Unsignedio
/
python-gnupg
mirror of https://github.com/markqvist/python-gnupg.git
1
0
Fork 0
Code Issues Releases Wiki Activity

Merge branch 'release/1.2.3'

fix/24-enc-to-file
Isis Lovecruft 2013-10-08 14:20:17 +00:00
parent abf981a1ca 33fd778137
commit 287f188340
No known key found for this signature in database
GPG Key ID: 5C17776E27F7E84D
10 changed files with 1508 additions and 105 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

832
docs/Complete-Guide-to-Publishing-OpenPGP-in-DNS.md 100644
View File

@ -0,0 +1,832 @@
<!--
Livejournal Introduction:
I recently wrestled with something, learned quite a lot, and came up with a document that I'm really rather proud of, that
shares knowledge that's not all out there in one place anywhere else. Along the way I've written some software that I'm
releasing, that makes all of what I've learned a lot easier, and may help make the world a little more secure. I'd like to
share it here.
This is going to be a technical post. For that I apologize. The target of this post is anyone who has a GPG key that they'd
like to expand to a greater audience, and who controls DNS for any of the email domains they publish. Anyone that I host DNS
or mail for is also welcome to do this, if you use PGP, as part of the goal of writing this is to encourage adoption and use
of these methods
<lj-cut text="This will be long and technical">
-->
# The complete guide to publishing PGP keys in DNS
## Introduction
Publishing PGP keys is a pain. There are many disjoint keyservers, three or
four _networks_ of which, which do (or don't) share information with each
other. Some are corporate, some are private. And it's a crapshoot as to
whose key is going to be on which, or worse, which will have the latest copy
of a person's key.
For a long time, GPG has had a way to publish keys in DNS, but it hasn't been
well documented. This document hopes to change that.
After reading this, you should:
* Know the three ways to publish a key
* Have at least a couple tools to do so
* Have learned a bit more about DNS
The target audience for this guide is a technical one. It's expected you
understand what DNS is, and what an RFC and a resource record is.
There are three ways to publish a PGP key in DNS. Most modern versions of GPG
can retrieve from all three, although it's not enabled by default. There are
no compile-time options you need to enable it, and it's simple to turn on. Of
the three key-publishing methods, there are two that you probably shouldn't
use at the same time, and there are advantages and disadvantages to each,
which I hope to outline below, both in general and for each method.
### Advantages to DNS publishing of your keys
* It's universal. Your DNS is your own, and you don't have to worry about
which network of vastly-disconnectedkeyservers is caching your key.
* Using DNS does not stop you from publishing via other means.
* If you run an organization, you can easily publish all your employee-keys
via this method, and in the same step,define a signing-policy, such that a
person need only assign trust to your organization's "keysigning key" (or
theCEO's key, or the CTO's), without the trouble of running a keyserver.
* DNSSEC can be (somewhat) used as an additional trust-path vector. More on
this in the notes at the bottom.
* You do not have to be searching DNS for keys in order to publish. On the
same note, you do not have to be publishing in this manner to search
forothers there.
### Disadvantages to DNS publishing
* If you don't control your own DNS (or have a good relationship with your DNS
admin), this isn't going to beas easy or even possible. Ideally, you want
to be running BIND.
* With two of the three methods listed here, you're going to need to be able
to put a CERT record into your DNS. Mostweb-enabled DNS tools probably will
not give you this ability. The third uses TXT records, which SPF has caused
to befairly universal in web-interfaces. However, it's also the least
standards-defined of the three.
* Using at least some of these methods, it's not always a "set it and forget
it" procedure. You may need toperiodically re-export your key and
re-publish it, especially if you gain new signatures.
* Using some of these methods, you're going to be putting some pretty large,
pretty unwiedly lines in your DNS zones. Not everyone will easily be able
to retrieve them, but again, you can still publish other ways.
* Using some of these methods, DNS is just a means to an end: you still need
to publish your key elsewhere, like a webpage,and the DNS records just point
at it.
* Initial verifications of most of these seem to imply that only DSA keys are
supported, although I welcome feedback. Itseems the community is trying to
get RSA keys to make a comeback. They're the only type supported by the
gpg2.0 card, andthey are the default keytype. There was a while where they
weren't, though. Since writing this document, I've discoveredthat "new" RSA
keys work, but ancient RSA keys with no subkeys tend to misbehave.
### Turning on key-fetching via DNS
Inside your GPG "options" file, find the "auto-key-locate" line, and add
"cert" and/or "pka" to the options.
auto-key-locate cert pka (as well as other methods, like keyserver URLs)
Don't be surprised if a lot of people don't use this method.
Note that you can also turn on two options during signature verification.
They are specified in a "verify-options" clause in your config file, or on the
command line, and they are (right from the GPG manpage):
pka-lookups
Enable PKA lookups to verify sender addresses. Note that
PKA is based on DNS, and so enabling this option may dis-
close information on when and what signatures are veri-
fied or to whom data is encrypted. This is similar to the
"web bug" described for the auto-key-retrieve feature.
And:
pka-trust-increase
Raise the trust in a signature to full if the signature
passes PKA validation. This option is only meaningful if
pka-lookups is set.
You can also use the same options on the command line (as you'll see in this
document).
## Types of PGP Key Records
### DNS PKA Records
Relevant RFCs: None that I can find.
Other Docs: The GPG source and mailing lists.
#### Advantages
* It's a TXT record. Easy to put in a zonefile with most management software.
* No special tools required to generate, just three simple pieces of data.
* Since it uses a special subzone, you can manage the _pka namespace in a
separate zonefile.
* GPG has an option, when verifying a signature, to look up these records
(--verify-options pka-lookups), so it's doubly useful, both from a
distribution and a verification point.
#### Disadvantages
* As with IPGP certs, you're at the mercy of the URL. This doesn't put your
key in DNS, just the location of it, and the fingerprint. Some clients may
not be able to support https or http 1.1.
* Not RFC standard.
#### Howto
1. Figure out which key you want to export:
%gpg --list-keys danm@prime.gushi.org
Warning: using insecure memory!
pub 1024D/624BB249 2000-10-02 &lt;-- I'm going to use this one.
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
uid Daniel Mahoney (Secondary Email) &lt;gushi@gushi.org&gt;
sub 2048g/DE20C529 2000-10-02
pub 1024R/309C17C5 1997-05-08
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
2. Export the key to a file (I use keyid.pub.asc, but it can be anything)
%gpg --export --armor 624BB249 &gt; 624BB249.pub.asc
Warning: using insecure memory!
%
3. Get the fingerprint for your key:
%gpg --list-keys --fingerprint 624BB249
gpg: WARNING: using insecure memory!
gpg: please see http://www.gnupg.org/faq.html for more information
pub 1024D/624BB249 2000-10-02
Key fingerprint = C206 3054 5492 95F3 3490 37FF FBBE 5A30 624B B249 &lt;-- That bit is your fingerprint.
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
uid Daniel Mahoney (Secondary Email) &lt;gushi@gushi.org&gt;
sub 2048g/DE20C529 2000-10-02
4. Copy the file somewhere, like your webspace. It need not live on the same
server. It needs to be accessable by the url you create in the next step.
%cp 624BB249.pub.asc public_html/danm.pubkey.txt
5. Make up your text record. The format is:
danm._pka.prime.gushi.org. TXT "v=pka1;fpr=C2063054549295F3349037FFFBBE5A30624BB249;uri=http://prime.gushi.org/danm.pubkey.txt"
We'll take this in several parts. The record label is simply the email
address with "._pka." replacing the "@". danm@prime.gushi.org becomes
danm._pka.prime.gushi.org. Don't forget the trailing dot, if you're using the
fully qualified name. I recommend sticking with fully-qualified, for
simplicity.
The body of the record is also simple. The v portion is just a version.
There's only one version as far as I can tell, 'pka1'. The fpr is the
fingerprint, with all whitespace stripped, and in uppercase. The uri is the
location a key can be retrieved from. All the "names" are lowercase,
separated by semicolons.
6. Publish the above record in your DNS. Bump your serial number and reload
your nameserver. If you're using DNSSEC, re-sign your zone.
#### Testing
Most of the tests we're going to do for these are essentially the same
