Handle retries without returning errors for non-network errors

.gitignore

@@ -14,3 +14,6 @@
 # Dependency directories (remove the comment below to include it)
 # vendor/
 .env
+
+# For now ignore local VS Code settings
+.vscode/

.golangci.yml

@@ -0,0 +1,24 @@
+version: 2
+
+# See https://golangci-lint.run/usage/configuration/ for all options
+
+run:
+  timeout: 5m
+
+linters:
+  disable-all: true
+  enable:
+    - govet
+    - errcheck
+    - staticcheck
+    - unused
+    - ineffassign
+    - misspell
+
+formatters:
+  enable:
+    - gofmt
+    - goimports
+
+issues:
+  exclude-use-default: false

Makefile

@@ -0,0 +1,77 @@
+# Default target
+default: build
+
+# Build the application
+.PHONY: build
+
+build:
+	go build ./...
+
+# Test target to run all tests in the project
+.PHONY: test
+
+test: build
+	go test ./... -v
+
+# Lint target to run the linter
+.PHONY: lint
+
+lint:
+	golangci-lint run
+
+# Run linter with fixes
+.PHONY: lint-fix
+
+lint-fix:
+	golangci-lint run --fix
+
+# Check Go code formatting without modifying files
+.PHONY: fmt-check
+
+fmt-check:
+	@echo "Checking Go formatting..."
+	@if [ -n "$$(gofmt -l .)" ]; then \
+		echo "The following files need formatting:"; \
+		gofmt -l .; \
+		exit 1; \
+	fi
+	@if [ -n "$$(goimports -l .)" ]; then \
+		echo "The following files need import formatting:"; \
+		goimports -l .; \
+		exit 1; \
+	fi
+	@echo "All files are properly formatted!"
+
+# Format Go code
+.PHONY: fmt
+
+fmt:
+	gofmt -w .
+	goimports -w .
+
+# Install the application
+.PHONY: install
+
+install: build
+	go install
+
+# Check for Go modules updates
+.PHONY: update
+
+update:
+	go get -u ./...
+	go mod tidy
+
+
+# Ensure pre-commit hook is executable
+.PHONY: config-pre-commit
+
+config-pre-commit:
+	@echo "Setting up pre-commit hook..."
+	git config --local core.hooksPath .githooks/
+
+# Pre-commit checks (format, lint, test)
+.PHONY: pre-commit
+
+pre-commit: fmt lint test
+	@echo "Pre-commit checks passed!"

README.md

@@ -33,6 +33,9 @@ Or pass in the required secrets directly:
 
 ```go
 client, err := wx.NewClient(
+  wx.WithClientRetryConfig(wx.NewRetryConfig(
+    wx.WithReturnHTTPStatusAsErr(false)),
+  ),
   wx.WithWatsonxAPIKey(apiKey),
   wx.WithWatsonxProjectID(projectID),
 )
@@ -44,7 +47,7 @@ Generation:
 
 ```go
 result, _ := client.GenerateText(
-  "meta-llama/llama-3-1-8b-instruct",
+  "meta-llama/llama-3-3-70b-instruct",
   "Hi, who are you?",
   wx.WithTemperature(0.4),
   wx.WithMaxNewTokens(512),
@@ -57,7 +60,7 @@ Stream Generation:
 
 ```go
 dataChan, _ := client.GenerateTextStream(
-  "meta-llama/llama-3-1-8b-instruct",
+  "meta-llama/llama-3-3-70b-instruct",
   "Hi, who are you?",
   wx.WithTemperature(0.4),
   wx.WithMaxNewTokens(512),
@@ -142,6 +145,9 @@ Specify the Watsonx URL and IAM endpoint through the parameters of the NewClient
 
 ```go
 client, err := wx.NewClient(
+  	wx.WithClientRetryConfig(wx.NewRetryConfig(
+			wx.WithReturnHTTPStatusAsErr(false)),
+		),
   wx.WithURL("us-south.ml.test.cloud.ibm.com"),
   wx.WithIAM("iam.test.cloud.ibm.com"),
   wx.WithWatsonxAPIKey(apiKey),

pkg/internal/tests/models/embedding_test.go

@@ -1,9 +1,10 @@
 package test
 
 import (
-	wx "github.com/IBM/watsonx-go/pkg/models"
 	"reflect"
 	"testing"
+
+	wx "github.com/IBM/watsonx-go/pkg/models"
 )
 
 const (

pkg/internal/tests/models/generate_test.go

@@ -7,6 +7,11 @@ import (
 	wx "github.com/IBM/watsonx-go/pkg/models"
 )
 
