SanskritVerb - A verb form generator

[dhaval patel]

Dear scholars,

We are happy to present before you our project SanskritVerb - a computational simulation of step by step Sanskrit verb form generation (तिङन्तप्रक्रिया).

Expected outcome -

1. The code gives step by step derivation of all forms of a given verb in given tense.

2. The code gives a derived table at the end.

Source code - https://github.com/drdhaval2785/SanskritVerb

Online testing - http://www.sanskritworld.in/sanskrittool/SanskritVerb/tiGanta.html

N.B. -

1. Currently the machine doesn't create sanAdi verb forms. It will be included in later versions.

2. Bug reports are welcome at https://github.com/drdhaval2785/SanskritVerb/issues

3. We have relied upon the following books - (1) Ashtadhyayi sahajabodha by Dr. Pushpa Dikshit, (2) Siddhantakaumudi with Balamanorama commentary and (3) Kashikavritti.

4. Current version is version 1.1.0 (https://github.com/drdhaval2785/SanskritVerb/releases/tag/v1.1.0)

Thanking you,

Dr. Dhaval Patel, I.A.S and Dr. Shivakumari Katuri

[Mārcis Gasūns]

We are happy to see the results, words hardly describe my thankfulness.

Not only good, but open code. Documentation will follow, I'm sure.

If "कर्तृवाच्य Mandatory field. Wait for ajax to load." is mandatory, why do I have to click on it? Can you make it checked already? 

If I enter kalpate, do I use it correctly?

[Pradyumna Achar]

Dear Dr. Patel,

The tool is very useful and informative. Thank you.

Is there a facility for the reverse query, i.e., given a form like "आसम्" is there a facility to tell what are all the possible ways by which it can be arrived at?
I think that would be useful while reading an existing text and trying to understand the meaning. (Eg: I came across a word recently -- तेनैवासमितीरयन् -- तेन+एव+आसाम्+इति+ईरयन् ).

Regards,
Pradyumna

[dhaval patel]

Not right now.

Once I am sure about the output of the code, I plan to store the output in XML file or some kind of database and then the reverse query can be easily handled. That would take at least take 6 months looking at the bugs we deal right now. But definitely on the radar.

--
You received this message because you are subscribed to the Google Groups "sanskrit-programmers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sanskrit-program...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Dr. Dhaval Patel, I.A.S

Collector and District Magistrate, Anand

www.sanskritworld.in

[kancuka tattva]

Thanks you for this valuable DB.
Is there a DB (even partial) of Sanskrit verb forms that are actually occurring in Vedic and Epic Sanskrit?

[Pankajashree R]

Thank you for this resource. Very helpful.  

Also, I would like to know if there is an API to get verb forms and English meaning if we input a root? 

I have seen UOHYD's work regarding verb form generator:

http://sanskrit.uohyd.ac.in/scl/#

This is a sample output for a root

http://sanskrit.uohyd.ac.in/cgi-bin/scl/skt_gen/verb/verb_gen.cgi?prayoga=%E0%A4%95%E0%A4%B0%E0%A5%8D%E0%A4%A4%E0%A4%B0%E0%A4%BF&vb=%E0%A4%B9%E0%A4%A8%E0%A5%8D1_%E0%A4%B9%E0%A4%A8%E0%A4%81_%E0%A4%85%E0%A4%A6%E0%A4%BE%E0%A4%A6%E0%A4%BF%E0%A4%83_%E0%A4%B9%E0%A4%BF%E0%A4%82%E0%A4%B8%E0%A4%BE%E0%A4%97%E0%A4%A4%E0%A5%8D%E0%A4%AF%E0%A5%8B%E0%A4%83&encoding=Unicode&upasarga=-&paxI=parasmEpaxI

and this gives meanings and synonyms - 

http://sanskrit.uohyd.ac.in/cgi-bin/scl/SHMT/options1.cgi?word=%E0%A4%B9%E0%A4%A8%E0%A5%8D

Are these programs and databases open source? 

If not, are there any APIs which give JSON outputs for the verb forms and meanings? 

[विश्वासो वासुकिजः (Vishvas Vasuki)]

​-indology​ where I'm not member.

2017-06-22 23:29 GMT-07:00 Pankajashree R <pankaj...@gmail.com>:

Are these programs and databases open source? 

​Yes~ dhaval will say more.​

 

If not, are there any APIs which give JSON outputs for the verb forms and meanings? 

Not fully JSON (lakAra-s will be html wrapped in json), but you'll be able to lookup in this database: dict_entries / http://vedavaapi.org:5984/dict_entries . You're free to replicate it or use it remotely - details in https://github.com/sanskrit-coders/dict-api.

