Help with hebrew to latin transcription in XeLaTeX

Question

I am attempting to transliterate hebrew text to latin using SBL handbook's transcription convention. I am almost successful, but there are some more advanced transliteration rules which I struggle to implement. See the problem below.

Hebrew is written right to left, and the hebrew script consists of consonants such as ידלת (translit. yḏlṯ). Vowels are added below the consonant, such as ְ ַ ָ (ā a ĕ ). When written with a hebrew keyboard each consonant is usually typed first, then the vowel (if any). The last vowel in the previous example (ְ ) is called shva (translit. ĕ) . In transliteration it should in certain cases be left out, namely when

1. A shva is on the last letter (consonant).
2. There are two shvas at the end of a word (on the last two consonants).

Example: Hebrew text: יָדַלְתְּ

My transliteration: yāḏalĕtĕ

Wanted transliteration: yād̲alt

My transliteration scheme uses a custom Mapping in polyglossia in which I simply translate single hebrew unicode characters or sequences of such to latin characters in a .map-file, which is later compiled with teckit_compile by SIL and amended by the Mapping function in polyglossia. I am struggling to implement rule 2 above, as it is dependent on the existence of a shva on another previous consonant. Since the mapping file is based on sequences of unicode numbers, I would in practice have to list every combination of shva, consonant and shva that there is in order to implement rule 2.

Is there a trick to avoid this, and instead somehow make reference to any existing shva on the previous consonant, i.e. two unicode characters backwards?

MWE:

\documentclass[12pt]{article}
\usepackage{fontspec}
\usepackage{polyglossia}
\setmainlanguage[]{english}
\setotherlanguage{hebrew}
\setmainfont{Latin Modern Roman} 
\newfontfamily{\hebrewfont}{Narkisim}
\newfontface\heblatrans[Mapping=my-hebrew-to-latin,FakeSlant=0.3]{Latin Modern Roman} % defining font and mapping for the translitteration.
\DeclareTextFontCommand{\texthlt}{\heblatrans}
\begin{document}
\section*{Hebrew to latin transliteration}
Hebrew is written right to left, and the hebrew script consists of consonants such as \texthebrew{ידלת} (ּ\textit{translit.} \texthlt{ידלת}). Vowels are added below the consonant, such as \texthebrew{ָ ַ ְ } (\texthlt{ָ ַ ְ }). When written with a hebrew keyboard each consonant is usually typed first, then the vowel (if any). The last vowel in the previous example \mbox{(\texthebrew{ְ } )} is called \emph{shva}. In transliteration it should in certain cases be left out, namely when
\begin{enumerate}
    \item A \emph{shva} is on the last letter (consonant).
    \item A \emph{shva} is the second of two \emph{shva}s at the end of a word.
\end{enumerate}

\begin{description}
    \item[Hebrew text] \texthebrew{יָדַלְתְּ}
    \item[My transliteration] \texthlt{יָדַלְתְּ}
    \item[Wanted transliteration] \textit{yād̲alt}
\end{description}
My transliteration scheme uses a custom \verb*|Mapping| in \verb*|polyglossia| in which I simply translate single hebrew unicode characters or sequences of such to latin characters in a \verb*|.map|-file, which is later compiled with \verb*|teckit_compile| by SIL and amended by the \verb*|Mapping| function in \verb*|polyglossia|. I am struggling to implement rule 2 above, as it is dependent on the existence of a \emph{shva} on another previous consonant. Since the mapping file is based on sequences of unicode numbers, I would in practice have to list every combination of \emph{shva}, consonant and \emph{shva} that there is. Is there a trick to avoid this, and instead somehow make reference to any existing \emph{shva} on the previous consonant, i.e. two unicode characters backwards?
\end{document}

my-hebrew-to-latin.map

; TECkit mapping for TeX input conventions <-> Unicode characters

LHSName "hebrew-to-latin"
RHSName "UNICODE"

pass(Unicode)

; ligatures from Knuth's original CMR fonts
U+002D U+002D           <>  U+2013  ; -- -> en dash
U+002D U+002D U+002D    <>  U+2014  ; --- -> em dash

