Add custom context parsing

README.md

@@ -13,6 +13,8 @@ environment, demonstrating its practicality and effectiveness.
   history.
 * Sliding window history management: Automatically trims conversation history while maintaining context to stay within
   token limits, enabling seamless and efficient conversations with the GPT model across CLI calls.
+* Custom context from local files: Easily provide custom context through piping, allowing the GPT model to reference
+  specific data during the conversation.
 * Viper integration for robust configuration management.
 
 ## Development
@@ -69,10 +71,6 @@ For more options, see:
 ./bin/chatgpt --help
 ```
 
-## Coming Soon
-
-Enable piping custom context for seamless interaction with the ChatGPT API.
-
 ## Useful Links
 
 * [ChatGPT API Documentation](https://platform.openai.com/docs/introduction/overview)

client/client.go

@@ -23,9 +23,9 @@ const (
 )
 
 type Client struct {
+	History    []types.Message
 	caller     http.Caller
 	readWriter history.Store
-	history    []types.Message
 	capacity   int
 }
 
@@ -52,7 +52,8 @@ func NewDefault(caller http.Caller, rw history.Store) *Client {
 // call using the Post method. If the response is not empty, it decodes the
 // response JSON and returns the content of the first choice.
 func (c *Client) Query(input string) (string, error) {
-	c.initHistory(input)
+	c.initHistory()
+	c.addQuery(input)
 
 	body, err := c.createBody(false)
 	if err != nil {
@@ -88,7 +89,8 @@ func (c *Client) Query(input string) (string, error) {
 // input and then makes an API call using the Post method. The actual
 // processing of the streamed response is done in the Post method.
 func (c *Client) Stream(input string) error {
-	c.initHistory(input)
+	c.initHistory()
+	c.addQuery(input)
 
 	body, err := c.createBody(true)
 	if err != nil {
@@ -105,36 +107,55 @@ func (c *Client) Stream(input string) error {
 	return nil
 }
 
+// ProvideContext adds custom context to the client's history by converting the
+// provided string into a series of messages. This allows the ChatGPT API to have
+// prior knowledge of the provided context when generating responses.
+//
+// The context string should contain the text you want to provide as context,
+// and the method will split it into messages, preserving punctuation and special
+// characters.
+func (c *Client) ProvideContext(context string) {
+	c.initHistory()
+	messages := createMessagesFromString(context)
+	c.History = append(c.History, messages...)
+}
+
 func (c *Client) createBody(stream bool) ([]byte, error) {
 	body := types.Request{
 		Model:    GPTModel,
-		Messages: c.history,
+		Messages: c.History,
 		Stream:   stream,
 	}
 
 	return json.Marshal(body)
 }
 
-func (c *Client) initHistory(query string) {
+func (c *Client) initHistory() {
+	if len(c.History) != 0 {
+		return
+	}
+
+	c.History, _ = c.readWriter.Read()
+	if len(c.History) == 0 {
+		c.History = []types.Message{{
+			Role:    SystemRole,
+			Content: AssistantContent,
+		}}
+	}
+}
+
+func (c *Client) addQuery(query string) {
 	message := types.Message{
 		Role:    UserRole,
 		Content: query,
 	}
 
-	c.history, _ = c.readWriter.Read()
-	if len(c.history) == 0 {
-		c.history = []types.Message{{
-			Role:    SystemRole,
-			Content: AssistantContent,
-		}}
-	}
-
-	c.history = append(c.history, message)
+	c.History = append(c.History, message)
 	c.truncateHistory()
 }
 
 func (c *Client) truncateHistory() {
-	tokens, rolling := countTokens(c.history)
+	tokens, rolling := countTokens(c.History)
