												'sessionid': 'zntsj09d92tjos1b6ruqjthlzv60xdin'})
+	print(json.dumps(c.list(), indent=1))

docker-compose.yml

@@ -0,0 +1,56 @@
+version: "2.2"
+
+services:
+ app:
+  image: docker.clkl.de/ma/celery:0.3.3
+  build: ./selector
+  cpu_count: 4
+  volumes:
+    - ./:/app
+  working_dir: /app/selector
+  command: python3 webserver.py
+  environment:
+    - PYTHONPATH=/app
+  networks:
+   - default
+   - traefik_net
+  labels:
+   - "traefik.enable=true"
+   - "traefik.port=5000"
+   - "traefik.docker.network=traefik_net"
+   - "traefik.url.frontend.rule=Host:select.ma.potato.kinf.wiai.uni-bamberg.de"
+
+ celery:
+  image: docker.clkl.de/ma/celery:0.3.3
+  environment:
+    - PYTHONPATH=/app
+  volumes:
+    - ./:/app
+    - ./data/results:/data/results
+  working_dir: /app
+  command: celery -A tasks.tasks worker --loglevel=info
+
+
+ redis:
+  image: redis:4-alpine
+  volumes:
+    - ./data/redis:/data
+  command: redis-server --appendonly yes
+
+ nginx:
+  image: nginx:1.13-alpine
+  volumes:
+   - ./data/results:/usr/share/nginx/html:ro
+  networks:
+   - traefik_net
+  labels:
+   - "traefik.enable=true"
+   - "traefik.port=80"
+   - "traefik.docker.network=traefik_net"
+   - "traefik.url.frontend.rule=Host:results.ma.potato.kinf.wiai.uni-bamberg.de"
+
+
+networks:
+ traefik_net:
+  external:
+   name: traefik_net

requirements.txt

@@ -1,8 +1,13 @@
 requests==2.18.4
-numpy==1.13.1
+numpy==1.14.2
 matplotlib==2.1.0
-osmnx==0.6
+#osmnx==0.6
 networkx==2.0
-pydot==1.2.3
-scipy==1.0.0
-ipython==6.2.1
+#pydot==1.2.3
+scipy==1.0.1
+#ipython==6.2.1
+
+flask==0.12.2
+
+celery==4.1.0
+redis==2.10.6

selector/static/script.js

@@ -0,0 +1,4 @@
+function validateSettings() {
+    alert(document.getElementById('safety').checked);
+    return false;
+}

selector/static/style.css

@@ -0,0 +1,9 @@
+body {
+    background-color: aqua;
+}
+#data{
+    display: none;
+}
+li{
+    list-style-type: none;
+}

