feat: parse Tool Spec in building stage

.dockerignore

@@ -12,4 +12,6 @@ credentials/
 
 **/.next
 **/node_modules
-config
+config
+
+.dist

.gitignore

@@ -20,4 +20,6 @@ config/**.yaml
 
 screenshots
 
-**/tmp
+**/tmp
+
+.dist

npiai/app/github/npi.yml

@@ -6,9 +6,9 @@ env:
 dependencies:
   - name: python
     version: "^3.10"
+  - name: npiai
+    version: "0.1.1"
   - name: openai
     version: "^1.30.3"
-  - name: termcolor
-    version: "^2.4.0"
   - name: pygithub
     version: "^2.3.0"

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "npiai"
-version = "0.1.0"
+version = "0.1.1"
 description = ""
 authors = ["wenfeng <w@npi.ai>"]
 readme = "README.md"

server/model/tool.go

@@ -108,10 +108,10 @@ type ToolFunctionSpec struct {
 	Functions    []Function   `json:"functions" bson:"functions"`
 }
 
-type Language int
+type Language string
 
 const (
-	Python = Language(1)
+	Python = Language("python")
 )
 
 type Runtime struct {
@@ -126,17 +126,17 @@ type Dependency struct {
 }
 
 type Function struct {
-	Name        string      `json:"name" bson:"name"`
-	Description string      `json:"description" bson:"description"`
-	Parameters  []Parameter `json:"parameters" bson:"parameters"`
-	FewShots    []FewShot   `json:"few_shots" bson:"few_shots"`
+	Name        string     `json:"name" bson:"name"`
+	Description string     `json:"description" bson:"description"`
+	Parameters  Parameters `json:"parameters" bson:"parameters"`
+	FewShots    []FewShot  `json:"fewShots" bson:"few_shots"`
 }
 
 func (f *Function) Arguments(lan Language) string {
 	switch lan {
 	case Python:
 		args := ""
-		for _, para := range f.Parameters {
+		for _, para := range f.Parameters.Properties {
 			args += fmt.Sprintf("%s=event['%s'], ", para.Name, para.Name)
 		}
 		return strings.TrimRight(args, ", ")
@@ -146,15 +146,11 @@ func (f *Function) Arguments(lan Language) string {
 
 func (f *Function) Schema() map[string]interface{} {
 	properties := map[string]interface{}{}
-	var required []string
-	for _, para := range f.Parameters {
+	for _, para := range f.Parameters.Properties {
 		properties[para.Name] = map[string]string{
 			"type":        para.Type.Name(),
 			"description": para.Description,
 		}
-		if para.Required {
-			required = append(required, para.Name)
-		}
 	}
 
 	schema := map[string]interface{}{
@@ -165,7 +161,7 @@ func (f *Function) Schema() map[string]interface{} {
 			"parameters": map[string]interface{}{
 				"type":       "object",
 				"properties": properties,
-				"required":   required,
+				"required":   f.Parameters.Required,
 			},
 		},
 	}
@@ -190,7 +186,7 @@ func (f *Function) OpenAPISchema() *v3.PathItem {
 
 	content := orderedmap.New[string, *v3.MediaType]()
 	properties := orderedmap.New[string, *base.SchemaProxy]()
-	for _, para := range f.Parameters {
+	for _, para := range f.Parameters.Properties {
 		properties.Set(para.Name, base.CreateSchemaProxy(&base.Schema{
 			Type:        []string{para.Type.Name()},
 			Description: para.Description,
@@ -219,7 +215,7 @@ func (f *Function) OpenAPISchema() *v3.PathItem {
 	return item
 }
 
-type ParameterType int
+type ParameterType string
 
 func (pt ParameterType) Name() string {
 	switch pt {
@@ -240,20 +236,25 @@ func (pt ParameterType) Name() string {
 }
 
 const (
-	String = ParameterType(1)
-	Int    = ParameterType(2)
-	Float  = ParameterType(3)
-	Bool   = ParameterType(4)
-	Map    = ParameterType(5)
-	List   = ParameterType(6)
+	String = ParameterType("string")
+	Int    = ParameterType("integer")
+	Float  = ParameterType("number")
+	Bool   = ParameterType("bool")
+	Map    = ParameterType("object")
+	List   = ParameterType("array")
 )
 