--

--
Vishvas /विश्वासः

[dhaval patel]

Currently it is not exactly API. Currently it is data+presentation mixed. It was preliminarily designed as web application to display HTML. Now I have learnt that it has much wider application. So trying to separate data and presentation. Once that is done, I will be able to generate good API.

In the meanwhile,

http://www.sanskritworld.in/sanskrittool/SanskritVerb/panini.php?first=BU&upasarga=&lakAra=law&sanAdi=&tran=SLP1&vAcya=kartR&frontend=1&number=01.0001&type=tiGanta

This may give you some idea about the structure of parameters.

[Mārcis Gasūns]

On Friday, 23 June 2017 18:34:45 UTC+3, Pankajashree R wrote:

Also, I would like to know if there is an API to get verb forms and English meaning if we input a root? 

I have seen UOHYD's work regarding verb form generator:

http://sanskrit.uohyd.ac.in/scl/#

This is a sample output for a root

http://sanskrit.uohyd.ac.in/cgi-bin/scl/skt_gen/verb/verb_gen.cgi?prayoga=%E0%A4%95%E0%A4%B0%E0%A5%8D%E0%A4%A4%E0%A4%B0%E0%A4%BF&vb=%E0%A4%B9%E0%A4%A8%E0%A5%8D1_%E0%A4%B9%E0%A4%A8%E0%A4%81_%E0%A4%85%E0%A4%A6%E0%A4%BE%E0%A4%A6%E0%A4%BF%E0%A4%83_%E0%A4%B9%E0%A4%BF%E0%A4%82%E0%A4%B8%E0%A4%BE%E0%A4%97%E0%A4%A4%E0%A5%8D%E0%A4%AF%E0%A5%8B%E0%A4%83&encoding=Unicode&upasarga=-&paxI=parasmEpaxI

It's a closed system. If you want open - 

The corresponding database of XML Sanskrit morphology is now distributed as the Gitlab repository https://gitlab.inria.fr/huet/Heritage_Resources.

 

and this gives meanings and synonyms - 

http://sanskrit.uohyd.ac.in/cgi-bin/scl/SHMT/options1.cgi?word=%E0%A4%B9%E0%A4%A8%E0%A5%8D

That's nothing more than a dictionary. http://www.sanskrit-lexicon.uni-koeln.de/scans/awork/apidev/sample/list-0.2.html is far more advanced.

 

Are these programs and databases open source? 

No, only Gerard and Cologne are open.

 

If not, are there any APIs which give JSON outputs for the verb forms and meanings? 

No, there is no API. If you are ready to code it, it can be added to Cologne.

[Mārcis Gasūns]

On Wednesday, 29 June 2016 02:33:56 UTC+3, kancuka tattva wrote:

Thanks you for this valuable DB.
Is there a DB (even partial) of Sanskrit verb forms that are actually occurring in Vedic and Epic Sanskrit?

It's Whitney's roots at http://warnemyr.com/skrgram/grammar/syntax/cases.html 

[Pankajashree R]

Thank you. I will try to use this and build a small app :) 

Regarding the WebAPI - 

https://github.com/sanskrit-coders/dict-api#web-api

Why am I not getting the characters properly for the entries on Chrome (Version 58.54.3029.81)

If I use Firefox, I get it properly. 

[Pankajashree R]

On Saturday, 24 June 2017 13:37:21 UTC+5:30, Mārcis Gasūns wrote:

It's a closed system. If you want open - 

The corresponding database of XML Sanskrit morphology is now distributed as the Gitlab repository https://gitlab.inria.fr/huet/Heritage_Resources.

Thank you. 

I'm curious to know, does Dr. Amba Kulkarni's program use a similar way to generate the forms of the verbs? Or is there a way for dynamic generation of verb forms, instead of keying in all forms..

 

 

and this gives meanings and synonyms - 

http://sanskrit.uohyd.ac.in/cgi-bin/scl/SHMT/options1.cgi?word=%E0%A4%B9%E0%A4%A8%E0%A5%8D

That's nothing more than a dictionary. http://www.sanskrit-lexicon.uni-koeln.de/scans/awork/apidev/sample/list-0.2.html is far more advanced.

 

Are these programs and databases open source? 

No, only Gerard and Cologne are open.

 

If not, are there any APIs which give JSON outputs for the verb forms and meanings? 

No, there is no API. If you are ready to code it, it can be added to Cologne.

