Added runtime tests for indexed-nl similarity joins that also return the similarity value.
git-svn-id: https://asterixdb.googlecode.com/svn/branches/asterix_inline_vars@809 eaa15691-b419-025a-1212-ee371bd00084
diff --git a/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ngram-edit-distance-inline.aql b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ngram-edit-distance-inline.aql
new file mode 100644
index 0000000..f061143
--- /dev/null
+++ b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ngram-edit-distance-inline.aql
@@ -0,0 +1,48 @@
+/*
+ * Description : Fuzzy joins two datasets, Customers and Customers2, based on the edit-distance function of their names.
+ * Customers has a 3-gram index on name, and we expect the join to be transformed into an indexed nested-loop join.
+ * We test the inlining of variables that enable the select to be pushed into the join for subsequent optimization with an index.
+ * Success : Yes
+ */
+
+drop dataverse test if exists;
+create dataverse test;
+use dataverse test;
+
+create type AddressType as open {
+ number: int32,
+ street: string,
+ city: string
+}
+
+create type CustomerType as open {
+ cid: int32,
+ name: string,
+ age: int32?,
+ address: AddressType?,
+ interests: [string],
+ children: [ { name: string, age: int32? } ]
+}
+
+create dataset Customers(CustomerType) partitioned by key cid;
+
+create dataset Customers2(CustomerType) partitioned by key cid;
+
+load dataset Customers
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k_olist/customer.adm"),("format"="adm"));
+
+load dataset Customers2
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k_olist/customer.adm"),("format"="adm"));
+
+create index ngram_index on Customers(name) type ngram(3);
+
+write output to nc1:"rttest/index-join_inverted-index-ngram-edit-distance-inline.adm";
+
+for $a in dataset('Customers')
+for $b in dataset('Customers2')
+let $ed := edit-distance($a.name, $b.name)
+where $ed <= 4 and $a.cid < $b.cid
+order by $ed, $a.cid, $b.cid
+return { "aname": $a.name, "bname": $b.name, "ed": $ed }
\ No newline at end of file
diff --git a/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ngram-jaccard-inline.aql b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ngram-jaccard-inline.aql
new file mode 100644
index 0000000..fbe5a69
--- /dev/null
+++ b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ngram-jaccard-inline.aql
@@ -0,0 +1,49 @@
+/*
+ * Description : Fuzzy joins two datasets, DBLP and CSX, based on the similarity-jaccard function of their titles' 3-gram tokens.
+ * DBLP has a 3-gram index on title, and we expect the join to be transformed into an indexed nested-loop join.
+ * We test the inlining of variables that enable the select to be pushed into the join for subsequent optimization with an index.
+ * Success : Yes
+ */
+
+drop dataverse test if exists;
+create dataverse test;
+use dataverse test;
+
+create type DBLPType as closed {
+ id: int32,
+ dblpid: string,
+ title: string,
+ authors: string,
+ misc: string
+}
+
+create type CSXType as closed {
+ id: int32,
+ csxid: string,
+ title: string,
+ authors: string,
+ misc: string
+}
+
+create dataset DBLP(DBLPType) partitioned by key id;
+
+create dataset CSX(CSXType) partitioned by key id;
+
+load dataset DBLP
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/dblp-small/dblp-small-id.txt"),("format"="delimited-text"),("delimiter"=":")) pre-sorted;
+
+load dataset CSX
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/pub-small/csx-small-id.txt"),("format"="delimited-text"),("delimiter"=":"));
+
+create index ngram_index on DBLP(title) type ngram(3);
+
+write output to nc1:"rttest/index-join_inverted-index-ngram-jaccard-inline.adm";
+
+for $a in dataset('DBLP')
+for $b in dataset('CSX')
+let $jacc := similarity-jaccard(gram-tokens($a.title, 3, false), gram-tokens($b.title, 3, false))
+where $jacc >= 0.5f and $a.id < $b.id
+order by $jacc, $a.id, $b.id
+return { "arec": $a.title, "brec": $b.title, "jacc": $jacc }
\ No newline at end of file
diff --git a/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-olist-edit-distance-inline.aql b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-olist-edit-distance-inline.aql
new file mode 100644
index 0000000..4db5161
--- /dev/null
+++ b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-olist-edit-distance-inline.aql
@@ -0,0 +1,48 @@
+/*
+ * Description : Fuzzy joins two datasets, Customers and Customers2, based on the edit-distance function of their interest lists.
