From a1342ac0740a67d3e08f3aa8582646bcb364cf76 Mon Sep 17 00:00:00 2001
From: Bela Schaum <schaumb@gmail.com>
Date: Wed, 21 Aug 2024 18:35:29 +0200
Subject: [PATCH] Revert exists removing

---
 src/dataframe/impl/aggregators.cpp     |  9 +++++++++
 src/dataframe/impl/dataframe.cpp       |  4 +++-
 src/dataframe/interface.h              |  3 ++-
 test/unit/dataframe/interface_test.cpp | 12 ++++++++++++
 4 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/src/dataframe/impl/aggregators.cpp b/src/dataframe/impl/aggregators.cpp
index fbcb55d2d..a455afc50 100644
--- a/src/dataframe/impl/aggregators.cpp
+++ b/src/dataframe/impl/aggregators.cpp
@@ -114,6 +114,15 @@ get_aggregators() noexcept
 		                *std::get_if<const std::string *>(&cell))
 			        set.insert(v);
 		        return static_cast<double>(set.size());
+	        }},
+	    {aggrs[static_cast<std::size_t>(aggregator_type::exists)],
+	        empty_double,
+	        [](custom_aggregator::id_type &id,
+	            cell_reference const &cell) -> double
+	        {
+		        auto &b = *std::get_if<double>(&id);
+		        if (is_valid(cell)) b = 1.0;
+		        return b;
 	        }}}}};
 }
 
diff --git a/src/dataframe/impl/dataframe.cpp b/src/dataframe/impl/dataframe.cpp
index f125cf079..42bdab5f9 100644
--- a/src/dataframe/impl/dataframe.cpp
+++ b/src/dataframe/impl/dataframe.cpp
@@ -94,7 +94,9 @@ dataframe::dataframe(std::shared_ptr<const data_source> other,
 void valid_unexistent_aggregator(const std::string_view &series,
     const dataframe::any_aggregator_type &agg)
 {
-	if (!agg || *agg != aggregator_type::count)
+	if (!agg
+	    || (*agg != aggregator_type::count
+	        && *agg != aggregator_type::exists))
 		error(error_type::series_not_found, series);
 }
 
diff --git a/src/dataframe/interface.h b/src/dataframe/interface.h
index 04e29b181..d37456173 100644
--- a/src/dataframe/interface.h
+++ b/src/dataframe/interface.h
@@ -27,7 +27,8 @@ enum class aggregator_type : std::uint8_t {
 	max,
 	mean,
 	count,
-	distinct
+	distinct,
+	exists
 };
 
 enum class sort_type : std::uint8_t {
diff --git a/test/unit/dataframe/interface_test.cpp b/test/unit/dataframe/interface_test.cpp
index d1f94eb76..f552154f3 100644
--- a/test/unit/dataframe/interface_test.cpp
+++ b/test/unit/dataframe/interface_test.cpp
@@ -486,11 +486,13 @@ const static auto tests =
 	auto &&pure1c = df->set_aggregate({}, count);
 	auto &&d1c = df->set_aggregate("d1", count);
 	auto &&d1d = df->set_aggregate("d1", distinct);
+	auto &&d1e = df->set_aggregate("d1", exists);
 	auto &&m1s = df->set_aggregate("m1", sum);
 	auto &&m1mi = df->set_aggregate("m1", min);
 	auto &&m1ma = df->set_aggregate("m1", max);
 	auto &&m1me = df->set_aggregate("m1", mean);
 	auto &&m1c = df->set_aggregate("m1", count);
+	auto &&m1e = df->set_aggregate("m1", exists);
 
 	/*
 auto &&m1t = df->set_aggregate("m1",
@@ -524,6 +526,8 @@ auto &&m1t = df->set_aggregate("m1",
 	        d1c,
 	        m1c,
 	        d1d,
+	        d1e,
+	        m1e,
 	        m1s,
 	        m1ma,
 	        m1me,
@@ -537,6 +541,8 @@ auto &&m1t = df->set_aggregate("m1",
 	check->*df->get_data(std::size_t{0}, d1c) == 5.0;
 	check->*df->get_data(std::size_t{0}, m1c) == 5.0;
 	check->*df->get_data(std::size_t{0}, d1d) == 1.0;
+	check->*df->get_data(std::size_t{0}, d1e) == 1.0;
+	check->*df->get_data(std::size_t{0}, m1e) == 1.0;
 	check->*df->get_data(std::size_t{0}, m1ma) == 88.0;
 	check->*df->get_data(std::size_t{0}, m1me) == 21.85;
 	check->*df->get_data(std::size_t{0}, m1mi) == 2.0;
@@ -547,6 +553,8 @@ auto &&m1t = df->set_aggregate("m1",
 	check->*df->get_data(std::size_t{1}, d1c) == 4.0;
 	check->*df->get_data(std::size_t{1}, m1c) == 3.0;
 	check->*df->get_data(std::size_t{1}, d1d) == 1.0;
+	check->*df->get_data(std::size_t{0}, d1e) == 1.0;
+	check->*df->get_data(std::size_t{0}, m1e) == 1.0;
 	check->*df->get_data(std::size_t{1}, m1ma) == 7.25;
 	check->*df->get_data(std::size_t{1}, m1me) == 5.0;
 	check->*df->get_data(std::size_t{1}, m1mi) == 3.5;
@@ -557,6 +565,8 @@ auto &&m1t = df->set_aggregate("m1",
 	check->*df->get_data(std::size_t{2}, d1c) == 1.0;
 	check->*df->get_data(std::size_t{2}, m1c) == 0.0;
 	check->*df->get_data(std::size_t{2}, d1d) == 1.0;
+	check->*df->get_data(std::size_t{0}, d1e) == 1.0;
+	check->*df->get_data(std::size_t{0}, m1e) == 0.0;
 	check
 	            ->*std::isnan(std::get<double>(
 	                df->get_data(std::size_t{2}, m1ma)))
@@ -577,6 +587,8 @@ auto &&m1t = df->set_aggregate("m1",
 	check->*df->get_data(std::size_t{3}, d1c) == 0.0;
 	check->*df->get_data(std::size_t{3}, m1c) == 1.0;
 	check->*df->get_data(std::size_t{3}, d1d) == 0.0;
+	check->*df->get_data(std::size_t{0}, d1e) == 0.0;
+	check->*df->get_data(std::size_t{0}, m1e) == 1.0;
 }
 
     | "aggregate multiple dim" |