SSH2: update conditions under which _disconnect's code is executed
* terrafrost/disconnect-fix:
SSH2: update conditions under which _disconnect's code is executed
Currently, Net_SCP::put() has a callback parameter, allowing the consumer to monitor/log/report progress (or whatever they please).
This patch adds the same feature, working in the same way, to Net_SFTP::put()
the last _send_channel_packet was unnecessarily complex. this lead
to a number of problems. for example,
the window size was checked for adjustments in two places. in the
second place it receives a window adjustment packet and just assumes
that the window adjustment packet was big enough for the rest of the
data.
(thanks, pixall!)
Currently, the call to "scp -t" or "scp -f" just uses naive quoting - i.e. a couple of quote marks are thrown in.
But, this can easily be escaped from - if the filename has a quote mark of its own in it, for example.
e.g. if the filename is as follows, then bad things will happen:
";rm -rf /
Instead, escapeshellarg should be used, to make sure it gets escaped properly.
aside from the addition of OpenSSL support a few other changes have been included:
- setEngine(), as added by petrich, is depricated (not that it was ever in trunk to begin with)
it has been replaced with isValidEngine() and setPreferredEngine().
- replace _generate_xor() with increment_str()
_increment_str() had extra functionality that wasn't being used. ie. it could concatenate
multiple successive string increments to one another automatically. but not only was that
functionality not used - it also made the function less versatile. _increment_str() can be
used more easily for iterative brute forcing (for example)
- rename Crypt_Base::_stringShift to Crypt_Base::_string_shift (for consistency)
- more expansive unit test coverage
Sometimes SSH servers will null pad their RSA keys. this null padding
broke Net/SSH2.php's RSA implementation (Crypt/RSA.php's implementation
works just fine). Also, the -3 was counting the initial "\0" of $h
twice so adjust it to -2.
SFTP: Replace incorrect comment about filesize. There is no 4 GiB limit.
* bantu/sftp-large-file-comment:
SFTP: Replace incorrect comment about filesize. There is no 4 GiB limit.
RSA: make XML keys use unsigned integers
* terrafrost/rsa-xml-unsigned:
RSA: syntax error in unit tests
RSA: make unit tests perform string comparisons sans white space chars
RSA: update unit test
RSA: add unit test
RSA: make XML keys use unsigned integers
it doesn't need to be defined since the strlen($this->last_interactive_response)
check in _login_helper serves the same purpose that NET_SSH_MASK_LOGIN_INTERACTIVE
was intended to serve
ASN1: rewrite _decode_der
* terrafrost/asn1-handle-indef-lengths:
ASN1: one more unit test change
ASN1: another unit test update
ASN1: unit test adjustments
ASN1: cs adjustments to unit test
ASN1: add unit tests for indefinite length decoding
ASN1: CS adjustment (rm whitespace at eol)
ASN1: rewrite _decode_der
Hash: add sha256-96 and sha512-96
* terrafrost/hash-96-2:
Hash: apply fixes to sha512 unit tests as well
Hash: one more fix to unit test
Hash: fix unit tests
Hash: add unit tests
Hash: add sha256-96 and sha512-96
SSH2: make it so negotiated algorithms can be seen before login
* terrafrost/ssh2-show-methods-before-login:
SSH2: white space adjustment
SSH: update getServerPublicHostKey() to use new _connect() method
SSH2: fix if statement for conditional _connect() call
simplify calls to $this->_connect()
SSH2: make it so negotiated algorithms can be seen before login
This small patch adds hmac-sha2-256 support separately as requested in #423.
Some security standards now recommend to disable MD5 and SHA1, and use SHA2 instead. This change was tested using SHA2 against RHEL6's OpenSSH v5.3p1 and Solaris 11. And was also tested with RHEL5's OpenSSH 4.3p2 which doesn't include SHA2.
SFTP: Do not check filename over and over again, remove '.' and '..' from the map instead.
* bantu/sftp-skip-current-dir-by-removal:
Do not check filename over and over again, remove '.' and '..' from the map instead.