+ * Customers has a keyword index on interests, and we expect the join to be transformed into an indexed nested-loop join.
+ * We test the inlining of variables that enable the select to be pushed into the join for subsequent optimization with an index.
+ * Success : Yes
+ */
+
+drop dataverse test if exists;
+create dataverse test;
+use dataverse test;
+
+create type AddressType as open {
+ number: int32,
+ street: string,
+ city: string
+}
+
+create type CustomerType as open {
+ cid: int32,
+ name: string,
+ age: int32?,
+ address: AddressType?,
+ interests: [string],
+ children: [ { name: string, age: int32? } ]
+}
+
+create dataset Customers(CustomerType) partitioned by key cid;
+
+create dataset Customers2(CustomerType) partitioned by key cid;
+
+load dataset Customers
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k_olist/customer.adm"),("format"="adm"));
+
+load dataset Customers2
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k_olist/customer.adm"),("format"="adm"));
+
+create index interests_index on Customers(interests) type keyword;
+
+write output to nc1:"rttest/index-join_inverted-index-olist-edit-distance-inline.adm";
+
+for $a in dataset('Customers')
+for $b in dataset('Customers2')
+let $ed := edit-distance($a.interests, $b.interests)
+where len($a.interests) > 2 and len($b.interests) > 2 and $ed <= 1 and $a.cid < $b.cid
+order by $ed, $a.cid, $b.cid
+return { "arec": $a.interests, "brec": $b.interests, "ed": $ed }
\ No newline at end of file
diff --git a/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-olist-jaccard-inline.aql b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-olist-jaccard-inline.aql
new file mode 100644
index 0000000..f5b0c5a
--- /dev/null
+++ b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-olist-jaccard-inline.aql
@@ -0,0 +1,48 @@
+/*
+ * Description : Fuzzy joins two datasets, Customers and Customers2, based on the Jaccard similarity of their interest lists.
+ * Customers has a keyword index on interests, and we expect the join to be transformed into an indexed nested-loop join.
+ * We test the inlining of variables that enable the select to be pushed into the join for subsequent optimization with an index.
+ * Success : Yes
+ */
+
+drop dataverse test if exists;
+create dataverse test;
+use dataverse test;
+
+create type AddressType as closed {
+ number: int32,
+ street: string,
+ city: string
+}
+
+create type CustomerType as closed {
+ cid: int32,
+ name: string,
+ age: int32?,
+ address: AddressType?,
+ interests: [string],
+ children: [ { name: string, age: int32? } ]
+}
+
+create dataset Customers(CustomerType) partitioned by key cid;
+
+create dataset Customers2(CustomerType) partitioned by key cid;
+
+load dataset Customers
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k_olist/customer.adm"),("format"="adm"));
+
+load dataset Customers2
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k_olist/customer.adm"),("format"="adm"));
+
+create index interests_index on Customers(interests) type keyword;
+
+write output to nc1:"rttest/index-join_inverted-index-olist-jaccard-inline.adm";
+
+for $a in dataset('Customers')
+for $b in dataset('Customers2')
+let $jacc := /*+ indexnl */similarity-jaccard($a.interests, $b.interests)
+where $jacc >= 0.9f and $a.cid < $b.cid and len($a.interests) > 1 and len($b.interests) > 1
+order by $jacc, $a.cid, $b.cid
+return { "a": $a.interests, "b": $b.interests, "jacc": $jacc }
\ No newline at end of file
diff --git a/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ulist-jaccard-inline.aql b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ulist-jaccard-inline.aql
new file mode 100644
index 0000000..8eb173a
--- /dev/null
+++ b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-ulist-jaccard-inline.aql
@@ -0,0 +1,48 @@
+/*
+ * Description : Fuzzy joins two datasets, Customers and Customers2, based on the Jaccard similarity of their interest sets.
