more IPA (#9)

cldf.py

@@ -35,7 +35,7 @@
         write2.writerow([entry, headword, '', data[entry][0]['ref']])
     for row in read:
         write.writerow(row)
-        write2.writerow([row[0], row[2], row[3], row[4]])
+        write2.writerow([row[0], row[2], '', row[3]])
 
 a = set()
 with open('cldf/forms.csv', 'w') as fout, open('errors.txt', 'w') as errors:
@@ -64,22 +64,24 @@
                     write.writerow([num, lang, entry, word, desc, '', '', entry, '', 'CDIAL'])
                 else:
                     if word[0] == '': continue
-                    word[0] = word[0].strip('.,')
+                    word[0] = word[0].strip('.,;- ')
+                    word[0] = word[0].replace('<? >', '')
                     word[0] = word[0].lower()
 
+                    for i in superscript:
+                        word[0] = word[0].replace('ˊ', '́').replace(' --', '-').replace('-- ', '-')
+                        word[0] = word[0].replace(f'<superscript>{i}</superscript>', superscript[i])
+
                     oldest = unicodedata.normalize('NFD', word[0])
                     oldest = oldest.replace('̄˘', '̄̆')
                     oldest = oldest.replace('̆̄', '̄̆')
                     oldest = oldest.replace('̄̆', '̄̆')
                     if '̄̆' in oldest:
                         form['words'].append([oldest.replace('̄̆', '̄'), word[1]])
-                        word[0] = word[0].replace('̄̆', '')
+                        oldest = oldest.replace('̄̆', '')
+                        word[0] = oldest
                     word[0] = unicodedata.normalize('NFC', word[0])
 
-                    for i in superscript:
-                        word[0] = word[0].replace('ˊ', '́').replace(' --', '-').replace('-- ', '-')
-                        word[0] = word[0].replace(f'<superscript>{i}</superscript>', superscript[i])
-
                     if '˚' not in word[0]: reference = word[0]
                     else:
                         old = word[0]
@@ -95,7 +97,7 @@
                     if lang in tokenizers and '˚' not in word[0]:
                         ipa = tokenizers[lang](word[0], column='IPA').replace(' ', '').replace('#', ' ')
                         if '�' in ipa:
-                            if lang == 'S': errors.write(f'{lang} {oldest} {word[0]} {ipa}\n')
+                            if lang == 'M': errors.write(f'{lang} {oldest} {word[0]} {ipa}\n')
                             ipa = ''
 
                     write.writerow([num, lang, entry, word[0], word[1], '', ipa, entry, '', 'CDIAL'])

cldf/cognates.csv

@@ -15433,7 +15433,7 @@ e14,Indo-Aryan,*locis-kāṣṭhikā,,patyal5
 e15,Indo-Aryan,sapakṣa,'easy',patyal5
 e16,Indo-Aryan,*ubbāsī,'yawn',arora
 e17,Indo-Aryan,*kakkara,'cloud',arora
-e18,Indo-Aryan,*celha,'waist',arora
+e18,Indo-Aryan,*celhi,'waist',arora
 e19,Indo-Aryan,*cora,'asparagus beans',arora
 e20,Indo-Aryan,*cora,'hut' (perhaps related to cakrá?),arora
 e21,Indo-Aryan,*jhañjhaṭ,'problem',arora