activity. See if our DNS server is handing out an answer, and then see if GPG
can retrieve it.
1. A simple dig:
%dig +short danm._pka.prime.gushi.org. TXT
"v=pka1\;fpr=C2063054549295F3349037FFFBBE5A30624BB249\;uri=http://prime.gushi.org/danm.pubkey.txt"
(The backslashes before the semicolons are normal). Other than that, it seems
to make sense and match what I put in.)
2. Test it with GPG. Rather than messing around with, and adding-from and
deleting from live keyrings, you can do:
%echo "foo" | gpg --no-default-keyring --keyring /tmp/gpg-$$ --encrypt --armor --auto-key-locate pka -r you@you.com
(where you@you.com is the address of your primary key.) The /tmp/gpg-$$
creates a random file named after your PID. What you should see, and what I
see, is something like this:
gpg: WARNING: using insecure memory!
gpg: please see http://www.gnupg.org/faq.html for more information
gpg: keyring `/tmp/gpg-39996' created
gpg: requesting key 624BB249 from http server prime.gushi.org
gpg: key 624BB249: public key "Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;" imported
gpg: public key of ultimately trusted key CF45887D not found
gpg: 3 marginal(s) needed, 1 complete(s) needed, PGP trust model
gpg: depth: 0 valid: 1 signed: 0 trust: 0-, 0q, 0n, 0m, 0f, 1u
gpg: Total number processed: 1
gpg: imported: 1
gpg: automatically retrieved `danm@prime.gushi.org' via PKA
gpg: DE20C529: There is no assurance this key belongs to the named user
pub 2048g/DE20C529 2000-10-02 Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
Primary key fingerprint: C206 3054 5492 95F3 3490 37FF FBBE 5A30 624B B249
Subkey fingerprint: CE40 B786 81E2 5CB9 F7D3 1318 9488 EB58 DE20 C529
It is NOT certain that the key belongs to the person named
in the user ID. If you *really* know what you are doing,
you may answer the next question with yes.
Use this key anyway? (y/N) y
-----BEGIN PGP MESSAGE-----
Version: GnuPG v1.4.10 (FreeBSD)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=NbgW
-----END PGP MESSAGE-----
The "insecure memory" warning is a silly warning that the only way to turn off is to run GPG setuid root.
You can see in the output that the key comes from PKA.
The "it is NOT certain" warning has nothing to do with the fact that it came
from DNS. You will get that warning every time you use that key (or any gpg
key) until you have edited it and assigned ownertrust to it, or until the key
is signed with a trusted signature, either from your personal web of trust, or
from a signing service like the pgp.com directory.
3. Ask other people to run it for you and send you the resulting blob. You should be able to decrypt it with your private key.
### PGP CERT Records
Also known as: The "big" CERT record.
Relevant RFCs: [RFC 2538](http://www.faqs.org/rfcs/rfc2538.html),
[RFC 4398](http://www.faqs.org/rfcs/rfc4398.html), specifically sections 2.1
and 3.3
#### Advantages
* DNS is all you need. You don't have to host the key elsewhere. As a DNS
nerd, this strikes me as very cool.
* Suprisingly easy to verify with dig, if you have a base64 converter handy
(openssl includes one)
#### Disadvantages
* These records can get big. Really big. Especially if you have photo-ids on your keys. You can play with export-options to shrink it somewhat. Big dns packets may require EDNS, or dns-over-tcp, which not everyone supports, but support is becoming more widespread as a result of DNSSEC awareness.
* Requires the make-dns-cert tool, which isn't built by default.
* Requires you to have some control over your actual zonefile. Most control panels won't cut it.
* Make-dns-cert currently generates a very ugly record for this.
#### How to
1. As before, the first step is to figure out which key we want.
%gpg --list-keys danm@prime.gushi.org
Warning: using insecure memory!
pub 1024D/624BB249 2000-10-02 &lt;-- I'm going to use this one.
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
uid Daniel Mahoney (Secondary Email) &lt;gushi@gushi.org&gt;
sub 2048g/DE20C529 2000-10-02
pub 1024R/309C17C5 1997-05-08
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
2. We export the key, but this time, it needs to be binary.
%gpg --export 624BB249 &gt; 624BB249.pub.bin
Warning: using insecure memory!
3. We run make-dns-cert on it. make-dns-cert comes with no manual or docs,
but running with -h gives you all the clue you need.
make-dns-cert
-f fingerprint
-u URL
-k key file
-n DNS name
So then,
make-dns-cert -n danm.prime.gushi.org. -k 624BB249.pub.bin
<pre>`%make-dns-cert -n danm.prime.gushi.org. -k 624BB249.pub.bin
danm.prime.gushi.org. TYPE37 \# 1298 0003 0000 00 9901A20439D8DAF1110400F770EC6AA006076334BEC6DB6FBB237DC194BC0AB8
302C8953F04C28FC2085235D4F10EFA027234FBD63D142CCADD5213AD2B79A22C89ED9B4138370D8220D0F987F993A5364A4A7AC3D42F3765C384
71DDD0FF3372E4AE6F7BEE1E18EF464A0BEB5BBE860A08238891455EBE7CB53D567E981F78ADBD263206B0493ADCB74DD00A0FF0E9A1CD245415E
CEF59435162AFCE4CDD14BC70400EA38FF501256E773DEA299404854D99F4EDB2757AA911A9C77C68AB8D6622E517A556C43D21F0523C568F016C
D0DB89EF435F0D53B4E07434213F899E6578955DC2C147931E7B6901C9FD8A02705417D69A879B3CC196D2AC2EAEF311192EE89ABAF5A60942167
B4625735FCBDFB5DE0E3AC1236A53FA4D7CDD7D75F5DE85AF50400867D9546B28B79AF10541053CF4AB06A6171BFD21458BFD12AF1AE2B2401CAD
8851661F8AF6602F80EDAC99C79616BE1F910F4156242003779C68D7A079A8B18F89DD293E1B247E7420471300A4A0730AA61DE281CCC211FC405
A0A8A79877999FF9042AD892AB927DA371E8883BBB370AB7A97841408C3486BB18598CF2559BB42844616E69656C20502E204D61686F6E6579203
C64616E6D407072696D652E67757368692E6F72673E884E04101102000E050239D8DAF1040B030102021901000A0910FBBE5A30624BB249FA2E00
9B057503ED498695AE5ED73CA1B98EBAEE13F717E500A0921E0D92724459100266FEBBC29E911C8B0F530BB43244616E69656C204D61686F6E657
920285365636F6E6461727920456D61696C29203C67757368694067757368692E6F72673E8860041311020020050245D49FD7021B23060B090807
030204150208030416020301021E01021780000A0910FBBE5A30624BB249158400A082C8AF43DA8B85F740D6B1A6E9FF0B4490520B8C00A08F77D
21FBF86C842963E8090DC0646D1DD7F95C9B9020D0439D8DAF4100800F64257B7087F081772A2BAD6A942F305E8F95311394FB6F16EB94B3820DA
01A756A314E98F4055F3D007C6CB43A994ADF74C648649F80C83BD65E917D4A1D350F8F5595FDC76524F3D3D8DDBCE99E1579259CDFDB8AE744FC
5FC76BC83C5473061CE7CC966FF15F9BBFD915EC701AAD35B9E8DA0A5723AD41AF0BF4600582BE5F488FD584E49DBCD20B49DE49107366B336C38
0D451D0F7C88B31C7C5B2D8EF6F3C923C043F0A55B188D8EBB558CB85D38D334FD7C175743A31D186CDE33212CB52AFF3CE1B1294018118D7C84A
70A72D686C40319C807297ACA950CD9969FABD00A509B0246D3083D66A45D419F9C7CBD894B221926BAABA25EC355E9320B3B00020207FF5E1A3C
C5DA00E1E94EC8EF6C7FE9B49D944C71D8BBC817DD8E64A7344B9E48392E0B833B3B1DB7E6D5A38BE2826DEF0060F78C6417871EAF1CFBCBC47D2
7E93718D975E0A3A36D868C021D6B771740CE2918307D69D614BBF0632DC31932EA31397A7F3B04618C9A76C2F38265C7037E303EDD8AEF03D069
208E3FE9C4EA77D83E6311ED36C013D58C54E914B263A459E22D463A0288510C4752B99C163EEA0A55686979691AB0D9F9AA0C06C834446D7A723
EC534D819301382621ACF8930C74E9FD28C8797718AEC2C30CF601E24194B799234104A3D6239657B1D4AD545BDAA637F61541435CB51B4D138FB
F55E1A9FD2EED860E4459D6795B6FCCA23155A8846041811020006050239D8DAF4000A0910FBBE5A30624BB249415A009E37BCFDC64E76CBF6A86
82B85EA161BD1DFB793DF00A0C471BC7B9723535CD855D8FF1EB93F01E251B698
%
The program prints that all on **one line**.
Immediately, we notice a few things.
* The record type isn't "CERT", it's "TYPE37". This confused me for a while until I discovered [RFC3597](http://www.faqs.org/rfcs/rfc3597.html) Basically, it's a way that a DNS server can handle a resource record it doesn't know about, by giving it some special fields like the "#", as well as a length (which is the 1298 you see there).
* The rest of the record is on one line. I wrapped it for the purposes of brevity. If I were using this in a zonefile, I would need to be careful that I wrapped it on a byte-boundary (every two characters is a byte). If I miss the boundary, named will refuse to load it, dnssec-signzone won't touch it, etc.
4. So the thing is ugly and you don't want to touch it. The easiest way to work with it is to drop all that into a file:
%make-dns-cert -n danm.prime.gushi.org. -k 624BB249.pub.bin &gt; 624BB249.big.cert
5. And then either read it into your editor, or tack it on like this:
%cat 624BB249.big.cert &gt;&gt; your.zonefile
Be sure to make a backup first. Either way, you never have to copy/paste the raw hex and worry about newlines being inserted where you don't want them.
6. Before you reload your zone, you might want to use named-checkzone on it first:
prime# named-checkzone gushi.org gushi.org.hosts
zone gushi.org/IN: loaded serial 2009102909
OK
prime#
7. Voice of experience: You may want to dial the TTL (which controls how long servers will cache your data) way down on the record above. It's not hard, just put a number before the TYPE37, with a space, i.e:
danm.prime.gushi.org. 30 TYPE37
This way if it all goes terribly wrong, or you need to make changes, it won't be cached for very long.
8. If it looks okay, bump your serial number and reload.
#### Testing
1. As above, you can dig, but you won't be able to easily read the results:
prime# dig +short danm.prime.gushi.org CERT
;; Truncated, retrying in TCP mode.