+ * Customers has a keyword index on interests, and we expect the join to be transformed into an indexed nested-loop join.
+ * We test the inlining of variables that enable the select to be pushed into the join for subsequent optimization with an index.
+ * Success : Yes
+ */
+
+drop dataverse test if exists;
+create dataverse test;
+use dataverse test;
+
+create type AddressType as closed {
+ number: int32,
+ street: string,
+ city: string
+}
+
+create type CustomerType as closed {
+ cid: int32,
+ name: string,
+ age: int32?,
+ address: AddressType?,
+ interests: {{string}},
+ children: [ { name: string, age: int32? } ]
+}
+
+create dataset Customers(CustomerType) partitioned by key cid;
+
+create dataset Customers2(CustomerType) partitioned by key cid;
+
+load dataset Customers
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k/customer.adm"),("format"="adm"));
+
+load dataset Customers2
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/semistructured/co1k/customer.adm"),("format"="adm"));
+
+create index interests_index on Customers(interests) type keyword;
+
+write output to nc1:"rttest/index-join_inverted-index-ulist-jaccard-inline.adm";
+
+for $a in dataset('Customers')
+for $b in dataset('Customers2')
+let $jacc := /*+ indexnl */ similarity-jaccard($a.interests, $b.interests)
+where $jacc >= 0.9f and $a.cid < $b.cid and len($a.interests) > 1 and len($b.interests) > 1
+order by $jacc, $a.cid, $b.cid
+return { "a": $a.interests, "b": $b.interests, "jacc": $jacc }
\ No newline at end of file
diff --git a/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-word-jaccard-inline.aql b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-word-jaccard-inline.aql
new file mode 100644
index 0000000..8696101
--- /dev/null
+++ b/asterix-app/src/test/resources/runtimets/queries/index-join/inverted-index-word-jaccard-inline.aql
@@ -0,0 +1,49 @@
+/*
+ * Description : Fuzzy joins two datasets, DBLP and CSX, based on the similarity-jaccard function of their titles' word tokens.
+ * DBLP has a keyword index on title, and we expect the join to be transformed into an indexed nested-loop join.
+ * We test the inlining of variables that enable the select to be pushed into the join for subsequent optimization with an index.
+ * Success : Yes
+ */
+
+drop dataverse test if exists;
+create dataverse test;
+use dataverse test;
+
+create type DBLPType as closed {
+ id: int32,
+ dblpid: string,
+ title: string,
+ authors: string,
+ misc: string
+}
+
+create type CSXType as closed {
+ id: int32,
+ csxid: string,
+ title: string,
+ authors: string,
+ misc: string
+}
+
+create dataset DBLP(DBLPType) partitioned by key id;
+
+create dataset CSX(CSXType) partitioned by key id;
+
+load dataset DBLP
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/dblp-small/dblp-small-id.txt"),("format"="delimited-text"),("delimiter"=":")) pre-sorted;
+
+load dataset CSX
+using "edu.uci.ics.asterix.external.dataset.adapter.NCFileSystemAdapter"
+(("path"="nc1://data/pub-small/csx-small-id.txt"),("format"="delimited-text"),("delimiter"=":"));
+
+create index keyword_index on DBLP(title) type keyword;
+
+write output to nc1:"rttest/index-join_inverted-index-word-jaccard-inline.adm";
+
+for $a in dataset('DBLP')
+for $b in dataset('CSX')
+let $jacc := similarity-jaccard(word-tokens($a.title), word-tokens($b.title))
+where $jacc >= 0.5f and $a.id < $b.id
+order by $jacc, $a.id, $b.id
+return { "arec": $a.title, "brec": $b.title, "jacc": $jacc }
\ No newline at end of file