-type Parameter struct {
+type Parameters struct {
+	Description string              `json:"description" bson:"description"`
+	Type        ParameterType       `json:"type" bson:"type"`
+	Required    []string            `json:"required" bson:"required"`
+	Properties  map[string]Property `json:"properties" bson:"properties"`
+}
+
+type Property struct {
 	Name        string        `json:"name" bson:"name"`
 	Description string        `json:"description" bson:"description"`
 	Type        ParameterType `json:"type" bson:"type"`
-	Required    bool          `json:"required" bson:"required"`
-	Default     interface{}   `json:"default" bson:"default"`
 }
 
 type FewShot struct {

server/reconcile/tool.go

@@ -2,12 +2,19 @@ package reconcile
 
 import (
 	"context"
+	"encoding/json"
+	"errors"
+	"github.com/mattn/go-isatty"
+	"github.com/npi-ai/npi/server/api"
 	"github.com/npi-ai/npi/server/db"
 	"github.com/npi-ai/npi/server/log"
 	"github.com/npi-ai/npi/server/model"
 	"github.com/npi-ai/npi/server/service"
 	"go.mongodb.org/mongo-driver/bson"
 	"go.mongodb.org/mongo-driver/mongo"
+	"os"
+	"os/exec"
+	"strings"
 	"time"
 )
 
@@ -81,16 +88,63 @@ func (tc *toolReconciler) Start(ctx context.Context) error {
 	return nil
 }
 
-func (tc *toolReconciler) BuildingHandler(ctx context.Context, ws model.ToolInstance) error {
-	imageURI, err := tc.builderSvc.Build(ctx, ws.Metadata)
+func (tc *toolReconciler) BuildingHandler(ctx context.Context, toolInstance model.ToolInstance) error {
+	imageURI, err := tc.builderSvc.Build(ctx, toolInstance.Metadata)
 	if err != nil {
 		return err
 	}
+
+	args := []string{"run", "--rm", "--entrypoint", "poetry", imageURI, "run", "python", "main.py", "spec"}
+	// get tool spec
+	if isatty.IsTerminal(os.Stdin.Fd()) {
+		//args = append([]string{"-it"}, args...)
+	}
+	cmd := exec.Command("docker", args...)
+	output, err := cmd.CombinedOutput()
+	if err != nil {
+		log.Warn().Str("output", string(output)).Msg("Failed to build docker image")
+		return api.ErrInternal.
+			WithMessage("Failed to build docker image").WithError(err)
+	}
+
+	strings.Split(string(output), "\n")
+	success := false
+	var md model.ToolMetadata
+	var spec model.ToolFunctionSpec
+	for _, line := range strings.Split(string(output), "\n") {
+		if strings.HasPrefix(line, "{\"kind\"") {
+			md, spec, err = parseToolSpec([]byte(line))
+			if err != nil {
+				return api.ErrInternal.
+					WithMessage("Failed to parse tool spec").WithError(err)
+			}
+
+			success = true
+			break
+		}
+	}
+	if !success {
+		return api.ErrInternal.WithMessage("Failed to get tool spec")
+	}
+
+	toolInstance.Metadata.Name = md.Name
+	toolInstance.Metadata.Description = md.Description
+	toolInstance.FunctionSpec = spec
+	// upload image
+	log.Info().Str("image", imageURI).Msg("Uploading docker image")
+	cmd = exec.Command("docker", "push", imageURI)
+	if output, err = cmd.CombinedOutput(); err != nil {
+		log.Warn().Str("output", string(output)).Msg("Failed to upload docker image")
+		return api.ErrInternal.
+			WithMessage("Failed to upload docker image").WithError(err)
+	}
 	// update tool instance
 	_, err = tc.coll.UpdateOne(ctx,
-		bson.M{"_id": ws.ID},
+		bson.M{"_id": toolInstance.ID},
 		bson.M{
 			"$set": bson.M{
+				"metadata":   toolInstance.Metadata,
+				"spec":       toolInstance.FunctionSpec,
 				"image":      imageURI,
 				"updated_at": time.Now(),
 			},
@@ -112,3 +166,37 @@ func (tc *toolReconciler) PausingHandler(ctx context.Context, ws model.ToolInsta
 	println("pausing")
 	return nil
 }
+
+func parseToolSpec(data []byte) (model.ToolMetadata, model.ToolFunctionSpec, error) {
+	m := map[string]interface{}{}
+	md := model.ToolMetadata{}
+	spec := model.ToolFunctionSpec{}
+	if err := json.Unmarshal(data, &m); err != nil {
+		return md, spec, err
+	}
+	if v, ok := m["metadata"]; ok {
+		mdMap := v.(map[string]interface{})
+		value, exist := mdMap["name"]
+		if !exist {
+			return md, spec, errors.New("metadata.name not found")
+		}
+		md.Name = value.(string)
+
+		value, exist = mdMap["description"]
+		if !exist {
+			return md, spec, errors.New("metadata.description not found")
+		}
+		md.Description = value.(string)
+	}
+	if v, ok := m["spec"]; ok {
+		d, err := json.Marshal(v)
+		if err != nil {
+			return md, spec, errors.New("failed to parse function spec")
+		}
+		if err = json.Unmarshal(d, &spec); err != nil {
+			return md, spec, err
+		}
+		// TODO merge dependencies
+	}
+	return md, spec, nil
+}

