1+ package zingg .common .core .block ;
2+
3+ import org .junit .jupiter .api .Assertions ;
4+ import org .junit .jupiter .api .Test ;
5+ import zingg .common .client .Arguments ;
6+ import zingg .common .client .ArgumentsUtil ;
7+ import zingg .common .client .IArguments ;
8+ import zingg .common .client .ZFrame ;
9+ import zingg .common .client .ZinggClientException ;
10+ import zingg .common .client .util .DFObjectUtil ;
11+ import zingg .common .core .block .dataUtility .CsvReader ;
12+ import zingg .common .core .block .dataUtility .DataUtility ;
13+ import zingg .common .core .block .model .Customer ;
14+ import zingg .common .core .block .model .CustomerDupe ;
15+ import zingg .common .core .util .BlockingTreeUtil ;
16+ import zingg .common .core .util .HashUtil ;
17+
18+ import java .util .Iterator ;
19+ import java .util .List ;
20+ import java .util .Objects ;
21+
22+ import static java .lang .Math .max ;
23+
24+ public abstract class TestBlockingTreeUtil <S , D , R , C , T > {
25+
26+ protected String TEST_DATA_BASE_LOCATION ;
27+ private int maxDepth = 1 , totalNodes = 0 ;
28+ private static String TEST_FILE = "test.csv" ;
29+ private static String CONFIG_FILE = "config.json" ;
30+ private final DataUtility dataUtility ;
31+
32+ public TestBlockingTreeUtil () {
33+ setTestDataBaseLocation ();
34+ this .dataUtility = new DataUtility (new CsvReader ());
35+ }
36+
37+ @ Test
38+ public void testSameBlockingTreeWithoutVariance () throws Exception , ZinggClientException {
39+ List <Customer > testCustomers = dataUtility .getCustomers (TEST_DATA_BASE_LOCATION + "/" + TEST_FILE );
40+ //setting variance as false
41+ List <CustomerDupe > testCustomerDupes = dataUtility .getCustomerDupes (TEST_DATA_BASE_LOCATION + "/" + TEST_FILE , false );
42+ DFObjectUtil <S , D , R , C > dfObjectUtil = getDFObjectUtil ();
43+
44+ ZFrame <D , R , C > zFrameTest = dfObjectUtil .getDFFromObjectList (testCustomers , Customer .class );
45+ ZFrame <D , R , C > zFramePositives = dfObjectUtil .getDFFromObjectList (testCustomerDupes , CustomerDupe .class );
46+
47+ testSameBlockingTree (zFrameTest , zFramePositives );
48+ }
49+
50+ @ Test
51+ public void testSameBlockingTreeWithVariance () throws Exception , ZinggClientException {
52+ List <Customer > testCustomers = dataUtility .getCustomers (TEST_DATA_BASE_LOCATION + "/" + TEST_FILE );
53+ //setting variance as true
54+ List <CustomerDupe > testCustomerDupes = dataUtility .getCustomerDupes (TEST_DATA_BASE_LOCATION + "/" + TEST_FILE , true );
55+ DFObjectUtil <S , D , R , C > dfObjectUtil = getDFObjectUtil ();
56+
57+ ZFrame <D , R , C > zFrameTest = dfObjectUtil .getDFFromObjectList (testCustomers , Customer .class );
58+ ZFrame <D , R , C > zFramePositives = dfObjectUtil .getDFFromObjectList (testCustomerDupes , CustomerDupe .class );
59+
60+ testSameBlockingTree (zFrameTest , zFramePositives );
61+ }
62+
63+ public void testSameBlockingTree (ZFrame <D , R , C > zFrameTest , ZFrame <D , R , C > zFramePositives ) throws Exception , ZinggClientException {
64+ setTestDataBaseLocation ();
65+ BlockingTreeUtil <S , D , R , C , T > blockingTreeUtil = getBlockingTreeUtil ();
66+ HashUtil <S , D , R , C , T > hashUtil = getHashUtil ();
67+
68+
69+ IArguments args = new ArgumentsUtil (Arguments .class ).createArgumentsFromJSON (
70+ TEST_DATA_BASE_LOCATION + "/" + CONFIG_FILE ,
71+ "" );
72+ args .setBlockSize (8 );
73+
74+ long ts = System .currentTimeMillis ();
75+ Tree <Canopy <R >> blockingTreeOptimized = blockingTreeUtil .createBlockingTree (zFrameTest , zFramePositives , 1 , -1 ,
76+ args , hashUtil .getHashFunctionList (), HashUtility .CACHED );
77+ System .out .println ("************ time taken to create optimized blocking tree ************ " + (System .currentTimeMillis () - ts ));
78+