PGP 0 0
mQGiBDnY2vERBAD3cOxqoAYHYzS+xttvuyN9wZS8CrgwLIlT8Ewo/CCF
I11PEO+gJyNPvWPRQsyt1SE60reaIsie2bQTg3DYIg0PmH+ZOlNkpKes
PULzdlw4Rx3dD/M3Lkrm977h4Y70ZKC+tbvoYKCCOIkUVevny1PVZ+mB
94rb0mMgawSTrct03QCg/w6aHNJFQV7O9ZQ1Fir85M3RS8cEAOo4/1AS
Vudz3qKZQEhU2Z9O2ydXqpEanHfGirjWYi5RelVsQ9IfBSPFaPAWzQ24
nvQ18NU7TgdDQhP4meZXiVXcLBR5Mee2kByf2KAnBUF9aah5s8wZbSrC
6u8xEZLuiauvWmCUIWe0Ylc1/L37XeDjrBI2pT+k183X119d6Fr1BACG
fZVGsot5rxBUEFPPSrBqYXG/0hRYv9Eq8a4rJAHK2IUWYfivZgL4DtrJ
nHlha+H5EPQVYkIAN3nGjXoHmosY+J3Sk+GyR+dCBHEwCkoHMKph3igc
zCEfxAWgqKeYd5mf+QQq2JKrkn2jceiIO7s3CrepeEFAjDSGuxhZjPJV
m7QoRGFuaWVsIFAuIE1haG9uZXkgPGRhbm1AcHJpbWUuZ3VzaGkub3Jn
PohOBBARAgAOBQI52NrxBAsDAQICGQEACgkQ+75aMGJLskn6LgCbBXUD
7UmGla5e1zyhuY667hP3F+UAoJIeDZJyRFkQAmb+u8KekRyLD1MLtDJE
YW5pZWwgTWFob25leSAoU2Vjb25kYXJ5IEVtYWlsKSA8Z3VzaGlAZ3Vz
aGkub3JnPohgBBMRAgAgBQJF1J/XAhsjBgsJCAcDAgQVAggDBBYCAwEC
HgECF4AACgkQ+75aMGJLskkVhACggsivQ9qLhfdA1rGm6f8LRJBSC4wA
oI930h+/hshClj6AkNwGRtHdf5XJuQINBDnY2vQQCAD2Qle3CH8IF3Ki
utapQvMF6PlTETlPtvFuuUs4INoBp1ajFOmPQFXz0AfGy0OplK33TGSG
SfgMg71l6RfUodNQ+PVZX9x2Uk89PY3bzpnhV5JZzf24rnRPxfx2vIPF
RzBhznzJZv8V+bv9kV7HAarTW56NoKVyOtQa8L9GAFgr5fSI/VhOSdvN
ILSd5JEHNmszbDgNRR0PfIizHHxbLY7288kjwEPwpVsYjY67VYy4XTjT
NP18F1dDox0YbN4zISy1Kv884bEpQBgRjXyEpwpy1obEAxnIByl6ypUM
2Zafq9AKUJsCRtMIPWakXUGfnHy9iUsiGSa6q6Jew1XpMgs7AAICB/9e
GjzF2gDh6U7I72x/6bSdlExx2LvIF92OZKc0S55IOS4Lgzs7Hbfm1aOL
4oJt7wBg94xkF4cerxz7y8R9J+k3GNl14KOjbYaMAh1rdxdAzikYMH1p
1hS78GMtwxky6jE5en87BGGMmnbC84JlxwN+MD7diu8D0Gkgjj/pxOp3
2D5jEe02wBPVjFTpFLJjpFniLUY6AohRDEdSuZwWPuoKVWhpeWkasNn5
qgwGyDREbXpyPsU02BkwE4JiGs+JMMdOn9KMh5dxiuwsMM9gHiQZS3mS
NBBKPWI5ZXsdStVFvapjf2FUFDXLUbTROPv1Xhqf0u7YYORFnWeVtvzK
IxVaiEYEGBECAAYFAjnY2vQACgkQ+75aMGJLsklBWgCeN7z9xk52y/ao
aCuF6hYb0d+3k98AoMRxvHuXI1Nc2FXY/x65PwHiUbaY
It's still ugly, but it's not AS ugly because it's base64, which includes
spaces, at least, and is easier to search for a pattern. Base64 can also be
easily wrapped on any boundary, which is nice.
You can run your existing exported key through a base64 converter, like the
one built into the openssl binary, if you want to compare:
%cat 624BB249.pub.bin | openssl enc -base64
mQGiBDnY2vERBAD3cOxqoAYHYzS+xttvuyN9wZS8CrgwLIlT8Ewo/CCFI11PEO+g
JyNPvWPRQsyt1SE60reaIsie2bQTg3DYIg0PmH+ZOlNkpKesPULzdlw4Rx3dD/M3
Lkrm977h4Y70ZKC+tbvoYKCCOIkUVevny1PVZ+mB94rb0mMgawSTrct03QCg/w6a
(...etc...)
OPv1Xhqf0u7YYORFnWeVtvzKIxVaiEYEGBECAAYFAjnY2vQACgkQ+75aMGJLsklB
WgCeN7z9xk52y/aoaCuF6hYb0d+3k98AoMRxvHuXI1Nc2FXY/x65PwHiUbaY
Now, while you could compare things byte-by-byte here, what I've done as a
"casual check" is just pick random strings in the text and see if they match
up. For example, you can see that "reaIsie2" is present in both. They both
start with and end with similar strings on every line. The real test, of
course, is to see if GPG recognizes it as a valid key.
By the way, since I use DNSSEC, dnssec-signzone rewrites this record into the
proper "presentation format" for me, which is base64. If you want a similar
function, you can use named-compilezone to get some of the same effects, or
you can use the shell script I provide later in this document, with which you
don't even need make-dns-cert.
2. Testing with gpg
As above, the command to test this is remarkably simple:
%rm /tmp/gpg-*
%echo "foo" | gpg --no-default-keyring --keyring /tmp/gpg-$$ --encrypt --armor --auto-key-locate cert -r danm@prime.gushi.org
gpg: keyring `/tmp/gpg-39996' created
gpg: key 624BB249: public key "Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;" imported
gpg: Total number processed: 1
gpg: imported: 1
gpg: automatically retrieved `danm@prime.gushi.org' via DNS CERT
gpg: DE20C529: There is no assurance this key belongs to the named user
pub 2048g/DE20C529 2000-10-02 Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
Primary key fingerprint: C206 3054 5492 95F3 3490 37FF FBBE 5A30 624B B249
Subkey fingerprint: CE40 B786 81E2 5CB9 F7D3 1318 9488 EB58 DE20 C529
It is NOT certain that the key belongs to the person named
in the user ID. If you *really* know what you are doing,
you may answer the next question with yes.
Use this key anyway? (y/N) y
-----BEGIN PGP MESSAGE-----
Version: GnuPG v1.4.10 (FreeBSD)
hQIOA5SI61jeIMUpEAf/Sx7MKWm+e9EpUTSrDaBp4nJfDcBeqbYJulPRbDZz7eVW
2+ol6sG0jWjuirbG1YppZccEr9mgqaQujdSXb/bleD8POS0TEWuf3aPswFQvHf90
NLEzHt6BnfLoeobXXxyCflNaGX8zW+XgJtwZqAc2+jietuz8MOUhrf5m17CsW/wZ
IuEqwaek+K1irJp+w3rhaE08Jzb/S4CCifeW9J3mK57chQoPOu7Nz3rY666YKp/3
9T9StOgmFiNpvtFPNy4N7hHMHvbQwRsKlnkl+a7n0Aq2+OF4d1+/k2EE4uSGgcz0
oHvee8DnuOx3P92mO4Jz5/0O0lwBD7I51iOjzUurTAgAiIM5sHV8/QFCVzH9Ule+
gd8Wo5momcphkU/AXpce5Xgi/Vm4oGQ0x0queii8afUrzkpeN5SuwgQfAdOPiXW5
2bo527jBllxOxjeBasfky82XheTnLzbAQNvQNTEM9zE7zCl1LQJUZEJ1hVzcOevI
s+cm/AaGII9VkrAtSt3aLSRZuRJHFmhGvYd2Hz5WzcV1YFjXXP1eLwfetDBlaeB9
/K5v4hZBkIZPbHX0DcLVrP96mCIT4wCBYSJw+I6n0E6Fz3IfybQG2HMfqWp966/c
00ijx/aRDh42Dr/fTropuzzFzQr7weYDa1JnN3Zoftv6Zb/n+NcrmMiDCH8jJV6E
uMkaeeB5Mv7ssDQ9kPhO989CHFcznrE1lgOxjX8=
=NTLY
-----END PGP MESSAGE-----
Okay, as above, try to decrypt that with your private key.
### IPGP CERT Records
Also known as: The "little" or "short" CERT record. (These terms are purely my
own).
Relevant RFCs: [RFC 2538](http://www.faqs.org/rfcs/rfc2538.html),
[RFC 4398](http://www.faqs.org/rfcs/rfc4398.html), specifically sections 2.1
and 3.3
IPGP certs are interesting. It's basically the same pieces of infomation that
are in the PKA record, as above, except that it's supported by an RFC.
Despite the RFC compliance, I am not sure if any non-gpg client knows to look
for them. However, because it's a DNS cert, make-dns-cert encodes the
information in binary, and your DNS server will see it in base64. So
verifying it visually is harder than verifying either of the above.
#### Advantages
* Small, easy-to-transmit records.
* Can use the same uri as the PKA record.
#### Disadvantages
* Relies on the URI scheme. I haven't yet been able to get a definitive list
of what uri schemes are supported, although I've seen http and finger. I've
also seen reports that unless gpg is compiled against curl, http 1.1 is not
supported (what this actually means is that any host that supports SSL will
probably work, because of some of the nuances of SSL).
* With PGP certs and IPGP certs, GPG will only parse the first key it gets, so
if you publish both, and one doesn't work, there's no failover. I've argued
that this should be fixed.
* Requires make-dns-cert, which is not built in GPG by default. (But see "A
Better Way" below)
* Requires publication in your main DNS zone.
* Despite being RFC compliant, GPG has additional trust vectors for PKA but
not this, despite the fact that they share basically the same information.
* Harder to verify with dig.
#### Howto
1. Note that some of these steps are redundant. If you're already doing a PKA
key, skip to step 5.
2. Dig:
%gpg --list-keys danm@prime.gushi.org
Warning: using insecure memory!
pub 1024D/624BB249 2000-10-02 &lt;-- I'm going to use this one.
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
uid Daniel Mahoney (Secondary Email) &lt;gushi@gushi.org&gt;
sub 2048g/DE20C529 2000-10-02
pub 1024R/309C17C5 1997-05-08
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
3. Export the key to a file (I use keyid.pub.asc, but it can be anything)
%gpg --export --armor 624BB249 &gt; 624BB249.pub.asc
Warning: using insecure memory!
%
4. Get the fingerprint for your key:
%gpg --list-keys --fingerprint 624BB249
gpg: WARNING: using insecure memory!
gpg: please see http://www.gnupg.org/faq.html for more information
pub 1024D/624BB249 2000-10-02
Key fingerprint = C206 3054 5492 95F3 3490 37FF FBBE 5A30 624B B249 &lt;-- That bit is your fingerprint.
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
uid Daniel Mahoney (Secondary Email) &lt;gushi@gushi.org&gt;
sub 2048g/DE20C529 2000-10-02
5. As above, run make-dns-cert. This time we use the -n, -f, and -u options:
%make-dns-cert -n danm.prime.gushi.org. -f C2063054549295F3349037FFFBBE5A30624BB249 -u http://prime.gushi.org/danm.pubkey.txt
danm.prime.gushi.org. TYPE37 \# 64 0006 0000 00 14 C2063054549295F3349037FFFBBE5A30624BB249 687474703A2F2F7072696D652E67757368692E6F72672F64616E6D2E7075626B65792E747874
%
6. Put the above in DNS. All on one line. Optionally add a TTL.
7. IMPORTANT: make sure you don't have any other CERT records with the same
label (i.e. a "big" cert, as above). While it won't break things, you have
no control over which (of multiple) people will get.
8. Reload your zone, and test. Testing will probably look VERY MUCH like the
above, but here are the steps anyway:
#### Testing
1. Dig:
%dig +short danm.prime.gushi.org CERT
6 0 0 FMIGMFRUkpXzNJA3//u+WjBiS7JJaHR0cDovL3ByaW1lLmd1c2hpLm9y Zy9kYW5tLnB1YmtleS50eHQ=
Sadly, I haven't come across an easy way to decipher it yet, but there's
always gpg.
2. GPG:
Since we're fetching the same kind of record, the command is exactly the same
as before:
%echo "foo" | gpg --no-default-keyring --keyring /tmp/gpg-$$ --encrypt --armor --auto-key-locate cert -r danm@prime.gushi.org
gpg: WARNING: using insecure memory!
gpg: please see http://www.gnupg.org/faq.html for more information