selector/temp_config.py

@@ -0,0 +1,133 @@
+KML = """{
+  "logFormat": "zip",
+  "entryType": "@class",
+  "spatials": [
+    "de.findevielfalt.games.game2.instance.log.entry.LogEntryLocation"
+  ],
+  "actions": [],
+  "boards": [
+    "de.findevielfalt.games.game2.instance.log.entry.ShowBoardLogEntry"
+  ],
+  "analyzers": {
+    "analysis.analyzers": [
+      "BiogamesCategorizer",
+      "LocationAnalyzer"
+    ]
+  },
+  "sequences": {
+    "start": "de.findevielfalt.games.game2.instance.log.entry.LogEntryCache",
+    "end": {
+      "@class": "de.findevielfalt.games.game2.instance.log.entry.LogEntryInstanceAction",
+      "action.@class": "de.findevielfalt.games.game2.instance.action.CacheEnableAction"
+    }
+  },
+  "custom": {
+    "simulation_rounds": [
+      "de.findevielfalt.games.game2.instance.log.entry.LogEntryQuestion"
+    ],
+    "simu_data": [
+      "de.findevielfalt.games.game2.instance.data.sequence.simulation.SimulationBoardData"
+    ],
+    "instance_start": "de.findevielfalt.games.game2.instance.log.entry.LogEntryStartInstance",
+    "instance_id": "instance_id",
+    "instance_config_id": "config.@id",
+    "sequences2": {
+      "id_field": "sequence_id",
+      "start": {
+        "@class": "de.findevielfalt.games.game2.instance.log.entry.ShowSequenceLogEntry",
+        "action": "START"
+      },
+      "end": {
+        "@class": "de.findevielfalt.games.game2.instance.log.entry.ShowSequenceLogEntry",
+        "action": "PAUSE"
+      }
+    },
+    "coordinates": "location.coordinates",
+    "metadata": {
+      "timestamp": "timestamp",
+      "gamefield": "instance_id",
+      "user": "player_group_name"
+    }
+  },
+  "source": {
+    "type": "Biogames",
+   "username": "ba",
+    "password": "853451",
+    "host": "http://biogames.potato.kinf.wiai.uni-bamberg.de"
+  },
+  "render": [
+    "KMLRender"
+  ]
+}"""
+
+ACTIVITY = """{
+  "logFormat": "zip",
+  "entryType": "@class",
+  "spatials": [
+    "de.findevielfalt.games.game2.instance.log.entry.LogEntryLocation"
+  ],
+  "actions": [],
+  "boards": [
+    "de.findevielfalt.games.game2.instance.log.entry.ShowBoardLogEntry"
+  ],
+  "analyzers": {
+    "analysis.analyzers": [
+      "BiogamesCategorizer",
+      "ActivityMapper"
+    ]
+  },
+  "sequences": {
+    "start": "de.findevielfalt.games.game2.instance.log.entry.LogEntryCache",
+    "end": {
+      "@class": "de.findevielfalt.games.game2.instance.log.entry.LogEntryInstanceAction",
+      "action.@class": "de.findevielfalt.games.game2.instance.action.CacheEnableAction"
+    }
+  },
+  "custom": {
+    "simulation_rounds": [
+      "de.findevielfalt.games.game2.instance.log.entry.LogEntryQuestion"
+    ],
+    "simu_data": [
+      "de.findevielfalt.games.game2.instance.data.sequence.simulation.SimulationBoardData"
+    ],
+    "instance_start": "de.findevielfalt.games.game2.instance.log.entry.LogEntryStartInstance",
+    "instance_id": "instance_id",
+    "instance_config_id": "config.@id",
+    "sequences2": {
+      "id_field": "sequence_id",
+      "start": {
+        "@class": "de.findevielfalt.games.game2.instance.log.entry.ShowSequenceLogEntry",
+        "action": "START"
+      },
+      "end": {
+        "@class": "de.findevielfalt.games.game2.instance.log.entry.ShowSequenceLogEntry",
+        "action": "PAUSE"
+      }
+    },
+    "coordinates": "location.coordinates",
+    "metadata": {
+      "timestamp": "timestamp",
+      "gamefield": "instance_id",
+      "user": "player_group_name"
+    }
+  },
+  "source": {
+    "type": "Biogames",
+   "username": "ba",
+    "password": "853451",
+    "host": "http://biogames.potato.kinf.wiai.uni-bamberg.de"
+  },
+  "render": [
+    "ActivityMapper"
+  ]
+}"""
+
+CONFIGS = {  # TODO
+	"KML": KML,
+	"ActivityMapper": ACTIVITY,
+}
+
+URLS = {
+	"KML": "/",
+	"ActivityMapper": "#",
+}

selector/templates/base.html

@@ -0,0 +1,5 @@
+<!doctype html>
+<title>…</title>
+<link rel="stylesheet" type="text/css" href="{{ url_for('static', filename='style.css') }}">
+<script type="application/javascript" src="{{url_for('static', filename='script.js') }}"></script>
+{% block body %} {% endblock %}

selector/templates/games.html

@@ -0,0 +1,25 @@
+{% extends "base.html" %}
+{% block body %}
+<form action="/start" method="post">
+    <div id="data"> {{logs}}</div>
+    <ul>
+        {% for log in logs %}
+        <li>
+            <input type="checkbox" name="logs" value="{{log['@id']}}">
+            {{log.start_date}}: {{log.player_group_name}} (→{{log.end_date}})
+        </li>
+        <!--{{log}}-->
+        {% endfor %}
+    </ul>
+    <!--input type="checkbox" id="safety"><label for="safety">Confirm selection</label-->
+    <input type="text" id="name" maxlength="128" placeholder="name" name="name"/><br>
+    <select name="config">
+        {% for config in configs %}
+        <option>{{config}}</option>
+        {% endfor %}
+    </select>
+    <input type="submit">
+
+</form>
+<a href="/results">show analysis progress/results</a>
+{% endblock %}

