From 1c070bfe858fbbbcfacbbe81146146cd5e6befe9 Mon Sep 17 00:00:00 2001
From: avouacr <avouacr@pm.me>
Date: Sat, 6 Jan 2024 15:33:43 +0000
Subject: [PATCH] add scripts to configure R & Python duckdb

---
 scripts/configure-duckdb.R          |  3 +++
 scripts/configure-duckdb.py         |  6 ++++++
 scripts/install-duckdb-extensions.R | 11 ++++-------
 scripts/onyxia-init.sh              | 11 +++++++++++
 4 files changed, 24 insertions(+), 7 deletions(-)
 create mode 100644 scripts/configure-duckdb.R
 create mode 100644 scripts/configure-duckdb.py

diff --git a/scripts/configure-duckdb.R b/scripts/configure-duckdb.R
new file mode 100644
index 00000000..3ab962bc
--- /dev/null
+++ b/scripts/configure-duckdb.R
@@ -0,0 +1,3 @@
+con <- DBI::dbConnect(duckdb::duckdb())
+DBI::dbExecute(con, 'LOAD httpfs')
+DBI::dbExecute(con, glue::glue('SET s3_endpoint=\"{Sys.getenv("AWS_S3_ENDPOINT")}\"'))
diff --git a/scripts/configure-duckdb.py b/scripts/configure-duckdb.py
new file mode 100644
index 00000000..91ca9371
--- /dev/null
+++ b/scripts/configure-duckdb.py
@@ -0,0 +1,6 @@
+import os
+
+import duckdb
+
+duckdb.sql("LOAD httpfs")
+duckdb.sql(f"SET s3_endpoint=\"{os.getenv('AWS_S3_ENDPOINT')}\"")
diff --git a/scripts/install-duckdb-extensions.R b/scripts/install-duckdb-extensions.R
index cc13e9b6..b08882e1 100755
--- a/scripts/install-duckdb-extensions.R
+++ b/scripts/install-duckdb-extensions.R
@@ -1,8 +1,5 @@
-library(glue)
-library(duckdb)
+con <- DBI::dbConnect(duckdb::duckdb())
 
-con <- dbConnect(duckdb())
-
-dbExecute(con, glue('SET extension_directory=\"{Sys.getenv("HOME")}\"'))
-dbExecute(con, 'INSTALL httpfs')
-dbExecute(con, 'INSTALL aws')
+DBI::dbExecute(con, glue::glue('SET extension_directory=\"{Sys.getenv("HOME")}\"'))
+DBI::dbExecute(con, 'INSTALL httpfs')
+DBI::dbExecute(con, 'INSTALL aws')
diff --git a/scripts/onyxia-init.sh b/scripts/onyxia-init.sh
index 62683022..c6ed2e8d 100644
--- a/scripts/onyxia-init.sh
+++ b/scripts/onyxia-init.sh
@@ -166,6 +166,7 @@ if [[ "$DARK_MODE" == "true" ]]; then
     fi
 fi
 
+# Configure duckdb CLI
 if [[ -n $AWS_S3_ENDPOINT ]] && command -v duckdb ; then
 cat <<EOF > ${HOME}/.duckdbrc
 -- Duck head prompt
@@ -176,6 +177,16 @@ SET s3_endpoint='$AWS_S3_ENDPOINT';
 EOF
 fi
 
+# Configure duckdb Python API
+if [[ -n $AWS_S3_ENDPOINT ]] && command -v python ; then
+    mamba run -n base python opt/configure-dubkdb.py
+fi
+
+# Configure duckdb R API
+if [[ -n $AWS_S3_ENDPOINT ]] && command -v R ; then
+    Rscript opt/configure-duckdb.R
+fi
+
 if [[ -e "$HOME/work" ]]; then
   if [[ $(id -u) = 0 ]]; then
     echo "cd $HOME/work" >> /etc/profile