gpg: keyring `/tmp/gpg-39996' created
gpg: requesting key 624BB249 from http server prime.gushi.org
gpg: key 624BB249: public key "Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;" imported
gpg: public key of ultimately trusted key CF45887D not found
gpg: 3 marginal(s) needed, 1 complete(s) needed, PGP trust model
gpg: depth: 0 valid: 1 signed: 0 trust: 0-, 0q, 0n, 0m, 0f, 1u
gpg: Total number processed: 1
gpg: imported: 1
gpg: automatically retrieved `danm@prime.gushi.org' via DNS CERT
gpg: DE20C529: There is no assurance this key belongs to the named user
pub 2048g/DE20C529 2000-10-02 Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
Primary key fingerprint: C206 3054 5492 95F3 3490 37FF FBBE 5A30 624B B249
Subkey fingerprint: CE40 B786 81E2 5CB9 F7D3 1318 9488 EB58 DE20 C529
It is NOT certain that the key belongs to the person named
in the user ID. If you *really* know what you are doing,
you may answer the next question with yes.
Use this key anyway? (y/N) y
-----BEGIN PGP MESSAGE-----
Version: GnuPG v1.4.10 (FreeBSD)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=zY9G
-----END PGP MESSAGE-----
%
Strangely, the output doesn't say what PKA does (a PKA retrieval has a line
about fetching via HTTP), however, by checking my webserver logs, I can see it
retrieved it from there:
%tail -200 /usr/local/apache/logs/prime.gushi.org.log | grep pubkey | tail -1
prime.gushi.org 72.9.101.130 - - [28/Oct/2009:23:50:43 -0400] "GET /danm.pubkey.txt HTTP/1.1" 200 4337 "-" "-"
%
As usual, test decryption, etc. You're done.
## Further Steps
* Figure out which of these are useful to you, and use them.* When someone
asks for your public key, tell them to run the above command instead of
mailing them your key or sending them a keyserver URL.
* Consider using the pka-related verify-options.
* Look into embracing DNSSEC. With a signed root, there's a good trust-path
vector here. Who knows, maybe some day GPG will be dnssec-aware so it will
give more credit to a secure DNS transaction. Without a signed root, there
are still ways to have those who care about security use it, through
services such as [ISC's DLV registry](http://dlv.isc.org).
* On DNSSEC: At present, GPG cannot see the difference between an insecure
response (one from an unsigned zone) and a correctly validated one from a
signed zone. (In a signed zone, an unsigned or malformed will simply get a
SERVFAIL dns response). Look into sponsoring development of GPG to make it
as an application more aware of this.
## A better way to generate records
In reading over a lot of these commands, I've come across a few problems with
the tools involved. They either require you to assemble large records by
hand, or manipulate huge files.
DNS has also come a long way since these tools were written, and RFCs have
solidified that have determined the "presentation format" (i.e. the "master
file format") of what CERT records should look like.
On top of everything, the make-dns-cert tool is not built by default, and is
not present in most binary distributions (RPM's, deb packages, FreeBSD's
ports).
Thus, I took it upon myself to rewrite make-dns-cert as a shell script.
### Advantages
* Extracts your key for you (takes a keyid as the argument).
* Formats all three record types for you, you can pipe it right into your zone
file.
* Takes email address as an argument, generates record label.
* No compiling needed.
* Should work with most systems. Requires openssl and sed, a few other
standard utilities.
* Generates base64-ified CERT records, split into easy, manageable pieces.
* Generates DNS-friendly comments, so repeating tasks are easy to reference.
* (Eventually) available as a tarball, or as a paste-and-go script.
* Arguments are in logical DNS record order `emailaddress keyid [url]`.
* Will generate an IPGP CERT record without a URI (this is legal per RFC4398).
You can see sample output
[here](http://www.gushi.org/make-dns-cert/sample-output.txt), and you can view
the script itself
[here](http://www.gushi.org/make-dns-cert/make-dns-cert.sh.txt). Depending on
your MIME settings, you can probably get a download link if you go
[here](http://www.gushi.org/make-dns-cert/make-dns-cert.sh). If you see the
script rather than getting a download prompt, you can just save-as.
README, Changelog, TODO coming soon.
## Other notes
I'm not 100 percent sure (mainly because I haven't tried), but with IPGP cert,
and PKA, I believe I could in theory point at a keyserver directly, for
example, specify a uri of
[http://pgp.mit.edu:11371/pks/lookup?op=get&amp;search=0xB0307039309C17C5](http://pgp.mit.edu:11371/pks/lookup?op=get&amp;search=0xB0307039309C17C5).
I'm a bit dubious about the question marks and equals-signs, or if I might
have to uri-encode things. It's something to be tried.
I'm trying to convince the GPG people that this would be much better adopted
if the make-dns-cert tool was built/included by default, or if its function
were included in gpg rather than a third-party tool. This is analagous as to
how dnssec-keygen is used to generate SSHFP DNS records.
It doesn't do any actual cryptography, just some binary conversion, so in
theory it could be rewritten in pure-perl, so there's nothing to compile.
I've made the argument to the GPG developers that if multiple CERT records are
available, all should be tried if one fails. So far, if multiple exist, only
the first received is parsed, and of course, DNS round-robins the answers by
default.
It took me quite a lot of trial and error to realize that there's a difference
between "modern" RSA keys, like this:
%gpg --list-keys --fingerprint gushi@prime.gushi.org
pub 2048R/CF45887D 2009-10-29
Key fingerprint = FCB0 485E 050D DDFA 83C6 76E3 E722 3C05 CF45 887D
uid Gushi Test &lt;gushi@prime.gushi.org&gt;
sub 2048R/C9761244 2009-10-29
and ancient RSA keys like this pgp2.6.2 monster:
%gpg --list-keys --fingerprint danm@prime.gushi.org
pub 1024R/309C17C5 1997-05-08
Key fingerprint = 04 4B 1A 2E C4 62 95 73 73 A4 EA D0 08 A4 45 76
uid Daniel P. Mahoney &lt;danm@prime.gushi.org&gt;
Note the lack of a subkey there. Note the weird fingerprint. I have not been
able to get this key to properly export with gpg. If someone knows the Deep
Magic, let me know.
## References
### Blog posts and list threads
While researching this I came across little more than a few blog posts, and a
few short discussions on the gpg-devel mailing list.
* [A blog entry](http://www.df7cb.de/blog/2007/openpgp-dns.html) that seems to
have things mostly right.
* [GPG Mailing List Discussion](http://lists.gnupg.org/pipermail/gnupg-users/2006-April/028314.html)
which seems to date towhen these features were first added.
* [My own thread](http://www.mail-archive.com/gnupg-users@gnupg.org/msg12336.html)
on the gnupg-users mailing list that led upto this doc.
* [A slideshow of a talk given on PKA](ftp://ftp.g10code.com/people/werner/talks/pka-intro.ps.gz)
(really the only doc I couldfind with regard to PKA). Note that this is a
postscript doc, for reasons I cannot fathom.
### RFCs
* [RFC 3597](http://www.faqs.org/rfcs/rfc3597.html) defines the odd format of
the records that make-dns-cert generates, if itconfuses you.
* [RFC 2538](http://www.faqs.org/rfcs/rfc2538.html), which was superseded by
[RFC4398](http://www.faqs.org/rfcs/rfc4398.html), defines the format for a
CERT record.
## Todo
* At least one GPG enthusiast has suggested to me that any tools I write to
handle keys should simply be able to insert themusing nsupdate. I don't
disagree, but there's a complicated metric there as some of these require
manipulation of a site'smain zone, or at the very least, many subzones. In
doing this I'd also like to find out a bit about how to do nsupdate
withsig(0) and KEY records, which with the right policies would mean I could
do this without touching named.conf. That may be the subject of a whole
other howto.
* (Done) I need to get the shell script cleaned up a bit more, and generate
proper docs, and start tracking it with version control.
* I should probably get the gumption up to formally license all this stuff.
For right now, I declare it under the
[ISCLicense](http://en.wikipedia.org/wiki/ISC_license).
* I'd like to track down the full list of supported URI types for PKA/IPGP
CERT records. There doesn't seem to be a defined standard for it.
## Epilogue
### About the author
Dan Mahoney is a Systems Admin in the Bay Area, California. In his spare time
he enjoys thinking for those brief fleeting moments what he would do if he had
more free time. Keyid 624BB249, or email address danm@prime.gushi.org.
### About this Document
This document was written in [gnu nano](http://nano-editor.org), and HTML was
generated using [Markdown](http://daringfireball.net/projects/markdown).
Markdown rocks.
Originally published on my livejournal at
[http://gushi.livejournal.com/524199.html](http://gushi.livejournal.com/524199.html),
its main home is at
[http://www.gushi.org/make-dns-cert/HOWTO.html](http://www.gushi.org/make-dns-cert/HOWTO.html),
which is where later versions will be published.
Free to use, comments to the above email address are welcome.

