me/offline-twitter
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix tweet URL parsing for urls with the stupid '?s=20' etc

Browse Source
This commit is contained in:
Alessio 2022-01-10 11:41:03 -05:00
parent 25fddaef9d
commit dc24ce5ad4
2 changed files with 12 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
scraper/user.go
View File

@ -91,7 +91,7 @@ func ParseHandleFromTweetUrl(tweet_url string) (UserHandle, error) {
tweet_url = ExpandShortUrl(tweet_url)
}
r := regexp.MustCompile(`^https://twitter.com/(\w+)/status/\d+$`)
r := regexp.MustCompile(`^https://twitter.com/(\w+)/status/\d+(?:\?.*)?$`)
matches := r.FindStringSubmatch(tweet_url)
if len(matches) != 2 { // matches[0] is the full string
return "", fmt.Errorf("Invalid tweet url: %s", tweet_url)

13
scraper/user_test.go
View File

@ -143,14 +143,23 @@ func TestParseHandleFromTweetUrl(t *testing.T) {
t.Errorf("Expected handle %q, got %q", expected_user, result)
}
// Test url with GET params
result, err = scraper.ParseHandleFromTweetUrl("https://twitter.com/NerdNoticing/status/1263192389050654720?s=20")
expected_user = scraper.UserHandle("NerdNoticing")
if err != nil {
t.Errorf("Unexpected error while parsing handle: %s", err)
}
if result != expected_user {
t.Errorf("Expected handle %q, got %q", expected_user, result)
}
// Test invalid url
_, err = scraper.ParseHandleFromTweetUrl("awjgwekf")
_, err = scraper.ParseHandleFromTweetUrl("https://twitter.com/NerdNoticing/status/1263192389050654720s=20")
if err == nil {
t.Errorf("Should have produced an error for invalid URL")
}
// Test empty string
_, err = scraper.ParseHandleFromTweetUrl("awjgwekf")
_, err = scraper.ParseHandleFromTweetUrl("")
if err == nil {
t.Errorf("Should have produced an error for invalid URL")
}