IRC Bots: Difference between revisions

From Lojban
Jump to navigation Jump to search
No edit summary
m (remove gerna)
Line 45: Line 45:
Also, its lujvo-breaking algorithm for ''(components)'' is backed by [[camxes]], so it will call you out on a badly-constructed lujvo (e.g. a [[tosmabru]]), unlike '''valsi'''.
Also, its lujvo-breaking algorithm for ''(components)'' is backed by [[camxes]], so it will call you out on a badly-constructed lujvo (e.g. a [[tosmabru]]), unlike '''valsi'''.


==='''gerna'''===
<!--==='''gerna'''===
This is a grammar checker based on [[jbofi'e]].
This is a grammar checker based on [[jbofi'e]].


Line 54: Line 54:
The message is just a Lojban sentence. If it parses, you'll get the structure. If it doesn't parse or jbofi'e crashes, it will say "not grammatical" and try to show you which word broke the parse.
The message is just a Lojban sentence. If it parses, you'll get the structure. If it doesn't parse or jbofi'e crashes, it will say "not grammatical" and try to show you which word broke the parse.


There are no options.
There are no options
 
-->
==='''camxes'''===
==='''camxes'''===
A grammar checking bot based on [[camxes]].
A grammar checking bot based on [[camxes]].
Line 76: Line 76:


==='''mensi''', '''livla'''===
==='''mensi''', '''livla'''===
These bots are parts of [[ilmentufa]] project and start talking when the channel ##jboselbau is dead for 15 minutes, so that there are no long stretches of time without Lojban being transmitted over the Internet.
Kitchen sink bot run by [[la gleki]] ([https://github.com/lagleki/glekitufa/tree/master/ircbot source]).


Run by [[la gleki]] ([https://github.com/lagleki/glekitufa/tree/master/ircbot source]).
Commands recognized by mensi:
 
'''livla''' repeats sentences from [[Tatoeba]], while '''mensi''' has a list of canned replies.
 
If you catch any mistakes, you should probably tell someone, since it means a Lojban entry on Tatoeba is wrong.
 
Command line flags used by mensi:
*"'''off:'''", "'''exp:'''"
*"'''off:'''", "'''exp:'''"
**These are parser options based on [[camxes]]. '''off''' parses regular Lojban while '''exp''' has (most of?) the [[ELG]] implemented.
**These are parser options based on [[camxes]]. '''off''' parses regular Lojban while '''exp''' has (most of?) the [[ELG]] implemented.
**You send this bot a sentence by prefixing it with the bot's name and a colon (the colon is mandatory, unlike other bots), e.g. "off: coi". You can also private-message it with no prefix.
**You send this bot a sentence by prefixing it with the parser name and a colon, e.g. "off: coi". You can also private-message it (the prefixes are also needed there).
**Accepted options (put them before the sentence, like "off:+s coi"):
**Accepted options (put them before the sentence, like "off:+s coi"):
*** +s
*** +s
Line 93: Line 87:
*** -f
*** -f
**** Don't show terminators.
**** Don't show terminators.
*** -f+s (not +s-f)
*** -f+s
**** Both of the above.
**** Both of the above.
*[[jbofihe]] parser
**"gerna:" parses using jbofihe, which was the de facto standard parser before camxes.
*[[John Cowan]]'s parser
**"cowan:" parses using the (very old) [[Official LLG Parser|official LLG parser]], based on the YACC grammar.
*Searching jbovlaste with '''mensi''':
*Searching jbovlaste with '''mensi''':
**Entering "en:coi" into the chat will output the English definition, notes and the author.
**Entering "en:coi" into the chat will output the English definition, notes and the author.
***lujvo are expanded and its structure is shown. If no lujvo is found it searches for other lujvo with the same se rafsi.
***lujvo are expanded and its structure is shown. If no lujvo is found it searches for other lujvo with the same se rafsi.
**Other languages are supported: "jbo:", "ja:" (Japanese), "ru:" (Russian), "fr-facile:" (Easy French), "de:" (German), "es:" (Spanish), "eo:" (Esperanto). The databases are autoupdated once every 24 hours.
**Other languages are supported: "jbo:", "ja:" (Japanese), "ru:" (Russian), "fr-facile:" (Easy French), "de:" (German), "es:" (Spanish), "eo:" (Esperanto). The databases are autoupdated once every 3 days.
**"/full " flag performs a search for a given sequence in definitions and notes. E.g. "en:/full greets".
**"/full " flag performs a search for a given sequence in definitions and notes. E.g. "en:/full greets".
*"Tatoeba:" outputs an arbitrary Lojban sentence from Tatoeba with a substring given
*"Tatoeba:" outputs an arbitrary Lojban sentence from Tatoeba with a substring given
Line 115: Line 113:
*lujvo building
*lujvo building
** "lujvo:" combines a given tanru into a list of lujvo and cmevla with their scores, using a method similar to [[jvocuhadju]]. The first (with the lowest score) lujvo are preferable in speech.
** "lujvo:" combines a given tanru into a list of lujvo and cmevla with their scores, using a method similar to [[jvocuhadju]]. The first (with the lowest score) lujvo are preferable in speech.
*[[John Cowan]]'s parser
 