U+0027          <>  U+2019  ; ' -> right single quote
U+0027 U+0027   <>  U+201D  ; '' -> right double quote
U+0022           >  U+201D  ; " -> right double quote

U+0060          <>  U+2018  ; ` -> left single quote
U+0060 U+0060   <>  U+201C  ; `` -> left double quote

U+0021 U+0060   <>  U+00A1  ; !` -> inverted exclam
U+003F U+0060   <>  U+00BF  ; ?` -> inverted question

; additions supported in T1 encoding
U+002C U+002C   <>  U+201E  ; ,, -> DOUBLE LOW-9 QUOTATION MARK
U+003C U+003C   <>  U+00AB  ; << -> LEFT POINTING GUILLEMET
U+003E U+003E   <>  U+00BB  ; >> -> RIGHT POINTING GUILLEMET

; hebrew to latin consonants
U+05D0 <> U+02BE ; alef
U+05D1 <> U+1E07 ; vet
U+05D1 U+05BC <> U+0062 ; bet
U+05D2 <> U+1E21 ; gimel
U+05D2 U+05BC <> U+0067; gimel dagesh
U+05D3 <> U+1E0F; dalet
U+05D3 U+05BC <> U+0064; dalet dagesh
U+05D4 <> U+0068; he
U+05D4 U+05BC <> U+0068; he dagesh
U+05D5 <> U+0077; vav
U+05D6 <> U+007A; zayin
U+05D6 U+05BC <> U+007A; zayin dagesh
U+05D7 <> U+1E25; chet
U+05D8 <> U+1E6D; tet
U+05D8 U+05BC <> U+1E6D; tet dagesh
U+05D9 <> U+0079; yod
U+05D9 U+05BC <> U+0079; yod dagesh
U+05DB <> U+1E35; chaf
U+05DA <> U+1E35; chaf sofit
U+05DB U+05BC <> U+006B; kaf
U+05DA U+05BC <> U+006B; kaf sofit
U+05DC <> U+006C ; lamed
U+05DC U+05BC <> U+006C ; lamed dagesh
U+05DE <> U+006D; mem
U+05DE U+05BC <> U+006D; mem dagesh
U+05DD <> U+006D; mem sofit
U+05E0 <> U+006E; nun
U+05E0 U+05BC <> U+006E; nun dagesh
U+05DF <> U+006E; nun sofit
U+05E1 <> U+0073; samek
U+05E1 U+05BC <> U+0073; samek dagesh
U+05E2 <> U+02BF; ayin
U+05E4 <> U+0070 U+0304; fei
U+05E4 U+05BC <> U+0070; pei
U+05E3 <> U+0070 U+0304; fei sofit
U+05E3 U+05BC <> U+0070; pei sofit
U+05E6 <> U+1E63; tzadi
U+05E6 U+05BC <> U+1E63; tzadi dagesh
U+05E5 <> U+1E63; tzadi sofit
U+05E7 <> U+0071; kuf
U+05E7 U+05BC <> U+0071; kuf dagesh
U+05E8 <> U+0072; reish
U+05E8 U+05BC <> U+0072; reish dagesh
U+05E9 U+05C1 <> U+0161; shin
U+FB2A <> U+0161; shin alt
U+05E9 U+05C1 U+05BC <> U+0161; shin dagesh
U+FB2C <> U+0161; shin dagesh alt
U+05E9 U+05BC U+05C1 <> U+0161; shin dagesh 2
U+05E9 U+05C2 <> U+015B; sin
U+05E9 U+05C2 U+05BC <> U+015B; sin dagesh
U+05E9 U+05BC U+05C2 <> U+015B; sin dagesh 2
U+05EA <> U+1E6F; tav 
U+05EA U+05BC <> U+0074; tav dagesh 

