Transcribe vocab up to brandt 10-3, allow comment header in TSV, hand…

…le edge case of variants used within one text where group is introduced

src/app/prebuild.ts

@@ -77,10 +77,12 @@ async function fillInMissingReadingsInTsvs() {
     const isBrandtPassage = textId.startsWith("brandt-");
     if (isBrandtPassage) brandtPassagesVisited += 1;
     const vocabFileContents = getPassageVocabFileContents(textId);
-    const { vocab, variants } = parsePassageVocabList(
+
+    const { vocab, variants, comment } = parsePassageVocabList(
       textId,
       vocabFileContents
     );
+
     const mainToSecondaryVariants: {
       [mainVariant: string]: string[];
     } = Object.entries(variants).reduce(
@@ -104,7 +106,20 @@ async function fillInMissingReadingsInTsvs() {
       Object.keys(vocab).concat(isBrandtPassage ? newCharsInPassage : [])
     );
 
-    for (const char of featuredChars) {
+    if (textId.includes("10-3")) {
+      console.log({ variants, vocab, newCharsInPassage, featuredChars });
+    }
+
+    const featuredCharsMainVariants = new Set(
+      [...featuredChars].flatMap((char) => {
+        const mainVariants = [
+          ...(lexicon.variants[char] || []),
+          ...(variants[char] || []),
+        ];
+        return mainVariants.length ? mainVariants : [char];
+      })
+    );
+    for (const char of featuredCharsMainVariants) {
       if (
         !lexicon.vocab[char] ||
         vocab[char]?.some((e) => vocabFileColumns.some((k) => !e[k.key]))
@@ -168,7 +183,7 @@ async function fillInMissingReadingsInTsvs() {
         }));
       }
     }
-    const newVocabFileContents = makeVocabTsvContent(vocab, variants);
+    const newVocabFileContents = makeVocabTsvContent(comment, vocab, variants);
 
     if (
       Object.keys(vocab).length &&
@@ -218,7 +233,10 @@ function getEmbeddedChineseSegments(text: string) {
 function writePassageVocabularyJsons(lexicon: PassageVocabWithVariants) {
   for (const textId of getTextsIds()) {
     const vocabFileContents = getPassageVocabFileContents(textId);
-    const { vocab } = parsePassageVocabList(textId, vocabFileContents);
+    const { vocab, variants: _passageVariants } = parsePassageVocabList(
+      textId,
+      vocabFileContents
+    );
     const passage = parsePassage(getPassageFileContents(textId));
 
     const vocabJsonPath = path.join(
@@ -229,9 +247,7 @@ function writePassageVocabularyJsons(lexicon: PassageVocabWithVariants) {
 
     const variants: PassageTermVariants = {};
     for (const char of passageChars) {
-      if (!vocab[char]) {
-        vocab[char] = lexicon.vocab[char];
-      }
+      vocab[char] ||= lexicon.vocab[char];
       const mainVariants = lexicon.variants[char];
       if (mainVariants) {
         variants[char] = mainVariants;
@@ -266,10 +282,12 @@ function getMandarinReadings(
 }
 
 function makeVocabTsvContent(
+  comment: string | null,
   vocab: Partial<Record<string, LexiconEntry[]>>,
   variants: PassageTermVariants
 ): string | NodeJS.ArrayBufferView {
   return [
+    ...(comment ? [comment.trim()] : []),
     `Traditional\tQieyun\tHanyu Pinyin\tJyutping\tKorean\tVietnamese\tEnglish`,
     ...Object.entries(vocab).flatMap(
       ([char, ee]) =>

src/app/texts/Passage.ts

@@ -42,8 +42,14 @@ export function parsePassageVocabList(
 ) {
   const vocab: PassageVocab = {};
   const variants: PassageTermVariants = {};
+  let comment: string | null = null;
+
   if (vocabFileContents) {
-    const lines = vocabFileContents.split("\n");
+    const [, commentText, body] =
+      vocabFileContents?.match(/^(#.*\n)?([\s\S]*)$/) || [];
+    comment = commentText || null;
+
+    const lines = body.split("\n");
 
