bb2cf version V: TCH Key is stored in encoded hex strings, should be bytes

[hrz6976]

When I was developing the new python driver, I was surprised to find that bb2cf never worked for me.
getValues won't work given any blob sha as well:

(base) ~ echo 125afaaefe99189eb2cec5aafc470770b79abbd0 | ~/lookup/getValues bb2cf
No 125afaaefe99189eb2cec5aafc470770b79abbd0 in /fast/bb2cfFullV
(base) ~ echo 125afaaefe99189eb2cec5aafc470770b79abbd0 | ~/lookup/getValues obb2cf
125afaaefe99189eb2cec5aafc470770b79abbd0;5664f2622743898c5ca094670f16b4fb8fb4b74f;4071672afedd71d8997f427ee8ffec5dd97a3a1c;src/templates/unitedstates/states/louisiana.ejs

Digging into the issue, I found out that the keys in tokyocabinet hashtables are encoded hex strings:

In [8]: from woc.tch import TCHashDB
   ...: db = TCHashDB('/fast/woc_azure/da3-fast/bb2cfFullV.18.tch'.encode(), ro=True)
   ...: for k in db:
   ...:     print(k, k.hex())
   ...:     break
   ...:
b'125afaaefe99189eb2cec5aafc470770b79abbd0' 31323561666161656665393931383965623263656335616166633437303737306237396162626430

Instead of bytes, as the ones of obb2cf:

In [7]: from woc.tch import TCHashDB
   ...: db = TCHashDB('/fast/woc_azure/da3-fast/obb2cfFullV.18.tch'.encode(), ro=True)
   ...: for k in db:
   ...:     print(k, k.hex())
   ...:     break
   ...:
b'\x12|O\x11\xcfS\x19+\xbeC\x93\xb2\x0c\xbd\xaf\xcfW\xe6<q' 127c4f11cf53192bbe4393b20cbdafcf57e63c71

And it works with the following walkaround:

# TODO: remove bb2cf quirk after fixing tch keys
# bb2cf: keys are stored as hex strings in tch db
if map_name == 'bb2cf':
    key = key.hex().encode('ascii')

Is it intended, or you are planning to fix that?

[audrism]

1. all sha1's are in binary form if they are keys. If they are mixed with strings in values, they may sometimes be in hex
2. 125afaaefe99189eb2cec5aafc470770b79abbd0 indeed does not have a parent, only a child, hence in obb2cf but not in bb2cf
3. I am surprised how bb2cf does not have binary keys:
   for i in {0..31}; do time zcat bb2cfFullV{$i,$((i+32)),$((i+64)),$((i+96))}.s | ~/lookup/h2fbbBinSorted.perl /fast/bb2cfFullV.$i.tch; done
4. recreating bb2cfFullV

[hrz6976]

Okay, I'll remove the quirk for bb2cf in python-woc after the fix.

[audrism]

Finished