selector/templates/index.html

@@ -0,0 +1,14 @@
+{% extends "base.html" %}
+{% block body %}
+<form action="/login" method="post">
+    <select name="game">
+        {% for game in clients %}
+        <option>{{ game }}</option>
+        {% endfor %}
+    </select>
+    <input type="text" name="username" placeholder="username"/>
+    <input type="password" name="password" placeholder="passwort"/>
+    <input type="submit">
+</form>
+
+{% endblock %}

selector/templates/results.html

@@ -0,0 +1,22 @@
+{% extends "base.html" %}
+{% block body %}
+
+<a href="/games">create new analysis</a>
+
+<div id="results">
+    <ul>
+        {% for job in jobs %}
+        <li> {{jobs[job].status}}: "{{job}}":
+            <ul>
+                {% for r in jobs[job].results %}
+                <li><a href="{{jobs[job] | get_prefix}}{{r | get_name}}">{{r|get_name}} {{jobs[job].start}}</a></li>
+                {% endfor %}
+            </ul>
+
+
+        </li>
+        {% endfor %}
+    </ul>
+</div>
+
+{% endblock %}

selector/webserver.py

@@ -0,0 +1,124 @@
+import json
+import logging
+import typing
+import uuid
+
+import time
+
+from clients.webclients import Client, CLIENTS
+
+from flask import Flask, render_template, request, redirect, session
+
+from tasks import tasks
+from selector.temp_config import CONFIGS, URLS
+
+BIOGAMES_HOST = "http://biogames.potato.kinf.wiai.uni-bamberg.de"
+#BIOGAMES_HOST = "http://www.biodiv2go.de"
+RESULT_HOST = "http://results.ma.potato.kinf.wiai.uni-bamberg.de/"
+
+app = Flask(__name__)
+clients: typing.Dict[str, Client] = {}
+log: logging.Logger = logging.getLogger(__name__)
+
+
+@app.route("/")
+def index():
+	return render_template("index.html", clients=CLIENTS)
+
+
+@app.route("/login", methods=["POST"])
+def login():
+	game = request.form["game"]
+	if not game in CLIENTS:
+		return redirect("/?invalid_game")
+	client = CLIENTS[game](host=BIOGAMES_HOST, username=request.form['username'], password=request.form['password'])
+	if client.login():
+		session['logged_in'] = True
+		session['uid'] = str(uuid.uuid4())
+		session['username'] = request.form['username']
+		session['cookies'] = client.cookies
+		session['game'] = game
+		session['host'] = BIOGAMES_HOST
+		clients[session['uid']] = client
+		return redirect("/results")
+	return redirect("/?fail")
+
+
+@app.route("/results")
+def results():
+	if not ('logged_in' in session and session['logged_in']):
+		return redirect("/")
+	if session['logged_in'] and not session['uid'] in clients:
+		clients[session['uid']] = CLIENTS[session['game']](host=session['host'], **session['cookies'])
+	status = tasks.redis.get(session['username'])
+	if status:
+		job_status = json.loads(status)
+	else:
+		job_status = {}
+	#for job in job_status:
+	#	results = []
+	#	for path in job_status[job]['results']:
+	#		results.append(path.replace(tasks.DATA_PATH, RESULT_HOST))
+	#	print(results) #TODO???
+	return render_template("results.html",  jobs=job_status)
+
+
+@app.route("/games")
+def games():
+	if not ('logged_in' in session and session['logged_in']):
+		return redirect("/")
+	if session['logged_in'] and not session['uid'] in clients:
+		clients[session['uid']] = CLIENTS[session['game']](host=session['host'], **session['cookies'])
+	return render_template("games.html", logs=clients[session['uid']].list(), configs=CONFIGS)
+
+
+@app.route("/start", methods=['POST'])
+def start():
+	print(str(request.form['logs']))
+	status = {
+		"status": "PENDING",
+		"submit": time.strftime("%c"),
+		"log_ids": request.form.getlist('logs'),
+		"config": request.form['config'],
+	}
+	params = {
+		"log_ids": request.form.getlist('logs'),
+		"config": CONFIGS[request.form['config']],
+		"username": session['username'],
+		"cookies": session['cookies'],
+		"host": session['host'],
+		"clientName": session['game'],
+		"name": request.form['name'],
+	}
+	tasks.status_update(session['username'], request.form['name'], status)
+	tasks.analyze.delay(**params)
+	return redirect("/results")
+
+
+@app.route("/status")
+def status():
+	return json.dumps(json.loads(tasks.redis.get(session['username'])), indent=2)
+
+
+@app.template_filter('get_url')
+def get_url(path: str):
+	return path.replace(tasks.DATA_PATH, RESULT_HOST)
+
+@app.template_filter('get_name')
+def get_url(path: str):
+	return path.replace(tasks.DATA_PATH, "")
+
+
+@app.template_filter('get_prefix')
+def get_prefix(job):
+	print(job)
+	try:
+		return RESULT_HOST + URLS[job['config']]
+	except:
+		return RESULT_HOST + "#"
+
+
+
+if __name__ == '__main__':
+	app.config.update({"SECRET_KEY": "59765798-2784-11e8-8d05-db4d6f6606c9"})
+	app.run(host="0.0.0.0", debug=True)

