%% This is part of the OpTeX project, see http://petr.olsak.net/optex \_codedecl \loadmath {Unicode Math fonts <2022-07-22>} % preloaded in format \_doc ----------------------------- \`\loadmath` `{}` loads the given font. It does: \begitems * define \`\_unimathfont` as ``, * redefine `\normalmath` and `\boldmath` macros to their Unicode counterparts, * load the `\_unimathfont` by `\normalmath`, * print information about the loaded font on the terminal, * redefine all encoding dependent setting by `\input unimath-codes.opm`, * protect new loading by setting \`\_ifmathloading` to false. \enditems \`\noloadmath` disallows Unicode-math loading by \`\_mathloadingfalse`.\nl \`\doloadmath` allows Unicode-math loading by \`\_mathloadingtrue`. \_cod ----------------------------- \_newifi \_ifmathloading \_mathloadingtrue \_def\_noloadmath{\_mathloadingfalse} \_def\_doloadmath{\_mathloadingtrue} \_def\_loadmath#1{% \_ifmathloading \_initunifonts \_isfont{#1}\_iffalse \_opwarning{Math font "#1" not found, skipped...}% \_else \_def\_unimathfont{#1}% \_let\_normalmath = \_normalunimath \_let\_boldmath = \_boldunimath \_normalmath \_wterm {MATH-FONT: "#1" -- unicode math prepared.}% \_ifx\_ncharrmA\_undefined \_opinput {unimath-codes.opm}\_fi \_mathloadingfalse \_fi\_fi} \_public \loadmath \noloadmath \doloadmath ; \_doc ----------------------------- \`\loadboldmath` `{} \to {}` defines \`\_unimathboldfont` as `` only if `\_unimathfont` is defined as ``. It is used when \^`\boldmath` macro is run. When no `\_unimathboldfont` is defined then the `\boldmath` macro use \"fake bold" generated by `embolden` \LuaTeX/ font feature. \_cod ----------------------------- \_def\_loadboldmath#1#2\to #3{% \_def\_tmp{#3}\_ifx\_unimathfont\_tmp % do work only if #3 is loaded as normal Math \_isfont{#1}\_iffalse \_opwarning{Bold-Math font "#1" not found, skipped...} \_else \_def\_unimathboldfont{#1}% \_wterm {MATH-FONT: "#1" -- unicode math bold prepared.}% \_fi\_fi} \_public \loadboldmath ; \_doc ----------------------------- The Unicode version of the \^`\normalmath` and \^`\boldmath` macros are defined here as \`\_normalunimath` and \`\_boldunimath` macros. They are using \`\_setunimathdimens` in a similar sense as \^`\_setmathdimens`. \nl You can combine more fonts if you register them to another math families (5, 6, 7, etc.) in the \^`\normalmath` macro. \nl The default value of \^`\_normalunimath` shows a combination of base Unicode-math font with 8bit Math font at family 4. See definition of `\script` macro where `\fam4` is used. \_cod ----------------------------- \_def\_normalunimath{% \_loadumathfamily 1 {\_unimathfont}{} % Base font \_loadmathfamily 4 rsfs % script \_setunimathdimens }% \_def\_boldunimath{% \_ifx\_unimathboldfont \_undefined \_loadumathfamily 1 {\_unimathfont}{embolden=1.7;} % Base faked bold \_else \_loadumathfamily 1 {\_unimathboldfont}{} % Base real bold font \_fi \_loadmathfamily 4 rsfs % script \_setunimathdimens }% \_def\_setunimathdimens{% PlainTeX sets these dimens for 10pt size only: \_delimitershortfall=0.5\_fontdimen6\_textfont1 \_nulldelimiterspace=0.12\_fontdimen6\_textfont1 \_setmathparam\_Umathspaceafterscript \_scriptspacefactor \_setbox0=\_hbox{\_everymath{}$\_fam1\_displaystyle{0\_atop0}$}% \_Umathfractiondelsize\_displaystyle = \_dimexpr(\_ht0-\_Umathaxis\_displaystyle)*2\_relax \_setbox0=\_box\_voidbox } \_doc ----------------------------- If you try the example above about \~`\loadboldmath``{[xitsmath-bold]} \to {[xitsmath-regular]}` then you can find a bug in XITSMath-Bold font: the symbols for norm $\|x\|$ are missing. So, we have to define `\_boldmath` macro manually. The missing symbol is loaded from family 5 as no-bold variant in our example: \begtt \loadmath{[xitsmath-regular]} \def\_boldmath{% \_loadumathfamily 1 {[xitsmath-bold]}{} % Base font \_loadmathfamily 4 rsfs % script \_loadumathfamily 5 {[xitsmath-regular]}{} \_def\|{\_Udelimiter 0 5 "02016 }% % norm delimiter from family 5 \_setmathdimens } \endtt \`\_loadumathfamily` ` {}{}` loads the given Unicode-math fonts in three sizes using single with different `mathsize=1,2,3` font features. The math font family is set with given ``. The `` are added to the default \`\_mfontfeatures` and to the size-dependent features `ssty=1` if script size is asked or `ssty=2` if scriptscriptsize is asked.\nl \`\_mparams``` inserts additional font feature `nomathparam` if the of the family is greater than 3. Lua\TeX/ sets math parameters (thickness of fraction rules etc., see section 7.4 in Lua\TeX/ documentation) repeatedly from loaded math fonts if `nomathparam` is not given. We want to load these parameters only from fonts at families 0--3 (and actually we are using only family 1 as main math font).\nl The \^`\_corrmsize` `` can be used just before `\_loadumathfamily`, see section~\ref[math-preload] for more information.\nl The \`\_textmff`, \`\_scriptmff` and \`\_sscriptmff` are font features for text, script and sscript sizes respectively. They are locally re-defined in \^`\mathbox` macro. \_cod ----------------------------- \_def\_umathname#1#2{"#1:\_mfontfeatures#2"} \_def\_mfontfeatures{mode=base;script=math;} \_def\_loadumathfamily #1 #2#3 {% \_font\_mF=\_umathname{#2}{\_textmff \_mparams{#1}#3} at\_sizemtext \_textfont #1=\_mF \_font\_mF=\_umathname{#2}{\_scriptmff \_mparams{#1}#3} at\_sizemtext \_scriptfont #1=\_mF \_font\_mF=\_umathname{#2}{\_sscriptmff\_mparams{#1}#3} at\_sizemtext \_scriptscriptfont#1=\_mF \_ptmunit=\_ptunit } \_def\_textmff {ssty=0;mathsize=1;} \_def\_scriptmff {ssty=1;mathsize=2;} \_def\_sscriptmff{ssty=2;mathsize=3;} \_def\_mparams#1{\_ifnum#1>3 nomathparam;\_fi} \_doc ----------------------------- Unicode math font includes all typical math alphabets together, user needs not to load more \TeX/ math families. These math alphabets are encoded by different parts of Unicode table. We need auxiliary macros for setting mathcodes by selected math alphabet. \nl \`\_umathrange` `{-}\` sets `\Umathcode`s of the characters in the interval `-` to `\`, `\+1`, `\+2` etc., but \`\_umathcharholes` are skipped (`\_umathcharholes` are parts of the Unicode table not designed for math alphabets, they cause that the math alphabets are not continuously spread out in the table; I mean that the designers were under the influence of drugs when they created this part of the Unicode table). The `-` clause includes characters like~`A-Z`. \nl \`\_umahrangegreek` `\` is the same as `\_umathrange {-}\`. \nl \`\_umahrangeGREEK` `\` is the same as `\_umathrange {-}\`. \nl \`\_greekdef` ` \_relax` defines each control sequence as a normal character with codes `\_umathnumB`, `\_umathnumB+1`, `\_umathnumB+2` etc. It is used for