133
docs/OpenPGP-keys-in-DNS.md 100644
View File

@ -0,0 +1,133 @@
[Christoph Berg's Blog](../index.html)/
[2007](../2007.html)/
</span>
<span class="title">
OpenPGP keys in DNS
</span>
</span>
</div>
<div class="actions">
* [RecentChanges](../recentchanges.html)
* [History](http://svn.df7cb.de/viewcvs.cgi/trunk/2007/openpgp-dns.mdwn?root=blog&view=log)
</div>
</div>
<div id="pagebody">
<div id="content">
The latest addition to the mutt CVS tree is PKA support via gpgme. While trying
to figure out how that works in mutt (I haven't yet...) I configured my DNS
server for PKA and CERT records.
## PKA
PKA (public key association) puts a pointer where to obtain a key into a TXT
record. At the same time that can be used to verify that a key belongs to a
mail address. The documentation is at the
[g10code website](http://www.g10code.de/docs/pka-intro.de.pdf)
(only in German so far). I put the following into the df7cb.de zone:
<p>
cb._pka IN TXT "v=pka1;fpr=D224C8B07E63A6946DA32E07C5AF774A58510B5A;uri=finger:cb@df7cb.de"
<pre>
$ host -t TXT cb._pka.df7cb.de
cb._pka.df7cb.de descriptive text "v=pka1\;fpr=D224C8B07E63A6946DA32E07C5AF774A58510B5A\;uri=finger:cb@df7cb.de"
</pre>
Now gpg can be told to use PKA to find the key:
<pre>
$ echo foo | gpg --auto-key-locate pka --recipient cb@df7cb.de --encrypt -a
gpg: no keyserver known (use option --keyserver)
gpg: requesting key 58510B5A from finger:cb@df7cb.de
gpg: key 58510B5A: public key "Christoph Berg " imported
gpg: Total number processed: 1
gpg: imported: 1
gpg: automatically retrieved `cb@df7cb.de' via PKA
</pre>
## CERT
CERT records work similarly. Records are generated by make-dns-cert (from the
tools directory in the gnupg source). cb.gpg is a stripped-down gpg keyring
(created with pgp-clean -s and converting from .asc to .gpg).
<pre>
$ ./make-dns-cert -f D224C8B07E63A6946DA32E07C5AF774A58510B5A -n cb
cb TYPE37 \# 26 0006 0000 00 14 D224C8B07E63A6946DA32E07C5AF774A58510B5A
$ ./make-dns-cert -k cb.gpg -n cb
cb TYPE37 \# 1338 0003 0000 00 9901A20440 [...] 509C96D4BFF17B7
</pre>
With a new bind and host (backports.org!) the format looks a bit nicer, that's
also what I copied into the zone file:
<pre>
$ host -t CERT cb.df7cb.de
;; Truncated, retrying in TCP mode.
cb.df7cb.de has CERT record PGP 0 0 mQGiBECBGdAR [...] UDlCcltS/8Xtw==
cb.df7cb.de has CERT record 6 0 0 FNIkyLB+Y6aUbaMuB8Wvd0pYUQta
</pre>
Again, gpg can be told to use that:
<pre>
$ echo foo | gpg --auto-key-locate cert --recipient cb@df7cb.de --encrypt -a
gpg: key 58510B5A: public key "Christoph Berg " imported
gpg: Total number processed: 1
gpg: imported: 1
gpg: automatically retrieved `cb@df7cb.de' via DNS CERT
</pre>
Thanks to weasel for some hints on using CERT.
## SSHFP
I'm also mentioning SSHFP records here since it fits in the topic - I have been
using them for some months now:
<pre>
$ host -t SSHFP tesla.df7cb.de
tesla.df7cb.de has SSHFP record 1 1 EE49B803541293656C33B86ECD781BD8F1D78AB5
tesla.df7cb.de has SSHFP record 2 1 3E82FB5EE8AA0205305F0D0186F94D6FB3E0E744
$ ssh -o 'VerifyHostKeyDNS yes' tesla.df7cb.de
The authenticity of host 'tesla.df7cb.de (88.198.227.218)' can't be established.
RSA key fingerprint is 5a:c9:38:ca:c0:2b:11:c1:c8:fb:f1:ad:73:a1:9c:8b.
Matching host key fingerprint found in DNS.
Are you sure you want to continue connecting (yes/no)?
</pre>
The records are generated with ssh-keygen -r.
</div>
</div>
<div id="footer" class="pagefooter">
<div id="pageinfo">
<div class="tags">
Tags:
[debian](../tag/debian.html)
</div>
<div class="pagedate">
Last edited <span class="date">Do 17 Feb 2011 13:21:52 CET</span>
<!-- Created <span class="date">Do 01 Mär 2007 20:01:27 CET</span> -->
</div>
</div>
<!-- from Christoph Berg's Blog -->