I am interested in looking into Cologne ..

[dhaval patel]

Thank you. 

I'm curious to know, does Dr. Amba Kulkarni's program use a similar way to generate the forms of the verbs? Or is there a way for dynamic generation of verb forms, instead of keying in all forms..

Last time I checked, it is manually entered. It is verb form lookup method, and not verb form generator method.

I am interested in looking into Cologne  

https://github.com/sanskrit-lexicon/Cologne/issues

And

https://github.com/sanskrit-lexicon/CORRECTIONS/issues

maybe good starting point.

[विश्वासो वासुकिजः (Vishvas Vasuki)]

2017-07-09 3:00 GMT-07:00 Pankajashree R <pankaj...@gmail.com>:

Thank you. I will try to use this and build a small app :) 

Regarding the WebAPI - 

https://github.com/sanskrit-coders/dict-api#web-api

 

Why am I not getting the characters properly for the entries on Chrome (Version 58.54.3029.81)

​​That's odd - I am able to see it fine on Version 60.0.3112.32 (Official Build) beta (64-bit) on ubuntu.

​

​Are you able to correctly read other devanAgarI pages fine? How about the devanAgarI in https://raw.githubusercontent.com/sanskrit-coders/dict-api/master/README.md ?

​

There's an outdated post on a similar issue at stack exchange - https://stackoverflow.com/questions/16658463/utf-8-characters-not-displaying-in-chrome . Please post your question there so that someone might help. (And copy the link here so that we might follow along.)

If I use Firefox, I get it properly. 

On Friday, 23 June 2017 20:45:43 UTC+5:30, विश्वासो वासुकिजः wrote:

​

Not fully JSON (lakAra-s will be html wrapped in json), but you'll be able to lookup in this database: dict_entries / http://vedavaapi.org:5984/dict_entries . You're free to replicate it or use it remotely - details in https://github.com/sanskrit-coders/dict-api.

--

--
Vishvas /विश्वासः

--
You received this message because you are subscribed to the Google Groups "sanskrit-programmers" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sanskrit-programmers+unsub...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[विश्वासो वासुकिजः (Vishvas Vasuki)]

(readding bvp and samskrita since it was cross posted)

2017-07-09 5:12 GMT-07:00 dhaval patel <drdhav...@gmail.com>:

Thank you. 

I'm curious to know, does Dr. Amba Kulkarni's program use a similar way to generate the forms of the verbs? Or is there a way for dynamic generation of verb forms, instead of keying in all forms..

Last time I checked, it is manually entered. It is verb form lookup method, and not verb form generator method.

​I believe it ​uses a Finite State Transducer, built and specified using http://wiki.apertium.org/wiki/Lttoolbox technology (a dix file is written and compiled to a bin file). This approach can potentially have wider uses in sanskrit. smt ambA, who might be watching this thread may have more details.

[Shreevatsa R]

On Sun, Jul 9, 2017 at 3:00 AM, Pankajashree R <pankaj...@gmail.com> wrote:

Thank you. I will try to use this and build a small app :) 

Regarding the WebAPI - 

https://github.com/sanskrit-coders/dict-api#web-api

Why am I not getting the characters properly for the entries on Chrome (Version 58.54.3029.81)

If I use Firefox, I get it properly. 

It's just a character encoding issue; the bytes are there and will be usable when you parse the json.

The server sends its response as Content-Type:application/json to your browser (and indeed the response is valid JSON and intended to be consumed by programs, not read in your browser), then your browser has the task of interpreting those bytes and displaying them in the DOM. When it does that it tries to guess the character encoding, and in this case the server has returned UTF-8 but your browser has guessed ISO-8859-15. If you're writing a program you can specify the encoding and everything will be decoded into the proper Unicode characters. 

Actually JSON strings are required to be Unicode, but the browser in this case isn't smart enough to realize that what it's seeing is JSON and factor that into its encoding-guessing. (For what it's worth, I get correct output in Chrome and similar to yours in Firefox.) I guess in this case if the server returns Content-type: application/json; charset=utf-8 then there's a greater chance of the browser guessing correctly. It should be a non-issue as the server responses aren't designed to be viewed in the browser. Or maybe the server could do that too, return either html or json depending on the request's "accept" headers.

[विश्वासो वासुकिजः (Vishvas Vasuki)]

2017-07-09 3:00 GMT-07:00 Pankajashree R <pankaj...@gmail.com>:

Thank you. I will try to use this and build a small app :) 

​Also, please host your work-in-progress on github, so that we may observe and possibly offer suggestions as you proceed.​