misc/llppac

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