208
docs/gpg-migrate.txt 100644
View File

@ -0,0 +1,208 @@
# taken from http://atom.smasher.org/gpg/gpg-migrate.txt on 8 Aug 2013
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
HOW TO MIGRATE A (SUB)KEY INTO A NEW KEY
this document [is intended to] explain how to migrate a key or subkey from
one OpenPGP key into another OpenPGP key.
here i'll walk you through the steps of migrating my old primary signing
key (3D7D41E3) into my new key (D9F57808). the process of migrating
encryption (and signing) subkeys is nearly identical. for the adventurous,
you can even migrate several keys at once using this method (i recommend
going through it once or twice with only one key).
============================================================================
doing this requires:
1) basic knowledge of a *nix command line and how to use it
2) advanced knowledge of gpg and how to use it
3) common sense (BACKUP YOUR DATA!!)
please note:
* this works for me. that does not necessarily mean that it will
work for you
* if you screw something up, it's *YOUR* problem, not mine
* this was tested with GnuPG 1.2.4, and written on or about
12 May 2004
* updates, if there are any, will probably be noted above
* comments and suggestions about this tutorial should be sent to:
<atom {at} smasher.org>
* questions about gpg should be sent to the gnupg-users mailing
list: http://lists.gnupg.org/mailman/listinfo/gnupg-users
============================================================================
* old key: 3EBE 2810 30AE 601D 54B2 4A90 9C28 0BBF 3D7D 41E3
pub 1024D/3D7D41E3 2003-10-04 Atom Smasher <atom@suspicious.org>
uid Atom Smasher <atom@smasher.org>
sub 2048g/1E88BF71 2003-10-04 [expires: 2005-01-26]
================
* new key: 762A 3B98 A3C3 96C9 C6B7 582A B88D 52E4 D9F5 7808
pub 4096R/D9F57808 2004-05-11 Atom Smasher <atom@smasher.org>
uid Atom Smasher <atom@suspicious.org>
sub 1024D/3D7D41E3 2003-10-04 [expires: 2006-02-13]
sub 2048g/1E88BF71 2003-10-04 [expires: 2006-01-26]
===================================
backup the new keys:
$ gpg --export D9F57808 > D9F57808_original.txt
$ gpg --export-secret-key D9F57808 > D9F57808_original_secret.txt
and the old keys:
$ gpg --export 3D7D41E3 > 3D7D41E3_original.txt
$ gpg --export-secret-key 3D7D41E3 > 3D7D41E3_original_secret.txt
break the old secret key into pieces:
$ gpg --export-secret-key 3D7D41E3 | gpgsplit -vp OLD_SEC
===================================
in this case, the old primary key needs to be converted into a subkey.
pgpdump shows that it's a primary key:
$ pgpdump OLD_SEC000001-005.secret_key
Old: Secret Key Packet(tag 5)
<<snip>>
only do this if you're converting a PRIMARY KEY into a SUBKEY: open that
file in a hex editor and refer to RFC2440 4.2 & 4.3. i recommend setting
the hex editor into a binary display. in this example the first byte is
"10010101" and it needs to be changed to "10011101". the change can be
confirmed with pgpdump:
$ pgpdump OLD_SEC000001-005.secret_key
Old: Secret Subkey Packet(tag 7)
<<snip>>
we've now converted the old primary key into a subkey. if you're moving a
subkey from one key to another, you don't have to do that.
===================================
use "edit-key" and add a subkey of the same type (DSA, for this example)
and size to the new key. this subkey will be discarded, but we need to
generate it for now: this seems to be the quickest way to generate a
keybinding signature with the correct features.
exit from "edit-key" and save.
===================================
split the current version of the new public key:
$ gpg --export D9F57808 | gpgsplit -vp TEMP_KEY1
the only part we need from that split is the binding signature that was
just generated.
delete (from the keyring) both the private and public copies of the new
key:
$ gpg --delete-secret-key D9F57808
$ gpg --delete-key D9F57808
also delete (from the keyring) both the private and public copies of the
old key.
because the subkey that we're adding to the new key does not correspond to
the subkey binding signature that was created for it, gpg will not allow
the key to be imported. the way around that is to "force feed" the key into
the keyring, bypassing the normal sanity checks. once it's in the keyring
we can make it all work.
* note: the actual file names that you're using may differ somewhat from
mine. when in doubt (or rather, when you're not sure), use pgpdump to
examine the contents of files.
~import~ (directly into the secret keyring) the original copy of the new
key (D9F57808_original_secret.txt), the edited copy of the old primary key
(now a subkey, OLD_SEC000001-005.secret_key) and the binding signature of
the subkey that we just generated (TEMP_KEY1000007-002.sig):
$ cat D9F57808_original_secret.txt \
OLD_SEC000001-005.secret_key \
TEMP_KEY1000007-002.sig >> ~/.gnupg/secring.gpg
~import~ (directly into the public keyring) a public key by adding
"| gpgsplit --no-split --secret-to-public" to the above command like this:
$ cat D9F57808_original_secret.txt \
OLD_SEC000001-005.secret_key \
TEMP_KEY1000007-002.sig \
| gpgsplit --no-split --secret-to-public >> ~/.gnupg/pubring.gpg
now we have to make a *valid* keybinding signature for the subkey that we
just added. use "edit-key", select the newly added subkey, and reset it's
expiration date. that will generate a valid keybinding signature.
while in "edit-key", reset the password. otherwise you may inadvertently
create a key with multiple passwords, as described here -
http://atom.smasher.org/gpg/gpg-passwords.txt
exit from "edit-key" and save.
=========================
this last part makes no sense to me (but it doesn't seem to work
otherwise).
back up (export) the latest version of the public and private keys:
$ gpg --export-secret-key D9F57808 > new-key.sec
$ gpg --export D9F57808 > new-key.pub
delete (from the keyring) the private and public key:
$ gpg --delete-secret-key D9F57808
$ gpg --delete-key D9F57808
also, delete (from the keyring) all copies of the old (sub)key that was
just added to the new key.
import the new public and private keys:
$ gpg --import new-key.sec new-key.pub
=========================
before you publish your new key:
* make sure the key is "ultimately trusted". deleting and importing will
have removed it from the trust db. since you own the key, ultimate trust
seems reasonable.
* check all expiration dates and preferences. some of these operations may
have changed your expiration dates and preferences; reset as necessary.
* test out all key components for creating and verifying signatures, and
encryption/decryption. use the bang (!) to force each (sub)key:
create & verify signatures:
$ date | gpg -u 'D9F57808!' --clearsign | gpg -v --verify
$ date | gpg -u '3D7D41E3!' --clearsign | gpg -v --verify
encrypt/decrypt:
$ date | gpg -ear 'D9F57808!' | gpg -v --decrypt
$ date | gpg -ear '1E88BF71!' | gpg -v --decrypt
* after testing out the keys locally, send your new public key to one or
two people and test all key components (sending signed/encrypted messages
to each other using all key components). make sure that they first delete
(from their keyrings) your old key! and make sure that they understand that
the key should NOT be circulated until all functions are verified to be
working!
* when putting the new key into circulation, it's probably a good idea to
expire/revoke the old key. include a revocation comment that specifies the
new key ID and instructions to delete the old key from the keyring.
* note on key revocation: according to the OpenPGP standards a revocation
generated by a sub key will be ignored, unless that subkey has been
designated (by the primary key) as a revocation key. GnuPG seems to behave
correctly, but some versions of PGP(tm) may not. if someone is claiming
that your new key is revoked, have then remove all of your old and current
keys from their keyring: then re-import your current key(s).
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (FreeBSD)
iD8DBQFApwlpnCgLvz19QeMRAhrpAJ4rhLrmVDjABh8CpPdTZ5jNMi7LsgCgp35S
6qcUe4csx1p5AE2rAsvDi9c=
=y7bA
-----END PGP SIGNATURE-----

71
gnupg/_meta.py
View File

@ -90,7 +90,6 @@ class GPGBase(object):
"""Base class for property storage and to control process initialisation."""
__metaclass__ = GPGMeta
_decode_errors = 'strict'
_result_map = { 'crypt': _parsers.Crypt,
'delete': _parsers.DeleteResult,
@ -140,6 +139,12 @@ class GPGBase(object):
log.error("GPGBase.__init__(): %s" % ae.message)
raise RuntimeError(ae.message)
else:
if verbose is True:
# The caller wants logging, but we need a valid --debug-level
# for gpg. Default to "basic", and warn about the ambiguity.
# (garrettr)
verbose = "basic"
log.warning('GPG(verbose=True) is ambiguous, defaulting to "basic" logging')
self.verbose = verbose
self.use_agent = use_agent
@ -498,19 +503,24 @@ class GPGBase(object):
break
lines.append(line)
line = line.rstrip()
if line[0:9] == '[GNUPG:] ':
# Chop off the prefix
line = line[9:]
log.status("%s" % line)
L = line.split(None, 1)
keyword = L[0]
if len(L) > 1:
value = L[1]
else:
value = ""
if line.startswith('[GNUPG:]'):
line = _util._deprefix(line, '[GNUPG:] ', log.status)
keyword, value = _util._separate_keyword(line)
result._handle_status(keyword, value)
elif line[0:5] == 'gpg: ':
log.warn("%s" % line)
elif line.startswith('gpg:'):
line = _util._deprefix(line, 'gpg: ')
keyword, value = _util._separate_keyword(line)
# Log gpg's userland messages at our own levels:
if keyword.upper().startswith("WARNING"):
log.warn("%s" % value)
elif keyword.upper().startswith("FATAL"):
log.critical("%s" % value)
# Handle the gpg2 error where a missing trustdb.gpg is,
# for some stupid reason, considered fatal:
if value.find("trustdb.gpg") and value.find("No such file"):
result._handle_status('NEED_TRUSTDB', '')
else:
if self.verbose:
log.info("%s" % line)
@ -519,19 +529,29 @@ class GPGBase(object):
result.stderr = ''.join(lines)
def _read_data(self, stream, result):
"""Read the contents of the file from GPG's stdout."""
"""Incrementally read from ``stream`` and store read data.
All data gathered from calling ``stream.read()`` will be concatenated
and stored as ``result.data``.
:param stream: An open file-like object to read() from.
:param result: An instance of one of the result parsing classes from
:attr:`GPGBase._result_mapping`.
"""
chunks = []
log.debug("Reading data from stream %r..." % stream.__repr__())
while True:
data = stream.read(1024)
if len(data) == 0:
break
log.debug("read from stdout: %r" % data[:256])
chunks.append(data)
if _util._py3k:
# Join using b'' or '', as appropriate
result.data = type(data)().join(chunks)
else:
result.data = ''.join(chunks)
log.debug("Read %4d bytes" % len(data))
# Join using b'' or '', as appropriate
result.data = type(data)().join(chunks)
log.debug("Finishing reading from stream %r..." % stream.__repr__())
log.debug("Read %4d bytes total" % len(result.data))
def _collect_output(self, process, result, writer=None, stdin=None):
"""Drain the subprocesses output streams, writing the collected output
@ -600,7 +620,8 @@ class GPGBase(object):
return result
def _sign_file(self, file, default_key=None, passphrase=None,
clearsign=True, detach=False, binary=False):
clearsign=True, detach=False, binary=False,
digest_algo='SHA512'):
"""Create a signature for a file.
:param file: The file stream (i.e. it's already been open()'d) to sign.
@ -609,6 +630,10 @@ class GPGBase(object):
:param bool clearsign: If True, create a cleartext signature.
:param bool detach: If True, create a detached signature.
:param bool binary: If True, do not ascii armour the output.
:param str digest_algo: The hash digest to use. Again, to see which
hashes your GnuPG is capable of using, do:
``$ gpg --with-colons --list-config digestname``.
The default, if unspecified, is ``'SHA512'``.
"""
log.debug("_sign_file():")
if binary:
@ -629,6 +654,8 @@ class GPGBase(object):
if default_key:
args.append(str("--default-key %s" % default_key))
args.append(str("--digest-algo %s" % digest_algo))
## We could use _handle_io here except for the fact that if the
## passphrase is bad, gpg bails and you can't write the message.
result = self._result_map['sign'](self)
@ -797,5 +824,5 @@ class GPGBase(object):
% (data, type(data)))
self._handle_io(args, data, result,
passphrase=passphrase, binary=True)
log.debug('GPG.encrypt_file(): Result: %r', result.data)
log.debug("\n%s" % result.data)
return result

