IRSA-7700: Add support for SQL WINDOW function

src/Query/ADQL_parser.hxx

@@ -553,4 +553,8 @@ struct ADQL_parser : boost::spirit::qi::grammar<std::string::const_iterator,
                             ADQL::Coord_Sys::Reference_Position(),
                             boost::spirit::ascii::space_type>
             reference_position;
+
+    boost::spirit::qi::rule<std::string::const_iterator, ADQL::Window_Function(),
+                            boost::spirit::ascii::space_type>
+            window_function;
 };

src/Query/ADQL_parser/ADQL_parser/init_factor.cxx

@@ -256,17 +256,36 @@ void ADQL_parser::init_factor() {
                           -('(' > unsigned_integer > ')') >> ')'];
     cast_function.name("cast_function");
 
+    window_function =
+            hold[whitelisted_function_name >> lit('(') >> lit(')') >>
+                 &(ascii::no_case[lit("OVER")])][at_c<0>(_val) = _1] >>
+            lexeme[ascii::no_case[lit("OVER")] >> &boost::spirit::qi::space] >>
+            lit('(') >>
+            lexeme[ascii::no_case[lit("PARTITION")] >> &boost::spirit::qi::space] >>
+            lexeme[ascii::no_case[lit("BY")] >> &boost::spirit::qi::space] >>
+            (value_expression % lit(','))[at_c<1>(_val) = _1] >>
+            lexeme[ascii::no_case[lit("ORDER")] >> &boost::spirit::qi::space] >>
+            lexeme[ascii::no_case[lit("BY")] >> &boost::spirit::qi::space] >>
+            (value_expression % lit(','))[at_c<2>(_val) = _1] >> lit(')');
+    window_function.name("window_function");
+
     position_function %=
             hold[ascii::no_case["POSITION"] >> '(' >> character_string_literal >>
                  &no_skip[boost::spirit::qi::space] >> ascii::no_case["IN"] >>
                  &no_skip[boost::spirit::qi::space] >> column_reference >> ')'];
     position_function.name("position_function");
 
     // FIXME: numeric_value_function should have
-    // numeric_geometry_function
+    // numeric_geometry_function.
+
+    // Note that window_function must come before whitelisted_function
+    // since ROW_NUMBER would otherwise be consumed by the whitelisted
+    // function rule first.
+
     numeric_value_function %= trig_function | math_function | cast_function |
                               position_function | non_predicate_geometry_function |
-                              whitelisted_function | sql_no_arg_function;
+                              window_function | whitelisted_function |
+                              sql_no_arg_function;
 
     numeric_value_function.name("numeric_value_function");
     // Flipped the order here, because a value_expression can match a
@@ -304,6 +323,7 @@ void ADQL_parser::init_factor() {
     BOOST_SPIRIT_DEBUG_NODE(whitelisted_function_name);
     BOOST_SPIRIT_DEBUG_NODE(whitelisted_function_param);
     BOOST_SPIRIT_DEBUG_NODE(whitelisted_function);
+    BOOST_SPIRIT_DEBUG_NODE(window_function);
     BOOST_SPIRIT_DEBUG_NODE(cast_function);
     BOOST_SPIRIT_DEBUG_NODE(position_function);
     BOOST_SPIRIT_DEBUG_NODE(numeric_value_function);

src/Query/ADQL_parser/ADQL_parser/init_reserved_words.cxx

@@ -82,8 +82,9 @@ void ADQL_parser::init_reserved_words() {
                             ascii::no_case["PRIVILEGES"] | ascii::no_case["PRIOR"] |
                             ascii::no_case["PRIMARY"] | ascii::no_case["PRESERVE"] |
                             ascii::no_case["PREPARE"] | ascii::no_case["PRECISION"] |
-                            ascii::no_case["POSITION"] | ascii::no_case["PARTIAL"] |
-                            ascii::no_case["PAD"] | ascii::no_case["OVERLAPS"] |
+                            ascii::no_case["POSITION"] | ascii::no_case["PARTITION"] |
+                            ascii::no_case["PARTIAL"] | ascii::no_case["PAD"] |
+                            ascii::no_case["OVERLAPS"] | ascii::no_case["OVER"] |
                             ascii::no_case["OUTPUT"] | ascii::no_case["OUTER"] |
                             ascii::no_case["ORDER"] | ascii::no_case["OR"];
 