cldf/parameters.csv

@@ -15416,33 +15416,33 @@ ID,Name,Concepticon_ID,Description
 14843,hárati:,,"<number>14843</number> <b>hárati:</b> Paš. <i>ar</i> -- ʻ to accept, agree, hear ʼ rather (like <i>ār</i> -- ʻ to bring ʼ) &lt; <smallcaps>āˊharati</smallcaps>. <p></p>"
 14844,hásta- :,,"<number>14844</number> <b>hásta -- :</b> Wg. (Gambir) <i>dōš</i> ʻ hand and forearm ʼ, <i>doš- &amp;atildedotdot ʻ palm of hand ʼ.</i> <p></p>"
 14845,himna- :,,"<number>14845</number> *<b>himna -- :</b> Sh.gil. <i>hĭn</i> m., gur. koh. <i>hinn</i> f., jij. <i>hĭ̄n</i> ʻ snow ʼ, Sv. <i>hina</i> rather than &lt; <smallcaps>sníh</smallcaps> -- ?<hr/><div id=""responsive_footer_table""><div id=""responsive_footer_tr""><div id=""responsive_footer_td""><a href=""/dictionaries/soas/"">Back to the Search Page </a>  |          <a href=""/dictionaries/"">Back to the DDSA Page</a>"
-e1,su-vāṭa,'good enclosure',patyal3
-e2,anna-dhāra,'stream of food',patyal2
-e3,ku-varṇa,'of bad colour',patyal2
-e4,gāvyáya,'belonging to or coming from cow or cattle' RV 9.70.7; 10.48.4,patyal2
-e5,dantā-valī,'row of teeth; teeth collectively',patyal2
-e6,saptā-nnādya,'seven grains',patyal2
-e7,*kalyā-hāra,'breakfast',patyal5
-e8,*citra-karbura,'spotted',patyal5
-e9,*catur-akṣa,'four-eyed',patyal5
-e10,jánitr̥,,patyal5
-e11,*dēva-sthala,,patyal5
-e12,*vādya-tantra,,patyal5
-e13,*mr̥taka-sthāna,,patyal5
-e14,*locis-kāṣṭhikā,,patyal5
-e15,sapakṣa,'easy',patyal5
-e16,*ubbāsī,'yawn',arora
-e17,*kakkara,'cloud',arora
-e18,*celha,'waist',arora
-e19,*cora,'asparagus beans',arora
-e20,*cora,'hut' (perhaps related to cakrá?),arora
-e21,*jhañjhaṭ,'problem',arora
-e22,*padopānah,"'shoe' [padá1, upānáh]",fritz
-e23,*yavanakadala,"'banana' [yavaná, kadala]",fritz
-e24,*yavanavālukā,"'sand' [yavaná, vālukā]",fritz
-e25,*varṣadhanus,"'rainbow' [varṣá, dhánus]",fritz
-e26,*ayasgaṇḍa,"'iron' [áyas, gaṇḍa]",fritz
-e27,*pādadīrghamarkaṭa,"'long-legged (?) spider' [pā́da, dīrghá, markaṭa]",fritz
-e28,*bhakkhamukha,"'lump-faced' [bhakkha, múkha]",fritz
-e29,*rūpāsti,"'has form?' [rūpá, ásti]",fritz
-e30,*anyaika,"'another' [anyá, ḗka]",fritz
+e1,su-vāṭa,,'good enclosure'
+e2,anna-dhāra,,'stream of food'
+e3,ku-varṇa,,'of bad colour'
+e4,gāvyáya,,'belonging to or coming from cow or cattle' RV 9.70.7; 10.48.4
+e5,dantā-valī,,'row of teeth; teeth collectively'
+e6,saptā-nnādya,,'seven grains'
+e7,*kalyā-hāra,,'breakfast'
+e8,*citra-karbura,,'spotted'
+e9,*catur-akṣa,,'four-eyed'
+e10,jánitr̥,,
+e11,*dēva-sthala,,
+e12,*vādya-tantra,,
+e13,*mr̥taka-sthāna,,
+e14,*locis-kāṣṭhikā,,
+e15,sapakṣa,,'easy'
+e16,*ubbāsī,,'yawn'
+e17,*kakkara,,'cloud'
+e18,*celhi,,'waist'
+e19,*cora,,'asparagus beans'
+e20,*cora,,'hut' (perhaps related to cakrá?)
+e21,*jhañjhaṭ,,'problem'
+e22,*padopānah,,"'shoe' [padá1, upānáh]"
+e23,*yavanakadala,,"'banana' [yavaná, kadala]"
+e24,*yavanavālukā,,"'sand' [yavaná, vālukā]"
+e25,*varṣadhanus,,"'rainbow' [varṣá, dhánus]"
+e26,*ayasgaṇḍa,,"'iron' [áyas, gaṇḍa]"
+e27,*pādadīrghamarkaṭa,,"'long-legged (?) spider' [pā́da, dīrghá, markaṭa]"
+e28,*bhakkhamukha,,"'lump-faced' [bhakkha, múkha]"
+e29,*rūpāsti,,"'has form?' [rūpá, ásti]"
+e30,*anyaika,,"'another' [anyá, ḗka]"

data/extensions_ia.csv

@@ -15,7 +15,7 @@ e14,Indo-Aryan,*locis-kāṣṭhikā,,patyal5
 e15,Indo-Aryan,sapakṣa,'easy',patyal5
 e16,Indo-Aryan,*ubbāsī,'yawn',arora
 e17,Indo-Aryan,*kakkara,'cloud',arora
-e18,Indo-Aryan,*celha,'waist',arora
+e18,Indo-Aryan,*celhi,'waist',arora
 e19,Indo-Aryan,*cora,'asparagus beans',arora
 e20,Indo-Aryan,*cora,'hut' (perhaps related to cakrá?),arora
 e21,Indo-Aryan,*jhañjhaṭ,'problem',arora

data/ipa/cdial/G.txt

@@ -0,0 +1,68 @@
+Grapheme	IPA
+a	ə
+ã	ə̃
+ā	aː
+ã̄	ãː
+ᵃ	ᵊ
+ᶦ	ɪ
+i	ɪ
+ï	ɪ
+ĩ	ɪ̃
+ī	iː
+ī˜	ĩː
+u	ʊ
+ũ	ʊ̃
+ū	uː
+ū̃	ũː
+e	eː
+ē	eː
+ẽ	ẽː
+ē̃	ẽː
+o	oː
+ō	oː
+õ	õː
+ō̃	õː
+ai	ɛː
+aĩ	ɛ̃ː
+au	ɔː
+ɔ	ɔ
+ɔ̃	ɔ̃
+aũ	ɔ̃ː
+k	k
+kh	kʰ
+g	g
+gh	gʱ
+ṅ	ŋ
+c	t͡ʃ
+ch	t͡ʃʰ
+j	d͡ʒ
+jh	d͡ʒʱ
+ñ	ɲ
+t	t̪
+th	t̪ʰ
+d	d̪
+dh	d̪ʱ
+n	n
+ṭ	ʈ
+ṭh	ʈʰ
+ḍ	ɖ
+ḍh	ɖʱ
+ṇ	ɳ
+p	p
+ph	pʰ
+b	b
+bh	bʱ
+y	j
+r	ɾ
+l	l
+v	ʋ
+w	ʋ
+s	s
+ś	ʃ
+ṣ	ʂ
+h	ɦ
+ṛ	ɽ
+x	x
+m	m
+–	
+ḷ	ɭ

