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     Joe Armstrong  [J oe (rip :()]
 116
 117 almost kerning
 118     hope doesn't
 119     Illegal
 120     automatically effective
 121     Ilium is the Latin for Ilion.
 122
 123 (from UTF8-demo.txt)
 124 Box drawing alignment tests:                                          █
 125                                                                       ▉
 126   ╔══╦══╗  ┌──┬──┐  ╭──┬──╮  ╭──┬──╮  ┏━━┳━━┓  ┎┒┏┑   ╷  ╻ ┏┯┓ ┌┰┐    ▊ ╱╲╱╲╳╳╳
 127   ║┌─╨─┐║  │╔═╧═╗│  │╒═╪═╕│  │╓─╁─╖│  ┃┌─╂─┐┃  ┗╃╄┙  ╶┼╴╺╋╸┠┼┨ ┝╋┥    ▋ ╲╱╲╱╳╳╳
 128   ║│╲ ╱│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╿ │┃  ┍╅╆┓   ╵  ╹ ┗┷┛ └┸┘    ▌ ╱╲╱╲╳╳╳
 129   ╠╡ ╳ ╞╣  ├╢   ╟┤  ├┼─┼─┼┤  ├╫─╂─╫┤  ┣┿╾┼╼┿┫  ┕┛┖┚     ┌┄┄┐ ╎ ┏┅┅┓ ┋ ▍ ╲╱╲╱╳╳╳
 130   ║│╱ ╲│║  │║   ║│  ││ │ ││  │║ ┃ ║│  ┃│ ╽ │┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▎
 131   ║└─╥─┘║  │╚═╤═╝│  │╘═╪═╛│  │╙─╀─╜│  ┃└─╂─┘┃  ░░▒▒▓▓██ ┊  ┆ ╎ ╏  ┇ ┋ ▏
 132   ╚══╩══╝  └──┴──┘  ╰──┴──╯  ╰──┴──╯  ┗━━┻━━┛  ▗▄▖▛▀▜   └╌╌┘ ╎ ┗╍╍┛ ┋  ▁▂▃▄▅▆▇█
 133                                                ▝▀▘▙▄▟
 134
 135 https://mirrors.zju.edu.cn/CTAN/fonts/gofonts/doc/gofonts.pdf (page 2)
 136 """
 137 DIN Legibility Standard
 138
 139 The recent German DIN 1450 legibility standard
 140 recommends several features for font legibility, including
 141 differentiation of letter shapes to reduce confusion. The Go fonts
 142 conform to the 1450 standard by carefully differentiating zero from
 143 capital O; numeral 1 from capital I (eye) and lowercase l (ell);
 144 numeral 5 from capital S; and numeral 8 from capital B. The shapes of
 145 bowls of b d p q follow the natural asymmetries of legible Renaissance
 146 handwriting, aiding differentiation to reduce confusion. [5]
 147 """
 148
 149 Mistaking S for a 5 (at certain sizes) is more of a problem... but... Ze Germans
 150
 151 (https://en.wikipedia.org/wiki/Jabberwocky
 152  Idea from the aforementioned gofonts paper)
 153
 154     Бе сгладне и честлинните комбурси  Það leið að stekju, og slýgir greðlar
 155     търляха се и сврецваха във плите;  sig snældu og böluðu um slöffruna,
 156     съвсем окласни бяха тук щурпите    og angurvært sungu sópfiðrungar
 157     и отма равапсатваха прасурси.      við sífgelt týðmana svíræna.
 158
 159 (https://en.wikipedia.org/wiki/Typeface)
 160
 161 For the rational mind, type design can be a maddening game
 162 of drawing things differently in order to make them appear the
 163 same.
 164 </pre>
 165 EOF
 166 }
 167
 168 maketext() {
 169     # https://github.com/react-boilerplate/react-boilerplate/issues/1340
 170     test -d "$cachedir/fonts" || mkdir "$cachedir/fonts"
 171     filename="$(basename "$1")"
 172     textfilename="$2"
 173     cat >"$cachedir/fonts/text.html" <<EOF
 174 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
 175  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
 176 <html>
 177   <head>
 178     <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
 179     <title>$filename</title>
 180     <style type="text/css">
 181       @font-face { font-family: moo; src: url('$1'); }
 182       pre { font-family: moo; }
 183     </style>
 184   </head>
 185   $filename
 186   <body>
 187 EOF
 188     for size in 6 8 10 12 4 2; do
 189         text $size $textfilename >>"$cachedir/fonts/text.html"
 190     done
 191     cat >>"$cachedir/fonts/text.html" <<EOF
 192     <br/>
 193     $(fc-scan -f '%{fullname} %{style} %{slant} %{file} %{fontversion} %{capability} %{lang}' "$1" || file "$1")
 194   </body>
 195 </html>
 196 EOF
 197 }
 198
 199 trap 'test -n "$casp" && rm -f "$casp"' 0
 200
 201 while getopts s:c:m:t:fu opt; do
 202     case $opt in
 203         m) mime=$OPTARG;;
 204         t) type=$OPTARG;;
 205         f) force=true;;
 206         c) css="-s $OPTARG";;
 207         s) textfilename="$OPTARG";;
 208         ?) die "usage: $0 [-c css] [-m mime/type] [-t filter] [-f] [-s textfile] [path|url]";;
 209     esac
 210 done
 211 shift $(($OPTIND - 1))
 212 test -z "$1" && die "usage $0: path"
 213
 214 origin="$1"
 215 if ${force-test ! -e "$1"} && expr >/dev/null "$1" : "\(ftp\|https\?\)://"; then
 216     if executable_p wget; then
 217         dl() { wget -q $1 -O $2; }
 218     elif executable_p curl; then
 219         dl() { curl -L $1 -o $2; }
 220     else
 221         die "no program to fetch remote urls found"
 222     fi
 223
 224     hashof="$cachedir/$(basename "$1")"
 225     dl "$1" "$hashof" || test -e "$md5of"
 226     shift
 227     set -- "$hashof" "$@"
 228 else
 229     hashof="$1"
 230 fi
 231
