[ett] Configuring and scraping Etruscan.

CHANGELOG.md

@@ -34,6 +34,7 @@ Unreleased
 -   Scraped Kashmiri (`kas`). (\#431)
 -   Added Malayalam (`mal`) scrape. (\#434)
 -   Configuration and initial scrape for Dhivehi (`div`, Maldivian). (\#437)
+-   Configuring and scraping Etruscan (`ett`). (\#444)
 
 #### Changed
 

data/scrape/README.md

@@ -86,6 +86,7 @@
 | [TSV](tsv/enm_latn_broad.tsv) | enm | Middle English (1100-1500) | Middle English | Latin |  | False | Broad | True | 6,855 |
 | [TSV](tsv/epo_latn_broad.tsv) | epo | Esperanto | Esperanto | Latin |  | False | Broad | True | 14,990 |
 | [TSV](tsv/est_latn_broad.tsv) | est | Estonian | Estonian | Latin |  | False | Broad | True | 429 |
+| [TSV](tsv/ett_ital_broad.tsv) | ett | Etruscan | Etruscan | Old Italic |  | False | Broad | False | 130 |
 | [TSV](tsv/ewe_latn_broad.tsv) | ewe | Ewe | Ewe | Latin |  | False | Broad | True | 120 |
 | [TSV](tsv/fao_latn_broad.tsv) | fao | Faroese | Faroese | Latin |  | False | Broad | True | 1,740 |
 | [TSV](tsv/fao_latn_narrow.tsv) | fao | Faroese | Faroese | Latin |  | False | Narrow | True | 1,120 |
@@ -198,7 +199,7 @@
 | [TSV](tsv/mah_latn_broad.tsv) | mah | Marshallese | Marshallese | Latin |  | False | Broad | True | 900 |
 | [TSV](tsv/mah_latn_narrow.tsv) | mah | Marshallese | Marshallese | Latin |  | False | Narrow | True | 1,502 |
 | [TSV](tsv/mak_latn_narrow.tsv) | mak | Makasar | Makasar | Latin |  | False | Narrow | True | 405 |
-| [TSV](tsv/mal_mlym_narrow.tsv) | mal | Malayalam | Malayalam | Malayalam |  | False | Narrow | None | 141 |
+| [TSV](tsv/mal_mlym_narrow.tsv) | mal | Malayalam | Malayalam | Malayalam |  | False | Narrow | False | 141 |
 | [TSV](tsv/mar_deva_broad.tsv) | mar | Marathi | Marathi | Devanagari |  | False | Broad | False | 588 |
 | [TSV](tsv/mar_deva_narrow.tsv) | mar | Marathi | Marathi | Devanagari |  | False | Narrow | False | 118 |
 | [TSV](tsv/may_arab_ara_broad.tsv) | may | Malay (macrolanguage) | Malay | Arabic |  | False | Broad | True | 628 |

data/scrape/lib/languages.json

@@ -555,6 +555,15 @@
             "latn": "Latin"
         }
     },
+    "ett": {
+        "iso639_name": "Etruscan",
+        "wiktionary_name": "Etruscan",
+        "wiktionary_code": "ett",
+        "casefold": false,
+        "script": {
+            "ital": "Old Italic"
+        }
+    },
     "ewe": {
         "iso639_name": "Ewe",
         "wiktionary_name": "Ewe",

