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