From fe14e6a9d7cca7d2ee68c2248c50b6d8528fd789 Mon Sep 17 00:00:00 2001
From: Darren Shepherd <darren@acorn.io>
Date: Tue, 5 Nov 2024 05:49:24 -0700
Subject: [PATCH] chore: package datasets and aws encryption provider

---
 Dockerfile                  | 23 ++++++++---------------
 pkg/api/handlers/version.go | 31 ++++++++-----------------------
 pkg/services/config.go      | 10 ++++++----
 tools/package-tools.sh      | 20 +++++++++++++++++++-
 4 files changed, 41 insertions(+), 43 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index 4c53163b5..e545468a7 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,33 +1,20 @@
-# First Stage: Builder
 FROM cgr.dev/chainguard/wolfi-base AS builder
 
-# Install build dependencies
 RUN apk add --no-cache go npm make git pnpm curl
-
-# Set the working directory
 WORKDIR /app
-
-# Copy the source code
 COPY . .
-
 RUN --mount=type=cache,id=pnpm,target=/root/.local/share/pnpm/store \
     --mount=type=cache,target=/root/.cache/go-build \
     --mount=type=cache,target=/root/go/pkg/mod \
     make in-docker-build
 
-# Second Stage: Final
 FROM cgr.dev/chainguard/wolfi-base AS final
-
-# Install build dependencies
 RUN apk add --no-cache git python-3.13 py3.13-pip openssh-server npm bash tini
 COPY --chmod=0755 /tools/package-chrome.sh /
 RUN /package-chrome.sh && rm /package-chrome.sh
 RUN sed -E 's/^#(PermitRootLogin)no/\1yes/' /etc/ssh/sshd_config -i
 RUN ssh-keygen -A
 RUN mkdir /run/sshd && /usr/sbin/sshd
-
-
-# Copy the compiled application from the builder stage
 COPY --from=builder /app/bin/otto8 /bin/
 COPY --link --from=builder /app/otto8-tools /otto8-tools
 RUN <<EOF
@@ -41,9 +28,14 @@ set -e
 mkdir -p /run/sshd
 /usr/sbin/sshd -D &
 mkdir -p /data/cache
+# This is YAML
 export OTTO_SERVER_VERSIONS="$(cat <<VERSIONS
-Tools: $(cd /otto8-tools && git rev-parse HEAD)
-WorkspaceProvider: $(cd /otto8-tools/workspace-provider && git rev-parse HEAD)
+"github.com/otto8-ai/tools": "$(cd /otto8-tools && git rev-parse HEAD)"
+"github.com/gptscript-ai/workspace-provider": "$(cd /otto8-tools/workspace-provider && git rev-parse HEAD)"
+"github.com/gptscript-ai/datasets": "$(cd /otto8-tools/datasets && git rev-parse HEAD)"
+"github.com/kubernetes-sigs/aws-encryption-provider": "$(cd /otto8-tools/aws-encryption-provider && git rev-parse HEAD)"
+# double echo to remove trailing whitespace
+"chrome": "$(echo $(/opt/google/chrome/chrome --version))"
 VERSIONS
 )"
 exec tini -- otto8 server
@@ -54,6 +46,7 @@ ENV HOME=/data
 ENV XDG_CACHE_HOME=/data/cache
 ENV GPTSCRIPT_SYSTEM_TOOLS_DIR=/otto8-tools/
 ENV OTTO_SERVER_WORKSPACE_TOOL=/otto8-tools/workspace-provider
+ENV OTTO_SERVER_DATASETS_TOOL=/otto8-tools/datasets
 ENV OTTO_SERVER_TOOL_REGISTRY=/otto8-tools
 WORKDIR /data
 VOLUME /data
diff --git a/pkg/api/handlers/version.go b/pkg/api/handlers/version.go
index 8676a380a..0fa44a239 100644
--- a/pkg/api/handlers/version.go
+++ b/pkg/api/handlers/version.go
@@ -2,39 +2,24 @@ package handlers
 
 import (
 	"os"
-	"strings"
 
 	"github.com/otto8-ai/otto8/pkg/api"
 	"github.com/otto8-ai/otto8/pkg/version"
+	"sigs.k8s.io/yaml"
 )
 
-type versionResponse struct {
-	Otto              string `json:"otto,omitempty"`
-	Tools             string `json:"tools,omitempty"`
-	WorkspaceProvider string `json:"workspaceProvider,omitempty"`
-}
-
 func GetVersion(req api.Context) error {
 	return req.Write(getVersionResponse())
 }
 
