def main(args):
if len(args) < 3:
print(get_usage())
sys.exit(1)
votes = sys.argv[1]
outFile = sys.argv[2]
t = testutils.Timer(ENABLE_TIMER)
context = snap.TTableContext()
VoteS = snap.Schema()
VoteS.Add(snap.TStrTAttrPr("UserId", snap.atInt))
VoteS.Add(snap.TStrTAttrPr("AdminId", snap.atInt))
TVotes = snap.TTable.LoadSS("WikiVotes", VoteS, votes, context, '\t', snap.TBool(False))
t.show("load Votes", TVotes)
GroupBy = snap.TStrV()
GroupBy.Add("UserId")
JointTable = TVotes.SelfSimJoinPerGroup(GroupBy, "AdminId", DISTANCE_ATTRIBUTE, snap.Jaccard, 0.5)
t.show("SimJoinPerGroup complete", JointTable)
JointTable.SelectAtomic("WikiVotes_1.UserId", "WikiVotes_2.UserId", snap.NEQ)
t.show("Select complete", JointTable)
testutils.dump(JointTable, 20);
JointTable.SaveSS(outFile)
def main():
S = snap.Schema()
context = snap.TTableContext()
S.Add(snap.TStrTAttrPr("Animal", snap.atStr))
S.Add(snap.TStrTAttrPr("Size", snap.atStr))
S.Add(snap.TStrTAttrPr("Location", snap.atStr))
S.Add(snap.TStrTAttrPr("Number", snap.atInt))
Animals = snap.TTable.LoadSS("Animals", S, "/dfs/ilfs2/0/ringo/tests/animals.txt", context, '\t', snap.TBool(False))
# Gets animals with size=big
pred_size = snap.TAtomicPredicate(snap.atStr, snap.TBool(True), snap.EQ, "Size", "", 0, 0, "big")
node_size = snap.TPredicateNode(pred_size)
# Get animals with location=Australia
pred_location = snap.TAtomicPredicate(snap.atStr, snap.TBool(True), snap.EQ, "Location", "", 0, 0, "Australia")
node_location = snap.TPredicateNode(pred_location)
# size=big and location=Australia
node1 = snap.TPredicateNode(snap.AND)
node1.AddLeftChild(node_size)
node1.AddRightChild(node_location)
# Get animals with name==location (fabricated to show a non const case
pred_animal_location = snap.TAtomicPredicate(snap.atStr, snap.TBool(False), snap.EQ, "Animal","Location")
node2 = snap.TPredicateNode(pred_animal_location)
# (size=big and location=Australia) or Animal==Location
node_root = snap.TPredicateNode(snap.OR)
node_root.AddLeftChild(node1)
node_root.AddRightChild(node2)
pred = snap.TPredicate(node_root)
Animals.Select(pred)
testutils.dump(Animals)
def main(args):
if len(args) < 3:
print(get_usage())
sys.exit(1)
votes = sys.argv[1]
outFile = sys.argv[2]
t = testutils.Timer(ENABLE_TIMER)
context = snap.TTableContext()
VoteS = snap.Schema()
VoteS.Add(snap.TStrTAttrPr("UserId", snap.atInt))
VoteS.Add(snap.TStrTAttrPr("AdminId", snap.atInt))
TVotes = snap.TTable.LoadSS("WikiVotes", VoteS, votes, context, '\t',
snap.TBool(False))
t.show("load Votes", TVotes)
GroupBy = snap.TStrV()
GroupBy.Add("UserId")
JointTable = TVotes.SelfSimJoinPerGroup(GroupBy, "AdminId",
DISTANCE_ATTRIBUTE, snap.Jaccard,
0.5)
t.show("SimJoinPerGroup complete", JointTable)
JointTable.SelectAtomic("WikiVotes_1.UserId", "WikiVotes_2.UserId",
snap.NEQ)
t.show("Select complete", JointTable)
testutils.dump(JointTable, 20)
JointTable.SaveSS(outFile)
def main(args):
if len(args) < 3:
print(get_usage())
sys.exit(1)
yelp = sys.argv[1]
outFile = sys.argv[2]