src/Query/Query_Specification/Factor/Numeric_Primary/Numeric_Value_Function.hxx

@@ -4,6 +4,7 @@
 
 #include "../../Non_Predicate_Geometry_Function.hxx"
 #include "../../Whitelisted_Function_Wrap.hxx"
+#include "../../Window_Function_Wrap.hxx"
 #include "Numeric_Value_Function/Cast_Function.hxx"
 #include "Numeric_Value_Function/Math_Function.hxx"
 #include "Numeric_Value_Function/Position_Function.hxx"
@@ -14,7 +15,7 @@ class Numeric_Value_Function {
 public:
     typedef boost::variant<Trig_Function, Math_Function, Cast_Function,
                            Position_Function, Non_Predicate_Geometry_Function,
-                           Whitelisted_Function_Wrap, std::string>
+                           Window_Function_Wrap, Whitelisted_Function_Wrap, std::string>
             Variant;
     Variant variant;
     bool empty() const;

src/Query/Query_Specification/Factor/Numeric_Primary/Numeric_Value_Function/empty.cxx

@@ -1,5 +1,6 @@
 #include "../../../../empty_variant.hxx"
 #include "../../../Whitelisted_Function.hxx"
+#include "../../../Window_Function.hxx"
 #include "../Numeric_Value_Function.hxx"
 
 bool ADQL::Numeric_Value_Function::empty() const { return empty_variant(variant); }

src/Query/Query_Specification/Factor/Numeric_Primary/Numeric_Value_Function/ostream.cxx

@@ -1,4 +1,5 @@
 #include "../../../Whitelisted_Function.hxx"
+#include "../../../Window_Function.hxx"
 #include "../Numeric_Value_Function.hxx"
 
 namespace ADQL {

src/Query/Query_Specification/Select.hxx

@@ -4,6 +4,7 @@
 #include "Non_As.hxx"
 #include "Value_Expression_Primary.hxx"
 #include "Whitelisted_Function.hxx"
+#include "Window_Function.hxx"
 
 namespace ADQL {
 class Select {

src/Query/Query_Specification/Whitelisted_Function_Wrap.hxx

@@ -6,5 +6,6 @@
 namespace ADQL {
 class Whitelisted_Function;
 typedef boost::recursive_wrapper<Whitelisted_Function> Whitelisted_Function_Wrap;
+
 std::ostream &operator<<(std::ostream &os, const Whitelisted_Function_Wrap &s);
 }  // namespace ADQL

src/Query/Query_Specification/Window_Function.hxx

@@ -0,0 +1,45 @@
+#pragma once
+
+#include <iostream>
+#include <string>
+#include <vector>
+
+#include <boost/fusion/include/adapt_struct.hpp>
+#include <boost/fusion/include/io.hpp>
+
+#include "Value_Expression.hxx"
+
+namespace ADQL {
+class Window_Function {
+public:
+    std::string function_name;
+    std::vector<ADQL::Value_Expression> partition_by;
+    std::vector<ADQL::Value_Expression> order_by;
+    bool empty() const { return function_name.empty(); }
+};
+
+inline std::ostream &operator<<(std::ostream &os,
+                                const ADQL::Window_Function &window_function) {
+    os << window_function.function_name << "() OVER (PARTITION BY ";
+    for (auto p = window_function.partition_by.begin();
+         p != window_function.partition_by.end();) {
+        os << *p;
+        ++p;
+        if (p != window_function.partition_by.end()) os << ", ";
+    }
+    os << " ORDER BY ";
+    for (auto o = window_function.order_by.begin();
+         o != window_function.order_by.end();) {
+        os << *o;
+        ++o;
+        if (o != window_function.order_by.end()) os << ", ";
+    }
+    os << ")";
+    return os;
+}
+}  // namespace ADQL
+
+BOOST_FUSION_ADAPT_STRUCT(ADQL::Window_Function,
+    (std::string, function_name)
+    (std::vector<ADQL::Value_Expression>, partition_by)
+    (std::vector<ADQL::Value_Expression>, order_by))