redefinig the contol sequences for math Greek `\alpha`, `\beta`, `\gamma` etc. \_cod ----------------------------- \_newcount\_umathnumA \_newcount\_umathnumB \_def\_umathcorr#1#2{\_ea#1\_ea{\_the#2}} \_def\_umathprepare#1{\_def\_umathscanholes##1[#1]##2##3\_relax{##2}} \_def\_umathvalue#1{\_ea\_umathscanholes\_umathcharholes[#1]{#1}\_relax} \_def\_umathcharholes{% holes in math alphabets: [119893]{"210E}[119965]{"212C}[119968]{"2130}[119969]{"2131}% [119971]{"210B}[119972]{"2110}[119975]{"2112}[119976]{"2133}[119981]{"211B}% [119994]{"212F}[119996]{"210A}[120004]{"2134}% [120070]{"212D}[120075]{"210C}[120076]{"2111}[120085]{"211C}[120093]{"2128}% [120122]{"2102}[120127]{"210D}[120133]{"2115}[120135]{"2119} [120136]{"211A}[120137]{"211D}[120145]{"2124}% } \_def\_umathrange#1#2#3#4{\_umathnumB=#4\_def\_tmp{#2 #3 }\_umathrangeA#1} \_def\_umathrangeA#1-#2{\_umathnumA=`#1\_relax \_loop \_umathcorr\_umathprepare\_umathnumB \_Umathcode \_umathnumA = \_tmp \_umathcorr\_umathvalue{\_umathnumB} \_ifnum\_umathnumA<`#2\_relax \_advance\_umathnumA by1 \_advance\_umathnumB by1 \_repeat } \_def\_umathrangeGREEK{\_umathrange{^^^^0391-^^^^03a9}} \_def\_umathrangegreek{\_umathrange{^^^^03b1-^^^^03d6}} \_def\_greekdef#1{\_ifx#1\_relax \_else \_begingroup \_lccode`X=\_umathnumB \_lowercase{\_endgroup \_def#1{X}}% \_advance\_umathnumB by 1 \_ea\_greekdef \_fi } \_doc ----------------------------- \`\addUmathfont` ` {[]}{} {[]}{} {}` allocates new using \^`\newfam` and adds loading this font to the \~`\normalmath` and \~`\boldmath` macros. Note that allocationos using \^`\newfam` starts from 43 because numbers 1--42 are reserved for direct usage without \^`\newfam`. \_cod ----------------------------- \_def\_addUmathfont #1#2#3#4#5#6{% #1: fam (will be set), #2#3: normal font, #4#5: bold font \_ifx\_ncharrmA\_undefined \_errmessage{basic Unicode math font must be loaded first}% \_else \_isfont{#2}\_iffalse \_opwarning{font #2 is unavailable}% \_else \_newfam#1\_relax \_global\_addto\_normalmath{\_corrmsize#6 \_loadumathfamily #1 {#2}{#3} }% \_ifx\_relax#4\_relax \_global\_addto\_boldmath{\_corrmsize#6 \_loadumathfamily #1 {#2}{embolden=1.7;} }% \_else \_global\_addto\_boldmath{\_corrmsize#6 \_loadumathfamily #1 {#4}{#5} }% \_fi \_normalmath \_wterm{add-MATH-FONT: #1=\the#1, "#2", \ifx"#4"\else bold: "#4"\fi}% \_fi \_fi } \_doc ----------------------------- \`\resetmathchars` ` ;` saves to \`\_mafam` and applies \`\_resetmathcharsX` to each sequence from . The `\directlua` chunk prints the 8-digits hexadecimal code of the sequence followed by the sequence itself. The digts are scanned by \`\_resetmathcharsY` and new `\Umathchardef` and `\Umathcode` is declared. Note that the used primitives have the syntax: \begtt \catcode`<=13 \Umathchardef \Umathcode \endtt \_cod ----------------------------- \_def\_resetmathchars #1{\_chardef\_mafam=#1\_relax \_xargs \_resetmathcharX} \_def\_resetmathcharX#1{\_ea\_resetmathcharY \_directlua{tex.print(string.format("\_pcent08X", \_the#1))}#1} \_def\_resetmathcharY#1#2#3#4#5#6#7#8#9{% #9 is given \math-sequence \_Umathchardef #9\_numexpr"#3/2\_relax \_mafam "#4#5#6#7#8 \_Umathcode "#4#5#6#7#8=\_numexpr"#3/2\_relax \_mafam "#4#5#6#7#8 } \_public \addUmathfont \resetmathchars ; \_endcode The \^`\loadmath` `{}` macro loads math fonts and redefines all default math-codes using `\input