As already started in #88304 (but only for NormalizedParams)
and later reverted in #89312 (because of cgi-bin problems),
PATH_INFO is no longer considered as a preferable SCRIPT_NAME
alternative. All known server configurations set SCRIPT_NAME
these days to a proper value when cgi.fix_pathinfo is set.
The fallback to PATH_INFO has been introduced with
the initial revision of TYPO3 and isn't needed at all nowadays,
it's actually wrong, as a REQUEST_URI like /index.php/foo/bar
would incorrectly be interpreted as $scriptName == "/foo/bar",
which let's all calculations on $scriptName fail and
even leads to XSS where values derived from $scriptName are
printed without being escaped.
Also any ORIG_SCRIPT_NAME evaluation is dropped, as this variable
contains the SCRIPT_NAME that was set by the webserver configuration
before PHP applied cgi.fix_pathinfo. Using ORIG_SCRIPT_NAME
effectively meant bypassing PHP's pathinfo fix. It usually contains
the cgi-wrapper paths, which is why PATH_INFO was used to overrule
wrong ORIG_SCRIPT_NAME values.
GeneralUtility::getIndpEnv('PATH_INFO') is adapted to trust the
servers PATH_INFO information, now that we no longer allow
servers to send SCRIPT_NAME as PATH_INFO (we enforce
cgi.fix_pathinfo=1 for CGI installations).
The normalized SCRIPT_NAME is now adapted to be encoded as a URL
path by default, as all TYPO3 usages expect this to be an URL path.
Note that $_SERVER['SCRIPT_NAME'] refers to the servers file
system path, not the URL encoded value.
This SCRIPT_NAME sanitization actually enables:
a) TYPO3 to be run in a subfolder that contains characters
that need URL encoding
e.g. `/test:site/` – url encoded that'd be `/test3Asite/`.
b) prevention of XSS in case third party extensions
missed to escape any URL that is derived from SCRIPT_NAME
(while making sure that properly escaped
output is not double escaped)