Since nilling the pubkey curve will lead to a nil-pointer exception if
the key is later used for signature verification, we make sure to make a
copy before nilling and spewing.
Before this commit, if an invoice encoded multiple `r` fields, we would
decode them as one single route. We fix this by allowing an invoice to
store multiple routes.
In this commit, we fix a logic error in our routine for converting a
uint64 to/from base32. Before this commit, we assumed that the max
number of groups was 12. However, the math.MaxUint64 (1<<64 - 1) can
actually consume more than 12 groups with an extra set of bits. Before
this commit, we would panic when attempting to parse an invoice
generated like so:
* addinvoice --amt 1337000 --expiry 99999999999999999
To fix this issue, we modify our logic to expect at most 13 groups.
Additionally, we've added a new test that would panic before applying
this commit.
Fixes#972.
New tests are added for creating, decoding, and re-encoding
litecoin invoices for both mainnet and testnet, as well as a test
that expects an error when the active network mismatches the
invoice.
This change fixes a bug when an invoice is decoded for a network
whose bech32 segwit prefix is longer than 2 characters. The length
of the Bech32HRPSegwit network parameter is used to determine
where in the human-readable portion of the invoice the amount
begins, rather than assuming it begins after the first four
characters.
Decode() now throws an error when the encoded invoice does
not match the active network.
Changes the minimum hrp length check to >= 3 instead of >= 4.
Also removes a redundant "if ...; err != nil check" that was raising
a warning in invoice.go.
When accessing a value from a byte slice, the value is returned as a
byte, which is just a uint8. When the first byte takes more than 3 bits
of space, shifting 5 bits left results in data loss.
This commit allows parseRoutingInfo to return an error when parsing a
routing info field whose length is not a multiple of 51 bytes, rather
than crash.
This commit refactors parsing each of the tagged fields of an invoice
into their own method. This makes the code easier to read and will allow
us to introduce unit tests for each parsing method.
This commit renames the invoice field Expiry to expiry, and changes
the type from time.Time to time.Duration. Getting the value of the
field will now have to be done using the getter Expiry(), which
will also return the default invoice expiry (3600s) if it is not set
explicitly by the the invoice.
This commit fixes a panic that can result when a zpay32 payment request
that is too short (and possibly invalid) is attempted to be decoded.
To fix this bug, we now simply ensure that that after we decode the
zbase32 encoding, the resulting set of bytes is _exactly_ the length we
expect. A new error has been introduced to handle this case, and a
simple test has been added which ensures proper handling of short
payment requests.
Fixes#127.
This commit fixes a possible panic which can arise within the library
due to poor user-data. We now explicitly check for an empty string,
exiting early if found within the Decode method.
This commit adds a new package “zpay32”: which is used within the
daemon to encode/decode payment requests. A payment request currently
consists of: the public key of the payee, the payment hash to use, and
finally the amount to send over the network. The encoded payment
request consists of the mentioned fields concatenated to each other, a
cc32 checksum is added, then the blob is finally encoded using zbas32.
I call the resulting scheme “zpay32”.
A number of extensions may be explored in future commits including
adding a version byte, adding “hint” routing information,
cryptographically signed receipts and more,