t = testutils.Timer(ENABLE_TIMER)
context = snap.TTableContext()
YelpS = snap.Schema()
YelpS.Add(snap.TStrTAttrPr("Name", snap.atStr))
YelpS.Add(snap.TStrTAttrPr("City", snap.atStr))
YelpS.Add(snap.TStrTAttrPr("State", snap.atStr))
YelpS.Add(snap.TStrTAttrPr("Latitude", snap.atFlt))
YelpS.Add(snap.TStrTAttrPr("Longitude", snap.atFlt))
TYelp = snap.TTable.LoadSS("Yelp", YelpS, yelp, context, '\t', snap.TBool(True));
t.show("load Yelp", TYelp)
Cols = snap.TStrV()
Cols.Add("Latitude")
Cols.Add("Longitude")
# Get all business within 5 kilometers of each other
JointTable = TYelp.SelfSimJoin(Cols, DISTANCE_ATTRIBUTE, snap.Haversine, 2)
t.show("SimJoin complete", JointTable)
ProjectionV = snap.TStrV()
ProjectionV.Add("Yelp_1.Name")
ProjectionV.Add("Yelp_1.City")
ProjectionV.Add("Yelp_1.State")
ProjectionV.Add("Yelp_2.Name")
ProjectionV.Add("Yelp_2.City")
ProjectionV.Add("Yelp_2.State")
ProjectionV.Add(DISTANCE_ATTRIBUTE)
JointTable.ProjectInPlace(ProjectionV)
t.show("Project complete")
testutils.dump(JointTable, 100);
JointTable.SaveSS(outFile)
def main():
S = snap.Schema()
context = snap.TTableContext()
S.Add(snap.TStrTAttrPr("Animal", snap.atStr))
S.Add(snap.TStrTAttrPr("Size", snap.atStr))
S.Add(snap.TStrTAttrPr("Location", snap.atStr))
S.Add(snap.TStrTAttrPr("Number", snap.atInt))
Animals = snap.TTable.LoadSS("Animals", S,
"/dfs/ilfs2/0/ringo/tests/animals.txt",
context, '\t', snap.TBool(False))
# Gets animals with size=big
pred_size = snap.TAtomicPredicate(snap.atStr, snap.TBool(True), snap.EQ,
"Size", "", 0, 0, "big")
node_size = snap.TPredicateNode(pred_size)
# Get animals with location=Australia
pred_location = snap.TAtomicPredicate(snap.atStr, snap.TBool(True),
snap.EQ, "Location", "", 0, 0,
"Australia")
node_location = snap.TPredicateNode(pred_location)
# size=big and location=Australia
node1 = snap.TPredicateNode(snap.AND)
node1.AddLeftChild(node_size)
node1.AddRightChild(node_location)
# Get animals with name==location (fabricated to show a non const case
pred_animal_location = snap.TAtomicPredicate(snap.atStr, snap.TBool(False),
snap.EQ, "Animal", "Location")
node2 = snap.TPredicateNode(pred_animal_location)
# (size=big and location=Australia) or Animal==Location
node_root = snap.TPredicateNode(snap.OR)
node_root.AddLeftChild(node1)
node_root.AddRightChild(node2)
pred = snap.TPredicate(node_root)
Animals.Select(pred)
testutils.dump(Animals)
S.Add(snap.TStrTAttrPr("PostId", snap.atInt))
S.Add(snap.TStrTAttrPr("Tag", snap.atStr))
tags = snap.TTable.LoadSS("t2", S, tagsFile, context, '\t', snap.TBool(False))
t.show("load tags", tags)
# Select
# >>> tags.select('Tags = "python"')
tags.SelectAtomicStrConst("Tag", "python", snap.EQ)
t.show("select", tags)
# Join
# >>> questions = posts.join(tags)
questions = posts.Join("PostId", tags, "PostId")
t.show("join", questions)
testutils.dump(questions, 1)
# Project
# >>> questions.project(['PostId', 'UserId', 'AcceptedAnswerId'], in_place = True)
V = snap.TStrV()
V.Add("t1.PostId")
V.Add("t1.UserId")