src/Query/Query_Specification/Window_Function_Wrap.hxx

@@ -0,0 +1,10 @@
+#pragma once
+#include <boost/variant/recursive_wrapper.hpp>
+#include <iostream>
+
+namespace ADQL {
+class Window_Function;
+typedef boost::recursive_wrapper<Window_Function> Window_Function_Wrap;
+bool empty(const Window_Function_Wrap &s);
+std::ostream &operator<<(std::ostream &os, const Window_Function_Wrap &s);
+}  // namespace ADQL

src/Query/Query_Specification/Window_Function_Wrap/empty.cxx

@@ -0,0 +1,6 @@
+#include "../Window_Function.hxx"
+#include "../Window_Function_Wrap.hxx"
+
+namespace ADQL {
+bool empty(const Window_Function_Wrap &s) { return s.get().empty(); }
+}  // namespace ADQL

src/Query/Query_Specification/Window_Function_Wrap/ostream.cxx

@@ -0,0 +1,9 @@
+#include "../Window_Function.hxx"
+#include "../Window_Function_Wrap.hxx"
+
+namespace ADQL {
+std::ostream &operator<<(std::ostream &os,
+                         const ADQL::Window_Function_Wrap &wrap) {
+    return os << wrap.get();
+}
+}  // namespace ADQL

