Add files via upload

check_undirected.py

@@ -0,0 +1,37 @@
+'''
+
+@author: mmrahman,tksaha
+'''
+import sys
+if len(sys.argv) != 2: #program name and 1 arg
+    sys.exit("usage: check_undirected in")
+
+f_in=open(sys.argv[1]+'.txt','r')
+f_out=open(sys.argv[1]+'UND.txt','w')
+#f_dup=open(sys.argv[1]+'DUP.txt','w')
+
+edges=set()
+for line in f_in:
+    splits=line.split()
+    #print(line)
+    if len(splits)==0:
+        continue
+    v1=int(splits[0])
+    v2=int(splits[1])
+    if v1<v2:
+        e=str(v1)+'\t'+str(v2)
+    elif v2<v1:
+        e=str(v2)+'\t'+str(v1)
+    else:
+        print(str(v1)+' '+str(v2))
+        continue   
+    if e not in edges:
+        edges.add(e)
+        f_out.write(e)
+        f_out.write('\n')
+    #else:
+    #    f_dup.write(e)
+    #    f_dup.write('\n')
+
+
+

make_connected_BFS_2.py

@@ -0,0 +1,86 @@
+#!usr/bin/python
+#input graph must be in edge list. No duplicate edge. No self edge.
+import sys
+import queue
+import operator
+
+def BFS():
+    while vc.empty()==False:
+        u=vc.get_nowait()
+        #print('out:')
+        #print(u)
+
+        for y in adj[u]:
+            if v[y]==0:
+                v[y]=1
+                con.add(y)
+                vc.put(y)
+                #print('in:')
+                #print(y)
+        v[u]=2    
+
+
+if len(sys.argv) != 2: #program name and infile
+    sys.exit("usage: check_undirected in")
+
+f_in=open(sys.argv[1],'r')
+f_out=open(sys.argv[1]+'BFSCON','w')
+
+v={}
+adj={}
+vc=queue.Queue()
+
+con=set()
+
+t=0
+for line in f_in:
+    t+=1
+    #if operator.mod(t,100000) == 0:
+    #    print(t)
+
+    splits=line.split()
+    if (len(splits)==0):
+        continue;
+    if splits[0] not in v:
+        v[splits[0]]=0
+        adj[splits[0]]=set()
+    if splits[1] not in v:
+        v[splits[1]]=0
+        adj[splits[1]]=set()
+    adj[splits[0]].add(splits[1])
+    adj[splits[1]].add(splits[0])
+
+print('graph loaded')
+
+for x in v:
+    if v[x]==1:
+        print('error')
+    if v[x]==2:
+        #vc=set()
+        continue
+    if v[x]==0:
+        con.clear()
+        con.add(x)
+        vc=queue.Queue()
+        vc.put(x)
+        v[x]=1
+        BFS()
+    print(len(con))
+    print(len(v))    
+    if len(con) > len(v)/2:
+        break
+f_in.close()
+f_in=open(sys.argv[1],'r')
+
+for line in f_in:
+    splits=line.split()
+    if (len(splits)==0):
+        continue;
+    if splits[0] not in con:
+        continue
+    if splits[1] not in con:
+        continue
+    f_out.write(line)
+    #print(len(con))
+
+print('connect largest ready')

make_proper_input_file_for_mcmc_sample.py

@@ -0,0 +1,48 @@
+import os;
+import sys;
+
+# @author: Tanay Kumar Saha
+# @email:  [email protected]
+
+
+# Parameter 1: input file (graph file)
+# Each line contains a pair (st tab end);
+inputfile = open (sys.argv[1]);
+
+#Output file:  
+output_file = open (sys.argv[1]+"-mcmc-format","w");
+input_mapping_file = open (sys.argv[1]+"-mappings","w");
+
+node_list=[];
+
+# read the file for the first pass
+for	line	in	inputfile:
+	lineelems = line.strip().split("\t");
+	node_list.append (int(lineelems[0]));
+	node_list.append (int(lineelems[1]));
+
+
+uniq_nodeid_list = list (sorted(set (node_list)));
+
+
+# read the input file for the second pass, build the correct input file
+# write the mapping
+
+inputfile = open (sys.argv[1]);
+inputfile_mappings ={};
+
+#populate the mapping
+
+for	pos	in	range (0,len(uniq_nodeid_list)):
+	inputfile_mappings [ uniq_nodeid_list[pos]]= pos;
+	input_mapping_file.write (str(uniq_nodeid_list[pos])+"\t"+str(pos)+os.linesep);
+
+
+
+for	line	in	inputfile:
+	lineelems = line.strip().split("\t");
+	st = inputfile_mappings[int(lineelems[0])];
+	end = inputfile_mappings [int (lineelems [1])];
+	output_file.write (str(st)+"\t"+str(end)+"\t"+"1"+"\t"+"1"+"\t"+"1"+os.linesep);
+
+