+const (
+	modelLlama3  = "meta-llama/llama-3-3-70b-instruct"
+	modelFlanUL2 = "google/flan-ul2"
+)
+
 func TestClientCreationWithEnvVars(t *testing.T) {
 	_, err := wx.NewClient()
 
@@ -26,6 +31,9 @@ func TestClientCreationWithPassing(t *testing.T) {
 	}
 
 	_, err := wx.NewClient(
+		wx.WithClientRetryConfig(wx.NewRetryConfig(
+			wx.WithReturnHTTPStatusAsErr(false),
+		)),
 		wx.WithWatsonxAPIKey(apiKey),
 		wx.WithWatsonxProjectID(projectID),
 	)
@@ -51,7 +59,7 @@ func TestNilOptions(t *testing.T) {
 	client := getClient(t)
 
 	_, err := client.GenerateText(
-		"meta-llama/llama-3-70b-instruct",
+		modelLlama3,
 		"What day is it?",
 		nil,
 	)
@@ -64,7 +72,7 @@ func TestValidPrompt(t *testing.T) {
 	client := getClient(t)
 
 	_, err := client.GenerateText(
-		"meta-llama/llama-3-70b-instruct",
+		modelLlama3,
 		"Test prompt",
 	)
 	if err != nil {
@@ -76,7 +84,7 @@ func TestGenerateText(t *testing.T) {
 	client := getClient(t)
 
 	result, err := client.GenerateText(
-		"meta-llama/llama-3-70b-instruct",
+		modelLlama3,
 		"Hi, who are you?",
 		wx.WithTemperature(0.9),
 		wx.WithTopP(.5),
@@ -95,7 +103,7 @@ func TestGenerateTextStream(t *testing.T) {
 	client := getClient(t)
 
 	dataChan, err := client.GenerateTextStream(
-		"google/flan-ul2",
+		modelFlanUL2,
 		"Hi, who are you?",
 		wx.WithTemperature(0.9),
 		wx.WithTopP(.5),
@@ -126,7 +134,7 @@ func TestGenerateTextWithNoPrompt(t *testing.T) {
 	client := getClient(t)
 
 	dataChan, err := client.GenerateTextStream(
-		"google/flan-ul2",
+		modelFlanUL2,
 		"",
 		wx.WithTemperature(0.9),
 		wx.WithTopP(.5),
@@ -158,7 +166,7 @@ func TestGenerateTextWithNilOptions(t *testing.T) {
 	client := getClient(t)
 
 	result, err := client.GenerateText(
-		"meta-llama/llama-3-70b-instruct",
+		modelLlama3,
 		"Who are you?",
 		nil,
 	)

pkg/internal/tests/models/retry_test.go

@@ -35,10 +35,12 @@ func TestRetryWithSuccessOnFirstRequest(t *testing.T) {
 
 	resp, err := wx.Retry(
 		sendRequest,
-		wx.WithOnRetry(func(n uint, err error) {
-			retryCount = n
-			log.Printf("Retrying request after error: %v", err)
-		}),
+		wx.NewRetryConfig(
+			wx.WithOnRetry(func(n uint, err error) {
+				retryCount = n
+				log.Printf("Retrying request after error: %v", err)
+			}),
+		),
 	)
 
 	if err != nil {
@@ -60,8 +62,8 @@ func TestRetryWithSuccessOnFirstRequest(t *testing.T) {
 	}
 }
 