test/parse_adql.cxx

@@ -811,6 +811,97 @@ int main(int argc, char *argv[]) {
             "tap_ancillary.irsa_directory_datatypes v ON d.collection=v.collection "
             "WHERE d.semantics like '%primary%' ORDER BY "
             "LOWER(facility_name),d.collection,instrument",
+
+			// IRSA-7700: support for WINDOW/PARTITION/OVER
+            "SELECT ROW_NUMBER()  Over (PARTITION BY facility, instrument ORDER BY "
+            "dist) FROM my_table",
+
+            "SELECT * FROM (SELECT SIA2_MINIMAL_JOIN.*, ROW_NUMBER() OVER (PARTITION "
+            "BY SIA2_MINIMAL_JOIN.facility, SIA2_MINIMAL_JOIN.instrument, "
+            "SIA2_MINIMAL_JOIN.upload_row_id ORDER BY "
+            "SIA2_MINIMAL_JOIN.dist_to_point_meters) as rownum FROM SIA2_MINIMAL_JOIN) "
+            "ranked WHERE rownum = 1 AND DISTANCE(POINT('ICRS', ranked.ra, "
+            "ranked.dec), POINT('ICRS', 1.0, 2.0)) < 0.1",
+
+            "SELECT * FROM (SELECT SIA2_MINIMAL_JOIN.*, ROW_NUMBER() OVER (PARTITION "
+            "BY SIA2_MINIMAL_JOIN.facility, SIA2_MINIMAL_JOIN.instrument, "
+            "SIA2_MINIMAL_JOIN.upload_row_id ORDER BY "
+            "SIA2_MINIMAL_JOIN.dist_to_point_meters) as rownum, DISTANCE(POINT('ICRS', "
+            "SIA2_MINIMAL_JOIN.ra, SIA2_MINIMAL_JOIN.dec), POINT('ICRS', 1.0, 2.0)) as "
+            "dist FROM SIA2_MINIMAL_JOIN) ranked WHERE rownum = 1",
+
+            "WITH SIA2_MINIMAL_JOIN AS (SELECT o.obsid as obsid, p.planeid  as "
+            "planeid, o.telescope_name as facility_name, o.instrument_name as "
+            "instrument_name, CAST ('1' AS BIGINT) as upload_row_id FROM "
+            "(caom.simulated_observation o JOIN caom.simulated_plane p ON o.obsid = "
+            "p.obsid)  ) SELECT coord1(p.pt) as s_ra, coord2(p.pt) as s_dec, "
+            "SIA2_CLOSEST_SUBQUERY.dist_to_point_meters / 111194.68229846345 as "
+            "min_dist_to_point FROM (SELECT "
+            "obsid,planeid,facility_name,instrument_name,upload_row_id FROM (SELECT "
+            "SIA2_MINIMAL_JOIN, ROW_NUMBER() OVER (PARTITION BY "
+            "SIA2_MINIMAL_JOIN.facility, SIA2_MINIMAL_JOIN.instrument, "
+            "SIA2_MINIMAL_JOIN.upload_row_id ORDER BY "
+            "SIA2_MINIMAL_JOIN.dist_to_point_meters) as rownum FROM SIA2_MINIMAL_JOIN) "
+            "ranked WHERE rownum = 1) SIA2_CLOSEST_SUBQUERY, ((caom.observation o JOIN "
+            "caom.plane p ON o.obsid = p.obsid) JOIN caom.artifact a ON p.planeid = "
+            "a.planeid) WHERE SIA2_CLOSEST_SUBQUERY.obsid = o.obsid AND "
+            "SIA2_CLOSEST_SUBQUERY.planeid = p.planeid AND "
+            "(SIA2_CLOSEST_SUBQUERY.facility IS NULL OR SIA2_CLOSEST_SUBQUERY.facility "
+            "= o.telescope_name) AND (SIA2_CLOSEST_SUBQUERY.instrument IS NULL OR "
+            "SIA2_CLOSEST_SUBQUERY.instrument = o.instrument_name)",
+
+            "WITH SIA2_MINIMAL_JOIN AS (SELECT o.obsid as obsid, p.planeid as planeid, "
+            "CAST ('1' AS BIGINT) as upload_row_id FROM (caom.simulated_observation o "
+            "JOIN caom.simulated_plane p ON o.obsid = p.obsid) WHERE "
+            "(((p.dataproducttype = 'image') OR (p.dataproducttype = 'cube'))) ) "
+            "SELECT coord1(p.pt) as s_ra, coord2(p.pt) as s_dec, o.telescope_name as "
+            "facility_name, o.instrument_name as instrument_name, a.producttype as "
+            "dataproduct_subtype, p.calibrationlevel as calib_level, p.dataproducttype "
+            "as dataproduct_type, p.energy_bandpassname as energy_bandpassname, "
+            "p.energy_emband as energy_emband, o.observationid as obs_id, "
+            "p.position_resolution as s_resolution, p.energy_bounds_lower as em_min, "
+            "p.energy_bounds_upper as em_max, p.energy_resolvingpower as em_res_power, "
+            "o.proposal_title as proposal_title, CASE WHEN position('http' IN a.uri) > "
+            "0 AND position('ipac.caltech.edu' IN a.uri) = 0 THEN a.uri ELSE "
+            "'HOST:PORT/' || strip_url_prefix(a.uri, 'https.*edu/') END as access_url, "
+            "a.contenttype as access_format, CAST (CEIL(a.contentlength/1000.0) AS "
+            "BIGINT) as access_estsize, p.time_exposure as t_exptime, CASE WHEN "
+            "(p.poly IS NOT NULL) THEN poly_to_region(p.poly) ELSE pt_to_region(p.pt) "
+            "END as s_region, o.collection as obs_collection, o.intent as obs_intent, "
+            "o.algorithm_name as algorithm_name, o.telescope_keywords as "
+            "facility_keywords, o.instrument_keywords as instrument_keywords, "
+            "o.environment_photometric as environment_photometric, o.proposal_id as "
+            "proposal_id, o.proposal_pi as proposal_pi, o.proposal_project as "
+            "proposal_project, o.target_name as target_name, o.target_type as "
+            "target_type, o.target_standard as target_standard, o.target_moving as "
+            "target_moving, o.target_keywords as target_keywords, p.datarelease as "
+            "obs_release_date, p.position_dimension_naxis1 as s_xel1, "
+            "p.position_dimension_naxis2 as s_xel2, p.position_samplesize as "
+            "s_pixel_scale, p.position_timedependent as position_timedependent, "
+            "p.time_bounds_lower as t_min, p.time_bounds_upper as t_max, "
+            "p.time_resolution as t_resolution, p.time_dimension as t_xel, "
+            "'ivo://irsa.ipac/' || o.collection || '?' || o.observationid || '/' || "
+            "p.productID as obs_publisher_did, "
+            "sqrt(p.position_dimension_naxis1*p.position_dimension_naxis2)*p.position_"
+            "samplesize/3600 as s_fov, p.energy_dimension as em_xel, "
+            "p.polarization_states as pol_states, p.polarization_dimension as pol_xel, "
+            "SIA2_cloud_access_column(o.collection, a.uri) as cloud_access, NULL::char "
+            "as o_ucd, SIA2_CLOSEST_SUBQUERY.upload_row_id as upload_row_id, CASE "
+            "a.producttype WHEN 'science' THEN 'aaa' WHEN 'calibration' THEN 'aab' "
+            "ELSE producttype END as irsa_temp_producttype, "
+            "SIA2_CLOSEST_SUBQUERY.dist_to_point_meters / 111194.68229846345 as "
+            "min_dist_to_point FROM (SELECT * FROM (SELECT SIA2_MINIMAL_JOIN.*, "
+            "ROW_NUMBER() OVER (PARTITION BY SIA2_MINIMAL_JOIN.facility, "
+            "SIA2_MINIMAL_JOIN.instrument, SIA2_MINIMAL_JOIN.upload_row_id ORDER BY "
+            "SIA2_MINIMAL_JOIN.dist_to_point_meters) as rownum FROM SIA2_MINIMAL_JOIN) "
+            "ranked WHERE rownum = 1) SIA2_CLOSEST_SUBQUERY, ((caom.observation o JOIN "
+            "caom.plane p ON o.obsid = p.obsid) JOIN caom.artifact a ON p.planeid = "
+            "a.planeid) WHERE SIA2_CLOSEST_SUBQUERY.obsid = o.obsid AND "
+            "SIA2_CLOSEST_SUBQUERY.planeid = p.planeid AND "
+            "(SIA2_CLOSEST_SUBQUERY.facility IS NULL OR SIA2_CLOSEST_SUBQUERY.facility "
+            "= o.telescope_name) AND (SIA2_CLOSEST_SUBQUERY.instrument IS NULL OR "
+            "SIA2_CLOSEST_SUBQUERY.instrument = o.instrument_name)",
+
 #endif  // RUN_ALL
     };
 
