From f6175bcd1803327eda1f479db3d07648742579e8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Daniel=20Mart=C3=AD?= <mvdan@mvdan.cc>
Date: Tue, 22 Nov 2022 12:08:50 +0000
Subject: [PATCH] cmd/xurls: show parse, request, and dial errors

These were being entirely ignored, which isn't right for a "fix" feature
to ensure that URLs are good.
---
 cmd/xurls/main.go                   | 11 +++++++----
 cmd/xurls/sequencer.go              |  8 ++++----
 cmd/xurls/testdata/script/fix.txtar |  9 ++++++---
 3 files changed, 17 insertions(+), 11 deletions(-)

diff --git a/cmd/xurls/main.go b/cmd/xurls/main.go
index 88c362e..46edbac 100644
--- a/cmd/xurls/main.go
+++ b/cmd/xurls/main.go
@@ -97,6 +97,7 @@ func scanPath(re *regexp.Regexp, path string) error {
 				match := line[pair[0]:pair[1]]
 				origURL, err := url.Parse(match)
 				if err != nil {
+					r.appendBroken(match, err.Error())
 					continue
 				}
 				fixed := origURL.String()
@@ -128,15 +129,17 @@ func scanPath(re *regexp.Regexp, path string) error {
 					}
 					req, err := http.NewRequest(http.MethodHead, fixed, nil)
 					if err != nil {
+						r.appendBroken(match, err.Error())
 						continue
 					}
 					req.Header.Set("User-Agent", userAgent)
 					resp, err := client.Do(req)
 					if err != nil {
+						r.appendBroken(match, err.Error())
 						continue
 					}
-					if resp.StatusCode >= 400 {
-						r.appendBroken(match, resp.StatusCode)
+					if code := resp.StatusCode; code >= 400 {
+						r.appendBroken(match, fmt.Sprintf("%d %s", code, http.StatusText(code)))
 					}
 					resp.Body.Close()
 				}
@@ -148,7 +151,7 @@ func scanPath(re *regexp.Regexp, path string) error {
 					atomic.AddUint32(&atomicFixedCount, 1)
 				}
 			}
-			io.WriteString(r, line)
+			io.WriteString(r, line) // add the fixed line to outBuf
 			return nil
 		})
 		if err := scanner.Err(); err != nil {
@@ -173,7 +176,7 @@ func scanPath(re *regexp.Regexp, path string) error {
 		var s strings.Builder
 		fmt.Fprintf(&s, "found %d broken urls in %q:\n", len(state.brokenURLs), path)
 		for _, broken := range state.brokenURLs {
-			fmt.Fprintf(&s, "  * %d: %s\n", broken.statusCode, broken.url)
+			fmt.Fprintf(&s, "  * %s - %s\n", broken.url, broken.reason)
 		}
 		return errors.New(s.String())
 	}
diff --git a/cmd/xurls/sequencer.go b/cmd/xurls/sequencer.go
index ae7ea00..9a5e4e1 100644
--- a/cmd/xurls/sequencer.go
+++ b/cmd/xurls/sequencer.go
@@ -114,8 +114,8 @@ type reporterState struct {
 }
 
 type brokenURL struct {
-	url        string
-	statusCode int
+	url    string
+	reason string
 }
 
 // getState blocks until any prior reporters are finished with the reporter
@@ -135,9 +135,9 @@ func (r *reporter) Write(p []byte) (int, error) {
 	return r.getState().out.Write(p)
 }
 
-func (r *reporter) appendBroken(url string, statusCode int) {
+func (r *reporter) appendBroken(url, reason string) {
 	state := r.getState()
-	state.brokenURLs = append(state.brokenURLs, brokenURL{url, statusCode})
+	state.brokenURLs = append(state.brokenURLs, brokenURL{url, reason})
 }
 
 // Report emits a non-nil error to the reporter's error stream,
diff --git a/cmd/xurls/testdata/script/fix.txtar b/cmd/xurls/testdata/script/fix.txtar
index 90a6642..7b3874f 100644
--- a/cmd/xurls/testdata/script/fix.txtar
+++ b/cmd/xurls/testdata/script/fix.txtar
@@ -27,9 +27,10 @@ cmp redirects redirects.golden
 
 ! exec xurls -fix broken
 stdout -count=1 '^broken$'
-stderr -count=1 '4 broken urls'
-stderr -count=2 '/404'
-stderr -count=2 '/500'
+stderr -count=1 '5 broken urls'
+stderr -count=2 '/404 - 404 Not Found'
+stderr -count=2 '/500 - 500 Internal Server Error'
+stderr -count=1 'totallydoesnotexist.localhost/ - Head .* dial tcp'
 cmp broken broken.golden
 
 -- nothing --
@@ -74,7 +75,9 @@ Temporary redirect codes:
 One redirect: ${SERVER}/redir-1
 404 errors: ${SERVER}/404 ${SERVER}/404
 500 errors: ${SERVER}/500 ${SERVER}/500
+Dial error: http://totallydoesnotexist.localhost/
 -- broken.golden --
 One redirect: ${SERVER}/plain
 404 errors: ${SERVER}/404 ${SERVER}/404
 500 errors: ${SERVER}/500 ${SERVER}/500
+Dial error: http://totallydoesnotexist.localhost/