From 9ab22f4c2728f75dc7bbc8112b09caad3b65be96 Mon Sep 17 00:00:00 2001
From: Adriano Cunha <35786489+adrcunha@users.noreply.github.com>
Date: Fri, 6 Dec 2019 08:16:06 -0800
Subject: [PATCH] Report 4XX and 3XX links as link errors (#24)

* Report 4XX and 3XX links as link errors

If a link points to a file that 4XXs or 5XXs, it should be treated as a failure. Most obvious case is a typo in a link leading to 404 when clicked.

Also deal with the "small read buffer" error, so all tests pass.

* Make lint happy
---
 url_checker.go      | 21 +++++++++++++++------
 url_checker_test.go |  2 ++
 2 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/url_checker.go b/url_checker.go
index f3895a0..3104ce2 100644
--- a/url_checker.go
+++ b/url_checker.go
@@ -1,7 +1,8 @@
 package main
 
 import (
-	"errors"
+	"fmt"
+	"net/http"
 	"net/url"
 	"os"
 	"path"
@@ -42,12 +43,20 @@ func (c urlChecker) Check(u string, f string) error {
 	c.semaphore.Request()
 	defer c.semaphore.Release()
 
+	var sc int
 	if c.timeout == 0 {
-		_, _, err := fasthttp.Get(nil, u)
-		return err
+		sc, _, err = fasthttp.Get(nil, u)
+	} else {
+		sc, _, err = fasthttp.GetTimeout(nil, u, c.timeout)
+	}
+	if sc >= http.StatusBadRequest {
+		return fmt.Errorf("%s (HTTP error %d)", http.StatusText(sc), sc)
+	}
+	// Ignore errors from fasthttp about small buffer for URL headers,
+	// the content is discarded anyway.
+	if _, ok := err.(*fasthttp.ErrSmallBuffer); ok {
+		err = nil
 	}
-
-	_, _, err = fasthttp.GetTimeout(nil, u, c.timeout)
 	return err
 }
 
@@ -83,7 +92,7 @@ func (c urlChecker) resolveURL(u string, f string) (string, bool, error) {
 	}
 
 	if c.documentRoot == "" {
-		return "", false, errors.New("document root directory is not specified")
+		return "", false, fmt.Errorf("document root directory is not specified")
 	}
 
 	return path.Join(c.documentRoot, uu.Path), true, nil
diff --git a/url_checker_test.go b/url_checker_test.go
index 435976b..5d53912 100644
--- a/url_checker_test.go
+++ b/url_checker_test.go
@@ -15,6 +15,8 @@ func TestURLCheckerCheck(t *testing.T) {
 		assert.Equal(t, nil, c.Check(u, "README.md"))
 	}
 
+	assert.NotEqual(t, nil, c.Check("http://www.google.com/README-I-DONT-EXIST.md", "README.md"))
+
 	for _, u := range []string{"https://hey-hey-hi-google.com", "READYOU.md", "://"} {
 		assert.NotEqual(t, nil, c.Check(u, "README.md"))
 	}