misc/llppac

   1 #!/bin/sh
   2 set -e
   3
   4 font=$HOME/x/fnt/LastResort.ttf
   5 ! test -r $font || export LLPP_FALLBACK_FONT=$font
   6
   7 die() {
   8     echo "$1" >&2
   9     exit 1
  10 }
  11
  12 cachedir="${XDG_CACHE_HOME:-$HOME/.cache}/llpp"
  13 test -d "$cachedir" || die "cache directory '$cachedir' does not exist"
  14
  15 caspsuf=
  16 type=
  17
  18 executable_p() { command -v "$1" >/dev/null 2>&1; }
  19
  20 missing() {
  21     executable_p $1 || \
  22         eval "$1() { die \"$2 is needed for \$type conversion\"; }"
  23 }
  24
  25 text() {
  26     cat <<EOF
  27 <pre style="font-size: ${1}pt;">
  28 $1 <hr>
  29 EOF
  30     test -n "$2" && cat $2 || cat <<EOF
  31 S9 З3Ӡʒ I|l1 O0 0O o0 0o OD DO Oo Do m rn m -~ ~-  8B B8 BB 88B8 88B8
  32
  33 abcdefghijklmnopqrstuvwxyz
  34 ABCDEFGHIJKLMNOPQRSTUVWXYZ
  35 абвгдеёжзийклмнопрстуфхцчшщэюя
  36 АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЭЮЯ
  37 инпипнишпнипшнипнпншпнпинпипнпипщнпипипин
  38 иннициатива шиорокошарикоподшипниковый взбзднуть гидроаэроионизация
  39 телегаммааппарат фельдъегерь четырёхсотпятидесятисемимиллиметровое
  40 превысокомногорассмотрительствующий вскользь забулдыжничество
  41 цецецница
  42 https://ru.wikipedia.org/wiki/Панграмма https://en.wikipedia.org/wiki/Pangram
  43 Экс-граф? Плюш изъят. Бьём чуждый цен хвощ!
  44 Waltz, bad nymph, for quick jigs vex
  45 ',.ფyfგcრლაოეუიდჰთნსყჯქxბმwვზ
  46 Σομε πσθδο γρεεκ
  47 Il1| Il|1 I1l| I1|l I|1l I|l1 lI1| lI|1 l1I| l1|I l|1I l|I1
  48 1lI| 1l|I 1Il| 1I|l 1|Il 1|lI |l1I |lI1 |1lI |1Il |I1l |Il1
  49 \`1234567890-=\\
  50 ~!@#\$$%&permil;^&amp;*()_+|
  51 ₹₽£$€
  52 [];',.//|/\ &gt;- -&gt; ~~&lt; &lt;-- --&gt; == === ==&gt; &lt;==
  53 {}:"&lt;&gt;?"
  54 S in S9 59 S9 is like 5 also
  55 final fjord flair ft fff
  56 3ЗЭэзээз3ЗqgO0o ijklI1| bgqpykvwlliliiijil1 lLLl
  57 01234567890 3ДС ЗДС ӠДС ʒДС
  58 rnrmnnmrrrnmmr <i>Nürnberg</i> (rn -> m)
  59 Apay Арау
  60 örök für A̋ a̋ ő ű https://en.wikipedia.org/wiki/Diacritic
  61 https://en.wikipedia.org/wiki/Double_acute_accent
  62 Latin
  63 A̋ a̋
  64 E̋ e̋
  65 I̋ i̋
  66 M̋ m̋
  67 Ő ő
  68 Ű ű
  69 Cyrillic
  70 Ӳ ӳ
  71 молодо́й
  72 - - (~-) ~ ~ (-~) -- ~~ ~&mdash; &mdash;~ ''"''""., 1.65Mb
  73 curly vs plain braces {} () {} ({ {( }) )} →
  74 Bjarne Stroustrup (/ˈbjɑːrnə ˈstraʊstrʊp/; Danish: [ˈbjɑːnə ˈsdʁʌwˀsdʁɔb]
  75 ○●🅼🐫🐧﷽🙋✨🌸
  76 😁 😂 😃 😅 😆 😇 😈 😉 😊 😋 😌 😍 😎 😏 😐 😑
  77 😒 😓 😕 😗 😘 😙 😚 😛 😜 😞 😠 😡 😢 😥 😧 😨
  78 😩 😪 😫 😬 😭 😮 😯 😰 😱 😲 😳 😵 😶 😸 😹 😻
  79 😼 😽 😾 😿 🙀 🙁 🙂 🙃 🙄 🙆 🙇 🙉 🙊 🙌 🙍 🙎
  80 🙏 😀 😀 😡 😤 😦 😖 😄 😔 🙅 🙈 🙋 😠 😧 😟 😴
  81 😷 😺 😉 😌 😝 😣 ✉ 📧
  82
  83 kerning
  84     echo \${@:5}
  85     life           [l i fe]
  86     lambdabot      [l a mb]
  87     Commutative    [Co mm utative]
  88     weird          [we i rd]
  89     Before         [Bef ore, Be for e]
  90     higher         [gh, h i gher, hi gher]
  91     multi          [mu lti]
  92     till           [ti ll]
  93     language       [an]
  94     plan           [la]
  95     when           [wh]
  96     kerning        [rn | ng]
  97     мне            [мн е]
  98     oiled          [oi led]
  99     mkfifo         [mk f i fo, mk f i fo]
 100     didn't         [di dn't]
 101     argumnet (sic) []
 102     more           [mo re]
 103     CmmCall        [CmmC a l l]
 104     -Wvla          [-Wv la]
 105     align          [a l i gn]
 106     Compiling      [Co mp iling]
 107     print          [pr int]
 108     fcmono         [f cmono]
 109     Mail           [M ail]
 110     xsrc           [x src]