     const [, ...columnHeaders] = lines[0].trim().split("\t");
     const invalidColumnHeaders = columnHeaders.filter(
@@ -92,7 +98,7 @@ export function parsePassageVocabList(
       }
     }
   }
-  return { vocab, variants };
+  return { vocab, variants, comment };
 }
 
 export function parsePassage(passageFileContents: string) {

texts/brandt-ch03-3.vocab.tsv

@@ -15,6 +15,5 @@ Traditional	Qieyun	Hanyu Pinyin	Jyutping	Korean	Vietnamese	English
 觀	見合一寒平?, 見合一寒去?	guān	gun¹	관	quan	to gaze at; to look; to inspect
 等	端開一咍上?, 端開一登上?	děng	dang²	등	đẳng	to wait; a class; a rank
 虚		xū	heoi¹		hư	
-分	幫三文平?, 並三文去?	fēn	fan¹	분	phân	
 量	來開三陽平?, 來開三陽去?	liàng	loeng⁴	량?, 양?		
 低	端開四齊平	dī	dai¹	저	đây	

texts/brandt-ch04-3.vocab.tsv

@@ -37,7 +37,7 @@ Traditional	Qieyun	Hanyu Pinyin	Jyutping	Korean	Vietnamese	English
 起	溪開三之上	qǐ	hei²	기	khởi	to rise up. to raise; to start
 切	清開四齊去?, 清開四先入?	qiè	cit³	절?, 체?	thiết	to cut; urgent; pressing; very
 兵	幫三庚平	bīng	bing¹	병	binh	a soldier, a weapon; military
-器	溪開三B脂去	qì	hei³	기	khí	a vessel. implements; capacity
+器	溪開三B脂去	qì	hei³	기	khí	a vessel; implements; a disih; an apparatus. capacity; ability.
 概	見開一咍去	gài	koi³			to level: to adjust; all
 違	云合三微平	wéi	wai⁴	위	vi	to oppose; to disobey
 禁	見開三B侵平?, 見開三B侵去?	jìn	gam³	금	cấm	to forbid; to prohibit

texts/brandt-ch08-1.vocab.tsv

@@ -13,7 +13,7 @@ Traditional	Qieyun	Hanyu Pinyin	Jyutping	Korean	Vietnamese	English
 虐	疑開三陽入	nüè	joek⁶	학	ngược	to be cruel; cruelty; to oppress.
 施	書開三支平?, 書開三支去?	shī	si¹	시	thi	to give; to bestow; to apply.
 女	孃開三魚上?, 孃開三魚去?	nǚ	neoi⁵	녀?, 여?	nữa	woman; female; daughter.
-便	並三A仙平?, 並三A仙去?	biàn	bin⁶	변?, 편?	tiện	convenient; then.
+便	並三A仙平?, 並三A仙去?	biàn	bin⁶	변?, 편?	tiện	convenient; cheap. ordinary; plain. then; in that case.
 狃	孃三尤上?, 孃三尤去?	niǔ	nau²		nữu	to be accustomed to; to be used to.
 俗	邪三鍾入	sú	zuk⁶	속	tục	common; vulgar. a custom.
 衛	云合三祭去	wèi	wai⁶	위	vệ	to escort; to guard.

