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