;hebrew to latin vowels
U+05B0 <> U+0115; shva
U+05B1 <> U+0115; chataf segol
U+05B2 <> U+0103; chataf patach
U+05B3 <> U+014F; chataf kamatz
U+05B4 <> U+0069; chirik
U+05B5 <> U+0113; tzeire
U+05B6 <> U+0065; segol
U+05B7 <> U+0061; patach
U+05B8 <> U+0101; kamatz gadol
U+05B9 <> U+014D; cholam
U+05BB <> U+0075; kubutz
U+05D5 U+05BC <> U+00FB; shuruk
U+05D5 U+05B9 <> U+00F4; full holem
U+05B4 U+05D9 <> U+00EE; hireq yod
U+05B4 U+05D9 U+05BC <> U+00EE U+0079; hireq yod2
U+05B8 U+05D4 <> U+00E2; final qamets he

;hebrew diphtongs
U+05B5 U+05D9 <> U+00EA; tzeire yud
U+05B6 U+05D9 <> U+00EA; segol yud
U+05B7 U+05D9 <> U+0061 U+0069; patach yud
U+05B7 U+05D9 <> U+0061 U+0069; patach yud
U+05B8 U+05D9 U+05B0 <> U+0101 U+0069; kamatz gadol yud
U+05B8 U+05D9 <> U+0101 U+0069; kamatz katan yud
U+05B9 U+05D9 U+05B0 <> U+014D U+0069; cholam yud
U+05B9 U+05D9 <> U+014D U+0069; cholam yud
U+05BB U+05D9 U+05B0 <> U+0075 U+0069; kubutz yud
U+05BB U+05D9 <> U+0075 U+0069; kubutz yud
U+05D5 U+05BC U+05D9 U+05B0 <> U+00FB U+0020; shuruk yud

;hebrew qamets chatuph
;U+05B8

;hebrew misc symbols
U+05BE <> U+002D; maqaf (hyphen) 
U+05BC <> U+2060; dagesh alone replaced with word joiner
U+05C3  <> U+2060; sof pasuq replaced with word joiner

;hebrew to latin accents
U+0591 <> U+2060; accent replaced with word joiner
U+0592 <> U+2060; accent replaced with word joiner
U+0593 <> U+2060; accent replaced with word joiner
U+0594 <> U+2060; accent replaced with word joiner
U+0595 <> U+2060; accent replaced with word joiner
U+0596 <> U+2060; accent replaced with word joiner
U+0597 <> U+2060; accent replaced with word joiner
U+0598 <> U+2060; accent replaced with word joiner
U+0599 <> U+2060; accent replaced with word joiner
U+059A <> U+2060; accent replaced with word joiner
U+059B <> U+2060; accent replaced with word joiner
U+059C <> U+2060; accent replaced with word joiner
U+059D <> U+2060; accent replaced with word joiner
U+059E <> U+2060; accent replaced with word joiner
U+059F <> U+2060; accent replaced with word joiner
U+05A0 <> U+2060; accent replaced with word joiner
U+05A1 <> U+2060; accent replaced with word joiner
U+05A2 <> U+2060; accent replaced with word joiner
U+05A3 <> U+2060; accent replaced with word joiner
U+05A4 <> U+2060; accent replaced with word joiner
U+05A5 <> U+2060; accent replaced with word joiner
U+05A6 <> U+2060; accent replaced with word joiner
U+05A7 <> U+2060; accent replaced with word joiner
U+05A8 <> U+2060; accent replaced with word joiner
U+05A9 <> U+2060; accent replaced with word joiner
U+05AA <> U+2060; accent replaced with word joiner
U+05AB <> U+2060; accent replaced with word joiner
U+05AC <> U+2060; accent replaced with word joiner
U+05AD <> U+2060; accent replaced with word joiner
U+05AE <> U+2060; accent replaced with word joiner
U+05AF <> U+2060; accent replaced with word joiner

my-hebrew-to-latin.tec (generated by teckit_compile of my-hebrew-to-latin.map)

