def import_file(database_name, rectangle_file):
# read rectangles from file
# limit = 500
limit = None
(rectangles, count) = rectangle_file_reader.read_rectangles(rectangle_file, limit=limit)
print("Read %d rectangles" % (count))
# drop existing database
r = requests.delete(couch_url + database_name)
print("Trying to delete existing database: " + r.text)
# create database
r = requests.put(couch_url + database_name)
print("Creating database '" + database_name + "': " + r.text)
if r.status_code != 201:
raise Exception("Could not create database")
# insert rectangles in chunks
chunk = []
current_chunk_size = 0
index = 0
for rectangle in rectangles:
chunk.append((index, rectangle))
current_chunk_size += 1
index += 1
if current_chunk_size >= chunk_size:
send(chunk)
chunk = []
current_chunk_size = 0
# send remaining rectangles
if current_chunk_size >= 0:
send(chunk)
# add spatial view that simply returns the document id
r = requests.put(
couch_url + database_name + "/_design/benchmark",
data='{"spatial":{"justid":"(function(doc) { if (doc.geometry) { emit(doc.geometry, doc._id);}})"}}',
)
print("Adding spatial view '" + database_name + "/_design/benchmark" + "': " + r.text)
if r.status_code != 201:
raise Exception("Could not add spatial view")
# measure time to build the R-Tree
print("Started building R-Tree at " + time.strftime("%Y-%m-%d %H:%M:%S", time.gmtime()) + " ...")
start = time.time()
r = requests.get(couch_url + database_name + "/_design/benchmark/_spatial/justid?bbox=0,0,0,0&count=true")
time_difference = time.time() - start
print("Building R-Tree took %.5f seconds" % time_difference)
print("Finished at " + time.strftime("%Y-%m-%d %H:%M:%S", time.gmtime()))
def run_queries(database_name, dataset):
queries_file = "data/framework-queries/" + query_type + "/" + dataset + ".rec"
# Leaf-accesses log file
file = "/tmp/LeafAccesses.txt"
limit = None
(query_rectangles, count) = rectangle_file_reader.read_rectangles(queries_file, limit=limit)
print("Read %d query rectangles" % (count))
# delete previous leaf-accesses file
if os.path.exists(file):
os.remove(file)
# measure time to make the queries
print("Making queries ...")
start = time.time()
for rectangle in query_rectangles:
(xMin, yMin, xMax, yMax) = rectangle
queryBbox = "%.10f,%.10f,%.10f,%.10f" % (xMin, yMin, xMax, yMax)
r = requests.get(couch_url + database_name + "/_design/benchmark/_spatial/justid?bbox=" + queryBbox + "&count=true")
time_difference = time.time() - start
print("%.f queries took %.5f seconds" % (count, time_difference))
# check if leaf accesses have been logged
try:
with open(file) as f:
# print time of last modification
mod_time = os.stat(file)[stat.ST_MTIME]
print("File '" + file + "' was last modified at " +
time.strftime("%Y-%m-%d %H:%M:%S", time.gmtime(mod_time)))
# calculate statistics
count = 0
leafAccessesTotal = 0
for line in f:
leafAccesses = int(line.strip())
leafAccessesTotal += leafAccesses
count += 1
print("Number of queries: %d" % count)
print("Total leaf accesses: %d" % leafAccessesTotal)
print("Avg. leaf accesses per query: %.2f" % (float(leafAccessesTotal)/count))
except IOError as e:
print("Leaf accesses have not been tracked in '" + file + "'")