79+ ts = System .currentTimeMillis ();
80+ Tree <Canopy <R >> blockingTreeDefault = blockingTreeUtil .createBlockingTree (zFrameTest , zFramePositives , 1 , -1 ,
81+ args , hashUtil .getHashFunctionList (), HashUtility .DEFAULT );
82+ System .out .println ("************ time taken to create blocking tree ************ " + (System .currentTimeMillis () - ts ));
83+
84+ int depth = 1 ;
85+ //assert both the trees are equal
86+ Assertions .assertTrue (dfsSameTreeValidation (blockingTreeDefault , blockingTreeOptimized , depth ));
87+
88+ System .out .println ("-------- max depth of trees -------- " + maxDepth );
89+ System .out .println ("-------- total nodes in a trees -------- " + totalNodes );
90+ }
91+
92+
93+ private boolean dfsSameTreeValidation (Tree <Canopy <R >> node1 , Tree <Canopy <R >> node2 , int depth ) {
94+ totalNodes ++;
95+ maxDepth = max (maxDepth , depth );
96+
97+ //if both the node1 and node2 are null, return true
98+ if (node1 == null && node2 == null ){
99+ return true ;
100+ }
101+ //if only one of node1 or node2 is null, return false
102+ if (node1 == null || node2 == null ){
103+ return false ;
104+ }
105+
106+ if (!performValidationOnNode1AndNode2 (node1 , node2 )) {
107+ return false ;
108+ }
109+
110+ Iterator <Tree <Canopy <R >>> canopyIterator1 = node1 .getSubTrees ().iterator ();
111+ Iterator <Tree <Canopy <R >>> canopyIterator2 = node2 .getSubTrees ().iterator ();
112+
113+ boolean isEqual = true ;
114+
115+ //recurse through sub-trees
116+ while (canopyIterator1 .hasNext () && canopyIterator2 .hasNext ()) {
117+ isEqual &= dfsSameTreeValidation (canopyIterator1 .next (), canopyIterator2 .next (), depth + 1 );
118+ }
119+
120+ return isEqual ;
121+ }
122+
123+
124+ private boolean performValidationOnNode1AndNode2 (Tree <Canopy <R >> node1 , Tree <Canopy <R >> node2 ) {
125+ boolean functionEqual = isNodeFunctionEqual (node1 .getHead (), node2 .getHead ());
126+ boolean contextEqual = isNodeContextEqual (node1 .getHead (), node2 .getHead ());
127+ boolean hashEqual = isNodeHashEqual (node1 .getHead (), node2 .getHead ());
128+ boolean subtreeSizeEqual = isNodeSubTreesSizeEqual (node1 , node2 );
129+
130+ return functionEqual && contextEqual && hashEqual && subtreeSizeEqual ;
131+ }
132+ private boolean isNodeFunctionEqual (Canopy <R > node1Head , Canopy <R > node2Head ) {
133+ if (node1Head .getFunction () == null && node2Head .getFunction () == null ) {
134+ return true ;
135+ } else if (node1Head .getFunction () == null || node2Head .getFunction () == null ) {
136+ return false ;
137+ } else {
138+ return Objects .equals (node1Head .getFunction ().getName (), node2Head .getFunction ().getName ());
139+ }
140+ }
141+
142+ private boolean isNodeHashEqual (Canopy <R > node1Head , Canopy <R > node2Head ) {
143+ return Objects .equals (node1Head .getHash (), node2Head .getHash ());
144+ }
145+
146+ private boolean isNodeContextEqual (Canopy <R > node1Head , Canopy <R > node2Head ) {
147+
148+ if (node1Head .getContext () == null && node2Head .getContext () == null ) {
149+ return true ;
150+ } else if (node1Head .getContext () == null || node2Head .getContext () == null ) {
151+ return false ;
152+ } else {
153+ return Objects .equals (node1Head .getContext ().getName (), node2Head .getContext ().getName ());
154+ }
155+ }
156+
157+ private boolean isNodeSubTreesSizeEqual (Tree <Canopy <R >> node1 , Tree <Canopy <R >> node2 ) {
158+ return node1 .getSubTrees ().size () == node2 .getSubTrees ().size ();
159+ }
160+
161+
162+ protected abstract DFObjectUtil <S , D , R , C > getDFObjectUtil ();
163+ protected abstract BlockingTreeUtil <S , D , R , C , T > getBlockingTreeUtil ();
164+ protected abstract HashUtil <S , D , R , C , T > getHashUtil ();
165+ protected abstract void setTestDataBaseLocation ();
166+ }
0 commit comments