zQmp  "ðxÚíYIpUE½oJ@!$@~H„A‘1„€€œ¨²
‰@AŠÂT
Å‚….P£¢¢‚¢¢8 2$•º`ÁBF™D&Rå‚…UH}Ïíwþÿ_Q¬4]uþ;·O÷}Ýýú
÷þEe5âŠ#"O›_… FâÖ `¢Ñ¥ÃE3«*–VV-]X´°jvÍœÅK*+ªµeêô©“ÆM+/Ó‹J¦‹§n:ÔJXà#uPh+Ú¬iÛè›óD¯;Q‰þ‹k¦ÞZnT’¡2L†Ë)£¤X–-cä+ã¤TÆËy»t’L–ÇdŠ”ÉT™&Ëò¤<%å~ìÀÁâKo¹WúH )ð™*â÷ø9†üéÙûIºt”é„]ž)ù’%¥«t‘l)nrŸt—Ò_r¤§\é%ìþ¼Ûº>÷Hß8ogiæ>þóÒ^oQi÷ÏÞ¶uƒÞq¥(j]†»¡\¢xF]ñ£)’õ$
v:Ž™@ÔwzDÉr^@ÈC}o Ð( ú…À@`ð 0
FÅ@ 0”€‰ÀdœsŠH$¯ç"¹údY à%ïî„=/û
¢3·½3-Üq‘ò¦ôä]PÌï—™@€oÑï’@#°8\ÅpœŽ@0À:8Ï 5ÀÀj ýôsö G€sÀÕð“ÉE_7à[Ç<`<.Îë®^Ö €-À·À Îï^~ñðüö²€> üxðãaÆïÍž0sðÖ›ŒÇÃ<<øñàÇû]ÄO²ÌÅÇ»Í/0ãñ1ãñ1ãñ1¿Ø«ßd£ üË5 ‚:1Më,»^×Ï¡
þœg¸Å€îY®U;3l\v…>
«NÏ¿+©n7ÛÍHj·[ë\ËŸö³ìÝì£sÛÂyé® ËÙÎÔkçë#•ìS—XG¿‰+ÙGëÂ>ãZíbmè§Nmú©·ü”ÑO¼>VÇ~õj³_ƒu-–Yý8æeV¿µ¹Öù
½—Úàݨ•+`÷¤¦<›Z^hËzÚ£iï¤]˜aî­H.íâŒð>ÛD»D»ë´ëý,ǸϑTjû©Í¥¶ZµÔfS; -ÚAjó¨éu˜Gí0µÔ+§v$ÔrP;¢œZl_,§vÔÚǩͧvý¢v,I;fi'¨UQÓgMµ“Ôbc9©œÚ)jÕÔN)§všÚj§•S;G­Æ ÷È9åx&øÔÏ&ég“ôó\ŸYô}^9µì»ˆÚåÔ.R[LMcÅÅÔš¨-¥ÖŸ©må:7QÛŠC&µmIÚ6ð9ÜûÛ©UX÷C³:öÙ®v"&Íj“ˆIÛ
mEž¹í÷À¾‰¯ã•öa›nÿár?w†Ý¦Ï*Qš >Ó(¯+b´Á8æ1Ô|E"·¾xi,Øèe¾n^éÐB]Æ
útâ1ý&EÍÍËPí¨°yÔ›jq܈›ÿºhô<ð6.ÌIµTX¼kdqËK›xìŸ&÷0§ÐN:Þ#ÞzqZË]Z2¥ß6¢¬$;ûoôéÂcÎ
ÚEná¸ó[¨Ëe^ µüÝœ”grRmM^Ê“tÔuŠ:’èV>Žz­Ëqì
dÝ€î€æ¬ðUü ìÓüŽû5‡…ã“Çrõú‡4…ã-æ²<Ö<Úœ×\Ö`T4‡aE}ü×<ø×À7bbx'‚mÐ&“4¿{G4Os\¦täÔoŠ1À3G+ÕÀZ@s{ 1/ó¯±4 óD¥@9óD«€Z`€~Î÷ÀA+O¤)ÜO.úºC¬<ÑL扞gž¨–y"Äü.b'w/óD™'ò™'ÂØ=ìgñ·7x˜
hL‡±x/k€ Æãa<Æã!žô®†Ÿy>üø¹øXcñ±þRà –9"ô÷¿Õ<€¸ÙÇœü_ß±¦ˆ¡ø    ôzb<|X— sà/€¿ þŒ)À˜Œ)€OÍihl`}5†×87À˜c
—‚ë“‚=”R a~¡&ÃÌ?8—dŸ
cZg5î
-û¬ÚhïÁ¾,çþ8jµ»«ãI© *ÅÄ-×1wR¡vèÛq¨×Ñ÷æ°_óú°Ÿæ‰‚:ö+£^ŸÔ¯Y=û•©Í¹,£Þм_ózŽs™Úã‚_ËðÞãú›¾\‡k±úÐŽdòß„"Çœ3’³Y§mrø¼Í§­Ñ”æŒÊés½‰¿öNðÑ    ;{.RßÄÜm1mÍkiN©„ö³bòHñk7WLî(nÏ“/2¶Î]÷ßAjó¸×bÚ
ðÃÔOh¹À„ZÈ­~˹¶nŒ[ý}Ç-~Œíæ‹ÉÅÇ©ö1Ë'žÁ   j†[šžÿ$5Ó´ŸÈ«õ>b»jÞS–ö3ùðÓlg¸5þYàç9oÓ´_ècøú0Ü:×bð‹ÔOÌ[ÿ?.»¸§a_ë–ðeôÂ$½‘ýÕßÖíf›¸Mv]#Ûé¸ô™Ôd§3ïƒÿŠí®òž¨§}†÷rlO5ñ]sÔ²·%î=ý_"vÿ³øËÅâ¯Z¼Öâ¯Yüu‹¿añ5ÓâoYüm‹¯µø;×âïY|Å×[ü}‹`ñ
ÿÐâYüc‹o´ø'ÿÔâ›,þ™Å?·øÿòÂ[Kd

Answer 1

I found the best solution to be outside of LaTeX. I instead parse the .tex file in a python script, and then compile this document.

See python script below:

# This python script takes in a string (either word or sentence(s)) and removes the sheva on the last consonant, and second last consonant (if there is a sheva on the last consonant). It does so irrespective of any dagesh which might throw it off.
import re

def remove_sheva(text):
    # Defining dagesh and sheva
    sheva = '\u05B0'
    dagesh = '\u05BC'

    # Splitting text in single words
    words = text.split()

    # Treating each word separately
    new_words = []
    for word in words:
        # Finding indicies for all consonants in the word
        consonant_indices = [m.start() for m in re.finditer(r'[\u05D0-\u05EA]', word)]

        # If no consonants in the word, adding them in list over nye words without any change
        if not consonant_indices:
            new_words.append(word)
            continue

        # Finding indicies of last and second last consonant.
        last_consonant_index = consonant_indices[-1]
        second_last_consonant_index = consonant_indices[-2] if len(consonant_indices) > 1 else None

        # Checking for any sheva after last consonant
        sheva_after_last_consonant = False
        if last_consonant_index + 1 < len(word) and word[last_consonant_index + 1] == sheva:
            sheva_after_last_consonant = True
            word = word[:last_consonant_index + 1] + word[last_consonant_index + 2:]
        elif last_consonant_index + 2 < len(word) and word[last_consonant_index + 1] == dagesh and word[last_consonant_index + 2] == sheva:
            sheva_after_last_consonant = True
            word = word[:last_consonant_index + 2] + word[last_consonant_index + 3:]

        # Removing sheva between the two last consonants, but only if there is a sheva on last consonant.
        if second_last_consonant_index is not None and sheva_after_last_consonant:
            if second_last_consonant_index + 2 < len(word) and word[second_last_consonant_index + 1] == dagesh and word[second_last_consonant_index + 2] == sheva:
                word = word[:second_last_consonant_index + 2] + word[second_last_consonant_index + 3:]
            elif second_last_consonant_index + 1 < len(word) and word[second_last_consonant_index + 1] == sheva:
                word = word[:second_last_consonant_index + 1] + word[second_last_consonant_index + 2:]

        new_words.append(word)

    # Joining words again to create sentence.
    new_text = ' '.join(new_words)
    return new_text