 111     Pokorna
 112     buildglyphs.js [buil dg lyphs]
 113     Dollar         [Do ll ar]
 114     signed         [s i gned]
 115     Pine           [Pi ne]
 116     right          [r i ght]
 117     ugliness
 118     Nigeria        [Ni g eria]
 119     Nothing        [Nothi ng]
 120     modify         [mo dify]
 121     mplus          [mp lus]
 122     Apostolic      [A postolic]
 123     коде           [к о де]
 124     Прокофьевну    [Про ко фьевну]
 125     Хэлп           [X элп]
 126     коснулся       [ко снулся]
 127     обьявляным     [об ья вленым]
 128     Queen          [Qu een]
 129     etymological   [et ymological]
 130     sigh           [s i gh]
 131     git            [gi t]
 132     некочевого     [не ко чевого]
 133     Розовое        [Ро зо вое]
 134     Trump          [Tru mp]
 135     покрытие       [по кр ытие]
 136     Singapore      [Si ng apore]
 137     Georgy         [Ge orgy]
 138     with           [wi th]
 139     Yokozuna       [Y okozuna]
 140     подтверждение  [под тверждение]
 141     Xwayland       [X wayland]
 142     callable       [ca l l ab le]
 143     Noah           [N oa h]
 144     Los            [L os]
 145     refused        [ref used]
 146     Julia          [J ulia | Jul i a]
 147     Joe Armstrong  [J oe (rip :()]
 148     James          [J ames]
 149     CAPTCHA        [CA PTCHA]
 150     KIZER          [KI ZER]
 151     Kavanugh       [K avaunaugh]
 152     DejaVu         [D e ja V u]
 153     Never          [N ever]
 154
 155 almost kerning
 156     hope doesn't
 157     Illegal
 158     automatically effective
 159     Ilium is the Latin for Ilion.
 160     jtanx
 161
 162 (from UTF8-demo.txt)
 163 Box drawing alignment tests:                                          █
 164                                                                       ▉
 165   ╔══╦══╗  ┌──┬──┐  ╭──┬──╮  ╭──┬──╮  ┏━━┳━━┓  ┎┒┏┑   ╷  ╻ ┏┯┓ ┌┰┐    ▊ ╱╲╱╲╳╳╳
 166   ║┌─╨─┐║  │╔═╧═╗│  │╒═╪═╕│  │╓─╁─╖│  ┃┌─╂─┐┃  ┗╃╄┙  ╶┼╴╺╋╸┠┼┨ ┝╋┥    ▋ ╲╱╲╱╳╳╳
 167   ║│╲ ╱│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╿ │┃  ┍╅╆┓   ╵  ╹ ┗┷┛ └┸┘    ▌ ╱╲╱╲╳╳╳
 168   ╠╡ ╳ ╞╣  ├╢   ╟┤  ├┼─┼─┼┤  ├╫─╂─╫┤  ┣┿╾┼╼┿┫  ┕┛┖┚     ┌┄┄┐ ╎ ┏┅┅┓ ┋ ▍ ╲╱╲╱╳╳╳
 169   ║│╱ ╲│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╽ │┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▎
 170   ║└─╥─┘║  │╚═╤═╝│  │╘═╪═╛│  │╙─╀─╜│  ┃└─╂─┘┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▏
 171   ╚══╩══╝  └──┴──┘  ╰──┴──╯  ╰──┴──╯  ┗━━┻━━┛  ▗▄▖▛▀▜   └╌╌┘ ╎ ┗╍╍┛ ┋  ▁▂▃▄▅▆▇█
 172                                                ▝▀▘▙▄▟
 173
 174 https://mirrors.zju.edu.cn/CTAN/fonts/gofonts/doc/gofonts.pdf (page 2)
 175 """
 176 DIN Legibility Standard
 177
 178 The recent German DIN 1450 legibility standard
 179 recommends several features for font legibility, including
 180 differentiation of letter shapes to reduce confusion. The Go fonts
 181 conform to the 1450 standard by carefully differentiating zero from
 182 capital O; numeral 1 from capital I (eye) and lowercase l (ell);
 183 numeral 5 from capital S; and numeral 8 from capital B. The shapes of
 184 bowls of b d p q follow the natural asymmetries of legible Renaissance
 185 handwriting, aiding differentiation to reduce confusion. [5]
 186 """
 187
 188 (https://en.wikipedia.org/wiki/Jabberwocky
 189  Idea from the aforementioned gofonts paper)
 190
 191     Бе сгладне и честлинните комбурси  Það leið að stekju, og slýgir greðlar
 192     търляха се и сврецваха във плите;  sig snældu og böluðu um slöffruna,