texts/brandt-ch10-1.vocab.tsv

@@ -1,35 +1,62 @@
+# nao (428) the brain. 腦 無數 wu-shu-innumerable. 知 chih-chüch-perception 知血器 hsieh-blood. shen² (824)-spirits; gods. The soul; the mind. Force; expression. ch'i-a vessel; a dish; an apparatus. Capacity; ability. 神經 shen²-ching-nerves. ch'ian-all; the whole; 全 complete; perfect. nao-ti-the brain sub- stance. 充ch'ung to fill; to satisfy: to fulfil. 塞 sai (se¹) to stop up; to block. 充塞 ch'ung-sai-to fill. 電線 tien-lightning; electri- city. hsien a thread; a wire. 電線tien-hsien wires. telegraph 分fen to divide; to dis- tribute. lu²(207) the skull; the forehead. 分布 fen-put-to distribute. 互通 hu-mutual; together. 頭顱 l'ou-lu² the head. shu-a number; some. 通 t'ung to go through; to circulate. General; wholly; complete. hsiao (124) to melt; to consume; to disperse. 消息 hsiao-hsi news. rumours; 五人wu-jen-we. 動作 tung-tso to move; move. a 傅合 chuan-ling-to orders. issue 揮huil to move; to directi to shake. 指揮 chih-hui-to direct. 百 po-ti-the whole body; the mechanism of body. tsung (380) to unite: to 咸 sum up. To control. 同 chu²-position; circum- stances office. A board; an 總局 tsung-chi-a head of fice. 分局fen-chi-a branch-office. 痛楚 t'ung-ch'u³-pain; sore. 痾ko (01) (699)-sickness; pain. 癢 yang³ (151)-to itch. 荷癢 k'o-yang-itching. 觸 chut (788) to butt; to strike against. 肌膚 chi¹ (396)-the flesh. fu the skin; the flesh. 肌膚 chi-fu the flesh; the skin. 臭 hsiu-to smell. Read ch'ous strong smelling; stinking. 味 wei (495)-test; flavor. chich¹ (85) to receive; to take. 接鼻 pi-the nose. 感 kan³ to touch; to in- fluence; to excite. kan-chuch-sensation. 耗 hao-to waste; to destroy. A rat. 消耗 hsiao-hao-to spend; to waste. 睡 shui to sleep. mien (358) to close the eyes; to sleep. 睡眠 shui-mien-to sleep. t'unga boy under 15 years of age, A girl. 小時 hsiao-shih-an hour. tuta measure; a limit. To cross over.
 Traditional	Qieyun	Hanyu Pinyin	Jyutping	Korean	Vietnamese	English
-腦	泥開一豪上?, 泥開一豪去?	nǎo	nou⁵	뇌	não	
+腦	泥開一豪上	nǎo	nou⁵	뇌	não	the brain
+腦	泥開一豪去	?	?	?	?	tanned animal hide; leather. smooth, glossy. [morohashi]
+知覺						perception
 本	幫一魂上	běn	bun²	본	bản	
+腦體						the brain substance
 心	心開三侵平	xīn	sam¹	심	tâm	
-血	曉合四先入	xuè	hyut³	혈	huyết	
-充	昌三東平	chōng	cung¹	충	sung	
-顱	來一模平	lú	lou⁴	로		
+血	曉合四先入	xuè	hyut³	혈	huyết	blood
+充	昌三東平	chōng	cung¹	충	sung	to fill; to satisfy; to fulfil
+顱	來一模平	lú	lou⁴	로		the skull; the forehead
+頭顱						the head
 部	並一模上?, 並一侯上?	bù	bou⁶	부	bộ	
-數	生三虞上?, 生三虞去?, 生二江入?	shù	sou³	삭?, 수?	số	
-神	船開三眞平	shén	san⁴	신	thần	
-全	從合三仙平	quán	cyun⁴	전	toàn	
-電	定開四先去	diàn	din⁶	전		
-線	心開三仙去	xiàn	sin³	선		
-互	匣一模去	hù	wu⁶	호	hỗ	
