26278 2014-07-07 20:48:01 +0000 getElementText - no info about U+200E, U+200F 2014-07-16 20:09:16 +0000 1 1 1 Unclassified Browser Test/Tools WG WebDriver unspecified All All RESOLVED FIXED P2 normal --- 20860 1 botalov.andrey public-browser-tools-testing dburns mike public-browser-tools-testing oldest_to_newest 108765 0 botalov.andrey 2014-07-07 20:48:01 +0000 Atoms contain some treatment for those characters - https://github.com/SeleniumHQ/selenium/blob/master/javascript/atoms/dom.js#L1208 This change seems to be done in August, 2013 after the algorithm for Webdriver W3C spec was written. So I think README.md of Webdriver project should have a not that when someone makes changes to API of remote end he should also file a bug against Webdriver spec (or even file a bug against Webdriver spec prior to making a change) so Webdriver spec and Selenium won't become out-of-sync. P.S.: Also I haven't noticed in lines near L1208 code that removes \f, \v 109003 1 dburns 2014-07-16 13:32:28 +0000 (In reply to Andrey Botalov from comment #0) > > P.S.: Also I haven't noticed in lines near L1208 code that removes \f, \v Step 2 -> 1 -> 2nd bullet -> 1 handles this scenario 109006 2 dburns 2014-07-16 13:53:18 +0000 https://dvcs.w3.org/hg/webdriver/rev/4e8c789c7f54 109025 3 botalov.andrey 2014-07-16 20:02:26 +0000 There are other whitespace and BiDi characters in http://www.unicode.org/Public/6.3.0/ucd/PropList.txt and http://en.wikipedia.org/wiki/Space_(punctuation)#Spaces_in_Unicode. I think that if only \u200b, \u200e, \u200f, \v, \f should be removed by getElementText() from the string, then the spec should also contain an explanation (note) about what makes those characters special and why other invisible "spaces" shouldn't be removed. I don't know much about Unicode but IMO those "spaces" also look like zero-width: U+180E U+200C U+2060 U+061C etc. I also found this line in gecko-dev repository: https://github.com/mozilla/gecko-dev/blob/master/browser/base/content/browser.js#L2205: > value = value.replace(/[\u00ad\u034f\u061c\u115f-\u1160\u17b4-\u17b5\u180b-\u180d\u200b\u200e-\u200f\u202a-\u202e\u2060-\u206f\u3164\ufe00-\ufe0f\ufeff\uffa0\ufff0-\ufff8]|\ud834[\udd73-\udd7a]|[\udb40-\udb43][\udc00-\udfff]/g, encodeURIComponent); It seems that implementation in Firefox is a bit more complicated. 109026 4 dburns 2014-07-16 20:09:16 +0000 I would rather have this either mimick the current implementation, which this bug initial was about. Adding other spaces would need to be in a new bug with specific use cases so that we can discuss. I suggest bringing this issue up on the mailing list once you have a bug.