 	effectiveTokenSize := calculateEffectiveTokenSize(c.capacity, MaxTokenBufferPercentage)
 
 	if tokens <= effectiveTokenSize {
@@ -153,15 +174,15 @@ func (c *Client) truncateHistory() {
 		}
 	}
 
-	c.history = append(c.history[:1], c.history[index+1:]...)
+	c.History = append(c.History[:1], c.History[index+1:]...)
 }
 
 func (c *Client) updateHistory(response string) {
-	c.history = append(c.history, types.Message{
+	c.History = append(c.History, types.Message{
 		Role:    AssistantRole,
 		Content: response,
 	})
-	_ = c.readWriter.Write(c.history)
+	_ = c.readWriter.Write(c.History)
 }
 
 func calculateEffectiveTokenSize(maxTokenSize int, bufferPercentage int) int {
@@ -192,3 +213,25 @@ func countTokens(messages []types.Message) (int, []int) {
 
 	return result, rolling
 }
+
+func createMessagesFromString(input string) []types.Message {
+	words := strings.Fields(input)
+	var messages []types.Message
+
+	for i := 0; i < len(words); i += 100 {
+		end := i + 100
+		if end > len(words) {
+			end = len(words)
+		}
+
+		content := strings.Join(words[i:end], " ")
+
+		message := types.Message{
+			Role:    UserRole,
+			Content: content,
+		}
+		messages = append(messages, message)
+	}
+
+	return messages
+}

client/client_test.go

@@ -278,6 +278,23 @@ func testClient(t *testing.T, when spec.G, it spec.S) {
 			})
 		})
 	})
+	when("ProvideContext()", func() {
+		it("updates the history with the provided context", func() {
+			context := "This is a story about a dog named Kya. Kya loves to play fetch and swim in the lake."
+			mockStore.EXPECT().Read().Return(nil, nil).Times(1)
+			subject.ProvideContext(context)
+
+			Expect(len(subject.History)).To(Equal(2)) // The system message and the provided context
+
+			systemMessage := subject.History[0]
+			Expect(systemMessage.Role).To(Equal(client.SystemRole))
+			Expect(systemMessage.Content).To(Equal("You are a helpful assistant."))
+
+			contextMessage := subject.History[1]
+			Expect(contextMessage.Role).To(Equal(client.UserRole))
+			Expect(contextMessage.Content).To(Equal(context))
+		})
+	})
 }
 
 func createBody(messages []types.Message, stream bool) ([]byte, error) {

cmd/chatgpt/main.go

@@ -8,6 +8,7 @@ import (
 	"github.com/kardolus/chatgpt-cli/http"
 	"github.com/spf13/cobra"
 	"github.com/spf13/viper"
+	"io/ioutil"
 	"os"
 	"strings"
 )
@@ -20,9 +21,11 @@ var clearHistory bool
 func main() {
 	var rootCmd = &cobra.Command{
 		Use:   "chatgpt",
-		Short: "ChatGPT Proof of Concept",
-		Long:  "A Proof of Concept for building ChatGPT clients.",
-		RunE:  run,
+		Short: "ChatGPT CLI Tool",
+		Long: "A powerful ChatGPT client that enables seamless interactions with the GPT model. " +
+			"Provides multiple modes and context management features, including the ability to " +
+			"pipe custom context into the conversation.",
+		RunE: run,
 	}
 
 	rootCmd.PersistentFlags().BoolVarP(&queryMode, "query", "q", false, "Use query mode instead of stream mode")
@@ -60,6 +63,16 @@ func run(cmd *cobra.Command, args []string) error {
 	}
 	client := client.NewDefault(http.New().WithSecret(secret), history.NewDefault())
 
+	// Check if there is input from the pipe (stdin)
+	stat, _ := os.Stdin.Stat()
+	if (stat.Mode() & os.ModeCharDevice) == 0 {
+		pipeContent, err := ioutil.ReadAll(os.Stdin)
+		if err != nil {
+			return fmt.Errorf("failed to read from pipe: %w", err)
+		}
+		client.ProvideContext(string(pipeContent))
+	}
+
 	if queryMode {
 		result, err := client.Query(strings.Join(args, " "))
 		if err != nil {