data/ipa/cdial/M.txt

@@ -0,0 +1,72 @@
+Grapheme	IPA
+a	ə
+ã	ə̃
+ā	aː
+ã̄	ãː
+ᵃ	ᵊ
+ᶦ	ɪ
+i	ɪ
+ï	ɪ
+ĩ	ɪ̃
+ī	iː
+ī˜	ĩː
+u	ʊ
+ũ	ʊ̃
+ū	uː
+ū̃	ũː
+e	eː
+ē	eː
+ẽ	ẽː
+ē̃	ẽː
+o	oː
+ō	oː
+õ	õː
+ō̃	õː
+ai	ɛː
+aĩ	ɛ̃ː
+au	ɔː
+ɔ	ɔ
+ɔ̃	ɔ̃
+aũ	ɔ̃ː
+k	k
+kh	kʰ
+g	g
+gh	gʱ
+ṅ	ŋ
+c	t͡ʃ
+ch	t͡ʃʰ
+j	d͡ʒ
+jh	d͡ʒʱ
+ñ	ɲ
+t	t̪
+th	t̪ʰ
+d	d̪
+dh	d̪ʱ
+n	n
+ṭ	ʈ
+ṭh	ʈʰ
+ḍ	ɖ
+ḍh	ɖʱ
+ṇ	ɳ
+p	p
+ph	pʰ
+b	b
+bh	bʱ
+y	j
+r	ɾ
+l	l
+v	ʋ
+w	ʋ
+s	s
+ś	ʃ
+ṣ	ʂ
+h	ɦ
+ṛ	ɽ
+x	x
+m	m
+–	
+ḷ	ɭ
+ċ	t͡s
+ċh	t͡sʰ
+j̈	d͡z
+j̈h	d͡zʱ

data/ipa/cdial/P.txt

@@ -0,0 +1,66 @@
+Grapheme	IPA
+a	ə
+ã	ə̃
+ā	aː
+ã̄	ãː
+ᵃ	ᵊ
+ᶦ	ɪ
+i	ɪ
+ï	ɪ
+ĩ	ɪ̃
+ī	iː
+ī˜	ĩː
+u	ʊ
+ũ	ʊ̃
+ū	uː
+ū̃	ũː
+e	eː
+ē	eː
+ẽ	ẽː
+ē̃	ẽː
+o	oː
+ō	oː
+õ	õː
+ō̃	õː
+ai	ɛː
+aĩ	ɛ̃ː
+au	ɔː
+aũ	ɔ̃ː
+k	k
+kh	kʰ
+g	g
+gh	gʱ
+ṅ	ŋ
+c	t͡ʃ
+ch	t͡ʃʰ
+j	d͡ʒ
+jh	d͡ʒʱ
+ñ	ɲ
+t	t̪
+th	t̪ʰ
+d	d̪
+dh	d̪ʱ
+n	n
+ṭ	ʈ
+ṭh	ʈʰ
+ḍ	ɖ
+ḍh	ɖʱ
+ṇ	ɳ
+p	p
+ph	pʰ
+b	b
+bh	bʱ
+y	j
+r	ɾ
+l	l
+v	ʋ
+w	ʋ
+s	s
+ś	ʃ
+ṣ	ʂ
+h	ɦ
+ṛ	ɽ
+x	x
+m	m
+–	
+ḷ	ɭ

data/ipa/cdial/Pk.txt

@@ -0,0 +1,52 @@
+Grapheme	IPA
+ṁ	̃
+ṃ	
+a	ə
+ā	aː
+i	ɪ
+ï	ɪ
+ī	iː
+u	ʊ
+ü	ʊ
+ũ	ʊ̃
+ū	uː
+e	e
+ē	eː
+o	o
+ō	oː
+k	k
+kh	kʰ
+g	g
+gh	gʱ
+ṅ	ŋ
+c	t͡ʃ
+ch	t͡ʃʰ
+j	d͡ʒ
+jh	d͡ʒʱ
+ñ	ɲ
+t	t̪
+th	t̪ʰ
+d	d̪
+dh	d̪ʱ
+n	n
+ṭ	ʈ
+ṭh	ʈʰ
+ḍ	ɖ
+ḍh	ɖʱ
+ṇ	ɳ
+p	p
+ph	pʰ
+b	b
+bh	bʱ
+y	j
+r	ɾ
+l	l
+v	ʋ
+w	ʋ
+s	s
+ś	ʃ
+ṣ	ʂ
+h	ɦ
+ṛ	ɽ
+m	m
+–