initial

.dockerignore

@@ -0,0 +1 @@
+example

.gitignore

@@ -0,0 +1 @@
+example

Dockerfile

@@ -0,0 +1,37 @@
+# First stage: Use the golangci-lint image to run the linter
+FROM golangci/golangci-lint:latest as lint
+
+# Set the Current Working Directory inside the container
+WORKDIR /app
+
+# Copy the go.mod file and the rest of the application code
+COPY go.mod ./
+COPY . .
+
+# Run golangci-lint
+RUN golangci-lint run
+
+# Second stage: Use the official Golang image to run tests
+FROM golang:1.22 as test
+
+# Set the Current Working Directory inside the container
+WORKDIR /app
+
+# Copy the go.mod file and the rest of the application code
+COPY go.mod ./
+COPY . .
+
+# Run tests
+RUN go test -v ./...
+
+# Final stage: Combine the linting and testing stages
+FROM golang:1.22 as final
+
+# Ensure that the linting stage succeeded
+WORKDIR /app
+COPY --from=lint /app .
+COPY --from=test /app .
+
+# Set the final CMD to something minimal since we only needed to verify lint and tests during build
+CMD ["echo", "Build and tests passed successfully!"]
+

Makefile

@@ -0,0 +1,22 @@
+# Targets
+.PHONY: all run test clean
+
+all: run
+
+example: *.go ./cmd/example/*.go
+	go build -o $@ ./cmd/example/main.go
+
+run: example
+	./example
+
+test:
+	go test -v ./...
+
+clean:
+	rm -f example
+
+docker:
+	docker build --progress plain .
+
+lint:
+	golangci-lint run

alexatop.go

@@ -0,0 +1,119 @@
+package alexatop
+
+import (
+	"archive/zip"
+	"bytes"
+	_ "embed"
+	"encoding/csv"
+	"fmt"
+	"io"
+	"math/rand"
+	"strings"
+	"sync"
+)
+
+//go:embed backup/alexa.zip
+var alexaZip []byte
+
+var (
+	urls       []string
+	fetchError error
+	once       sync.Once
+)
+
+// FetchAlexaTop1M fetches the Alexa Top 1M websites from the embedded zip file and returns a list of URLs
+func FetchAlexaTop1M() ([]string, error) {
+	once.Do(func() {
+		// Step 1: Open the zip file from the embedded data
+		r, err := zip.NewReader(bytes.NewReader(alexaZip), int64(len(alexaZip)))
+		if err != nil {
+			fetchError = fmt.Errorf("failed to open zip file: %v", err)
+			return
+		}
+
+		// Step 2: Locate and read the CSV file within the zip
+		var csvFile *zip.File
+		for _, f := range r.File {
+			if strings.HasSuffix(f.Name, ".csv") {
+				csvFile = f
+				break
+			}
+		}
+		if csvFile == nil {
+			fetchError = fmt.Errorf("csv file not found in zip archive")
+			return
+		}
+
+		rc, err := csvFile.Open()
+		if err != nil {
+			fetchError = fmt.Errorf("failed to open csv file: %v", err)
+			return
+		}
+		defer rc.Close()
+
+		// Step 3: Parse the CSV file and extract URLs
+		csvReader := csv.NewReader(rc)
+		var fetchedUrls []string
+		for {
+			record, err := csvReader.Read()
+			if err == io.EOF {
+				break
+			}
+			if err != nil {
+				fetchError = fmt.Errorf("error reading csv: %v", err)
+				return
+			}
+			if len(record) < 2 {
+				continue
+			}
+			fetchedUrls = append(fetchedUrls, fmt.Sprintf("http://%s", record[1]))
+		}
+
+		urls = fetchedUrls
+	})
+
+	return urls, fetchError
+}
+
+// RandomSite returns a random site from the list of Alexa Top 1M URLs
+func RandomSite() (string, error) {
+	urls, err := FetchAlexaTop1M()
+	if err != nil {
+		return "", err
+	}
+
+	return urls[rand.Intn(len(urls))], nil
+}
+
+// NthSite returns the nth site from the list of Alexa Top 1M URLs
+func NthSite(n int) (string, error) {
+	urls, err := FetchAlexaTop1M()
+	if err != nil {
+		return "", err
+	}
+
+	if n < 0 || n >= len(urls) {
+		return "", fmt.Errorf("index out of range")
+	}
+
+	return urls[n], nil
+}
+
+// RandomSites returns n random sites from the list of Alexa Top 1M URLs
+func RandomSites(n int) ([]string, error) {
+	urls, err := FetchAlexaTop1M()
+	if err != nil {
+		return nil, err
+	}
+
+	if n < 0 || n > len(urls) {
+		return nil, fmt.Errorf("invalid number of sites requested")
+	}
+
+	selected := make([]string, n)
+	for i := range selected {
+		selected[i] = urls[rand.Intn(len(urls))]
+	}
+
+	return selected, nil
+}

alexatop_test.go

@@ -0,0 +1,78 @@
+package alexatop
+
+import (
+	"testing"
+)
+
+func TestFetchAlexaTop1M(t *testing.T) {
+	urls, err := FetchAlexaTop1M()
+	if err != nil {
+		t.Fatalf("Expected no error, got %v", err)
+	}
+
+	if len(urls) < 700_000 {
+		t.Fatalf("Expected at least 700,000 URLs, got %d", len(urls))
+	}
+}
+
+func TestRandomSite(t *testing.T) {
+	site, err := RandomSite()
+	if err != nil {
+		t.Fatalf("Expected no error, got %v", err)
+	}
+
+	if site == "" {
+		t.Fatalf("Expected a valid URL, got an empty string")
+	}
+}
+
+func TestNthSite(t *testing.T) {
+	urls, err := FetchAlexaTop1M()
+	if err != nil {
+		t.Fatalf("Expected no error, got %v", err)
+	}
+
+	site, err := NthSite(99)
+	if err != nil {
+		t.Fatalf("Expected no error, got %v", err)
+	}
+
+	if site != urls[99] {
+		t.Fatalf("Expected %s, got %s", urls[99], site)
+	}
+
+	_, err = NthSite(len(urls))
+	if err == nil {
+		t.Fatalf("Expected error for out-of-range index, got none")
+	}
+}
+
+func TestRandomSites(t *testing.T) {
+	n := 5
+	randomSites, err := RandomSites(n)
+	if err != nil {
+		t.Fatalf("Expected no error, got %v", err)
+	}
+
+	if len(randomSites) != n {
+		t.Fatalf("Expected %d random sites, got %d", n, len(randomSites))
+	}
+
+	for _, site := range randomSites {
+		if site == "" {
+			t.Fatalf("Expected a valid URL, got an empty string")
+		}
+	}
+}
+
+func TestRandomSitesMoreThanAvailable(t *testing.T) {
+	urls, err := FetchAlexaTop1M()
+	if err != nil {
+		t.Fatalf("Expected no error, got %v", err)
+	}
+
+	_, err = RandomSites(len(urls) + 1)
+	if err == nil {
+		t.Fatalf("Expected error for requesting more sites than available, got none")
+	}
+}

go.mod

@@ -0,0 +1,3 @@
+module sneak.berlin/go/alexatop
+
+go 1.22.2