From 94032f4f9033ee052b68d1a9f87ce97210b4fda4 Mon Sep 17 00:00:00 2001
From: Matt Fellenz <matt@felle.nz>
Date: Mon, 25 Mar 2024 10:05:24 -0700
Subject: [PATCH] Implement external exchanges (#99)

* Implement external exchanges

* test: translateUrl test cases

* fix: double slash bug in translateUrl

---------

Co-authored-by: httpjamesm <github@httpjames.space>
---
 src/routes/home.go      | 58 +++++++++++++++++++++++++++--------------
 src/routes/home_test.go | 32 +++++++++++++++++++++++
 src/routes/question.go  |  4 ++-
 3 files changed, 73 insertions(+), 21 deletions(-)
 create mode 100644 src/routes/home_test.go

diff --git a/src/routes/home.go b/src/routes/home.go
index 306d41c..5c546cd 100644
--- a/src/routes/home.go
+++ b/src/routes/home.go
@@ -20,7 +20,41 @@ type urlConversionRequest struct {
 	URL string `form:"url" binding:"required"`
 }
 
-var stackExchangeRegex = regexp.MustCompile(`https://(.+).stackexchange.com/questions/`)
+var coreRegex = regexp.MustCompile(`(?:https?://)?(?:www\.)?([^/]+)(/(?:questions|q|a)/.+)`)
+
+// Will return `nil` if `rawUrl` is invalid.
+func translateUrl(rawUrl string) string {
+	coreMatches := coreRegex.FindStringSubmatch(rawUrl)
+	if coreMatches == nil {
+		return ""
+	}
+
+	domain := coreMatches[1]
+	rest := coreMatches[2]
+
+	exchange := ""
+	if domain == "stackoverflow.com" {
+		// No exchange parameter needed.
+	} else if sub, found := strings.CutSuffix(domain, ".stackexchange.com"); found {
+		if sub == "" {
+			return ""
+		} else if strings.Contains(sub, ".") {
+			// Anything containing dots is interpreted as a full domain, so we use the correct full domain.
+			exchange = domain
+		} else {
+			exchange = sub
+		}
+	} else {
+		exchange = domain
+	}
+
+	// Ensure we properly format the return string to avoid double slashes
+	if exchange == "" {
+		return rest
+	} else {
+		return fmt.Sprintf("/exchange/%s%s", exchange, rest)
+	}
+}
 
 func PostHome(c *gin.Context) {
 	body := urlConversionRequest{}
@@ -33,16 +67,9 @@ func PostHome(c *gin.Context) {
 		return
 	}
 
-	soLink := body.URL
+	translated := translateUrl(body.URL)
 
-	// remove the www.
-	soLink = strings.ReplaceAll(soLink, "www.", "")
-
-	// validate URL
-	isStackOverflow := strings.HasPrefix(soLink, "https://stackoverflow.com/questions/")
-	isShortenedStackOverflow := strings.HasPrefix(soLink, "https://stackoverflow.com/a/") || strings.HasPrefix(soLink, "https://stackoverflow.com/q/")
-	isStackExchange := stackExchangeRegex.MatchString(soLink)
-	if !isStackExchange && !isStackOverflow && !isShortenedStackOverflow {
+	if translated == "" {
 		c.HTML(400, "home.html", gin.H{
 			"errorMessage": "Invalid stack overflow/exchange URL",
 			"theme":        c.MustGet("theme").(string),
@@ -50,14 +77,5 @@ func PostHome(c *gin.Context) {
 		return
 	}
 
-	// if stack overflow, trim https://stackoverflow.com
-	if isStackOverflow || isShortenedStackOverflow {
-		c.Redirect(302, strings.TrimPrefix(soLink, "https://stackoverflow.com"))
-		return
-	}
-
-	// if stack exchange, extract the subdomain
-	sub := stackExchangeRegex.FindStringSubmatch(soLink)[1]
-
-	c.Redirect(302, fmt.Sprintf("/exchange/%s/%s", sub, strings.TrimPrefix(soLink, fmt.Sprintf("https://%s.stackexchange.com", sub))))
+	c.Redirect(302, translated)
 }
diff --git a/src/routes/home_test.go b/src/routes/home_test.go
new file mode 100644
index 0000000..eda6d64
--- /dev/null
+++ b/src/routes/home_test.go
@@ -0,0 +1,32 @@
+package routes
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestTranslateUrl(t *testing.T) {
+	assert := assert.New(t)
+
+	// Test with a Valid StackOverflow URL
+	assert.Equal("/questions/example-question", translateUrl("https://stackoverflow.com/questions/example-question"), "StackOverflow URL should not be modified")
+
+	// Test with Complex Subdomain
+	assert.Equal("/exchange/meta.math.stackexchange.com/q/example-question", translateUrl("https://meta.math.stackexchange.com/q/example-question"), "Complex StackExchange subdomain should be used as full exchange")
+
+	// Test with Non-StackExchange Domain
+	assert.Equal("/exchange/example.com/questions/example-question", translateUrl("https://example.com/questions/example-question"), "Non-StackExchange domain should be detected as exchange")
+
+	// Test with Invalid URL
+	assert.Equal("", translateUrl("This is not a URL"), "Invalid URL should return an empty string")
+
+	// Test with Empty String
+	assert.Equal("", translateUrl(""), "Empty string should return an empty string")
+
+	// Test with Missing Path
+	assert.Equal("", translateUrl("https://stackoverflow.com"), "URL without path should return an empty string")
+
+	// Test with Valid URL but Root Domain for StackExchange
+	assert.Equal("", translateUrl("https://stackexchange.com"), "Root StackExchange domain without subdomain should return an empty string")
+}
diff --git a/src/routes/question.go b/src/routes/question.go
index 5a70c7f..f4f99f3 100644
--- a/src/routes/question.go
+++ b/src/routes/question.go
@@ -47,7 +47,9 @@ func ViewQuestion(c *gin.Context) {
 
 	domain := "stackoverflow.com"
 
-	if params.Sub != "" {
+	if strings.Contains(params.Sub, ".") {
+		domain = params.Sub
+	} else if params.Sub != "" {
 		domain = fmt.Sprintf("%s.stackexchange.com", params.Sub)
 	}