-通	透一東平	tōng	tung¹	통	thông	
-消	心開三宵平	xiāo	siu¹	소	tiêu	
-揮	曉合三微平	huī	fai¹	휘	huy	
-總	精一東上	zǒng	zung²	총	tổng	
-局	羣三鍾入	jú	guk⁶	국	cục	
-疴	溪開二麻去	kē	o¹	아		
-癢	以開三陽上	yǎng	joeng⁵	양		
-觸	昌三鍾入	chù	zuk¹	촉	xúc	
-肌	見開三B脂平	jī	gei¹	기		
-膚	幫三虞平	fū	fu¹	부		
-臭	昌三尤去	chòu	cau³	취	xú	
-味	明三微去	wèi	mei⁶	미	vị	
-接	精開三鹽入	jiē	zip³	접	tiếp	
-鼻	並三A脂去	bí	bei⁶	비		
-感	見開一覃上	gǎn	gam²	감	cảm	
-耗		hào	hou³	모	hao	
-睡	常合三支去	shuì	seoi⁶	수		
-眠	明四先平	mián	min⁴	면?, 민?	miên	
-童	定一東平	tóng	tung⁴	동	đồng	
-八	幫二山入	bā	baat³	팔	bát	
-度	定一模去?, 定開一唐入?	dù	dou⁶	도?, 탁?	độ	
+數	生三虞上	shũ	?	수	?	count; enumerate. rebuke. permit. compare; be equal to. [kroll]
+數	生三虞去	shù	sou³	수	số	number; some
+數	生二江入	cù	?	삭	số	close together, dense; tight-meshed, close-stitched. [kroll]
+無數						innumerable
+神	船開三眞平	shén	san⁴	신	thần	spirits; gods. the soul; the mind. force; expression
+神經						nerves
+全	從合三仙平	quán	cyun⁴	전	toàn	all; the whole; complete; perfect
+電	定開四先去	diàn	din⁶	전		lightning; electricity
+線	心開三仙去	xiàn	sin³	선		a thread; a wire
+電線						telegraph wires
+分	幫三文平	fēn	fan¹	분	phân	to divide; to distribute
+分	並三文去	fèn	?	분	?	portion; rank; allotment; fate. suppose; infer. affection; goodwill. [kanjikai, kroll, hxwd]
+分佈						to distribute
+互	匣一模去	hù	wu⁶	호	hỗ	mutual; together
+通	透一東平	tōng	tung¹	통	thông	to go through; to circulate. general; wholly; complete
+消	心開三宵平	xiāo	siu¹	소	tiêu	to melt; to consume; to disperse
+消息						rumours; news
+吾人						we
+動作						to move; a move
+傳令						to issue orders
+揮	曉合三微平	huī	fai¹	휘	huy	to move; to direct; to shake
+指揮						to direct
+百體						the whole body; the mechanism of body
+總	精一東上	zǒng	zung²	총	tổng	to unite; to sum up. to control
+局	羣三鍾入	jú	guk⁶	국	cục	a head office
+分局						a branch-office
+痛楚						pain; sore
+疴	溪開二麻去	kē	o¹	아		sickness; pain
+癢	以開三陽上	yǎng	joeng⁵	양		to itch
+荷癢						itching
+觸	昌三鍾入	chù	zuk¹	촉	xúc	to butt; to strike against
+肌	見開三B脂平	jī	gei¹	기		the flesh
+膚	幫三虞平	fū	fu¹	부		the skin; the flesh
+肌膚						the flesh; the skin
+臭	昌三尤去	chòu	cau³	취	xú	strong smelling; stinking
+嗅,臭		xiù	cau³	후		to smell
+味	明三微去	wèi	mei⁶	미	vị	test; flavor
+接	精開三鹽入	jiē	zip³	접	tiếp	to receive; to take
+鼻	並三A脂去	bí	bei⁶	비		the nose
+感	見開一覃上	gǎn	gam²	감	cảm	to touch; to influence; to excite
+耗		hào	hou³	모	hao	to waste; to destroy. a rat
+消耗						to spend; to waste
+睡	常合三支去	shuì	seoi⁶	수		to sleep
+眠	明四先平	mián	min⁴	면?, 민?	miên	to close the eyes; to sleep
+童	定一東平	tóng	tung⁴	동	đồng	a boy under 15 years of age. a girl.
+小時						an hour
+度	定一模去	dù	dou⁶	도	độ	a measure; a limit. to cross over
+度	定開一唐入	duó	?	탁	?	measure; calculate; plan; conjecture. [kroll]
+八	幫二山入	bā	baat³	팔	bát	eight.