server/reconcile/tool.spec.json

@@ -0,0 +1 @@
+{"kind": "Function", "metadata": {"name": "github", "description": "Manage GitHub issues and pull requests", "provider": "private"}, "spec": {"runtime": {"language": "python", "version": "3.11"}, "dependencies": [{"name": "npiai", "version": "0.1.0"}], "functions": [{"description": "Add a comment to the target issue.", "name": "add_issue_comment", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "number": {"description": "Issue number", "type": "integer"}, "body": {"description": "Body of the comment in markdown format", "type": "string"}}, "required": ["repo", "number", "body"], "type": "object"}, "fewShots": null}, {"description": "Add a comment to the target pull request.", "name": "add_pull_request_comment", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "number": {"description": "Pull request number", "type": "integer"}, "body": {"description": "Body of the comment in markdown format", "type": "string"}}, "required": ["repo", "number", "body"], "type": "object"}, "fewShots": null}, {"description": "Create an issue under the given repository.", "name": "create_issue", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "title": {"description": "Title of the issue", "type": "string"}, "body": {"description": "Body of the issue in markdown format", "type": "string"}, "labels": {"default": null, "description": "List of labels to add to the issue", "items": {"type": "string"}, "type": "array"}, "assignees": {"default": null, "description": "List of users to assign to the issue", "items": {"type": "string"}, "type": "array"}}, "required": ["repo", "title", "body"], "type": "object"}, "fewShots": null}, {"description": "Create a pull request under the given repository.", "name": "create_pull_request", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "title": {"description": "Title of the pull request", "type": "string"}, "body": {"description": "Body of the pull request", "type": "string"}, "base": {"description": "Base branch of the pull request, i.e., the branch to merge the pull request into", "type": "string"}, "head": {"description": "Head branch of the pull request, i.e., the branch with your changes", "type": "string"}, "is_draft": {"default": false, "description": "Whether the pull request is a draft or not", "type": "boolean"}, "labels": {"default": null, "description": "List of labels to add to the pull request", "items": {"type": "string"}, "type": "array"}, "assignees": {"default": null, "description": "List of users to assign to the pull request", "items": {"type": "string"}, "type": "array"}}, "required": ["repo", "title", "body", "base", "head"], "type": "object"}, "fewShots": null}, {"description": "Edit an existing issue. You can also close or reopen an issue by specifying the state parameter.", "name": "edit_issue", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "number": {"description": "Issue number", "type": "integer"}, "title": {"default": null, "description": "Title of the issue", "type": "string"}, "body": {"default": null, "description": "Body of the issue in markdown format", "type": "string"}, "labels": {"default": null, "description": "List of labels to add to the issue", "items": {"type": "string"}, "type": "array"}, "assignees": {"default": null, "description": "List of users to assign to the issue", "items": {"type": "string"}, "type": "array"}, "state": {"default": null, "description": "Whether the issue is open or closed", "enum": ["open", "closed"], "type": "string"}}, "required": ["repo", "number"], "type": "object"}, "fewShots": null}, {"description": "Edit an existing pull request. You can also close or reopen a pull request by specifying the state parameter.", "name": "edit_pull_request", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "number": {"description": "Pull request number", "type": "integer"}, "title": {"default": null, "description": "Title of the pull request", "type": "string"}, "body": {"default": null, "description": "Body of the pull request in markdown format", "type": "string"}, "base": {"default": null, "description": "Base branch of the pull request, i.e., the branch with your changes", "type": "string"}, "labels": {"default": null, "description": "List of labels to add to the pull request", "items": {"type": "string"}, "type": "array"}, "assignees": {"default": null, "description": "List of users to assign to the pull request", "items": {"type": "string"}, "type": "array"}, "state": {"default": null, "description": "Whether the pull request is open or closed", "enum": ["open", "closed"], "type": "string"}}, "required": ["repo", "number"], "type": "object"}, "fewShots": null}, {"description": "Fork a repository on GitHub.", "name": "fork", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}}, "required": ["repo"], "type": "object"}, "fewShots": null}, {"description": "Get an issue from the given repository.", "name": "get_issue", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "number": {"description": "Issue number", "type": "integer"}}, "required": ["repo", "number"], "type": "object"}, "fewShots": null}, {"description": "Get a pull request from the given repository.", "name": "get_pull_request", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}, "number": {"description": "Pull request number", "type": "integer"}}, "required": ["repo", "number"], "type": "object"}, "fewShots": null}, {"description": "Search for issues or pull requests on GitHub.", "name": "search_issue_pr", "parameters": {"properties": {"query": {"description": "Search query to search for GitHub issues and pull requests. Below are some query examples:\n1. Search for all issues containing \"performance\" in the repository npi/npi: `is:issue repo:npi/npi performance`\n2. Search for open pull requests in repository npi/npi: `is:pr is:open repo:npi/npi`", "type": "string"}, "max_results": {"description": "Maximum number of results to return", "type": "integer"}}, "required": ["query", "max_results"], "type": "object"}, "fewShots": null}, {"description": "Search for repositories on GitHub.", "name": "search_repositories", "parameters": {"properties": {"query": {"description": "Search query to search for GitHub repositories. Below are some query examples:\n1. Search for all repositories that contain \"test\" and use Python: `test language:python`\n2. Search for repositories with more than 1000 stars: `stars:>1000`", "type": "string"}, "max_results": {"description": "Maximum number of results to return", "type": "integer"}}, "required": ["query", "max_results"], "type": "object"}, "fewShots": null}, {"description": "Star a repository on GitHub.", "name": "star", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}}, "required": ["repo"], "type": "object"}, "fewShots": null}, {"description": "Subscribe to notifications (a.k.a. watch) for activity in a repository on GitHub.", "name": "watch", "parameters": {"properties": {"repo": {"description": "Name of the repository in format {owner}/{repo}", "type": "string"}}, "required": ["repo"], "type": "object"}, "fewShots": null}]}}