@@ -921,6 +1012,9 @@ int main(int argc, char *argv[]) {
             "SELECT bar FROM foo where (ST(ST(ST(ST(ST(ST(ST(ST(ST())))))))))",
             "SELECT * FROM my_table1 where sys_context('USERENV','DB_NAME')='wise1'",
 
+            "SELECT DISTINCT ON (collection) ra, dec, collection FROM my_table ORDER "
+            "BY collection, dist",
+
             // IRSA-7735: retire support for table() function
             "WITH temp (collection, multi_type) AS (SELECT collection,mytype "
             "FROM table(tap_ancillary.DCE_DATATYPE('irsa_directory'))) "
@@ -937,6 +1031,7 @@ int main(int argc, char *argv[]) {
             "WHERE semantics like '%primary%' AND "
             "irsa_directory.collection=temp.collection "
             "ORDER BY facility_name,irsa_directory.collection,instrument_name",
+            "SELECT RANK() OVER (PARTITION BY collection ORDER BY dist) FROM my_table",
     };
 
     int result(0);

wscript

@@ -114,6 +114,8 @@ def build(ctx):
                  'src/Query/Query_Specification/Value_Expression_Primary/Case_Expression/Case_Abbreviation/Nullif/ostream.cxx',
                  'src/Query/Query_Specification/Value_Expression_Primary/Case_Expression/Case_Abbreviation/Coalesce/ostream.cxx',
                  'src/Query/Query_Specification/Whitelisted_Function_Wrap/ostream.cxx',
+                 'src/Query/Query_Specification/Window_Function_Wrap/empty.cxx',
+                 'src/Query/Query_Specification/Window_Function_Wrap/ostream.cxx',
                  'src/Query/Query_Specification/Value_Expression/empty.cxx',
                  'src/Query/Query_Specification/Value_Expression/ostream.cxx',
                  'src/Query/Query_Specification/Value_Expression_Wrap/ostream.cxx',