-// TestRetryWithNoSuccessStatusOnAnyRequest tests the retry mechanism with a server that always returns a 429 status code.
-func TestRetryWithNoSuccessStatusOnAnyRequest(t *testing.T) {
+// TestLegacyRetryWithNoSuccessStatusOnAnyRequest tests the retry mechanism with a server that always returns a 429 status code.
+func TestLegacyRetryWithNoSuccessStatusOnAnyRequest(t *testing.T) {
 	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		w.WriteHeader(http.StatusTooManyRequests)
 	}))
@@ -79,11 +81,13 @@ func TestRetryWithNoSuccessStatusOnAnyRequest(t *testing.T) {
 
 	resp, err := wx.Retry(
 		sendRequest,
-		wx.WithBackoff(backoffTime),
-		wx.WithOnRetry(func(n uint, err error) {
-			retryCount = n
-			log.Printf("Retrying request after error: %v", err)
-		}),
+		wx.NewRetryConfig(
+			wx.WithBackoff(backoffTime),
+			wx.WithOnRetry(func(n uint, err error) {
+				retryCount = n
+				log.Printf("Retrying request after error: %v", err)
+			}),
+		),
 	)
 
 	endTime := time.Now()
@@ -108,3 +112,72 @@ func TestRetryWithNoSuccessStatusOnAnyRequest(t *testing.T) {
 		t.Errorf("Expected minimum time of %v, but got %v", expectedMinimumTime, elapsedTime)
 	}
 }
+
+// TestRetryWithNoSuccessStatusOnAnyRequest tests the retry mechanism with a server that always returns a 429 status code.
+func TestRetryWithNoSuccessStatusOnAnyRequest(t *testing.T) {
+	expectedStatusCode := http.StatusTooManyRequests
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(expectedStatusCode)
+	}))
+	defer server.Close()
+
+	var backoffTime = 2 * time.Second
+	var retryCount uint = 0
+	var expectedRetries uint = 3
+
+	sendRequest := func() (*http.Response, error) {
+		return http.Get(server.URL + "/notfound")
+	}
+
+	startTime := time.Now()
+
+	resp, err := wx.Retry(
+		sendRequest,
+		wx.NewRetryConfig(
+			wx.WithReturnHTTPStatusAsErr(false), // Use new behavior: only return actual network errors
+			wx.WithBackoff(backoffTime),
+			wx.WithOnRetryV2(func(n uint, resp *http.Response, err error) {
+				retryCount = n
+				if err != nil {
+					t.Errorf("In OnRetry, expected nil, got error: %v", err)
+				}
+
+				if resp == nil {
+					t.Errorf("In OnRetry, expected non-nil response, got nil")
+				}
+
+				if resp != nil && resp.StatusCode != expectedStatusCode {
+					t.Errorf("Expected status code %d, got %d", expectedStatusCode, resp.StatusCode)
+				}
+
+				log.Printf("Retrying request after response with status code: %d", resp.StatusCode)
+			}),
+		),
+	)
+
+	endTime := time.Now()
+
+	elapsedTime := endTime.Sub(startTime)
+	expectedMinimumTime := backoffTime * time.Duration(expectedRetries)
+
+	if err != nil {
+		t.Errorf("Expected nil, got error: %v", err)
+	}
+
+	if resp == nil {
+		t.Errorf("Expected non-nil response, got nil")
+	}
+
+	if resp != nil && resp.StatusCode != expectedStatusCode {
+		t.Errorf("Expected status code %d, got %d", expectedStatusCode, resp.StatusCode)
+	}
+
+	if retryCount != expectedRetries {
+		t.Errorf("Expected 3 retries, but got %d", retryCount)
+	}
+
+	if elapsedTime < expectedMinimumTime {
+		t.Errorf("Expected minimum time of %v, but got %v", expectedMinimumTime, elapsedTime)
+	}
+}

pkg/internal/tests/models/test_utils.go

@@ -1,9 +1,10 @@
 package test
 
 import (
-	wx "github.com/IBM/watsonx-go/pkg/models"
 	"os"
 	"testing"
+
+	wx "github.com/IBM/watsonx-go/pkg/models"
 )
 
 func getClient(t *testing.T) *wx.Client {
@@ -17,6 +18,9 @@ func getClient(t *testing.T) *wx.Client {
 	}
 
 	client, err := wx.NewClient(
+		wx.WithClientRetryConfig(wx.NewRetryConfig(
+			wx.WithReturnHTTPStatusAsErr(false)),
+		),
 		wx.WithWatsonxAPIKey(apiKey),
 		wx.WithWatsonxProjectID(projectID),
 	)