SSH2: make isConnected return true if even if we're not logged in
* terrafrost/ssh2-connected:
SSH2: make isConnected return true if even if we're not logged in
SCP: add support for file names with spaces
* terrafrost/scp-spaces:
SCP: always encapsulate filenames within double quotes
SCP: add support for file names with spaces
SFTP: add readlink and symlink functions
* terrafrost/symlink:
SFTP: return $sftp object in unit tests
SFTP: fix unit test
SFTP: syntax error in unit test
SFTP: add unit tests for symlinks
SFTP: add readlink and symlink functions
Random: include the Crypt_* classes if they're available
* bantu/a-random-change:
phpseclib_is_includable() -> phpseclib_resolve_include_path()
Drop Net_SSH2::_is_includable() method.
Random: a few changes to the stream_resolve_include_path_function
Replace stream_resolve_include_path with phpseclib_is_includable.
Random: add explanatory comment
SSH2: replace _is_includable() with stream_resolve_include_path
Random: include the Crypt_* classes if they're available
SSH2: make it so you can verify server public host key without logging on
* terrafrost/server-host-key:
SSH2: rm redundant part of SSH2 unit test
SSH2: add unit test for getServerPublicHostKey
SSH2: make it so you can verify server public host key without logging on
RSA: add support for PKCS8 encoded private keys
* terrafrost/pkcs8:
Crypt/Base: readability improvement
RSA: CS adjustments
RSA: rename PUBLIC_FORMAT_PKCS1_RAW -> PUBLIC_FORMAT_PKCS8
RSA: add PKCS8 unit tests
RSA: add support for saving encrypted PKCS8 keys
Crypt/Base: adjust default key size for pbkdf1
RSA: add support for loading PKCS8 encrypted private keys
Crypt/Base: add support for pbkdf1
RSA: add support for saving to PKCS8 (unencrypted)
BigInteger: make it so you can do $min->random($max)
* terrafrost/another-random-change:
BigInteger: one more change to the comments
BigInteger: update comments
BigInteger: unit test update
BigInteger: change argument names for random / randomPrime
BigInteger: fix unit tests
BigInteger: fix syntax error
BigInteger: make it so you can do $min->random($max)
RSA: slight adjustment to CRYPT_RSA_PUBLIC_FORMAT_PKCS1_RAW format
* terrafrost/rsa-format-change:
RSA: slight adjustment to CRYPT_RSA_PUBLIC_FORMAT_PKCS1_RAW format
none of the other statically defined variables have the potential
to change as multiple Net_SSH2 objects are created but this one does.
ie. if you connect to one SSH-2.0-SSHD server then no subsequent server
will use hmac-sha1-96 even if it could
CRYPT_RSA_PUBLIC_FORMAT_PKCS1_RAW and CRYPT_RSA_PUBLIC_FORMAT_PKCS1
produce two very similar looking keys but they are not the same.
As dissection OpenSSL's asn1parse would reveal CRYPT_RSA_PUBLIC_FORMAT_PKCS1
has the fact that it is an RSA key embedded within it whereas
CRYPT_RSA_PUBLIC_FORMAT_PKCS1_RAW does not. phpseclib now resolves
this ambiguity in the same way that OpenSSH's ssh-keygen does.
Despite this change CRYPT_RSA_PUBLIC_FORMAT_PKCS1_RAW is still incompatible
with OpenSSL's rsautl (CRYPT_RSA_PUBLIC_FORMAT_PKCS1 is compatible). I guess
this incompatibility isn't just due to the headers but is also due to the
overall structure of the format.
SFTP: add the ability for nlist() and rawlist() to be sorted
* terrafrost/sftp-sort2:
SFTP: update one last comment
SFTP: update unit test comments
SFTP: add new line to end of unit test
SFTP: assertEquals -> assertSame
SFTP: reset sort options every time and update unit test
SFTP: rm whitespace
SFTP: define $sortOptions
SFTP: add the ability for nlist() and rawlist() to be sorted
SSH2: make it so callback functions can make exec() return early
* terrafrost/ssh2-exec-callback-termination:
SSH2: 0x7FFFFFFF -> $this->window_size in one more place
SSH2: make it so callback functions can make exec() return early
SFTP: add file_exists, is_dir and is_file functions
* terrafrost/sftp-changes-20140510:
SFTP: use_cache -> use_stat_cache
SFTP: switch from using file existence cache to stat cache, like PHP
SFTP: add support for recursive nlist and rawlist
SFTP: add file_exists, is_dir and is_file functions
The code to convert the regular integer type into the appropriate format could all just be in-line'd when an integer primitive is used but this approach is easier to read.