sources/__init__.py

@@ -1,5 +0,0 @@
-from .biogames import Biogames
-
-SOURCES = {
-	"Biogames": Biogames,
-}

sources/biogames.py

@@ -1,85 +0,0 @@
-import json
-import logging
-import typing
-from tempfile import TemporaryDirectory
-
-import os
-
-from sources.source import Source
-
-import shutil
-import requests
-
-log: logging.Logger = logging.getLogger(__name__)
-
-
-class Biogames(Source):
-	def __init__(self):
-		self.headers: typing.Dict[str, str] = {'Accept': 'application/json'}
-		self.cookies: typing.Dict[str, str] = {}
-		self.id2link: typing.Dict[str, str] = {}
-		self.host: str = None
-
-	def connect(self, **kwargs):
-		for i in ['username', 'password', 'url', 'login_url', 'host']:
-			if not i in kwargs:
-				raise ValueError("missing value " + i)
-		csrf_request = requests.get(kwargs['url'])
-		if csrf_request.status_code != 200:
-			raise ConnectionError("unable to obtain CSRF token (" + str(csrf_request) + ")")
-		self.cookies['csrftoken'] = csrf_request.cookies['csrftoken']
-		log.info("obtained CSRF token (" + self.cookies['csrftoken'] + ")")
-		login_payload = {
-			'username': kwargs['username'],
-			'password': kwargs['password'],
-			'next': '',
-			'csrfmiddlewaretoken': 'csrftoken'
-		}
-		login = requests.post(kwargs['login_url'], data=json.dumps(login_payload), cookies=self.cookies)
-		if login.status_code != 200:
-			raise ConnectionError("Unable to authenticate!", login, login.text)
-		self.cookies['sessionid'] = login.cookies['sessionid']
-		log.info("obtained sessionid (" + self.cookies['sessionid'] + ")")
-		self.url = kwargs['url']
-		self.host = kwargs['host']
-		log.info("stored url (" + self.url + ")")
-
-	def list(self):
-		logs = self.get_json(self.url)
-		log.info(len(logs))
-		for i in logs:
-			self.id2link[i["id"]] = i["link"]  # TODO
-		return logs
-
-	def get(self, ids: typing.Collection):
-		dir = TemporaryDirectory()
-		files = []
-		for i in ids:
-			url = self.id2link[i]
-			filename = os.path.join(dir.name, url.split("/")[-1])
-			file = self.download_file(url, filename)
-			if file:
-				files.append(file)
-		return dir
-
-	def download_file(self, url, filename):
-		with open(filename, "wb") as out:
-			try:
-				download = self._get(url)
-				shutil.copyfileobj(download.raw, out)
-				return filename
-			except Exception as e:
-				log.exception(e)
-				os.remove(filename)
-
-	def get_json(self, url):
-		http = self._get(url, stream=False)
-		if not http.ok:
-			raise ConnectionError("HTTP status is not OK", http.url)
-		return http.json()
-
-	def close(self):
-		pass
-
-	def _get(self, url, stream=True):
-		return requests.get(self.host + url, cookies=self.cookies, headers=self.headers, stream=stream)

