tool.help: relocate help texts to top of module
This commit is contained in:
parent
3c0c961385
commit
ec15a0b8e8
1 changed files with 48 additions and 48 deletions
|
|
@ -24,6 +24,50 @@ from .. import main_tool
|
|||
|
||||
from .common import tool_cmd_base, fstr
|
||||
|
||||
usage_hdr = """
|
||||
GENERAL USAGE INFORMATION FOR MMGEN-TOOL COMMANDS
|
||||
|
||||
Arguments with only type specified in square brackets are required
|
||||
|
||||
Arguments with both type and default value specified in square brackets are
|
||||
optional and must be specified in the form ‘name=value’
|
||||
|
||||
For more detailed usage information for a particular tool command, type
|
||||
‘mmgen-tool help <command name>’
|
||||
"""
|
||||
|
||||
usage_examples = """
|
||||
To force a command to read from STDIN instead of file (for commands taking
|
||||
a filename as their first argument), substitute "-" for the filename.
|
||||
|
||||
|
||||
EXAMPLES:
|
||||
|
||||
Generate a random LTC Bech32 public/private keypair:
|
||||
$ mmgen-tool -r0 --coin=ltc --type=bech32 randpair
|
||||
|
||||
Generate a DASH address with compressed public key from the supplied WIF key:
|
||||
$ mmgen-tool --coin=dash --type=compressed wif2addr XJkVRC3eGKurc9Uzx1wfQoio3yqkmaXVqLMTa6y7s3M3jTBnmxfw
|
||||
|
||||
Generate a well-known burn address:
|
||||
$ mmgen-tool hextob58chk 000000000000000000000000000000000000000000
|
||||
|
||||
Generate a random 12-word seed phrase:
|
||||
$ mmgen-tool -r0 mn_rand128 fmt=bip39
|
||||
|
||||
Same as above, but get additional entropy from user:
|
||||
$ mmgen-tool mn_rand128 fmt=bip39
|
||||
|
||||
Encode bytes from a file to base 58:
|
||||
$ mmgen-tool bytestob58 /etc/timezone pad=20
|
||||
|
||||
Reverse a hex string:
|
||||
$ mmgen-tool hexreverse "deadbeefcafe"
|
||||
|
||||
Same as above, but supply input via STDIN:
|
||||
$ echo "deadbeefcafe" | mmgen-tool hexreverse -
|
||||
"""
|
||||
|
||||
def main_help():
|
||||
|
||||
from ..util import capfirst
|
||||
|
|
@ -61,52 +105,8 @@ def gen_tool_usage():
|
|||
|
||||
from ..util import capfirst
|
||||
|
||||
m1 = """
|
||||
GENERAL USAGE INFORMATION FOR MMGEN-TOOL COMMANDS
|
||||
|
||||
Arguments with only type specified in square brackets are required
|
||||
|
||||
Arguments with both type and default value specified in square brackets are
|
||||
optional and must be specified in the form ‘name=value’
|
||||
|
||||
For more detailed usage information for a particular tool command, type
|
||||
‘mmgen-tool help <command name>’
|
||||
"""
|
||||
|
||||
m2 = """
|
||||
To force a command to read from STDIN instead of file (for commands taking
|
||||
a filename as their first argument), substitute "-" for the filename.
|
||||
|
||||
|
||||
EXAMPLES:
|
||||
|
||||
Generate a random LTC Bech32 public/private keypair:
|
||||
$ mmgen-tool -r0 --coin=ltc --type=bech32 randpair
|
||||
|
||||
Generate a DASH address with compressed public key from the supplied WIF key:
|
||||
$ mmgen-tool --coin=dash --type=compressed wif2addr XJkVRC3eGKurc9Uzx1wfQoio3yqkmaXVqLMTa6y7s3M3jTBnmxfw
|
||||
|
||||
Generate a well-known burn address:
|
||||
$ mmgen-tool hextob58chk 000000000000000000000000000000000000000000
|
||||
|
||||
Generate a random 12-word seed phrase:
|
||||
$ mmgen-tool -r0 mn_rand128 fmt=bip39
|
||||
|
||||
Same as above, but get additional entropy from user:
|
||||
$ mmgen-tool mn_rand128 fmt=bip39
|
||||
|
||||
Encode bytes from a file to base 58:
|
||||
$ mmgen-tool bytestob58 /etc/timezone pad=20
|
||||
|
||||
Reverse a hex string:
|
||||
$ mmgen-tool hexreverse "deadbeefcafe"
|
||||
|
||||
Same as above, but supply input via STDIN:
|
||||
$ echo "deadbeefcafe" | mmgen-tool hexreverse -
|
||||
"""
|
||||
|
||||
for line in m1.lstrip().split('\n'):
|
||||
yield line.lstrip('\t')
|
||||
for line in usage_hdr.lstrip().split('\n'):
|
||||
yield line
|
||||
|
||||
for clsname, cmdlist in main_tool.mods.items():
|
||||
cls = main_tool.get_mod_cls(clsname)
|
||||
|
|
@ -128,8 +128,8 @@ def gen_tool_usage():
|
|||
w = max_w)
|
||||
yield ''
|
||||
|
||||
for line in m2.rstrip().split('\n'):
|
||||
yield line.lstrip('\t')
|
||||
for line in usage_examples.rstrip().split('\n'):
|
||||
yield line
|
||||
|
||||
def gen_tool_cmd_usage(mod, cmdname):
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue