add SCSI bootstrap script for datasets (#1978)

gms/database/scripts/index/dataset-bootstrap.sql

@@ -0,0 +1,86 @@
+/*
+ First, create a two-column table composed of urn & urnParts
+ This table will only contain distinct URNs which only exist in metadata_aspect but not in metadata_index
+ urn: Actual dataset URN
+ urnParts: Stripped dataset URN which only contains fields of the dataset URN
+ Example dataset_urn_parts table:
++-----------------------------------------------------+------------------------------------+
+| urn                                                 | urn_parts                          |
++-----------------------------------------------------+------------------------------------+
+| urn:li:dataset:(urn:li:dataPlatform:ambry,aaa.2,EI) | urn:li:dataPlatform:ambry,aaa.2,EI |
++-----------------------------------------------------+------------------------------------+
+ */
+CREATE TEMPORARY TABLE dataset_urn_parts
+  SELECT urn,
+         substring(urn, 17, urnLength - 17) AS urn_parts # len("urn:li:dataset:(") = 16 and indexing in SQL starts at 1
+  FROM   (SELECT DISTINCT urn,
+                          length(urn) AS urnLength
+          FROM   metadata_aspect
+          WHERE  urn NOT IN (SELECT DISTINCT urn
+                             FROM   metadata_index
+                             WHERE  aspect="com.linkedin.common.urn.DatasetUrn")
+                 AND urn RLIKE "^urn:li:dataset:\\(urn:li:dataPlatform:[^:,]+,[^:,]+,[^:,]+\\)$"
+                 AND version=0)
+         AS
+         dataset_urn_parts;
+
+/*
+ Parse platform field of the dataset URN and insert these as rows into the index table
+ */
+INSERT INTO metadata_index
+            (urn,
+             aspect,
+             path,
+             stringVal)
+SELECT urn,
+       "com.linkedin.common.urn.DatasetUrn",
+       "/platform"                        AS path,
+       substring_index(urn_parts, ',', 1) AS stringVal
+FROM   dataset_urn_parts;
+
+/*
+ Parse datasetName field of the dataset URN and insert these as rows into the index table
+ */
+INSERT INTO metadata_index
+            (urn,
+             aspect,
+             path,
+             stringVal)
+SELECT urn,
+       "com.linkedin.common.urn.DatasetUrn",
+       "/datasetName"                                               AS path,
+       substring_index(substring_index(urn_parts, ',', 2), ',', -1) AS stringVal
+FROM   dataset_urn_parts;
+
+/*
+ Parse origin field of the dataset URN and insert these as rows into the index table
+ */
+INSERT INTO metadata_index
+            (urn,
+             aspect,
+             path,
+             stringVal)
+SELECT urn,
+       "com.linkedin.common.urn.DatasetUrn",
+       "/origin"                           AS path,
+       substring_index(urn_parts, ',', -1) AS stringVal
+FROM   dataset_urn_parts;
+
+/*
+ Parse platformName field of the data platform URN and insert these as rows into the index table
+ */
+INSERT INTO metadata_index
+            (urn,
+             aspect,
+             path,
+             stringVal)
+SELECT urn,
+       "com.linkedin.common.urn.DatasetUrn",
+       "/platform/platformName"                                     AS path,
+       substring_index(substring_index(urn_parts, ',', 1), ':', -1) AS stringVal
+FROM   dataset_urn_parts;
+
+/*
+ Get rid of dataset_urn_parts table
+ */
+DROP TEMPORARY TABLE dataset_urn_parts;