9
gnupg/_parsers.py
View File

@ -525,11 +525,13 @@ def _get_options_group(group=None):
'--default-recipient-self',
'--detach-sign',
'--export',
'--export-ownertrust',
'--export-secret-keys',
'--export-secret-subkeys',
'--fingerprint',
'--fixed-list-mode',
'--gen-key',
'--import-ownertrust',
'--list-config',
'--list-key',
'--list-keys',
@ -747,6 +749,11 @@ def _get_all_gnupg_options():
--merge-only --with-key-data
--min-cert-level --yes
""").split()
# These are extra options which only exist for GnuPG>=2.0.0
three_hundred_eighteen.append('--export-ownertrust')
three_hundred_eighteen.append('--import-ownertrust')
gnupg_options = frozenset(three_hundred_eighteen)
return gnupg_options
@ -1283,6 +1290,8 @@ class Crypt(Verify):
"MISSING_PASSPHRASE", "DECRYPTION_FAILED",
"KEY_NOT_CREATED"):
self.status = key.replace("_", " ").lower()
elif key == "NEED_TRUSTDB":
self._gpg._create_trustdb()
elif key == "NEED_PASSPHRASE_SYM":
self.status = 'need symmetric passphrase'
elif key == "BEGIN_DECRYPTION":

102
gnupg/_trust.py 100644
View File

@ -0,0 +1,102 @@
# -*- coding: utf-8 -*-
#
# This file is part of python-gnupg, a Python interface to GnuPG.
# Copyright © 2013 Isis Lovecruft, <isis@leap.se> 0xA3ADB67A2CDB8B35
# © 2013 Andrej B.
# © 2013 LEAP Encryption Access Project
# © 2008-2012 Vinay Sajip
# © 2005 Steve Traugott
# © 2004 A.M. Kuchling
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the Free
# Software Foundation, either version 3 of the License, or (at your option)
# any later version.
#
# This program is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
# FITNESS FOR A PARTICULAR PURPOSE. See the included LICENSE file for details.
'''trust.py
-----------
Functions for handling trustdb and trust calculations.
The functions within this module take an instance of :class:`gnupg.GPGBase` or
a suitable subclass as their first argument.
'''
from __future__ import absolute_import
import os
from . import _util
from ._util import log
def _create_trustdb(cls):
"""Create the trustdb file in our homedir, if it doesn't exist."""
trustdb = os.path.join(cls.homedir, 'trustdb.gpg')
if not os.path.isfile(trustdb):
log.info("GnuPG complained that your trustdb file was missing. %s"
% "This is likely due to changing to a new homedir.")
log.info("Creating trustdb.gpg file in your GnuPG homedir.")
cls.fix_trustdb(trustdb)
def export_ownertrust(cls, trustdb=None):
"""Export ownertrust to a trustdb file.
If there is already a file named 'trustdb.gpg' in the current GnuPG
homedir, it will be renamed to 'trustdb.gpg.bak'.
:param string trustdb: The path to the trustdb.gpg file. If not given,
defaults to 'trustdb.gpg' in the current GnuPG homedir.
"""
if trustdb is None:
trustdb = os.path.join(cls.homedir, 'trustdb.gpg')
try:
os.rename(trustdb, trustdb + '.bak')
except (OSError, IOError) as err:
log.debug(err.message)
export_proc = cls._open_subprocess('--export-ownertrust')
tdb = open(trustdb, 'wb')
_util._threaded_copy_data(export_proc.stdout, tdb)
def import_ownertrust(self, trustdb=None):
"""Import ownertrust from a trustdb file.
:param string trustdb: The path to the trustdb.gpg file. If not given,
defaults to 'trustdb.gpg' in the current GnuPG homedir.
"""
if trustdb is None:
trustdb = os.path.join(cls.homedir, 'trustdb.gpg')
import_proc = cls._open_subprocess('--import-ownertrust')
tdb = open(trustdb, 'rb')
_util._threaded_copy_data(tdb, import_proc.stdin)
def fix_trustdb(cls, trustdb=None):
"""Attempt to repair a broken trustdb.gpg file.
GnuPG>=2.0.x has this magical-seeming flag: '--fix-trustdb'. You'd think
it would fix the the trustdb. Hah! It doesn't. Here's what it does
instead:
(python-gnupg)∃!isisⒶwintermute:(testing/digest-algo *$=)~/code/python-gnupg ∴ gpg2 --fix-trustdb
gpg: You may try to re-create the trustdb using the commands:
gpg: cd ~/.gnupg
gpg: gpg2 --export-ownertrust > otrust.tmp
gpg: rm trustdb.gpg
gpg: gpg2 --import-ownertrust < otrust.tmp
gpg: If that does not work, please consult the manual
Brilliant piece of software engineering right there.
:param string trustdb: The path to the trustdb.gpg file. If not given,
defaults to 'trustdb.gpg' in the current GnuPG homedir.
"""
if trustdb is None:
trustdb = os.path.join(cls.homedir, 'trustdb.gpg')
export_proc = cls._open_subprocess('--export-ownertrust')
import_proc = cls._open_subprocess('--import-ownertrust')
_util._threaded_copy_data(export_proc.stdout, import_proc.stdin)

73
gnupg/_util.py
View File

@ -7,12 +7,12 @@
# © 2008-2012 Vinay Sajip
# © 2005 Steve Traugott
# © 2004 A.M. Kuchling
#
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the Free
# Software Foundation, either version 3 of the License, or (at your option)
# any later version.
#
#
# This program is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
# FITNESS FOR A PARTICULAR PURPOSE. See the included LICENSE file for details.
@ -219,6 +219,34 @@ def create_uid_email(username=None, hostname=None):
return uid
def _deprefix(line, prefix, callback=None):
"""Remove the prefix string from the beginning of line, if it exists.
:param string line: A line, such as one output by GnuPG's status-fd.
:param string prefix: A substring to remove from the beginning of
``line``. Case insensitive.
:type callback: callable
:param callback: Function to call if the prefix is found. The signature to
callback will be only one argument, the ``line`` without the ``prefix``, i.e.
``callback(line)``.
:rtype: string
:returns: If the prefix was found, the ``line`` without the prefix is
returned. Otherwise, the original ``line`` is returned.
"""
try:
assert line.upper().startswith(u''.join(prefix).upper())
except AssertionError:
log.debug("Line doesn't start with prefix '%s':\n%s" % (prefix, line))
return line
else:
newline = line[len(prefix):]
if callback is not None:
try:
callback(newline)
except Exception as exc:
log.exception(exc)
return newline
def _find_binary(binary=None):
"""Find the absolute path to the GnuPG binary.
@ -231,23 +259,35 @@ def _find_binary(binary=None):
:returns: The absolute path to the GnuPG binary to use, if no exceptions
occur.
"""
gpg_binary = None
found = None
if binary is not None:
if not os.path.isabs(binary):
try: binary = _which(binary)[0]
try:
found = _which(binary)
log.debug("Found potential binary paths: %s"
% '\n'.join([path for path in found]))
found = found[0]
except IndexError as ie:
log.error(ie.message)
if binary is None:
try: binary = _which('gpg')[0]
except IndexError: raise RuntimeError("GnuPG is not installed!")
log.info("Could not determine absolute path of binary: '%s'"
% binary)
if found is None:
try: found = _which('gpg')[0]
except IndexError as ie:
log.error("Could not find binary for 'gpg'.")
try: found = _which('gpg2')[0]
except IndexError as ie:
log.error("Could not find binary for 'gpg2'.")
if found is None:
raise RuntimeError("GnuPG is not installed!")
try:
assert os.path.isabs(binary), "Path to gpg binary not absolute"
assert not os.path.islink(binary), "Path to gpg binary is symlink"
assert os.access(binary, os.X_OK), "Lacking +x perms for gpg binary"
assert os.path.isabs(found), "Path to gpg binary not absolute"
assert not os.path.islink(found), "Path to gpg binary is symlink"
assert os.access(found, os.X_OK), "Lacking +x perms for gpg binary"
except (AssertionError, AttributeError) as ae:
log.error(ae.message)
else:
return binary
return found
def _has_readwrite(path):
"""
@ -406,6 +446,15 @@ def _now():
"""Get a timestamp for right now, formatted according to ISO 8601."""
return datetime.isoformat(datetime.now())
def _separate_keyword(line):
"""Split the line, and return (first_word, the_rest)."""
try:
first, rest = line.split(None, 1)
except ValueError:
first = line.strip()
rest = ''
return first, rest
def _threaded_copy_data(instream, outstream):
"""Copy data from one stream to another in a separate thread.