data/scrape/tsv/ett_ital_broad.tsv

@@ -0,0 +1,130 @@
+𐌀	a
+𐌀𐌂𐌀	a c a
+𐌀𐌂𐌀𐌅𐌉𐌔𐌄𐌓	a c a v i s e r
+𐌀𐌂𐌀𐌆𐌓	a c a z r
+𐌀𐌂𐌀𐌋𐌀	a c a l a
+𐌀𐌂𐌀𐌋𐌄	a c a l e
+𐌀𐌂𐌀𐌋𐌅𐌄	a c a l v e
+𐌀𐌂𐌀𐌋𐌉𐌀	a c a l i a
+𐌀𐌂𐌀𐌐	a c a p
+𐌀𐌂𐌀𐌓𐌉𐌀	a c a r i a
+𐌀𐌂𐌀𐌓𐌉𐌀𐌋	a c a r i a l
+𐌀𐌂𐌀𐌔	a c a s
+𐌀𐌂𐌀𐌔𐌀	a c a s a
+𐌀𐌂𐌀𐌔𐌂𐌄	a c a s c e
+𐌀𐌂𐌀𐌔𐌓𐌉	a c a s r i
+𐌀𐌂𐌄	a c e
+𐌀𐌂𐌄𐌉	a t͡ʃ e i̯
+𐌀𐌂𐌄𐌉𐌀𐌋	a t͡ʃ e i̯ a̯ l
+𐌀𐌂𐌄𐌋	a c e l
+𐌀𐌂𐌄𐌋𐌔	a c e l s
+𐌀𐌂𐌄𐌔	a c e s
+𐌀𐌂𐌉𐌋	a c i l
+𐌀𐌂𐌉𐌋𐌀	a c i l a
+𐌀𐌂𐌉𐌋𐌖	a c i l u
+𐌀𐌂𐌉𐌋𐌖𐌍𐌉𐌀	a c i l u n i a
+𐌀𐌂𐌋𐌖𐌔	a c l u s
+𐌀𐌂𐌍𐌀𐌉𐌂𐌄	a c n a i c e
+𐌀𐌂𐌍𐌀𐌉𐌍𐌄	a c n a i n e
+𐌀𐌂𐌍𐌀𐌍𐌀𐌔	a c n a n a s
+𐌀𐌂𐌍𐌀𐌑𐌅𐌄𐌓𐌔	a c n a s v e r s
+𐌀𐌂𐌍𐌉𐌍𐌀	a c n i n a
+𐌀𐌂𐌍𐌔	a c n s
+𐌀𐌅𐌄	a f e
+𐌀𐌅𐌉𐌋	ɑ w ɪ l
+𐌀𐌅𐌉𐌋𐌔	a v i l s
+𐌀𐌉𐌈𐌕𐌓𐌀	a i θ r a
+𐌀𐌉𐌔	a j s
+𐌀𐌉𐌔𐌍𐌀	a j s n a
+𐌀𐌉𐌖𐌔	a i u s
+𐌀𐌊𐌀	a c a
+𐌀𐌋𐌂𐌄	a l c e
+𐌀𐌋𐌂𐌖	a l c u
+𐌀𐌋𐌄	a l e
+𐌀𐌋𐌅	a l f
+𐌀𐌋𐌅𐌀	a l f a
+𐌀𐌋𐌅𐌀𐌋	a l f a l
+𐌀𐌋𐌅𐌀𐌔	a l f a s
+𐌀𐌋𐌅𐌉𐌄	a l f i e̯
+𐌀𐌋𐌉𐌂𐌄	a l i c e
+𐌀𐌋𐌉𐌊𐌄	a l i c e
+𐌀𐌋𐌉𐌗𐌀	a l i χ a
+𐌀𐌋𐌉𐌗𐌄	a l i χ e
+𐌀𐌋𐌑𐌀𐌔𐌄	a l z a s e
+𐌀𐌋𐌗𐌖	a l χ u
+𐌀𐌍𐌂𐌀𐌓𐌉𐌀	a n c a r i a
+𐌀𐌍𐌉𐌀𐌗	a n i a χ
+𐌀𐌍𐌉𐌀𐌗𐌄𐌑	a n i a χ e z
+𐌀𐌏𐌖𐌌𐌉𐌂𐌀	a θ u m i c a
+𐌀𐌏𐌖𐌌𐌉𐌂𐌔	a θ u m i c s
+𐌀𐌐𐌀	ɑ p ə
+𐌀𐌐𐌉𐌀𐌍𐌀	a p i a n a
+𐌀𐌐𐌉𐌓𐌄	a p i r e
+𐌀𐌐𐌉𐌓𐌄𐌔	a p i r e s
+𐌀𐌓𐌆𐌍𐌀	a r z n a
+𐌀𐌓𐌆𐌍𐌄𐌀𐌋	a r z n e a̯ l
+𐌀𐌓𐌆𐌍𐌄𐌉	a r z n e i̯
+𐌀𐌓𐌆𐌍𐌉	a r z n i
+𐌀𐌓𐌆𐌍𐌉𐌔	a r z n i s
+𐌀𐌕𐌉	ɑ t ɪ
+𐌀𐌕𐌉𐌀𐌋	a t i a̯ l
+𐌀𐌕𐌓𐌀𐌍𐌄	a t r a n e
+𐌀𐌕𐌓𐌀𐌍𐌄𐌔	a t r a n e s
+𐌀𐌕𐌓𐌉𐌖𐌌	a t r i u m
+𐌀𐌗𐌉𐌋𐌀𐌔	a χ i l a s
+𐌀𐌗𐌉𐌋𐌄	a χ i l e
+𐌀𐌗𐌉𐌋𐌄𐌉	a χ i l e i̯
+𐌀𐌗𐌋𐌀𐌄	a χ l a e̯
+𐌀𐌗𐌋𐌄	a χ l e
+𐌀𐌗𐌌𐌄𐌌𐌓𐌖𐌍	a χ m e m r u n
+𐌀𐌗𐌍𐌀𐌋	a χ n a l
+𐌀𐌗𐌍𐌄𐌉	a χ n e i
+𐌀𐌗𐌍𐌉	a χ n i
+𐌀𐌗𐌓𐌖𐌌	a χ r u m
+𐌀𐌗𐌖	a χ u
+𐌀𐌗𐌖𐌀𐌋	a χ u a l
+𐌀𐌗𐌖𐌉	a χ u i
+𐌀𐌗𐌖𐌋𐌄	a χ u l e
+𐌀𐌚𐌓	a f r
+𐌀𐌚𐌖𐌓	a f u r
+𐌁𐌀𐌋𐌕𐌄𐌀	b a l t e a
+𐌁𐌄𐌓𐌂𐌏𐌌𐌔𐌍𐌀	b e r c o m s n a
+𐌂𐌀𐌄	c a e
+𐌂𐌀𐌄𐌔	c a e s
+𐌂𐌀𐌅𐌄	c a v e
+𐌂𐌀𐌅𐌄𐌈	c a v e θ
+𐌂𐌀𐌅𐌔𐌀	c a v s a
+𐌂𐌀𐌍𐌋𐌀𐌔	c a n l a s
+𐌂𐌀𐌏𐌍𐌀	c a θ n a
+𐌂𐌀𐌏𐌍𐌀𐌋	c a θ n a l
+𐌂𐌀𐌏𐌍𐌉	c a θ n i
+𐌂𐌀𐌏𐌍𐌉𐌔	c a θ n i s
+𐌂𐌀𐌔𐌔𐌉𐌃𐌀	c a s s i d a
+𐌂𐌀𐌔𐌔𐌉𐌔	c a s s i s
+𐌂𐌀𐌔𐌕𐌂𐌄	c a s t c e
+𐌂𐌀𐌕𐌍𐌉𐌔	c a t n i s
+𐌂𐌉	k i
+𐌄𐌉𐌔	e j s
+𐌄𐌉𐌔𐌍𐌀	e j s n a
+𐌄𐌔𐌀𐌋	e s a l
+𐌅𐌄𐌓𐌔𐌄	v e r s e
+𐌅𐌄𐌕𐌖𐌔	v e t u s
+𐌅𐌉𐌊𐌖	v i c u
+𐌆𐌀𐌋	t͡s a l
+𐌆𐌀𐌌𐌀𐌈𐌉	t͡s a m a tʰ i
+𐌆𐌀𐌌𐌈𐌉𐌂	t͡s a m tʰ i k
+𐌇𐌀𐌋𐌊	h a l k
+𐌇𐌀𐌋𐌗𐌆𐌀	h a l χ z a
+𐌇𐌀𐌋𐌗𐌆𐌄	h a l χ z a
+𐌈𐌖	tʰ u
+𐌈𐌖𐌍	tʰ u n
+𐌉𐌂𐌀𐌐	i c a p
+𐌋𐌖𐌂𐌖𐌌𐌏𐌍𐌄𐌔	l u c u m o n e s
+𐌋𐌖𐌂𐌖𐌌𐌖	l u c u m u
+𐌋𐌖𐌄𐌀	l u e a
+𐌌𐌀𐌙	m a kʰ
+𐌍𐌄𐌓𐌉	n e ɾ ɪ
+𐌐𐌄𐌓𐌂𐌖𐌌𐌔𐌍𐌀	p e r c u m s n a
+𐌑𐌀	ʃ a
+𐌗𐌉𐌋𐌀𐌔	χ i l a s
+𐌘𐌄𐌓𐌔𐌖	pʰ e r s u