sources/source.py

@@ -1,18 +0,0 @@
-import typing
-
-
-class Source:
-	def connect(self, **kwargs):
-		raise NotImplementedError
-
-	def list(self):
-		raise NotImplementedError
-
-	def get(self, ids: typing.Collection):
-		raise NotImplementedError
-
-	def get_json(self, url:str) -> dict:
-		raise NotImplementedError
-
-	def close(self):
-		raise NotImplementedError

static/progress/my.js

@@ -1,8 +1,9 @@
-$.getJSON("data/ff8f1e8f-6cf5-4a7b-835b-5e2226c1e771_03b9b6b4-c8ab-4182-8902-1620eebe8889.json", function (data) {    //urach
+//$.getJSON("data/ff8f1e8f-6cf5-4a7b-835b-5e2226c1e771_03b9b6b4-c8ab-4182-8902-1620eebe8889.json", function (data) {    //urach
 //$.getJSON("data/ff8f1e8f-6cf5-4a7b-835b-5e2226c1e771_de7df5b5-edd5-4070-840f-68854ffab9aa.json", function (data) {  //urach
 //$.getJSON("data/90278021-4c57-464e-90b1-d603799d07eb_07da99c9-398a-424f-99fc-2701763a63e9.json", function (data) {  //eichstätt
 //$.getJSON("data/13241906-cdae-441a-aed0-d57ebeb37cac_d33976a6-8a56-4a63-b492-fe5427dbf377.json", function (data) {  //stadtökologie
-//$.getJSON("data/5e64ce07-1c16-4d50-ac4e-b3117847ea43_2f664d7b-f0d8-42f5-8731-c034ef86703e.json", function (data) {  //filderstadt
+$.getJSON("data/5e64ce07-1c16-4d50-ac4e-b3117847ea43_2f664d7b-f0d8-42f5-8731-c034ef86703e.json", function (data) {  //filderstadt
+//$.getJSON("data/17d401a9-de21-49a2-95bc-7dafa53dda64_98edcb70-03db-4465-b185-a9c9574995ce.json", function (data) {  //oeb2016
     var images = {};
     var tiles = {
         "openstreetmap": L.tileLayer('https://{s}.tile.openstreetmap.org/{z}/{x}/{y}.png', {

tasks/__init__.py

@@ -0,0 +1,65 @@
+from .tasks import analyze
+
+__log__ = ["/app/data/008cad400ab848f729913d034a.zip"]
+
+__config__ = """{
+  "logFormat": "zip",
+  "entryType": "@class",
+  "spatials": [
+    "de.findevielfalt.games.game2.instance.log.entry.LogEntryLocation"
+  ],
+  "actions": [
+    "...QuestionAnswerEvent",
+    "...SimuAnswerEvent"
+  ],
+  "boards": [
+    "de.findevielfalt.games.game2.instance.log.entry.ShowBoardLogEntry"
+  ],
+  "analyzers": {
+    "analysis.analyzers": [
+      "BiogamesCategorizer",
+      "LocationAnalyzer"
+    ]
+  },
+  "sequences": {
+    "start": "de.findevielfalt.games.game2.instance.log.entry.LogEntryCache",
+    "end": {
+      "@class": "de.findevielfalt.games.game2.instance.log.entry.LogEntryInstanceAction",
+      "action.@class": "de.findevielfalt.games.game2.instance.action.CacheEnableAction"
+    }
+  },
+  "custom": {
+    "simulation_rounds": [
+      "de.findevielfalt.games.game2.instance.log.entry.LogEntryQuestion"
+    ],
+    "simu_data": [
+      "de.findevielfalt.games.game2.instance.data.sequence.simulation.SimulationBoardData"
+    ],
+    "instance_start": "de.findevielfalt.games.game2.instance.log.entry.LogEntryStartInstance",
+    "instance_id": "instance_id",
+    "instance_config_id": "config.@id",
+    "sequences2": {
+      "id_field": "sequence_id",
+      "start": {
+        "@class": "de.findevielfalt.games.game2.instance.log.entry.ShowSequenceLogEntry",
+        "action": "START"
+      },
+      "end": {
+        "@class": "de.findevielfalt.games.game2.instance.log.entry.ShowSequenceLogEntry",
+        "action": "PAUSE"
+      }
+    },
+    "coordinates": "location.coordinates",
+    "metadata": {
+      "timestamp": "timestamp",
+      "gamefield": "instance_id",
+      "user": "player_group_name"
+    }
+  },
+  "source": {
+    "type": "Biogames",
+   "username": "ba",
+    "password": "853451",
+    "host": "http://biogames.potato.kinf.wiai.uni-bamberg.de"
+  }
+}"""