 232 test -z "$type" && { ft=$(file -L --mime-type -b "$1") || die "$ft"; }
 233
 234 case $ft in
 235     application/x-gzip | application/x-compress) dc=zcat;;
 236     application/x-xz) dc=xzcat;;
 237     application/x-bzip2) dc=bzcat;;
 238     *) unset dc || true;;
 239 esac
 240
 241 filt='"${dc-cat}" "$1" |'
 242
 243 typeofmime() {
 244     case "$1" in
 245         application/postscript) type=ps;;
 246         application/pdf) type=pdf;;
 247         image/vnd.djvu) type=djvu;;
 248         text/html) type=html;;
 249         text/plain) type=text;;
 250         application/msword) type=word;;
 251         application/vnd.openxmlformats-officedocument.*  \
 252             | application/vnd.ms-powerpoint              \
 253             | application/vnd.ms-excel                   \
 254             | application/vnd.oasis.opendocument.*) type=uno;;
 255         image/svg+xml) type=svg;;
 256         image/png | image/jpeg) test -n "$dc" && type="image" || type="image2";;
 257         image/*) type=image;;
 258         application/x-dvi) type=dvi;;
 259         application/x-font-ttf                  \
 260             | application/vnd.ms-opentype       \
 261             | application/font-sfnt) type=font;;
 262         *) return 1;;
 263     esac
 264     return 0
 265 }
 266
 267 if test -z "$type"; then
 268     test -z "$mime" &&                                          \
 269         mime=$(file -L --mime-type -b "$1" || die "$mime") ||   \
 270             $(file -L --mime-type -bz "$1" || die "$mime")
 271     typeofmime "$mime" || die "unhandled file type: '$mime'"
 272 fi
 273
 274 caspsuf=".pdf"
 275 case $type in
 276     ps) conv='ps2pdf - "$casp"';;
 277     image2|pdf) test -z "$dc" && exec llpp "$@" || conv='cat >"$casp"';;
 278     texi) {
 279         missing texi2html "texi2html(http://www.nongnu.org/texi2html/)"
 280         missing prince "PrinceXML(http://www.princexml.com/)"
 281         conv='texi2html - -o - | prince $css - -o "$casp"'
 282     };;
 283     djvu) {
 284         missing ddjvu "ddjvu(http://djvu.sourceforge.net/doc/man/ddjvu.html)"
 285         conv='ddjvu -format=pdf - "$casp"'
 286     };;
 287     html) {
 288         missing prince "Prince(http://www.princexml.com/)"
 289         conv='prince $css - -o "$casp"'
 290     };;
 291     html2epub) {
 292         missing pandoc "pandoc(http://pandoc.org)"
 293         caspsuf=".epub"
 294         conv='pandoc -r html - -w epub2 -o "$casp"'
 295     };;
 296     word) {
 297         if executable_p unoconv && test -z "$dc"; then
 298             unset filt
 299             conv='unoconv -o "$casp" "$1"'
 300         else
 301             missing antiword "antiword or unoconv"
 302             conv='antiword -m 8859-1.txt -a a4 - >"$casp"'
 303         fi
 304     };;
 305     uno) {
 306         test -n "$dc" && die "cannot convert compressed '$mime'"
 307         unset filt
 308         missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 309         conv='unoconv -o "$casp" "$1"'
 310     };;
 311     svg) {
 312         if executable_p inkscape && test -z "$dc"; then
 313             unset filt
 314             conv='inkscape -z -A "$casp" "$1"'
 315         else
 316             if executable_p rsvg-convert; then
 317                 conv='rsvg-convert -f pdf -o "$casp"'
 318             else
 319                 test -n "$dc" && die "cannot convert compressed '$mime'"
 320                 unset filt
 321                 missing unoconv "unoconv(http://dag.wiee.rs/home-made/unoconv/)"
 322                 conv='unoconv -o "$casp" "$1"'
 323             fi
 324         fi
 325     };;
 326     font) {
 327         maketext "$1" "$textfilename"
 328         exec llpp -origin $1 -layout-height 0 "$cachedir/fonts/text.html"
 329     };;
 330     image) {
 331         missing convert "convert(http://www.imagemagick.org/script/convert.php)"
 332         conv='convert - pdf:"$casp"'
 333     };;
 334     dvi) {
 335         test -n "$dc" && die "cannot convert compressed '$mime'"
 336         unset filt
 337         missing dvipdf "dvipdf(http://ghostscript.com/)"
 338         conv='dvipdf "$1" "$casp"'
 339     };;
 340     text) {
 341         missing pandoc "pandoc(http://pandoc.org/)"
 342         conv='pandoc -t epub2 - -o "$casp"'
 343         caspsuf=.epub
 344     };;
 345     *) die "unhandled filter type: '$type'";;
 346 esac
 347
 348 hash=$(cksum "$hashof") || die "$hash"
 349 casp=$cachedir/${hash%% *}$caspsuf
 350
 351 { test -n "$force" || test ! -e "$casp"; } && eval "$filt" "$conv"
 352 shift
 353
 354 exec llpp -origin $origin "$@" "$casp"