**"cowan:" parses using the (very old) [[Official LLG Parser|official LLG parser]], based on the YACC grammar.
In the channel ##jboselbau, '''livla''' repeats sentences from [[Tatoeba]] every few minutes. If you catch any mistakes, you should probably tell someone, since it means a Lojban entry on Tatoeba is wrong.
 
==Other==
==Other==
* '''sidju'''
* '''sidju'''
Line 129: Line 128:
:: Opens the most recent link pasted in the channel, and retrieves its title.
:: Opens the most recent link pasted in the channel, and retrieves its title.
* '''xorban'''
* '''xorban'''
: Xorban is a parser bot for another logical language (you guessed it, called Xorban). It's actually the same bot as '''camxes''' and '''cipra''' but with a different grammar.
: Xorban is a parser bot for another logical language (you guessed it, called Xorban). It's actually the same bot as '''camxes''' but with a different grammar.
: Run by [[la selpa'i]].
: Run by [[la selpa'i]].
* '''vreji'''
* '''vreji'''

Revision as of 13:25, 8 February 2015

A Guide to the IRC Bots of #lojban (also #ckule and ##jboselbau).

You might think all ~120 logged on users of #lojban are watching you and/or participating in the conversation. Well, no, we all live in different timezones and at any given moment of time most of them are inactive. But also, there are a bunch of robots! Some of them are even helpful. Read on for more.

Lojban assistance

valsi

This is a bidirectional Lojban-English dictionary. It works from a dump of Jbovlaste, so it may not be completely up to date.

Run by la donri (source).

You can invoke valsi in the channel by prefixing a message with "valsi ", or send a private message to the bot (with no prefix).

The content of the message is just a word. If it's a Lojban word, you'll get its definition. If it's an English word, you'll get a list of suggested translations.

Accepted options for Lojban word lookup: (you can give up to one option, in parentheses, e.g. "valsi lercu'aca'a (components)")

  • affix
Get the affix (rafsi) form of the word.
  • class
Get the selma'o (grammatical category) of the word.
  • type
Get the type (cmavo/gismu/lujvo/etc) of the word.
  • notes
Return the notes field of the definition (without this option it isn't shown).
  • cll
Get the link to the section of the CLL that discusses the word (not always accurate).
  • url
Get the link to Vlasisku's page for the word.
  • components
Break a lujvo into its component gismu (works on nonce lujvo). Beware: lujvo with missing hyphens are still parsed, so you can't rely on this to check your construction.
  • rafsi
Treat the word as a rafsi for the purposes of the search (many cmavo are also rafsi of other words).
  • I don't see this option in the source, so maybe what's running is a slightly different version? Durka42 (talk)
  • You can also send queries in vlasisku's internal query language, like "valsi class:BAI" or "valsi affix:ka'i".

vlaste

Another instance of valsi, run by la durka, which contains an updated jbovlaste database, bug fixes and new features.

It supports the following additional commands, which valsi does not (until it is updated):

  • lujvo
Combine a tanru into a lujvo, using jvocu'adju. If the canonical form is defined, the definition is shown.
Relatedly, if you ask vlaste for a lujvo not in canonical form, and no definition is found, it will try to help you by silently correcting it to canonical form.
  • finti
Show the jbovlaste user who created the word.