V.Add("t1.AcceptedAnswerId")
questions.ProjectInPlace(V)
t.show("copy & project", questions)
# Rename
# >>> questions.rename('UserId', 'Asker')
questions.Rename("t1.UserId", "Asker")
t.show("rename", questions)
S.Add(snap.TStrTAttrPr("PostId", snap.atInt))
S.Add(snap.TStrTAttrPr("Tag", snap.atStr))
tags = snap.TTable.LoadSS("t2", S, tagsFile, context, '\t', snap.TBool(False))
t.show("load tags", tags)
# Select
# >>> tags.select('Tags = "python"')
tags.SelectAtomicStrConst("Tag", "python", snap.EQ)
t.show("select", tags)
# Join
# >>> questions = posts.join(tags)
questions = posts.Join("PostId", tags, "PostId")
t.show("join", questions)
testutils.dump(questions, 1)
# Project
# >>> questions.project(['PostId', 'UserId', 'AcceptedAnswerId'], in_place = True)
V = snap.TStrV()
V.Add("t1.PostId")
V.Add("t1.UserId")
V.Add("t1.AcceptedAnswerId")
questions.ProjectInPlace(V)
t.show("copy & project", questions)
# Rename
# >>> questions.rename('UserId', 'Asker')
questions.Rename("t1.UserId", "Asker")
t.show("rename", questions)
import sys
sys.path.append("../use-cases")
import snap
import testutils
import pdb
P1 = snap.TStrTAttrPr("col1", snap.atInt)
P2 = snap.TStrTAttrPr("col2", snap.atInt)
S = snap.Schema()
S.Add(P1)
S.Add(P2)
Context = snap.TTableContext()
T1 = snap.TTable.LoadSS("1", S, "test2.tsv", Context)
testutils.dump(T1)
V = snap.TStrV()
V.Add("col1")
T2 = T1.Project(V, "2")
testutils.dump(T2)
V = snap.TStrV()
V.Add("col2")
T3 = T1.Project(V, "3")
testutils.dump(T3)
T3.Rename("col2", "col1")
T4 = T2.Union(T3, "4")
testutils.dump(T4)
t.show("graph", graph)
# Get authority scores
HTHub = snap.TIntFltH()
HTAuth = snap.TIntFltH()
snap.GetHits(graph, HTHub, HTAuth)
t.show("hits", graph)
t5 = snap.TTable.TableFromHashMap("t5", HTAuth, "UserId", "Authority", context,
snap.TBool(False))
t.show("authority score", t5)
# Select top entries
# >>> t.select('Authority > 0.0')
#t5.SelectAtomicFltConst("Authority", 0.0, snap.GT)
#t.show("select", t5)
# Order by final score (in descending order)
# >>> t5.order(['Authority'], desc = True)
V = snap.TStrV()
V.Add("Authority")
t5.Order(V, "", snap.TBool(False), snap.TBool(False))
t.show("order", t5)
# Save
if not dstdir is None:
t5.SaveSS(os.path.join(dstdir, OUTPUT_TABLE_FILENAME))
t.show("save", t5)
testutils.dump(t5, 20)
# >>> t5.Count('CommentScore', 'UserId')
# >>> t5.Unique()
t5.Count("UserId", "ExpertCount")
pdb.set_trace()
V = snap.TStrV()
V.Add("UserId")
t5.Unique(V)
t.show("count", t5)
pdb.set_trace()
# Select
# >>> t5.select('ExpertCount >= 5')
t5.SelectAtomicIntConst("ExpertCount", 5, snap.GTE)
t.show("select", t5)
# Project
# >>> t5 = t5.project(['UserId'])
V = snap.TStrV()
V.Add("UserId")
t5.ProjectInPlace(V)
t.show("project", t5)
# Save
if not destFile is None:
t5.SaveSS(destFile)
t.show("save", t5)
testutils.dump(t5)
S.Add(snap.TStrTAttrPr("Key", snap.atStr))
S.Add(snap.TStrTAttrPr("Year", snap.atInt))
year = snap.TTable.LoadSS("2", S, yearFile, context, '\t', snap.TBool(False))
t.show("load year table", year)
# Select
# >>> year.select('Year >= 2005')
year.SelectAtomicIntConst("Year", 2005, snap.GTE)
t.show("select", year)
# Join
# >>> table = authors.join(year, ['Key'], ['Key'])
table = authors.Join("Key", year, "Key")
t.show("join", table)
testutils.dump(table, 2)
# Self-join
# >>> table.selfjoin(table, ['Key'])
table = table.SelfJoin("Key")
t.show("join", table)
# Select
# >>> table.select('Author_1 != Author_2')
table.SelectAtomic("1_2_1.1.Author", "1_2_2.1.Author", snap.NEQ)
t.show("select", table)
# Create network
# >>> table.graph('Author_1', 'Author_2', directed=False)
table.SetSrcCol("1_2_1.1.Author")
table.SetDstCol("1_2_2.1.Author")
t.show("graph", graph)
#graph.Dump()
# Get authority scores
HTHub = snap.TIntFltH()
HTAuth = snap.TIntFltH()
snap.GetHits(graph, HTHub, HTAuth)
t.show("hits", graph)
t5 = snap.TTable.TableFromHashMap("t5", HTAuth, "UserId", "Authority", context, snap.TBool(False))
t.show("authority score", t5)
# Select top entries
# >>> t.select('Authority > 0.0')
#t5.SelectAtomicFltConst("Authority", 0.0, snap.GT)
#t.show("select", t5)
# Order by final score (in descending order)
# >>> t5.order(['Authority'], desc = True)
V = snap.TStrV()
V.Add("Authority")
t5.Order(V, "", snap.TBool(False), snap.TBool(False))
t.show("order", t5)
# Save
if not destFile is None:
t5.SaveSS(destFile)
t.show("save", t5)
testutils.dump(t5, 20)
import sys
sys.path.append("../use-cases")
import snap
import testutils
import pdb
P1 = snap.TStrTAttrPr("col1", snap.atInt)
P2 = snap.TStrTAttrPr("col2", snap.atInt)
S = snap.Schema()
S.Add(P1)
S.Add(P2)
Context = snap.TTableContext()
T1 = snap.TTable.LoadSS("1", S, "test2.tsv", Context)
testutils.dump(T1)
V = snap.TStrV()
V.Add("col1")
T2 = T1.Project(V, "2")
testutils.dump(T2)
V = snap.TStrV()
V.Add("col2")
T3 = T1.Project(V, "3")
testutils.dump(T3)
T3.Rename("col2","col1")
T4 = T2.Union(T3, "4")
testutils.dump(T4)
import sys
sys.path.append("../utils")
import snap
import testutils
if __name__ == '__main__':
srcfile = '/dfs/ilfs2/0/ringo/StackOverflow_joined/debug.tsv'
context = snap.TTableContext()
print "Loading table..."
schema = snap.Schema()
schema.Add(snap.TStrTAttrPr("Val", snap.atInt))
table = snap.TTable.LoadSS("1", schema, srcfile, context, "\t", snap.TBool(False))
print "Selecting rows with val == 0 in place..."
table.SelectAtomicIntConst("Val", 0, snap.EQ)
print "Number of rows in result: %d" % table.GetNumValidRows()
print "10 first rows of table:"
testutils.dump(table, 10)
# Count
# >>> t5.Count('CommentScore', 'UserId')
# >>> t5.Unique()
t5.Count("UserId", "ExpertCount")
pdb.set_trace()
V = snap.TStrV()
V.Add("UserId")
t5.Unique(V)
t.show("count", t5)
pdb.set_trace()
# Select
# >>> t5.select('ExpertCount >= 5')
t5.SelectAtomicIntConst("ExpertCount", 5, snap.GTE)
t.show("select", t5)
# Project
# >>> t5 = t5.project(['UserId'])
V = snap.TStrV()
V.Add("UserId")
t5.ProjectInPlace(V)
t.show("project", t5)
# Save
if not destFile is None:
t5.SaveSS(destFile)
t.show("save", t5)
testutils.dump(t5)