tasks/tasks.py

@@ -0,0 +1,86 @@
+import json
+import logging
+import shutil
+import uuid
+import os.path
+import os
+
+import redis as redis_lib
+import time
+from celery import Celery
+from analysis import log_analyzer as la
+from analysis.analyzers import KMLRender, ActivityMapperRender
+from clients.webclients import CLIENTS
+
+FLASK_DB = 1
+REDIS_HOST = "redis"
+DATA_PATH = "/app/data/results/"
+
+RENDERERS = {  # TODO
+	"KMLRender": KMLRender,
+	"ActivityMapper": ActivityMapperRender,
+}
+
+app = Celery('tasks', backend='redis://redis', broker='redis://redis')
+redis = redis_lib.StrictRedis(host=REDIS_HOST, db=FLASK_DB)
+log: logging.Logger = logging.getLogger(__name__)
+
+
+def update_status(username, name, state, **kwargs):
+	status = json.loads(redis.get(username))
+	status[name][state[0]] = time.strftime("%c")
+	status[name]['status'] = state[1]
+	for i in kwargs:
+		status[name][i] = kwargs[i]
+	redis.set(username, json.dumps(status))
+
+
+@app.task
+def analyze(config, log_ids, **kwargs):
+	update_status(kwargs['username'], kwargs['name'], ('load', 'LOADING'))
+
+	try:
+		log.info("start analysis")
+		client = CLIENTS[kwargs['clientName']](host=kwargs['host'], **kwargs['cookies'])
+		logs = client.list()
+		id_urls = {str(x['@id']): x['file_url'] for x in logs}
+		urls = [id_urls[i] for i in log_ids]
+		tmpdir = client.download_files(urls)
+		log.info(tmpdir.name, list(os.scandir(tmpdir.name)))
+
+		update_status(kwargs['username'], kwargs['name'], ('start', 'RUNNING'))
+
+		settings = la.parse_settings(config)
+		store = la.run_analysis([p.path for p in os.scandir(tmpdir.name)], settings, la.LOADERS)
+		render = RENDERERS[settings.render[0]]()  # TODO
+		files = render.render(store.get_all())
+
+		uid = str(uuid.uuid4())
+		results = []
+		log.error(files)
+		os.mkdir(os.path.join(DATA_PATH, uid))
+		for file in files:
+			try:
+				head, tail = os.path.split(file)
+				target = os.path.join(DATA_PATH, uid, tail)
+				shutil.move(file, target)
+				results.append(target)
+			except FileNotFoundError as e:
+				log.exception(e)
+		tmpdir.cleanup()
+
+		update_status(kwargs['username'], kwargs['name'], ('done', 'FINISHED'), results=results)
+	except Exception as e:
+		log.exception(e)
+		update_status(kwargs['username'], kwargs['name'], ('abort', 'ERROR'), exception=str(e))
+
+
+def status_update(key, status_key, status):
+	record = redis.get(key)
+	if not record:
+		redis.set(key, json.dumps({status_key: status}))
+	else:
+		data = json.loads(record)
+		data[status_key] = status
+		redis.set(key, json.dumps(data))
+	redis.save()