aboutsummaryrefslogtreecommitdiff
path: root/brokenlinks/brokenlinks_test.go
diff options
context:
space:
mode:
Diffstat (limited to 'brokenlinks/brokenlinks_test.go')
-rw-r--r--brokenlinks/brokenlinks_test.go231
1 files changed, 160 insertions, 71 deletions
diff --git a/brokenlinks/brokenlinks_test.go b/brokenlinks/brokenlinks_test.go
index db3775a..a04be7f 100644
--- a/brokenlinks/brokenlinks_test.go
+++ b/brokenlinks/brokenlinks_test.go
@@ -4,8 +4,11 @@
package brokenlinks_test
import (
+ "errors"
"log"
+ "net"
"net/http"
+ "net/url"
"os"
"path/filepath"
"testing"
@@ -13,7 +16,9 @@ import (
libnet "git.sr.ht/~shulhan/pakakeh.go/lib/net"
"git.sr.ht/~shulhan/pakakeh.go/lib/test"
+ "golang.org/x/net/html/atom"
+ "git.sr.ht/~shulhan/jarink"
"git.sr.ht/~shulhan/jarink/brokenlinks"
"git.sr.ht/~shulhan/jarink/internal"
)
@@ -212,8 +217,27 @@ func runServerSlow() {
func TestScan(t *testing.T) {
var testUrl = `http://` + testAddress
+ // Generate ParentUrl.
+
+ parsedTestUrl, err := url.Parse(testUrl)
+ if err != nil {
+ t.Fatal(err)
+ }
+ parsedUrlBrokenHtml, err := url.Parse(testUrl + `/broken.html`)
+ if err != nil {
+ t.Fatal(err)
+ }
+ parsedUrlPage2, err := url.Parse(testUrl + `/page2`)
+ if err != nil {
+ t.Fatal(err)
+ }
+
+ // Generate error for ErrScan.
+
+ _, errScanPort := url.Parse(`http://127.0.0.1:abc`)
+
type testCase struct {
- exp map[string][]brokenlinks.Broken
+ exp map[string][]jarink.Link
expError string
desc string
opts brokenlinks.Options
@@ -246,92 +270,133 @@ func TestScan(t *testing.T) {
},
expError: `Scan: Options: unknown status code "50"`,
}, {
- desc: `With ` + testUrl,
+ desc: `With Url=testUrl`,
opts: brokenlinks.Options{
Url: testUrl,
IgnoreStatus: `403`,
Insecure: true,
IsVerbose: true,
},
- exp: map[string][]brokenlinks.Broken{
- testUrl: []brokenlinks.Broken{
+ exp: map[string][]jarink.Link{
+ testUrl: []jarink.Link{
{
- Link: testUrl + `/broken.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedTestUrl,
+ Url: testUrl + `/broken.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
}, {
- Link: testUrl + `/brokenPage`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedTestUrl,
+ Url: testUrl + `/brokenPage`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.A),
}, {
- Link: `http://127.0.0.1:abc`,
- Error: `parse "http://127.0.0.1:abc": invalid port ":abc" after host`,
- Code: brokenlinks.StatusBadLink,
+ ParentUrl: parsedTestUrl,
+ Url: `http://127.0.0.1:abc`,
+ ErrScan: errScanPort,
+ Error: `parse "http://127.0.0.1:abc": invalid port ":abc" after host`,
+ StatusCode: brokenlinks.StatusBadLink,
+ Kind: int(atom.A),
}, {
- Link: `http:/127.0.0.1:11836`,
- Error: `Get "http:/127.0.0.1:11836": http: no Host in request URL`,
- Code: brokenlinks.StatusBadLink,
+ ParentUrl: parsedTestUrl,
+ Url: `http:/127.0.0.1:11836`,
+ ErrScan: &url.Error{
+ Op: `Get`,
+ URL: `http:/127.0.0.1:11836`,
+ Err: errors.New(`http: no Host in request URL`),
+ },
+ Error: `Get "http:/127.0.0.1:11836": http: no Host in request URL`,
+ StatusCode: brokenlinks.StatusBadLink,
+ Kind: int(atom.A),
+ IsExternal: true,
}, {
- Link: `https://domain`,
- Error: `Get "https://domain": dial tcp: lookup domain: no such host`,
- Code: 700,
+ ParentUrl: parsedTestUrl,
+ Url: `https://domain`,
+ ErrScan: &url.Error{
+ Op: `Get`,
+ URL: `https://domain`,
+ Err: &net.OpError{
+ Op: `dial`,
+ Net: `tcp`,
+ Err: &net.DNSError{
+ Err: `no such host`,
+ Name: `domain`,
+ IsNotFound: true,
+ },
+ },
+ },
+ Error: `Get "https://domain": dial tcp: lookup domain: no such host`,
+ StatusCode: 700,
+ Kind: int(atom.A),
+ IsExternal: true,
},
},
- testUrl + `/broken.html`: []brokenlinks.Broken{
+ testUrl + `/broken.html`: []jarink.Link{
{
- Link: testUrl + `/brokenPage`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlBrokenHtml,
+ Url: testUrl + `/brokenPage`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.A),
},
},
- testUrl + `/page2`: []brokenlinks.Broken{
+ testUrl + `/page2`: []jarink.Link{
{
- Link: testUrl + `/broken.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/broken.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
}, {
- Link: testUrl + `/page2/broken/relative`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/page2/broken/relative`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.A),
}, {
- Link: testUrl + `/page2/broken2.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/page2/broken2.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
},
},
},
}, {
- desc: `With ` + testUrl + `/page2`,
// Scanning on "/page2" should not scan the the "/" or other
// pages other than below of "/page2" itself.