 193     съвсем окласни бяха тук щурпите    og angurvært sungu sópfiðrungar
 194     и отма равапсатваха прасурси.      við sífgelt týðmana svíræna.
 195
 196 (https://en.wikipedia.org/wiki/Typeface)
 197
 198 For the rational mind, type design can be a maddening game
 199 of drawing things differently in order to make them appear the
 200 same.
 201 </pre>
 202 EOF
 203 }
 204
 205 maketext() {
 206     # https://github.com/react-boilerplate/react-boilerplate/issues/1340
 207     test -d "$cachedir/fonts" || mkdir "$cachedir/fonts"
 208     filename="$(basename "$1")"
 209     textfilename="$2"
 210     cat >"$cachedir/fonts/text.html" <<EOF
 211 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
 212  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 213 <html>
 214   <head>
 215     <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
 216     <title>$filename</title>
 217     <style type="text/css">
 218       @font-face { font-family: moo; src: url('$1'); }
 219       pre { font-family: moo; }
 220     </style>
 221   </head>
 222   $filename
 223   <body>
 224 EOF
 225     for size in 6 8 10 12 4 2; do
 226         text $size $textfilename >>"$cachedir/fonts/text.html"
 227     done
 228     cat >>"$cachedir/fonts/text.html" <<EOF
 229     <br/>
 230     $(fc-scan -f '%{fullname} %{style} %{slant} %{file} %{fontversion} %{capability} %{lang}' "$1" || file "$1")
 231   </body>
 232 </html>
 233 EOF
 234 }
 235
 236 trap 'test -n "$casp" && rm -f "$casp"' 0
 237
 238 while getopts s:c:m:t:fu opt; do
 239     case $opt in
 240         m) mime=$OPTARG;;
 241         t) type=$OPTARG;;
 242         f) force=true;;
 243         c) css="-s $OPTARG";;
 244         s) textfilename="$OPTARG";;
 245         ?) die "usage: $0 [-c css] [-m mime/type] [-t filter] [-f] [-s textfile] [path|url]";;
 246     esac
 247 done
 248 shift $(($OPTIND - 1))
 249 test -z "$1" && die "usage $0: path"
 250
 251 origin="$1"
 252 if ${force-test ! -e "$1"} && expr >/dev/null "$1" : "\(ftp\|https\?\)://"; then
 253     if executable_p wget; then
 254         dl() { wget -q $1 -O $2; }
 255     elif executable_p curl; then
 256         dl() { curl -L $1 -o $2; }
 257     else
 258         die "no program to fetch remote urls found"
 259     fi
 260
 261     hashof="$cachedir/$(basename "$1")"
 262     dl "$1" "$hashof" || test -e "$md5of"
 263     shift
 264     set -- "$hashof" "$@"
 265 else
 266     hashof="$1"
 267 fi
 268
 269 test -z "$type" && { ft=$(file -L --mime-type -b "$1") || die "$ft"; }
 270
 271 case $ft in
 272     application/x-gzip | application/x-compress) dc=zcat;;
 273     application/x-xz) dc=xzcat;;
 274     application/x-bzip2) dc=bzcat;;
 275     *) unset dc || true;;
 276 esac
 277
 278 filt='"${dc-cat}" "$1" |'
 279
 280 typeofmime() {
 281     case "$1" in
 282         application/postscript) type=ps;;
 283         application/pdf) type=pdf;;
 284         image/vnd.djvu) type=djvu;;
 285         text/html) type=html;;
 286         text/plain) type=text;;
 287         application/msword) type=word;;
 288         application/vnd.openxmlformats-officedocument.*  \
 289             | application/vnd.ms-powerpoint              \
 290             | application/vnd.ms-excel                   \
 291             | application/vnd.oasis.opendocument.*) type=uno;;
 292         image/svg+xml) type=svg;;
 293         image/png | image/jpeg) test -n "$dc" && type="image" || type="image2";;
 294         image/*) type=image;;
 295         application/x-dvi) type=dvi;;
