List panel -- the meaning of 'L'

funderburkjim commented 1 year ago

In the CDSL dictionaries (such as MW, Apte, etc), the list pane (in the webtc1 display) is found approximately as follows:

Find the first entry with the given headword
Get the 'L' value for this entry. Suppose this L0= 100.
Get list of L-values centered at L0
- Say this list is 90, 91,...,100, 101,... 110
- take into account that (esp. in MW) that neighboring L-values may have same key1, and discard duplicates
get the corresponding list of 'key1' values for the list of L-values
display that list of key1 values in the list pane.

This makes sense for the CDSL dictionaries, since the entries are, with some exceptions, in Sanskrit alphabetical order.

funderburkjim commented 1 year ago

But for harsa and similar dictionaries, this scheme does not work.

Currently we get typically just 1 element (the 'middle' headword) in the list pane.

funderburkjim commented 1 year ago

As I understand it, in the digitization (harsa.txt), L roughly corresponds to the verse.

<L>87<pc>156
<k1>deva-puM<meanings>rAjan,meGa,sura
<k1>nava-puM<meanings>navya,stuti
<k1>Bava-puM<meanings>Siva,saMsAra,BUti,sattA,Apti,janman
rAjYi meGe sure devo navo navye stutAvapi .
BavaH Sive ca saMsAre BUtisattAptijanmasu .. 90 ..
<LEND>

This one record of harsa.txt becomes 14 records of harsa.xml, one for each of the key1 values

deva,rAjan,meGa,sura
nava,navya,stuti
Bava,Siva,saMsAra,BUti,sattA,Apti,janman

Each of these 14 xml records has the same 'embedded' L-value (displayed as ID=87). e.g.

<H1><h><key1>deva</key1><key2>deva</key2></h> <body>
<hwdetails><hwdetail><hw><s>deva-puM</s></hw><meaning><s>rAjan,meGa,sura</s></meaning></hwdetail></hwdetails>
<entrydetails><entrydetail><s>rAjYi meGe sure devo navo navye stutAvapi .</s></entrydetail><entrydetail><s>BavaH Sive ca saMsAre BUtisattAptijanmasu .. 90 ..</s></entrydetail></entrydetails>
</body><tail><L>87</L><pc>156</pc></tail></H1>

That <L>87</L> becomes [ID=87] in the displays.

harsa.sqlite

The displays access the xml records via the database made from the xml. The sqlite.py program creates harsa.sqlite from harsa.xml. There is one database record for each xml record, and this database record has 3 fields:

 key VARCHAR(100)  NOT NULL,    <<<   from <key1>deva</key1> in the xml record
 lnum DECIMAL(10,2) NOT NULL,   <<<  from `<L>87</L>` in the xml record
 data TEXT NOT NULL   <<<< the full record from harsa.xml

This 'lnum' field is used ONLY for list display.

solution idea

Change sqlite.py so it computes the 'lnum' field differently. Namely,

sort all the xml records by key1 in sanskrit order
compute sqlite field lnum as the index of the record according to this sorted order.

Let's try this idea and see what happens!.

funderburkjim commented 1 year ago

version v1

This version of kosha-dev pushed to Github. And installed at sanskrit-lexicon; sanskrit-lexicon url for displays: https://sanskrit-lexicon.uni-koeln.de/work/kosha-dev/v1/apps/harsa/web/

funderburkjim / kosha-dev