-func getVersionResponse() *versionResponse {
-	// Retrieve the multi-line environment variable
-	envVar := os.Getenv("OTTO_SERVER_VERSIONS")
-	// Initialize a map to store the parsed key-value pairs
+func getVersionResponse() map[string]string {
 	values := make(map[string]string)
-
-	// Parse the environment variable into the map
-	for _, line := range strings.Split(envVar, "\n") {
-		if parts := strings.SplitN(line, ":", 2); len(parts) == 2 {
-			values[parts[0]] = parts[1]
+	versions := os.Getenv("OTTO_SERVER_VERSIONS")
+	if versions != "" {
+		if err := yaml.Unmarshal([]byte(versions), &values); err != nil {
+			values["error"] = err.Error()
 		}
 	}
-
-	// Populate the versionResponse struct using the map
-	return &versionResponse{
-		Otto:              version.Get().String(),
-		Tools:             strings.TrimSpace(values["Tools"]),
-		WorkspaceProvider: strings.TrimSpace(values["WorkspaceProvider"]),
-	}
+	values["otto"] = version.Get().String()
+	return values
 }
diff --git a/pkg/services/config.go b/pkg/services/config.go
index 035b9a72c..49c43fb42 100644
--- a/pkg/services/config.go
+++ b/pkg/services/config.go
@@ -53,6 +53,7 @@ type Config struct {
 	ToolRegistry          string `usage:"The tool reference for the tool registry" default:"github.com/otto8-ai/tools"`
 	WorkspaceProviderType string `usage:"The type of workspace provider to use for non-knowledge workspaces" default:"directory" env:"OTTO_WORKSPACE_PROVIDER_TYPE"`
 	WorkspaceTool         string `usage:"The tool reference for the workspace provider" default:"github.com/gptscript-ai/workspace-provider"`
+	DatasetTool           string `usage:"The tool reference for the dataset provider" default:"github.com/gptscript-ai/datasets"`
 	HelperModel           string `usage:"The model used to generate names and descriptions" default:"gpt-4o-mini"`
 
 	AuthConfig
@@ -78,11 +79,12 @@ type Services struct {
 	GatewayServer         *gserver.Server
 }
 
-func newGPTScript(ctx context.Context, workspaceTool string) (*gptscript.GPTScript, error) {
+func newGPTScript(ctx context.Context, workspaceTool, datasetsTool string) (*gptscript.GPTScript, error) {
 	if os.Getenv("GPTSCRIPT_URL") != "" {
 		return gptscript.NewGPTScript(gptscript.GlobalOptions{
-			URL:           os.Getenv("GPTSCRIPT_URL"),
-			WorkspaceTool: workspaceTool,
+			URL:             os.Getenv("GPTSCRIPT_URL"),
+			WorkspaceTool:   workspaceTool,
+			DatasetToolRepo: datasetsTool,
 		})
 	}
 
@@ -155,7 +157,7 @@ func New(ctx context.Context, config Config) (*Services, error) {
 		config.UIHostname = "https://" + config.UIHostname
 	}
 
-	c, err := newGPTScript(ctx, config.WorkspaceTool)
+	c, err := newGPTScript(ctx, config.WorkspaceTool, config.DatasetTool)
 	if err != nil {
 		return nil, err
 	}
diff --git a/tools/package-tools.sh b/tools/package-tools.sh
index c0fb223ab..d10780d27 100755
--- a/tools/package-tools.sh
+++ b/tools/package-tools.sh
@@ -26,4 +26,22 @@ if [ ! -e workspace-provider ]; then
 fi
 
 cd workspace-provider
-CGO_ENABLED=0 go build -ldflags="-s -w" -o bin/gptscript-go-tool .
+go build -ldflags="-s -w" -o bin/gptscript-go-tool .
+
+cd ..
+
+if [ ! -e datasets ]; then
+    git clone --depth=1 https://github.com/gptscript-ai/datasets
+fi
+
+cd datasets
+go build -ldflags="-s -w" -o bin/gptscript-go-tool .
+
+cd ../..
+
+if [ ! -e aws-encryption-provider ]; then
+    git clone --depth=1 https://github.com/kubernetes-sigs/aws-encryption-provider
+fi
+
+cd aws-encryption-provider
+go build -o ../otto8-tools/aws-encryption-provider/bin/aws-encryption-provider cmd/server/main.go