ACC alternate headword patterns

[drdhaval2785]

This issue is devoted to bring out alternate headwords / embedded headword patterns from revised acc.txt (with meta lines).

[drdhaval2785]

Pattern 1

if len(line.split('#}')) > 2 and u'¦' in line:

This is reasonable estimate of two headwords on the first line of any entry.

This may miss longish alternate headwords ending on next line, but a total of 1229 such cases exist.

[gasyoun]

miss longish alternate headwords ending on next line

You've got a sample?

[drdhaval2785]

You've got a sample?

Just a possibility. Not sure whether any instance exists or not.

[drdhaval2785]

Pattern 2 - Author - work relationship

This is more or less peculiar to ACC.
Most likely the works should be treated as sub headwords if it doesn't exist (or maybe even if it exists).
Debate welcome.

<L>28<pc>1-001,2<k1>aKaRqAnanda<k2>aKaRqAnanda
{#aKaRqAnanda#}¦
<HI1>Advaitaratnakośa, vedānta. Rice 130.
<HI1>Ratnakośaṭīkā, vedānta. Rice 166.
<HI1>Mantroddhāraprakaraṇa. NW. 186.
<HI1>Mahāviṣṇupūjāpaddhati. NW. 186.
<HI1>Muktisopāna. Ben. 41.
<LEND>

[drdhaval2785]

Pattern 3 - Book - Commentary relationship

There are some items which are just mentioned by C.
But some of them are individual entries with name of commentary too.

See Sauravāsanā, Kiraṇāvalī, Sūryasiddhāntodāharaṇa, Vāsanābhāṣya, Saurabhāṣya, Gūḍhārthaprakāśaka.
Many of these are not enumerated as separate headwords.
They should also form alternate headwords. This will enhance possibility of identifying commentaries by their own names.

<L>38834<pc>2-175,2<k1>sUryasidDAnta<k2>sUryasidDAnta
{#sUryasidDAnta#}¦ jy. IO. 312. 454. 580. 1510 (fr.). 1844.
<>2260. 2263. Rgb. 884. 885. 887. Stein 177. Sūrya- 
[Page2-176-a+ 48]
<>siddhānte Bhūgolādhyāya and Gaṇitādhyāya (?). Peters.
<>4, 38.
<HI1><symbol n="C.">C.</symbol> Oudh XXII, 76.
<HI1><symbol n="C.">C.</symbol> Sauravāsanā by Kamalākara. Bhau Dāji 30
<>(inc). Rgb. 885.
<HI1><symbol n="C.">C.</symbol> by Caṇḍeśvarācārya. Bhau Dāji 125. Rgb. 886.
<HI1><symbol n="C.">C.</symbol> Kiraṇāvalī by Dādābhāī. IO. 1122. 2261.
<HI1><symbol n="C.">C.</symbol> Sūryasiddhāntodāharaṇa by Divākara. Oudh
<>XXII, 76.
<HI1><symbol n="C.">C.</symbol> Siddhāntadīpikā by Nārāyaṇa. CU. add. 1602.
<HI1><symbol n="C.">C.</symbol> Vāsanābhāṣya or Saurabhāṣya by Nṛsiṃha,
<>son of Kṛṣṇa Gaṇaka. IO. 1755. 2264.
<HI1><symbol n="C.">C.</symbol> by Bhaṭṭotpala. Quoted by Divākara in Prau-
<>ḍhamanoramā.
<HI1><symbol n="C.">C.</symbol> by Bhūdhara, son of Devadatta. IO. 580.
<>2262. Oudh XX, 112. 124. Stein 177.
<HI1><symbol n="C.">C.</symbol> by Yallaya. Gov. Or. Libr. Madras 109.
<HI1><symbol n="C.">C.</symbol> Gūḍhārthaprakāśaka by Raṅganātha, son of
<>Ballāla. GB. 120. IO. 454. 1844. 2263.
<>Stein 177.
<HI1><symbol n="C.">C.</symbol> Gahanārthaprakāśikā, a <symbol n="C.">C.</symbol> and udāharaṇa, by
<>Viśvanātha, son of Divākara. Devīpr. 79, 16.
<>Rgb. 807. 887. Stein 177.
<LEND>

[gasyoun]

Most likely the works should be treated as sub headwords

Makes sense.

They should also form alternate headwords.

What's the extraction algo?

[drdhaval2785]

@funderburkjim and @gasyoun Pattern 1 is implemented.
Total of 1591 alternate headwords added by this method.
They have been formed into the format described for acc_hwextra.txt in this comment sanskrit-lexicon/COLOGNE#133 (comment).

The code is in the dev server pywork/correctionwork/issue-alt-21/pattern_finder.py

Output is this - altlines.txt

On my local machine the alternates are showing properly. So Jim's modifications in redo_hw.sh scripts and redo_xml.sh scripts is working fine as usual.

[drdhaval2785]

longish alternate headwords ending on next line

There are many such cases @gasyoun . Now the code can handle them properly. Now the total counter is 1591. Therefore 362 (1591-1229) such cases are there where this line crossing phenomenon happenned.

See e.g.

<L>236<pc>1-006,1<k1>ajYAnaboDinI<k2>ajYAnaboDinI
{#ajYAnaboDinI#}¦ or {#aDyAtmavidyopadeSaviDi#} or {#saMkziptavedA-#}
<>{#ntaSAstraprakriyA,#} a <symbol n="C.">C.</symbol> on the Ātmabodha, by Śaṅka-
<>rācārya. IO. 100. Paris (B 159 c. D 57 b). Hall
<>p. 105. L. 678. Bik. 554. K. 112. B. 4, 36. 38.
<>Report XXVII. Ben. 69. 81. Rādh 5. Oudh V, 22.
<>NP. V, 170. Poona 43. Peters. 3, 391.
<HI1><symbol n="C.">C.</symbol> by Amṛtānanda. K. 112.
<LEND>

{#saMkziptavedA-#}\n<>{#ntaSAstraprakriyA,#} has some leftover on second line too.

[drdhaval2785]

Pattern 2 and 3 can't be fully segregated. If there is some interest of members in this direction, I may try.
Too much markup mess there.

[gasyoun]

362 (1591-1229) such cases

Indeed many. Pattern 2,3 may be of interest if a general approach can be developed. If it's unique for ACC, is not urgent. Otherwise, if it can help change the way we work with all other dictionaries - then it's the real difference between a book and digital copy - it's hyperlinks.

[drdhaval2785]

sanskrit-lexicon/COLOGNE#316 is the placeholder.