59
gnupg/gnupg.py
View File

@ -31,6 +31,7 @@ from __future__ import absolute_import
from codecs import open as open
import encodings
import functools
import os
import re
import textwrap
@ -43,6 +44,7 @@ except ImportError:
## see PEP-328 http://docs.python.org/2.5/whatsnew/pep-328.html
from . import _parsers
from . import _util
from . import _trust
from ._meta import GPGBase
from ._parsers import _fix_unsafe
from ._util import _is_list_or_tuple
@ -76,6 +78,13 @@ class GPG(GPGBase):
and private keyrings. Default is whatever GnuPG
defaults to.
:param str,int,bool verbose: String or numeric value to pass to gpg's
``--debug-level`` option. See the gpg man
page for the list of valid options. If
False, debug output is not generated by
the gpg binary. If True, defaults to
``--debug-level basic.``
:param str keyring: Name of keyring file containing public key data, if
unspecified, defaults to 'pubring.gpg' in the
``homedir`` directory.
@ -152,6 +161,52 @@ class GPG(GPGBase):
self.binary_version = version_line.split('\n')[0]
log.debug("Using GnuPG version %s" % self.binary_version)
if _util._is_gpg2:
# Make GnuPG>=2.0.0-only methods public:
self.fix_trustdb = self._fix_trustdb
self.import_ownertrust = self._import_ownertrust
self.export_ownertrust = self._export_ownertrust
# Make sure that the trustdb exists, or else GnuPG will exit with
# a fatal error (at least it does with GnuPG>=2.0.0):
self._create_trustdb()
@functools.wraps(_trust._create_trustdb)
def _create_trustdb(self):
if self.is_gpg2():
_trust._create_trustdb(self)
else:
log.info("Creating the trustdb is only available with GnuPG>=2.x")
@functools.wraps(_trust.fix_trustdb)
def _fix_trustdb(self, trustdb=None):
if self.is_gpg2():
_trust.fix_trustdb(self)
else:
log.info("Fixing the trustdb is only available with GnuPG>=2.x")
@functools.wraps(_trust.import_ownertrust)
def _import_ownertrust(self, trustdb=None):
if self.is_gpg2():
_trust.import_ownertrust(self)
else:
log.info("Importing ownertrust is only available with GnuPG>=2.x")
@functools.wraps(_trust.export_ownertrust)
def _export_ownertrust(self, trustdb=None):
if self.is_gpg2():
_trust.export_ownertrust(self)
else:
log.info("Exporting ownertrust is only available with GnuPG>=2.x")
def is_gpg1(self):
"""Returns true if using GnuPG <= 1.x."""
return _util._is_gpg1(self.binary_version)
def is_gpg2(self):
"""Returns true if using GnuPG >= 2.x."""
return _util._is_gpg2(self.binary_version)
def sign(self, data, **kwargs):
"""Create a signature for a message string or file.
@ -179,6 +234,10 @@ class GPG(GPGBase):
:param bool clearsign: If True, create a cleartext signature.
:param bool detach: If True, create a detached signature.
:param bool binary: If True, do not ascii armour the output.
:param str digest_algo: The hash digest to use. Again, to see which
hashes your GnuPG is capable of using, do:
``$ gpg --with-colons --list-config digestname``.
The default, if unspecified, is ``'SHA512'``.
"""
if 'default_key' in kwargs.items():
log.info("Signing message '%r' with keyid: %s"

110
gnupg/test/test_gnupg.py
View File

@ -152,24 +152,9 @@ def compare_keys(k1, k2):
return k1 != k2
class ResultStringIO(io.StringIO):
def __init__(self, init_string):
super(ResultStringIO, self).__init__(init_string)
def write(self, data):
super(ResultStringIO, self).write(unicode(data))
class GPGTestCase(unittest.TestCase):
""":class:`unittest.TestCase <TestCase>`s for python-gnupg."""
@classmethod
def setUpClass(cls):
"""Setup ``GPGTestCase`` and runtime environment for tests.
This function must be called manually.
"""
pass
def setUp(self):
"""This method is called once per self.test_* method."""
print("%s%s%s" % (os.linesep, str("=" * 70), os.linesep))
@ -306,8 +291,9 @@ class GPGTestCase(unittest.TestCase):
def test_make_args_drop_protected_options(self):
"""Test that unsupported gpg options are dropped."""
self.gpg.options = ['--tyrannosaurus-rex', '--stegosaurus']
gpg_binary_path = _util._find_binary('gpg')
cmd = self.gpg._make_args(None, False)
expected = ['/usr/bin/gpg',
expected = [gpg_binary_path,
'--no-options --no-emit-version --no-tty --status-fd 2',
'--homedir "%s"' % self.homedir,
'--no-default-keyring --keyring %s' % self.keyring,
@ -340,17 +326,30 @@ class GPGTestCase(unittest.TestCase):
def test_copy_data_bytesio(self):
"""Test that _copy_data() is able to duplicate byte streams."""
message = "This is a BytesIO string string in memory."
message = "This is a BytesIO string."
instream = io.BytesIO(message)
self.assertEqual(unicode(message), instream.getvalue())
outstream = ResultStringIO(u'result:')
copied = outstream
out_filename = 'test-copy-data-bytesio'
# Create the test file:
outfile = os.path.join(os.getcwdu(), out_filename)
outstream = open(outfile, 'w+')
# _copy_data() will close both file descriptors
_util._copy_data(instream, outstream)
self.assertTrue(outstream.readable())
self.assertTrue(outstream.closed)
self.assertFalse(instream.closed)
self.assertTrue(copied.closed)
#self.assertEqual(instream.getvalue()[6:], outstream.getvalue())
self.assertTrue(os.path.isfile(outfile))
with open(outfile) as out:
out.flush()
out.seek(0)
output = out.read()
self.assertEqual(message, output)
os.remove(outfile)
def generate_key_input(self, real_name, email_domain, key_length=None,
key_type=None, subkey_type=None, passphrase=None):
@ -766,58 +765,59 @@ authentication."""
def test_encryption_multi_recipient(self):
"""Test encrypting a message for multiple recipients"""
self.gpg.homedir = _util._here
ian = { 'name_real': 'Ian Goldberg',
'name_email': 'gold@stein',
'key_type': 'RSA',
'key_length': 2048,
'key_usage': '',
'subkey_type': 'RSA',
'subkey_length': 2048,
'subkey_usage': 'encrypt,sign',
'passphrase': 'victorygin' }
riggio = { 'name_real': 'Riggio',
'name_email': 'ri@gg.io',
'key_type': 'RSA',
'key_length': 2048,
'key_usage': '',
'subkey_type': 'RSA',
'subkey_length': 2048,
'subkey_usage': 'encrypt,sign',
'passphrase': 'victorygin' }
## when we don't specify the subkey lengths and the keylength
## gets set automatically in gen_key_input(), gpg complains:
##
## gpg: keysize invalid; using 1024 bits
##
kat = { 'name_real': 'Kat Hannah',
'name_email': 'kat@pics',
'key_type': 'RSA',
'key_length': 2048,
'key_usage': '',
'subkey_type': 'RSA',
'subkey_length': 2048,
'subkey_usage': 'encrypt,sign',
'passphrase': 'overalls' }
sicari = { 'name_real': 'Sicari',
'name_email': 'si@ca.ri',
'key_type': 'RSA',
'key_length': 2048,
'key_usage': '',
'subkey_type': 'RSA',
'subkey_length': 2048,
'subkey_usage': 'encrypt,sign',
'passphrase': 'overalls' }
ian_input = self.gpg.gen_key_input(separate_keyring=True, **ian)
riggio_input = self.gpg.gen_key_input(separate_keyring=True, **riggio)
log.info("Key stored in separate keyring: %s" % self.gpg.temp_keyring)
ian_key = self.gpg.gen_key(ian_input)
ian_fpr = str(ian_key.fingerprint)
self.gpg.options = ['--keyring {}'.format(ian_key.keyring)]
riggio = self.gpg.gen_key(riggio_input)
self.gpg.options = ['--keyring {}'.format(riggio.keyring)]
riggio_key = self.gpg.export_keys(riggio.fingerprint)
self.gpg.import_keys(riggio_key)
kat_input = self.gpg.gen_key_input(separate_keyring=True, **kat)
sicari_input = self.gpg.gen_key_input(separate_keyring=True, **sicari)
log.info("Key stored in separate keyring: %s" % self.gpg.temp_keyring)
kat_key = self.gpg.gen_key(kat_input)
kat_fpr = str(kat_key.fingerprint)
self.gpg.options.append('--keyring {}'.format(kat_key.keyring))
self.gpg.import_keys(kat_key.data)
sicari = self.gpg.gen_key(sicari_input)
self.gpg.options.append('--keyring {}'.format(sicari.keyring))
sicari_key = self.gpg.export_keys(sicari.fingerprint)
self.gpg.import_keys(sicari_key)
message = """
In 2010 Riggio and Sicari presented a practical application of homomorphic
encryption to a hybrid wireless sensor/mesh network. The system enables
transparent multi-hop wireless backhauls that are able to perform statistical
analysis of different kinds of data (temperature, humidity, etc.) coming from
analysis of different kinds of data (temperature, humidity, etc.) coming from
a WSN while ensuring both end-to-end encryption and hop-by-hop
authentication."""
log.debug("kat_fpr type: %s" % type(kat_fpr))
log.debug("ian_fpr type: %s" % type(ian_fpr))
if self.gpg.is_gpg2:
self.gpg.fix_trustdb()
encrypted = str(self.gpg.encrypt(message, ian_fpr, kat_fpr))
encrypted = str(self.gpg.encrypt(message,
riggio.fingerprint,
sicari.fingerprint))
log.debug("Plaintext: %s" % message)
log.debug("Ciphertext: %s" % encrypted)

16
setup.py
View File

@ -18,22 +18,10 @@
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
# FITNESS FOR A PARTICULAR PURPOSE. See the included LICENSE file for details.
#______________________________________________________________________________
#
# NOTE: setuptools is currently (as of 27 May 2013) being merged back into its
# parent project, distribute. By using the included distribute_setup.py
# script, we make sure that we have a recent version of setuptools/distribute,
# which is the *only* Python packaging framework compatible at this point with
# both Python>=2.4 and Python3.x.
#
from __future__ import absolute_import
from __future__ import print_function
## Upgrade setuptools to a version which supports Python 2 and 3
#os.system('python ./distribute_setup.py')
## Upgrade pip to a version with proper SSL support
#os.system('python ./get-pip.py')
import setuptools
import versioneer
versioneer.versionfile_source = 'gnupg/_version.py'
@ -56,10 +44,6 @@ through file descriptors. Input arguments are strictly checked and sanitised, \
and therefore this module should be safe to use in networked applications \
requiring direct user input. It is intended for use with Python 2.6 or \
greater.
Documentation can be found on readthedocs_.
.. _readthedocs: https://python-gnupg.readthedocs.org/en/latest/
""",
license="GPLv3+",