server/reconcile/tool_test.go

@@ -0,0 +1,17 @@
+package reconcile
+
+import (
+	"os"
+	"testing"
+)
+
+func Test_ParseToolSpec(t *testing.T) {
+	data, err := os.ReadFile("tool.spec.json")
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, _, err = parseToolSpec(data)
+	if err != nil {
+		t.Fatal(err)
+	}
+}

server/scripts/entrypoint.py

@@ -2,11 +2,13 @@
 import sys
 import importlib
 import time
+import asyncio
+import json
 
 
 def print_tool_spec():
     # Add the directory containing the file to sys.path
-    module_dir, module_name = os.path.split('/npiai/tools/app.py')
+    module_dir, module_name = os.path.split('%s')
     module_name = os.path.splitext(module_name)[0]  # Remove the .py extension
 
     if module_dir not in sys.path:
@@ -15,9 +17,9 @@ def print_tool_spec():
     # Now you can import the module using its name
     module = importlib.import_module(module_name)
     # Create an instance of the class
-    tool_class = getattr(module, 'GitHub')
+    tool_class = getattr(module, '%s')
     instance = tool_class()
-    print(instance.export())
+    print(json.dumps(instance.export()))
 
 
 async def main():
@@ -39,4 +41,14 @@ async def main():
         time.sleep(1000)
 
 if __name__ == '__main__':
-    print_tool_spec()
+    if len(sys.argv) > 1:
+        if sys.argv[1] == 'spec':
+            print_tool_spec()
+        elif sys.argv[1] == 'server':
+            asyncio.run(main())
+        else:
+            print('Usage: python entrypoint.py spec|server')
+            sys.exit(1)
+    else:
+        print('Usage: python entrypoint.py spec|server')
+        sys.exit(1)