texts/brandt-ch10-2.passage.md

@@ -34,7 +34,7 @@ How could I for cupidity and for dread of death (lit. because I see the interest
 吾不能生。而使公子獨死矣。
 I cannot remain alive and let the prince die alone.
 
-突遂與公子俱逃澤中。
+遂與公子俱逃澤中。
 And then she, holding the prince in her arms (lit. together with the prince), jumped into a pool.
 
 秦軍見而射之。

texts/brandt-ch10-2.vocab.tsv

@@ -1,20 +1,26 @@
+# 魏 weit (512)-name of (403-241 feudal State В. С.) 129/520 200% + Vocabulary. 蓋 乳ju milk; to suckle. 乳母:ju-mu-a wet nurse. ch'in² name of a feudal State (897-221 P. C.) ho² an interrogative par- ticle,--why not? would it not be better to....? ying) (202)-proper; sui- table. Ouglht; must. Read ying to answer; to correspond; to fulfil. 0 must. Business; duty; function. 務 wut to be necessary; 田 weit to fear; to be dread- ed. 公子 kung-tzu-a son of a prince; a heir-apparent. 員匿 罪 tsui-a crime; Punishment. a 族 shang to bestow; to re- ward.. nit to hide; to abscond. 誅 chu² (499)-to punish; to put to death. sin. 廢義 fei-i-to neglect duty. the tsu-a tribe; a clan; a family. 詐 cha (150) to deceive; artful; false. 120 行詐 hsing-chat-to commit a treachery.. i (98)-the wings of a bird; to shelter; to as- sist. 翼 逃 t'aos (383) to flee; escape. to i-pi-to cover; to pro- tect. 池澤 时 tse-a marsh; a pool. 射shet-to shoot. chu-here: "to hit"; "to strike" shiha dart; an arrow. To take an oath.
 Traditional	Qieyun	Hanyu Pinyin	Jyutping	Korean	Vietnamese	English
-魏	疑合三微去	wèi	ngai⁶	위		
-乳	日三虞上	rǔ	jyu⁵	유	vú	
-秦	從開三眞平	qín	ceon⁴	진		
-賞	書開三陽上	shǎng	soeng²	상	thưởng	
+魏	疑合三微去	wèi	ngai⁶	위		name of a feudal State
+乳	日三虞上	rǔ	jyu⁵	유	vú	milk; to suckle
+乳母						wet nurse
+秦	從開三眞平	qín	ceon⁴	진		name of a feudal State
+公子						son of a prince; heir-apparent
+賞	書開三陽上	shǎng	soeng²	상	thưởng	to bestow; to reward
 千	清開四先平	qiān	cin¹	천	thiên	
-匿	孃開三蒸入	nì	nik¹	닉?, 익?	nặc	
-罪	從合一灰上	zuì	zeoi⁶	죄	tội	
-族	從一東入	zú	zuk⁶	족	tộc	
-應	影開三蒸平?, 影開三蒸去?	yīng	jing³	응	ứng	
-務	明三虞去	wù	mou⁶	무	vụ	
-畏	影合三微去	wèi	wai³	외	uý	
-誅	知三虞平	zhū	zyu¹	주	tru	
-詐	莊開二麻去	zhà	zaa³	사	trá	
-逃	定開一豪平	táo	tou⁴	도	đào	
-射	以開三麻去?, 船開三麻去?, 以開三清入?, 船開三清入?	shè	se⁶	사	xạ	
-翼	以開三蒸入	yì	jik⁶	익	dực	
-二	日開三脂去	èr	ji⁶	이	nhì	
-矢	書開三脂上	shǐ	ci²	시	ngao?, thỉ?	
-突	透合一魂入?, 定合一魂入?	tū	dat⁶	돌	đột	
+匿	孃開三蒸入	nì	nik¹	닉?, 익?	nặc	to hide; to abscond
+罪	從合一灰上	zuì	zeoi⁶	죄	tội	crime; punishment
+族	從一東入	zú	zuk⁶	족	tộc	tribe; clan; family
+應	影開三蒸平?, 影開三蒸去?	yīng	jing³	응	ứng	to answer; to correspond; to fulfil
+務	明三虞去	wù	mou⁶	무	vụ	business; duty; function
+畏	影合三微去	wèi	wai³	외	uý	to fear; to be dreaded
+誅	知三虞平	zhū	zyu¹	주	tru	to punish; to put to death
+廢義						to neglect the duty.
+詐	莊開二麻去	zhà	zaa³	사	trá	to deceive; artful; false.
+行詐						to commit a treachery.
+逃	定開一豪平	táo	tou⁴	도	đào	to flee; to escape.
+射	以開三麻去?, 船開三麻去?, 以開三清入?, 船開三清入?	shè	se⁶	사	xạ	to shoot.
+翼	以開三蒸入	yì	jik⁶	익	dực	the wings of a bird; to shelter; to assist.
+翼蔽						to cover; to protect.
+二	日開三脂去	èr	ji⁶	이	nhì	two
+矢	書開三脂上	shǐ	ci²	시	ngao?, thỉ?	a dart; an arrow. to take an oath.
+