unimath-codes.opm`. If Unicode-math font is loaded then \^`\_mathloadingfalse` is set, so the new Unicode-math font isn't loaded until \^`\doloadmath` is used. \^`\loadboldmath` `{} \to {}` loads bold variant only if `` was sucessully loaded by the previous `\loadmath`. For example: \begtt \loadmath {[xitsmath-regular]} \loadboldmath {[xitsmath-bold]} \to {[xitsmath-regular]} \endtt There are very few Unicode-math fonts with full \~`\boldmath` support. I know only XITSMath-Bold and KpMath-Bold. If \^`\loadboldmath` is not used then \"faked bold" created from \~`\normalmath` is used by default. The {\em main math font} is loaded by \^`\loadmath` (typically indirectly using \^`\fontfam`) and you can load more {\em additional math fonts} by \^`\addUmathfont`: \begtt \catcode`<=13 \addUmathfont \famname {[]}{} {[]}{} {} \endtt The `\famname` is a control sequence declared by \^`\addUmathfont` for later use. It gets math family number. The `` is decimal number for size corrections in view of the main math font. If it is empty then =1. If is empty, the \"faked bold" derived from is used. Example: \begtt \fontfam[lm] % does \lodmath{[latinmodern-math]} \addUmathfont \xits {[XITSMath-Regular]}{} {[XITSMath-Bold]}{} {} \endtt % declares `latinmodern-math` as main math font (its bold variant is \"faked bold") and the additional math font `\xits` is `XITSMath-Regular` for normal printing and `XITSMath-Bold` for bold printing. All characters used in math formula is printed from main math font by default. But you can redeclare a list of characters for printing from additional font by \^`\resetmathchars` `\famname ;`. For example: \begtt \resetmathchars \xits \stareq \triangleq \veeeq \wedgeq ; \endtt % sets the characters `\stareq`, `\triangleq`, `\veeeq`, `\wedgeq` from the `\xits` additional font. You can see all possible sequences by `\input print-unimath.opm`. If you want to use whole math alphabet from additional math font, use `\addto\selector{\fam\famname}`, for example `\addto\cal{\fam\xits}` declares all `\cal` characters from the `\xits` font loaded by \^`\addUmathfont`. \secc Unicode-math macros preloaded in the format \printdoc math-unicode.opm \secc[unimath-codes] Macros and codes set when `\loadmath` is processed firstly The file `unimath-codes.opm` is loaded when the \^`\loadmath` is used. The macros here redefines globally all encoding dependent settings declared in the section~\ref[math-macros]. \printdoc unimath-codes.opm \printdoctail unimath-codes.opm \secc Printing all Unicode math slots in used math font \printdoctail print-unimath.opm \printdoc print-unimath.opm \_endinput 2022-07-22 \Umathspaceafterscript setting added. 2022-06-10 \addUmathfont and \resetmathchars added. 2021-08-16 \_loadumathfamily simplified, used mathsize= font feature. 2021-08-16 nomathparam font feature used for families > 3 (bug fixed). 2021-08-02 ssty fontfeatures rewritten; fam2, fam3 unused, \scriptspace not set 2021-04-04 \_setunimathdimens: \setbox0=\box\voidbox added 2021-03-09 \_setunimathdimes: \_begin/end/group instead {}, bug fixed 2021-02-15 \_textmff, \_scriptmff and \_sscriptmff introduced 2021-02-15 \_expandafter -> \_ea 2020-06-07 \Umathfractiondelsize (for \choose brackets) corrected, see https://tug.org/pipermail/luatex/2020-June/007365.html 2020-04-15 \_setmathdimens -> \_setuniathdimens 2020-02-25 implemented