Also, its lujvo-breaking algorithm for (components) is backed by camxes, so it will call you out on a badly-constructed lujvo (e.g. a tosmabru), unlike valsi.

camxes

A grammar checking bot based on camxes.

Run by la durka (source).

You can invoke camxes in the channel by prefix a message with "camxes: ", or send a private message to the bot (with no prefix).

The message is just a Lojban sentence. If it parses, you'll get the structure. If it doesn't parse, it'll try to tell you where you went wrong.

Accepted flags for parsing: (put them before the sentence, like "camxes: +s-f coi"; you can put any number of flags in any order)

  • +s
Show the selma'o of each word in the sentence.
  • -f
Don't show elidible terminators.
  • +exp (or -std)
Parse using the experimental grammar.
  • -exp (or +std)
Parse using the standard grammar (default).

mensi, livla

Kitchen sink bot run by la gleki (source).

Commands recognized by mensi:

  • "off:", "exp:"
    • These are parser options based on camxes. off parses regular Lojban while exp has (most of?) the ELG implemented.
    • You send this bot a sentence by prefixing it with the parser name and a colon, e.g. "off: coi". You can also private-message it (the prefixes are also needed there).
    • Accepted options (put them before the sentence, like "off:+s coi"):
      • +s
        • Show the selma'o of each word.
      • -f
        • Don't show terminators.
      • -f+s
        • Both of the above.
  • jbofihe parser
    • "gerna:" parses using jbofihe, which was the de facto standard parser before camxes.
  • John Cowan's parser
  • Searching jbovlaste with mensi:
    • Entering "en:coi" into the chat will output the English definition, notes and the author.
      • lujvo are expanded and its structure is shown. If no lujvo is found it searches for other lujvo with the same se rafsi.
    • Other languages are supported: "jbo:", "ja:" (Japanese), "ru:" (Russian), "fr-facile:" (Easy French), "de:" (German), "es:" (Spanish), "eo:" (Esperanto). The databases are autoupdated once every 3 days.
    • "/full " flag performs a search for a given sequence in definitions and notes. E.g. "en:/full greets".
  • "Tatoeba:" outputs an arbitrary Lojban sentence from Tatoeba with a substring given
  • Alternative orthography with mensi:
    • "rot13:" prefix performs ROT13 encoding on a given string
    • "mensi: j " performs jbopomofo encoding on a given string
    • "mensi: r " performs Cyrillic encoding on a given string
  • Messaging
    • "mensi: doi <nick> <message>"
      • Delayed messaging service. Next time <nick> says something in the channel, mensi will relay the message. You can use regular expressions in the nick.
  • Loglan to Lojban conversion
    • "loi:" converts Lojban text to Loglan (not perfect!). Failed words are marked with "*".
    • "coi:" converts Loglan text to Lojban (not perfect!). Failed words are marked with "*".
  • Glossing
    • "gloss:" converts Lojban text to glossed English words.
  • lujvo building
    • "lujvo:" combines a given tanru into a list of lujvo and cmevla with their scores, using a method similar to jvocuhadju. The first (with the lowest score) lujvo are preferable in speech.

In the channel ##jboselbau, livla repeats sentences from Tatoeba every few minutes. If you catch any mistakes, you should probably tell someone, since it means a Lojban entry on Tatoeba is wrong.

Other

  • sidju
This is an instance of the Phenny IRC bot.
Run by Tene (source).
Common commands (see the source for full documentation):
    • "sidju: tell <nick> <message>"
Delayed messaging service. Next time <nick> says something in the channel, sidju will relay the message. You can use * in the nick as a wildcard.
    • Google/Wikipedia search: ".g <query>" or ".wik <query>"
Searches and returns a link to the first result.
    • Link title lookup: ".title"
Opens the most recent link pasted in the channel, and retrieves its title.
  • xorban
Xorban is a parser bot for another logical language (you guessed it, called Xorban). It's actually the same bot as camxes but with a different grammar.
Run by la selpa'i.
  • vreji
Logs Lojban utterances in the channel at http://www.lojban.org/irclogs/.
Run by a pile of Perl scripts.
Does not respond to any commands, as far as I know.