data/scrape/tsv_summary.tsv

@@ -84,6 +84,7 @@ eng_latn_us_narrow.tsv	eng	English	English	Latin	US, General American	False	Narr
 enm_latn_broad.tsv	enm	Middle English (1100-1500)	Middle English	Latin		False	Broad	True	6855
 epo_latn_broad.tsv	epo	Esperanto	Esperanto	Latin		False	Broad	True	14990
 est_latn_broad.tsv	est	Estonian	Estonian	Latin		False	Broad	True	429
+ett_ital_broad.tsv	ett	Etruscan	Etruscan	Old Italic		False	Broad	False	130
 ewe_latn_broad.tsv	ewe	Ewe	Ewe	Latin		False	Broad	True	120
 fao_latn_broad.tsv	fao	Faroese	Faroese	Latin		False	Broad	True	1740
 fao_latn_narrow.tsv	fao	Faroese	Faroese	Latin		False	Narrow	True	1120
@@ -196,7 +197,7 @@ mac_cyrl_narrow.tsv	mac	Macedonian	Macedonian	Cyrillic		False	Narrow	True	6878
 mah_latn_broad.tsv	mah	Marshallese	Marshallese	Latin		False	Broad	True	900
 mah_latn_narrow.tsv	mah	Marshallese	Marshallese	Latin		False	Narrow	True	1502
 mak_latn_narrow.tsv	mak	Makasar	Makasar	Latin		False	Narrow	True	405
-mal_mlym_narrow.tsv	mal	Malayalam	Malayalam	Malayalam		False	Narrow		141
+mal_mlym_narrow.tsv	mal	Malayalam	Malayalam	Malayalam		False	Narrow	False	141
 mar_deva_broad.tsv	mar	Marathi	Marathi	Devanagari		False	Broad	False	588
 mar_deva_narrow.tsv	mar	Marathi	Marathi	Devanagari		False	Narrow	False	118
 may_arab_ara_broad.tsv	may	Malay (macrolanguage)	Malay	Arabic		False	Broad	True	628

wikipron/languagecodes.py

@@ -460,4 +460,6 @@
     "dv": "Dhivehi",
     "divehi": "Dhivehi",
     "maldivian": "Dhivehi",
+    # Etruscan: ISO 639-3 only.
+    "ett": "Etruscan",
 }