 296         application/x-font-ttf                  \
 297             | application/vnd.ms-opentype       \
 298             | application/font-sfnt) type=font;;
 299         *) return 1;;
 300     esac
 301     return 0
 302 }
 303
 304 if test -z "$type"; then
 305     test -z "$mime" &&                                          \
 306         mime=$(file -L --mime-type -b "$1" || die "$mime") ||   \
 307             $(file -L --mime-type -bz "$1" || die "$mime")
 308     typeofmime "$mime" || die "unhandled file type: '$mime'"
 309 fi
 310
 311 caspsuf=".pdf"
 312 case $type in
 313     ps) conv='ps2pdf - "$casp"';;
 314     image2|pdf) test -z "$dc" && exec llpp "$@" || conv='cat >"$casp"';;
 315     texi) {
 316         missing texi2html "texi2html(http://www.nongnu.org/texi2html/)"
 317         missing prince "PrinceXML(http://www.princexml.com/)"
 318         conv='texi2html - -o - | prince $css - -o "$casp"'
 319     };;
 320     djvu) {
 321         missing ddjvu "ddjvu(http://djvu.sourceforge.net/doc/man/ddjvu.html)"
 322         conv='ddjvu -format=pdf - "$casp"'
 323     };;
 324     html) {
 325         missing prince "Prince(http://www.princexml.com/)"
 326         conv='prince $css - -o "$casp"'
 327     };;
 328     html2epub) {
 329         missing pandoc "pandoc(http://pandoc.org)"
 330         caspsuf=".epub"
 331         conv='pandoc -r html - -w epub2 -o "$casp"'
 332     };;
 333     word) {
 334         if executable_p unoconv && test -z "$dc"; then
 335             unset filt
 336             conv='unoconv -o "$casp" "$1"'
 337         else
 338             missing antiword "antiword or unoconv"
 339             conv='antiword -m 8859-1.txt -a a4 - >"$casp"'
 340         fi
 341     };;
 342     uno) {
 343         test -n "$dc" && die "cannot convert compressed '$mime'"
 344         unset filt
 345         missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 346         conv='unoconv -o "$casp" "$1"'
 347     };;
 348     svg) {
 349         if executable_p inkscape && test -z "$dc"; then
 350             unset filt
 351             conv='inkscape -z -A "$casp" "$1"'
 352         else
 353             if executable_p rsvg-convert; then
 354                 conv='rsvg-convert -f pdf -o "$casp"'
 355             else
 356                 test -n "$dc" && die "cannot convert compressed '$mime'"
 357                 unset filt
 358                 missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 359                 conv='unoconv -o "$casp" "$1"'
 360             fi
 361         fi
 362     };;
 363     font) {
 364         maketext "$1" "$textfilename"
 365         exec llpp -origin $1 -layout-height 0 "$cachedir/fonts/text.html"
 366     };;
 367     image) {
 368         missing convert "convert(http://www.imagemagick.org/script/convert.php)"
 369         conv='convert - pdf:"$casp"'
 370     };;
 371     dvi) {
 372         test -n "$dc" && die "cannot convert compressed '$mime'"
 373         unset filt
 374         missing dvipdf "dvipdf(http://ghostscript.com/)"
 375         conv='dvipdf "$1" "$casp"'
 376     };;
 377     text) {
 378         missing pandoc "pandoc(http://pandoc.org/)"
 379         conv='pandoc -t epub2 - -o "$casp"'
 380         caspsuf=.epub
 381     };;
 382     *) die "unhandled filter type: '$type'";;
 383 esac
 384
 385 hash=$(cksum "$hashof") || die "$hash"
 386 casp=$cachedir/${hash%% *}$caspsuf
 387
 388 { test -n "$force" || test ! -e "$casp"; } && eval "$filt" "$conv"
 389 shift
 390
 391 exec llpp -origin $origin "$@" "$casp"