+ desc: `With Url=/page2`,
opts: brokenlinks.Options{
Url: testUrl + `/page2`,
IsVerbose: true,
},
- exp: map[string][]brokenlinks.Broken{
- testUrl + `/page2`: []brokenlinks.Broken{
+ exp: map[string][]jarink.Link{
+ testUrl + `/page2`: []jarink.Link{
{
- Link: testUrl + `/broken.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/broken.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
+ IsExternal: true,
}, {
- Link: testUrl + `/page2/broken/relative`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/page2/broken/relative`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.A),
}, {
- Link: testUrl + `/page2/broken2.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/page2/broken2.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
},
},
},
}}
- var (
- result *brokenlinks.Result
- err error
- )
for _, tcase := range listCase {
t.Run(tcase.desc, func(tt *testing.T) {
internal.CacheFile = func() (string, error) {
return tt.TempDir() + `/cache.json`, nil
}
- result, err = brokenlinks.Scan(tcase.opts)
+ result, err := brokenlinks.Scan(tcase.opts)
if err != nil {
- test.Assert(tt, tcase.opts.Url+` error`,
- tcase.expError, err.Error())
+ test.Assert(tt, tcase.opts.Url+` error`, tcase.expError, err.Error())
return
}
@@ -348,8 +413,13 @@ func TestScan(t *testing.T) {
func TestScan_pastResult(t *testing.T) {
var testUrl = `http://` + testAddress
+ parsedUrlPage2, err := url.Parse(testUrl + `/page2`)
+ if err != nil {
+ t.Fatal(err)
+ }
+
type testCase struct {
- exp map[string][]brokenlinks.Broken
+ exp map[string][]jarink.Link
desc string
expError string
opts brokenlinks.Options
@@ -369,33 +439,36 @@ func TestScan_pastResult(t *testing.T) {
PastResultFile: `testdata/past_result.json`,
IgnoreStatus: `403`,
},
- exp: map[string][]brokenlinks.Broken{
- testUrl + `/page2`: []brokenlinks.Broken{
+ exp: map[string][]jarink.Link{
+ testUrl + `/page2`: []jarink.Link{
{
- Link: testUrl + `/broken.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/broken.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
+ IsExternal: true,
}, {
- Link: testUrl + `/page2/broken/relative`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/page2/broken/relative`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.A),
}, {
- Link: testUrl + `/page2/broken2.png`,
- Code: http.StatusNotFound,
+ ParentUrl: parsedUrlPage2,
+ Url: testUrl + `/page2/broken2.png`,
+ StatusCode: http.StatusNotFound,
+ Kind: int(atom.Img),
},
},
},
}}
- var (
- result *brokenlinks.Result
- err error
- )
for _, tcase := range listCase {
t.Run(tcase.desc, func(tt *testing.T) {
internal.CacheFile = func() (string, error) {
return tt.TempDir() + `/cache.json`, nil
}
- result, err = brokenlinks.Scan(tcase.opts)
+ result, err := brokenlinks.Scan(tcase.opts)
if err != nil {
test.Assert(tt, tcase.opts.Url+` error`, tcase.expError, err.Error())
return
@@ -410,14 +483,24 @@ func TestScan_pastResult(t *testing.T) {
func TestScan_slow(t *testing.T) {
const testUrl = `http://` + testAddressSlow
+ parsedUrlSlow1, err := url.Parse(testUrl + `/slow1`)
+ if err != nil {
+ t.Fatal(err)
+ }
+ parsedUrlSlow2, err := url.Parse(testUrl + `/slow2`)
+ if err != nil {
+ t.Fatal(err)
+ }
+ parsedUrlSlow3, err := url.Parse(testUrl + `/slow3`)
+ if err != nil {
+ t.Fatal(err)
+ }
+
var opts = brokenlinks.Options{
Url: testUrl,
IsVerbose: true,
}
-
- var gotResult *brokenlinks.Result
- var err error
- gotResult, err = brokenlinks.Scan(opts)
+ gotResult, err := brokenlinks.Scan(opts)
if err != nil {
t.Fatal(err)
}
@@ -426,18 +509,24 @@ func TestScan_slow(t *testing.T) {
//t.Logf(`got=%s`, got)
var expResult = &brokenlinks.Result{
- BrokenLinks: map[string][]brokenlinks.Broken{
- testUrl + `/slow1`: []brokenlinks.Broken{{
- Link: testUrl + `/slow3/sub`,
- Code: http.StatusForbidden,
+ BrokenLinks: map[string][]jarink.Link{
+ testUrl + `/slow1`: []jarink.Link{{
+ ParentUrl: parsedUrlSlow1,
+ Url: testUrl + `/slow3/sub`,
+ StatusCode: http.StatusForbidden,
+ Kind: int(atom.A),
}},
- testUrl + `/slow2`: []brokenlinks.Broken{{
- Link: testUrl + `/slow3/sub`,
- Code: http.StatusForbidden,
+ testUrl + `/slow2`: []jarink.Link{{
+ ParentUrl: parsedUrlSlow2,
+ Url: testUrl + `/slow3/sub`,
+ StatusCode: http.StatusForbidden,
+ Kind: int(atom.A),
}},
- testUrl + `/slow3`: []brokenlinks.Broken{{
- Link: testUrl + `/slow3/sub`,
- Code: http.StatusForbidden,
+ testUrl + `/slow3`: []jarink.Link{{
+ ParentUrl: parsedUrlSlow3,
+ Url: testUrl + `/slow3/sub`,
+ StatusCode: http.StatusForbidden,
+ Kind: int(atom.A),
}},
},
}
generated by cgit v1.3 (git 2.53.0) at 2026-04-15 04:03:03 +0000