Lady’s Gitweb - LesML/blob - sh/lesml.sh

   1 #!/usr/bin/env sh
   2 # @(#)💄📝 Les·M·L       sh/lesml.sh     2026-03-31T01:28:11Z
   3 # SPDX-FileCopyrightText: 2023, 2024, 2026 Lady <https://www.ladys.computer/about/#lady>
   4 # SPDX-License-Identifier: MPL-2.0
   5
   6 ## ⁌ The 💄📝 Les·M·L shell script
   7 ##
   8 ## ∎ Copyright © 2023–2024, 2026 Lady [@ Ladys Computer].
   9 ##
  10 ## ⋮ This Source Code Form is subject to the terms of the Mozilla
  11 ##   Public License, version 2.0.
  12 ## If a copy of the M·P·L was not distributed with this file, You can
  13 ##   obtain one at {🔗<https://mozilla.org/MPL/2.0/>}.
  14 ##
  15 ## Usage :⁠—
  16 ##
  17 ## »|sh$sh ./sh/lesml.sh [--] [filename]
  18 ##
  19 ## This script provides a simple solution for converting a Les·M·L file
  20 ##   into X·H·T·M·L.
  21
  22 set -o 'errexit'
  23
  24 ## `LANG´ and `LC_ALL´ are set to `C´ because this script assumes
  25 ##   working with U·T·F‐8 strings as opaque series of bytes.
  26
  27 LANG=C
  28 LC_ALL=C
  29
  30 ## § Configuration
  31 ##
  32 ## All of the commands used by this script are overridable with your
  33 ##   own implementations by setting the corresponding `cmd_COMMAND´
  34 ##   variable.
  35
  36 : "${cmd_DIRNAME:=dirname}"
  37 : "${cmd_PRINTF:=printf}"
  38 : "${cmd_REALPATH:=realpath}"
  39 : "${cmd_SED:=sed}"
  40 : "${cmd_TEST:=test}"
  41 : "${cmd_TR:=tr}"
  42 : "${cmd_XSLTPROC:=xsltproc}"
  43
  44 ## In order to run, this program needs to know the location of the
  45 ##   💄📝 Les·M·L transform, which is calculated below.
  46 ## This location can be manually configured by setting the `path_LESML´
  47 ##   or `path_TRANSFORM´ environment variables.
  48
  49 thisfile="$(
  50   "${cmd_REALPATH}" -- "${0}"
  51 )"
  52 thisdir="$(
  53   "${cmd_DIRNAME}" -- "${thisfile}"
  54 )"
  55
  56 defaultlesml="$(
  57   "${cmd_REALPATH}" -- "${thisdir}"'/..'
  58 )"
  59 : "${path_LESML:=${defaultlesml}}"
  60
  61 defaulttransform="${path_LESML}"'/xslt/lesml.xslt'
  62 : "${path_TRANSFORM:=${defaulttransform}}"
  63
  64 ## The (unregistered) media type for Les·M·L files is `text/lesml´.
  65 ## By overriding both the media type and the transform, one can re·use
  66 ##   this script for processing other kinds of files with X·S·L·T.
  67
  68 : "${name_MEDIATYPE:=text/lesml}"
  69
  70 ## § Parameter processing
  71 ##
  72 ## A leading `--´ is allowed (and ignored).
  73 ## