Bijlage 9 ArabTEX — Typesetting Arabic with Vowels and
Contents
1. 539 548 7 Hasim80 3 AJ ktl ael olkd ace ala HASIM MUHAMMAD AL HATTAT Qawd id al Hatti al Arabi Maktaba an Nahda Baghdad Dar al Qalam Beirut 1400 1980 ISO R233 ISO R 233 1961 International System for the Transliteration of Arabic Characters International Standards Institution 1961 Knuth84 Donald E KNUTH The TEXbook Volume A of Computers amp Typesetting Addison Wesley Reading Mass 1984 Knuth and MacKay87 Donald E KNUTH and Pierre A MACKAY Mixing right to left texts with left to right texts TUGboat Volume 8 1987 pp 14 25 Lagallyg2 Klaus LAGALLY ArabTEX a System for Typesetting Arabic User manual Report 6 92 Fakult t Informatik Universitat Stuttgart 1992 Lamport86 Leslie LAMPORT BIEX a Document Preparation System Addison Wesley Reading Mass 1986 MacKay77 Pierre MACKAY The KATIB System a revolutionary advancement in Arabic Script Typesetting by means of the Computer in Scholarly Publishing 8 2 Toronto 1977 pp 142 150 Schimmel70o Annemarie SCHIMMEL Islamic Calli graphy E J Brill Leiden Netherlands 1970 115 Bijlage 9 Appendix Installing ArabTEX ArabTEX uses no preprocessor and thus should be com patible with any TEX implementation that allows dynamic loading of additional macro files and fonts The ArabTEX distribution consists of the following com ponents o TEX macro files with extensions sty and tex these files are inst2. US ed dG wa qabla an yutimmu guha kalamahu badaa l him ru yanhaqu fi stablihi DENN EC EON E NE fa qala lahu sadiquhu Me Iie Cal tf fa q la lahu guha garibun amruka ya sadiqi atusaddiqu l him ra wa tukaddibuni en pae l KiS kdl Sii l gas amp d Figure 4 Arabic text with transliteration Acknowledgments The development of ArabTeX would not have been pos sible without the assistance of many people Apart from my local team helpful advice came among others from Ivan Derzhansky Wolfdietrich Fischer Ahmed El Hadi Ab delsalam Heddaya Iqbal Khan Tom Koornwinder Eber hard Krueger Asif Lakehsar Jan Lodder Richard Lorch Eberhard Mattes and Bernd Raichle I also have to thank the many users who sent bug reports and comments Voorjaar 1998 Bijlage 9 References DIN31635 DIN 31 635 Umschrift des Arabischen Alphabets Deutsches Institut f r Normung e V 1982 Endress82a Gerhard ENDRESS Die Arabische Schrift in Fischer82 p 165 ff Endress82b Gerhard ENDRESS Handschriftenkunde in Fischer82 p 271 ff Fischer82 Wolfdietrich FISCHER ed Grundrif der Arabischen Philologie Band 1 Sprachwissenschaft Dr Ludwig Reichert Verlag Wiesbaden 1982 Fischer87 Wolfdietrich FISCHER Grammatik des Klassischen Arabisch 2 Auflage Verlag Otto Har rassowitz Wiesbaden 1987 Haralambous9g1 Yannis HARALAMBOUS TEX and Those Other Languages TUGboat Volume 12 1991 pp
3. a subword a character might take on one of several shapes the isolated initial medial and final shape This forms might still be modified if the character enters into a ligature Fortunately as far as ligat ures are concerned all characters of a class perform alike thus the number of different cases although large remains manageable When we process a word we perform the following steps o we sequentially process the input representation to break it up into a sequence of individual characters each with accompanying diacritical information o we process this sequence in reverse order to determine the shape of each character depending on its position in the subword and on the surrounding characters o starting on the baseline we position these character shapes so that they join smoothly either directly or by means of connecting strokes To each character we add the appropriate diacritical marks there may be none or even more than one per character For an example see Figure 1 n Whenever the next character considered this is the preceding one when writing by hand cannot be joined to its logical successor we have reached a subword boundary we reposition this character so it will again sit on the baseline and add suitable spacing The resulting graphical representation of the word is passed back to the caller to be inserted into the output 5 User Interface In the following we shall only describe the main fea tu
4. has to be suppressed wasla guha wa himaruhu rata sadiqun ila uh yatlubu minhu himarahu li yarkabahu fi safratin qasiratin wa q la lahu sawfa gt u sduhu ilayka ft l masasi wa adfahu laka u ratan fa qala guha an asifun giddan anni l astati u an suhaqqiqa laka ragbataka fa lhimaru laysa hun l yawma wa qabla an yutimmu uha kalamahu bada a l himaru yanhaqu ft stablihi fa qala lahu sadiquhu inni asmau him raka y uh yanhaqu fa q la lahu guha aribun amruka ya sadiqi atusaddiqu l him ra wa tukaddibuni Figure 2 Arabic transliteration 6 4 The Parser Module The Parser Module has to break up the token sequence into a backward sequence of writing syllables A writ ing syllable is not to be confused with a syllable in the usual sense but consists of a single consonant or long vowel with additional diacritical information denoting e g a short vowel consonant doubling tanwin and hamza Whereas the basic algorithm is straightforward there is a surprisingly large number of special cases since the various languages supported by ArabTEX have different notational conventions and there are also some options not described here to locally modify the writing A typical example is 113 Bijlage 9 the handling of hamza the glottal stop Whereas denoting a distinctive sound it is not considered a letter and thus a carrier for it has to be determined which depends on
5. just a single letter we also use that letter n whenever the transliteration uses a letter with a diacritical mark we use the same letter and precede it 110 Klaus Lagally with the punctuation mark most closely resembling the diacritic This is easily remembered fairly readable and works well because punctuation marks except hyphen never occur within a word Using this coding scheme we get an additional bonus if for some reason we want to also typeset the standard transliteration of an arabic word we have to code the dia critical marks used and whereas this can be done in TEX using existing commands these look awkward and are not easy to learn and remember On the other hand it turned out not to be too difficult to derive the transliteration from our coding scheme and so we can use it for both purposes thereby avoiding the danger of constantly confusing two closely related but different notations In fact the description we gave is somewhat oversimpli fied There are some fortunately rare exceptions to the transliteration rules and sometimes words written differ ently are transcribed identically so in these cases we have to code additional information 4 Processing Arabic Text In the following we give a general overview of the tasks our system has to perform when typesetting Arabic We discuss this in the context of a simplified model viz that a text as seen by TEX is a sequence of paragraphs each of wh
6. on the other hand there exists a language independent standard the Interna tional Phonetic Alphabet Transliteration on the other hand means representing the writing of the given language by using a different set of characters In theory just a unique representation is needed in practice it is also required that the translitera tion be easily readable and also give some indication of the sounds Therefore some compromises are usually made with the consequence that deducing the writing from the transliteration requires some knowledge of the language in question For Arabic and some other languages using the arabic 109 Bijlage 9 script there exist two nearly identical international stand ards DIN31635 ISO R233 for transliteration in the given loose sense As there are more arabic letters than in the Latin alphabet these conventions make heavy use of dia critical marks and so we cannot use them directly for our purpose 3 2 Input notation If we want to typeset arabic texts with TEX we have two possibilities o either have a preprocessor transform our input text into some intermediate notation that can be processed by TEX o or enhancing the power of TEX by adding suitable macros so that it can process our input text directly The first possibility is extremely flexible as far as the possible input codings are concerned and can be made very efficient It has been used in some existing systems e g Scholar
7. reasons quotations have to be very short anyway Longer arabic passages are handled differently Here we process the individual words in their natural order ar range the results in reverse order and do the line breaking ourselves Inside an arabic paragraph we can again have insertions e g short quotations now of non arabic text or even in line mathematical formulas For the same reas ons as above we have again to forbid line breaks inside an insertion In both cases we have to take care of the fact that num bers in Arabic are written like in the european languages i e the sequence of digits is not reversed We could have put the responsibility for indicating what is considered to be a number on the user however we decided just to define a number as a sequence of characters starting with a digit and ending with a space and to typeset this sequence in the natural order 4 2 Numbers Words Subwords As we saw every arabic word or number is processed in dividually and the result is a description of its graphical representation given in terms of symbols from a given font arranged in a two dimensional pattern There is no unique correspondence between these symbols and arabic charac ters a character image might be built up from several sym bols and it also sometimes happens that a symbol repres ents more than one character The reason behind this is that the arabic characters may be collected into several classes whose mem
8. Bijlage 9 ArabTgX Typesetting Arabic with Vowels Klaus Lagally Universitat Stuttgart Institut f r Informatik BreitwiesenstraBe 20 22 D 70565 Stuttgart lagally informatik uni stuttgart de abstract We present a TEX macro package for generating the arabic writing from a standardized ASCII input notation It can handle partial or full vocalization and generates automatically most of the common ligatures There is limited support for Farsi Urdu and Pashto ArabTEX is compatible with Plain TEX and also most IATEX environments arabic and other material can be mixed freely For special purposes the standard transliteration can be additionally generated ArabTEX uses no preprocessor and thus should be compatible with any TEX implementation that allows dynamic loading of additional macro files and fonts 1 Introduction This is a personal story The author interested in the arabic language since he was a young boy some time ago by acci dent found out about an evening course on Arabic at a local school and decided to join in The course was designed for people wishing to visit an arabic country with some knowledge of Arabic and as the teacher would not recom mend any suitable and easily affordable textbook for that purpose he handed out his own handwritten notes This intrigued the author and so he bought an arabic grammar book from a renowned publisher Fischer87 Upon closer inspection the arabic examples looked somewhat str
9. TgX Haralambousg1 However every user now needs a version of the preprocessor tailored to her his computer system and cooperating well with the local TEX implementation Thus we may run into portability and maintenance problems and possibly a complicated install ation procedure The second possibility which we adopted by itself is as portable as TEX itself is but writing the needed algorithms in TEX macro language is no easy task and the macros might not run as efficiently as a preprocessor system Like everywhere here also is a tradeoff between generality and speed If as we did we choose the macro solution then TEX must be able to read our input notation directly therefore we should better use only the standard 7 bit ASCII char acters there are extensions to TEX using 8 bit characters but these are in no way standardised so we could run into severe compatibility problems We want to keep the input notation easily readable but we have the problem that we need about 30 different letters and some of them sound very much alike Even when also using the capital letters for coding Arabic needs no capitals we could not find a one to one correspondence between ASCII characters and arabic sounds that is easy to read and remember The solution we finally found was to use both one character and two character encodings and to adhere closely to the standard transliteration The rules are simple o whenever the transliteration uses
10. alled on the TEX input path for source files o Font metric files extension tfm and compressed pixel files extension pk for the fonts nash14 and nash14bf at several common magnification steps Installation of these files is strongly system dependent in case that they cannot be used the METAFONT sources are also available extension mf to rebuild the fonts locally n installation notes user manual answers to questions demos and the like ASCII and or TEX files for local printing The system is available from the author s institution an onymous FTP from ftp informatik uni stuttgart de directory pub arabtex from the CTAN archive and also from many other common servers At the time of this writing version 3 06h is current ArabTEX is copyrighted but free use for scientific ex perimental and other strictly private noncommercial pur poses is granted We appreciate receiving a compliment ary copy of serious scientific work using ArabTEX for our private collection Space and time requirements are not negligible how ever ArabTEX has been used frequently and successfully even on a PC XT standard configuration 116 Klaus Lagally Post Scriptum 1998 The above report was originally written in 1992 and has been presented at the EUROTEX 92 conference at Prague but was not widely circulated at that time When we were asked for an update to cover the present state we found that nearly everything described above is
11. and on covering the complete Ar abic segment of Unicode MAPS
12. ange and after contacting the author of the book it turned out that the latter had added the vowel signs to the arabic examples on the printing plates by hand This came as a great surprise especially when consider ing the fact that the underlying printed arabic text looked beautiful Apparently there remained some unsolved prob lems in the printers business and knowing the power of TEX Knuth84 the author decided to try doing something about it The result of that effort is now called ArabTEX a system consisting of a large macro package and several fonts 108 and Ligatures 2 Design goals The typical user of ArabTEX as we imagine her him o knows some Arabic n is interested in high quality writing o has little money to spare o cannot afford specialized equipment o is willing to learn some simple rules but o is not and is not willing to become a TEX expert This description fits well onto several linguists we know Alas not every one of them can even afford a simple PC From this projected user profile follow some require ments for the system it should be inexpensive it should not require specialized equipment it should be easily portable it should be sufficiently powerful to generate any reasonable arabic text with high quality o it should after some training be usable by a person who is not a computer expert nomunuu However o it need not be extremely efficient o it need not suppo
13. bers are closely related and differ only in a few features that can be separated out Fortunately the same is true for the ligatures and we can also handle the vocaliz ation by the same mechanism so that a single font of less than 256 characters is sufficient for expressing a much lar ger set of graphical symbols and combinations When we want to typeset a number in the arabic script we just arrange the isolated graphical symbols correspond ing to the digits from left to right and we are done Typesetting a word of text is more involved Logically a written word consists of a sequence of character images connected to each other as far as possible and possibly changing their shape depending on the context In addi tion these character images may carry diacritical marks Not all characters can be joined to their successors prob ably because the writing would become ambiguous other wise and thus we can consider a word being a sequence of subwords whose characters are all connected To each subword corresponds a graphical representation and these are arranged side by side In this step they are possibly dis Voorjaar 1998 Bijlage 9 placed vertically such that their last 1 e leftmost charac ter has its normal position on the baseline and horizontally such that their spacing looks pleasant porum A 0 Figure 1 Character assembly with components shown 4 3 Characters ties diacritics Depending on its position in
14. cter on a common baseline but try to join adjacent let ters into a softly flowing curve This makes for ease of writ ing and also for aesthetic beauty but has the consequence that the script although still arranging the individual words in a horizontal sequence is essentially two dimensional Another consequence is that the form of a letter depends on the context and if adjacent letters are combined into ligat ures a surprising manifold of different forms may emerge Most of these are not mandatory but their omission will lead to a serious loss of quality that can easily be noticed even by an outsider and quality has always been considered very important A script of that characteristic is not very convenient to print and indeed the arabic script has resisted mechan ization for a long time Endress82a The first attempts to print Arabic with movable type were undertaken about 1500 A D surprisingly in central Europe but the printing tradition of Arabic seriously started in 1727 when the Ot toman printing agency in Istanbul was founded It had the types made in the Netherlands where the technology existed and for several decades only official documents and scientific works were allowed to be printed Religious works like the Qur an and its commentaries still were re produced by hand writing and later by lithography from hand written originals thus the risk of misprints in the Holy Scriptures was avoided A second official prin
15. h is finished by the normal TEX paragraphing mechanism For an example see Figure 4 The Word Processor passes the input to the Scanner Module ascan sty to generate a standardized internal representation independent of the external coding This in ternal representation is then passed to the Transliteration Voorjaar 1998 Bijlage 9 Module atrans sty if the transliteration is wanted Oth erwise or additionally it is passed to the Parser Module aparse sty to isolate the individual graphical compon ents The output of the Parser Module is further processed by the Assembly Module awrite sty to generate the ar abic representation 6 2 The Scanner Module The main task of the Scanner Module is to break up the input stream into tokens denoting individual arabic charac ters should the input notation be changed then only the Scanner Module would have to be adapted accordingly There is one case handled in a special way for hamza the character preceding it is repeated after it to ease further pro cessing 6 3 The Transliteration Module This module has to transform the sequence of tokens into the external representation of the standard transliteration As the transliteration does not always follow the arabic writing closely some special cases have to be considered e g in connection with endings and with the definite art icle whose spelling depends on the first consonant of the following word Also sometimes an initial vowel
16. ich is a sequence of words TEX will transform each word into an internal representation and will arrange these word images into lines The sequence of lines thus gener ated will be broken up into pages which will be sent to a device independent output file later to be viewed or prin ted by a device dependent driver program There is indeed much more to it but the details are not relevant to our ex position 4 1 Overall structure Quotations Paragraphs If we want to typeset a document containing arabic text we will distinguish two different cases o short arabic quotations inside a line of text in some european language o longer arabic passages consisting of one or several paragraphs An in line quotation is handled as a whole We process the arabic words in reverse order one word at a time and insert the results into the normal output This could lead to prob lems if a quotation would be split across a line boundary because in that case the two parts should be individually reversed We ought to do the line breaking first and the reversal afterwards but we know of no easy way of doing MAPS ArabTEX Typesetting Arabic with Vowels and Ligatures that with TEX To handle this problem an extension of TEX TEX XET has been proposed Knuth and MacKay87 but it is not generally available and also not compatible with the standard printer driver programs So we have to forbid line breaking within a quotation and for technical
17. it is joined to its left neighbour either directly or by means of a connecting stroke whose form depends on the partners As the connection point of its left neighbour need not be on the baseline the skeleton possibly must be vertically adjusted and a new connection point for its right neighbour if that exists will be determined o After positioning the skeleton the modifier will be added to identify the character in question u Finally the diacritical information is added 6 6 The Ligature Module This module is called by the Assembly Module for each character It will receive as input information a description of a skeleton shape and the shape of its right neighbour and will return a possibly changed skeleton shape a possibly changed shape of the right neighbour and frequently also a connecting stroke With the exception of very few but im portant cases where ligatures are mandatory the Ligature Module might return its input information unchanged and indeed there is an option to switch most ligatures off How ever the art of forming ligatures evolved gradually during many centuries of writing and their inclusion will greatly improve the quality of the result and whereas a good many cases are handled already there is still room for improve ment 7 Experiences One of the reasons for implementing ArabTEX this way was to test the power of TEX on a large example We found that it could be done but we drastically underesti
18. l or expressed by diacritical marks placed above or below the characters A native speaker generally does not need this additional information as he can deduce it from the context it is only required when introducing new words for resolving ambiguities and in religious texts where the exact pronunciation is considered important Considering the already very large number of different letter forms in a printer s box also storing all the possible combinations would be prohibitively expensive and thus manual correc tions are necessary This is awkward and expensive so it is avoided whenever possible and thus the religious texts we have seen all have been reproduced from manuscripts 3 1 Transcription and Transliteration If we want to generate the arabic writing of a given text automatically we have to denote the text in a way that can readily be processed by our computer There exists no standard suitable for our purpose so we have to in vent one and since linguists always had related problems and also are among our prospective users we try to imitate their solutions as closely as possible In this context there exist two concepts that are closely related and therefore frequently confused transcription and transliteration Transcription means representing the sounds of the given language as closely as possible This can even be done in the language itself e g transcribing the sound of the english word enough as enuff
19. mated the amount of work involved The techniques used in the de scribed modules are comparatively straightforward even the full power of context free language analysis is rarely needed However due to the great number of special cases the complexity is considerable and the macro technique used is extremely vulnerable to trivial coding errors whose effects will propagate throughout the system very quickly and frequently will lead to very puzzling results Thus sys tematic structuring is a must and a complete redesign after having a working prototype payed off very well and led to a considerable increase of stability There are still some errors in the system but they seem to be well hidden and show up at a surprisingly low rate Furter plans besides correcting errors are designing a Nasta liq font that looks better for Persian and generally improving on the still very rudimentary support for non arabic languages using the same script MAPS ArabTEX Typesetting Arabic with Vowels and Ligatures ales E guha wa himaruhu sata sadiqun ila uha yatlubu minhu himarahu li yarkabahu ft safratin Pi GA le amp Il G Jb i qasiratin wa q la lahu s ix sawfa usduhu ilayka fi l masai wa adfahu laka u ratan a a clad 3 adl dsl Gye 1 on c fa qala guha om we 2 m Jus rand asifun giddan anni l astati u an uhaqqiqa laka ragbataka mk Oh J nani J fa lhimaru laysa hun l yawma ps
20. or normal text or in local macros This special behaviour is switched on by language selection and can be switched off again by setnormal An arabic paragraph is started by the com mand begin arabtext and ends with end arabtext This looks like and nearly operates like a IATEX envir onment even when working with Plain TEX However neither displayed mathematical text nor other ATEX environments may be nested in an arabic paragraph Inside an arabic paragraph we can have non arabic quo tations delimited by angle brackets and in line mathemat ical formulas delimited by single dollar signs These inser tions must fit on one output line 5 4 Input coding The input notation the arabic writing in the isolated form and the transliteration of the characters used for Arabic and Persian are given in Table 1 For Urdu Pashto and for 112 Klaus Lagally Table 1 Coding of arabic characters special purposes there are some additional codings Note also the following o T is tah marbouta lt N gt is tanwin Y is alif maqsoura o A I lt U gt denote the long vowels lt a gt u the short vowels if required o right quote is hamza glottal stop After setarab its carrier will be determined by the context according to the full hamza rules otherwise by a following short vowel o A generates madda Doubled consonants are written twice shadda n lt gt will break unwanted ligature
21. res for more details see the ArabTEX documentation Lagally92 111 Bijlage 9 5 1 Activating ArabTEX To use the ArabTEX package with a file to be processed by Plain TEX load it via input arabtex with IATEX in clude arabtex as a document style option In both cases several additional files and the default font will be installed 5 2 Mode control As there are several language dependent writing conven tions you have to select a language by one of the com mands setarab setfarsi seturdu setpashto or setverb no special processing in this case There are three different modes of handling short vow els o vocalize short vowels written in the input will be indicated in the output by diacritical marks o fullvocalize also the absence of a short vowel will be indicated o novocalize short vowels will show up in the transliteration but will be omitted in the arabic writing You can locally override this feature By arabtrue arabfalse transtrue transfalse you can switch on and off the generation of the arabic writing and or the standard transliteration By default the arabic writing is on and the transliteration is off Bold face can be selected by setbold setnormal will revert to normal 5 3 Arabic text Short arabic quotations in normal text are included in angle brackets These thus have a special significance outside of mathematical mode and can no more be used for other purposes e g f
22. rt everyday office use o it need not be interactive As it happens our starting point was TEX in fact IATEX Lamport86 and we noticed that there are two quite different populations of TEX users o the experts in full control of all specialized features constantly finding new applications and o the everyday users getting their work done by filling in some forms designed by a expert and letting TEX do the rest Our hypothetical user definitely belongs to the second cat egory Therefore for him it is extremely important to have a convenient user interface Devising such an interface turned out to be a major task MAPS ArabTEX Typesetting Arabic with Vowels and Ligatures 3 Characteristics of the Arabic script The arabic script like the scripts for all semitic languages runs from right to left This fact whereas leading to some complications in connection with line breaking whenever we want to mix arabic and non arabic texts turned out to be an absolutely minor problem in comparison with the fact that the arabic script is a cursive style extremely well adapted to hand writing As far as we know this has al ways been so Endress82b and contrary to common be lief the script is very easy to write even a motivated be ginner can acquire a fair hand writing style within a few weeks Calligraphic excellence of course is a different matter Schimmel7o In a cursive hand we do not assemble character after chara
23. s lt gt joins two words and will only show up in the transliteration and lt gt will elongate the connection between two adjacent letters kashida o The definite article is always written al with hyphen even if it precedes a double sun letter i H 5 5 Special features For Farsi Urdu Pashto and some other languages using the arabic script the coding conventions are slightly different and not described here Furthermore the language specific processing may be locally overridden and there is also a verbatim mode capable of representing unusual or archaic ways of writing Mode changing commands may also oc cur inside an arabic paragraph thus allowing local mode changes MAPS ArabTEX Typesetting Arabic with Vowels and Ligatures 6 Implementation The ArabTEX system consists of a large number of mac ros and their interaction is surprisingly complex They are grouped into several packages each devoted to a separate task As ArabTEX can be considered a translator we im itate the usual modularization of a compiler In that view ArabTEX consists of a Driver Module calling a number of auxiliary modules for specialized tasks and finally passing the output back to the normal TEX paragraph mechanism Thus arabic text can also appear inside most IATEX envir onments including moving arguments However IATEX is no prerequisite for running ArabTEX 6 1 The Driver Module The Driver Module arabtex
24. still true and we had to do little more than updating a few technical details about the distribution Therefore we decided not to rewrite the present report but to concentrate instead on preparing a new edition of the User Manual to cover the many new features added since What has happened in the meantime Home computers have meanwhile become large and fast and even though some modern Operating Systems tend to use up the additional resources very quickly the users will normally no more notice the large amount of processing that goes on within an ArabTEX job Large parts of Ar abTEX have been rewritten several times leading to in creased stability and also enabling many extensions for special purposes The basic user interface has not changed therefore even this report may still be used as a minimal introduction but the system is no more limited to Arabic in transliter ation input Several additional standard encodings are sup ported and the range of languages covered now also con tains Uighuric Old Malay Sindhi and Hebrew in trans literation encoding ISO 8859 8 and the machine readable CCAT format Several critical editions of Arabic manu scripts using ArabTEX have been completed and published and we know of some additional ongoing projects Ar abTEX has been used successfully in conjunction with other packages e g PicTEX EDMAC and Babel A project like ArabTEX is never finished we are still busy on the Urdu mode
25. sty is loaded by TEX or by a small Loader Module arabtex tex when using Plain TEX The latter module simulates the few ATEX features used by ArabTEX The Driver Module when executed defines and ini tializes some common variables and loads the remaining files constituting ArabTEX It also implements the mode changing commands and contains several local submod ules o the Insertion Processor for arabic quotations o the Paragraph Processor for arabic paragraphs n the Output Processor o the Word Processor Both the Insertion Processor and the Paragraph Processor pass single arabic words to the Word Processor to generate the graphical representation and or possibly the transliter ation and process the resulting output further The Insertion Processor breaks up short quotations into individual words and feeds both the resulting arabic rep resentation and the transliteration into the normal output stream The Paragraph Processor also breaks up the input into in dividual words the output of the Word Processor however is now handled differently The transliteration if generated is fed into the normal output stream the arabic representa tion is passed to the Output Processor The Output Processor lines up the arabic representations from right to left in a local buffer Whenever a line is com pleted it is interleaved with the normal output if any At the end of an arabic paragraph the buffer is flushed and the paragrap
26. the context in a rather complicated way PG PA d ia ie e oli Jii vo d 4 d s je Seal bb Bolg sled 9 A A ON ON ARE CAR J e Gi ti pues X E uj al j5s DENN EIE dio d GAS E 34e eel i E Je Figure 3 Vocalized Arabic text 6 5 The Assembly Module Finally from the reversed sequence of writing syllables produced by the Parser Module the graphical representa tion is determined Every writing syllable consists of a basic character and diacritical information Every character belongs to a character class represented by a skeleton and is locally identified by a modifier usually a pattern of dots The further processing of a writing syllable proceeds in several steps o The skeleton and the modifier are determined o Depending on context the appropriate joining form of the skeleton isolated initial medial final is determined o Also depending on the context the skeleton may take part in a ligature and thus get a different shape Generally and with very few exceptions ligature generation is optional and since it is also complicated though not difficult it has been delegated to a separate Ligature Module aligs sty 114 Klaus Lagally o After the definite form of the skeleton has been determined it is positioned in the output If it is an isolated or final shape it is generally put on the baseline with suitable spacing to its left neighbour if any Otherwise
27. ting agency was founded 1821 in Cairo others followed and in 1906 a new typeface standard was adopted with remark ably good results that is still in use today Of the several different writing styles that exist Naskhi was adopted for printing as it is very easily readable and mostly adheres to the baseline Still even printing Naskhi is a formidable task whereas a european printer s box con tains less than 100 different letter forms including capitals digits and special characters you need far more than 500 different forms for good quality arabic printing The situation improved in the 1970 s when photo typesetting equipment became available and the first Voorjaar 1998 Bijlage 9 computer programs to typeset Arabic were developped MacKay77 Now also other writing styles like Nasta liq as used mainly in Iran and the adjacent countries could be handled and many new typefaces e g for newspapers were developped But you can still find headlines which have obviously been reproduced from a hand written ori ginal The calligrapher s profession is still alive see e g Hasim8o Even if the technology for printing arabic texts nowadays exists some problems remain In the Arabic language as in all semitic languages the main information resides in the consonants and the long vowels and usually only these are written explicitly Short vowels the doub ling of a consonant and the like are either not indicated at al
Download Pdf Manuals
Related Search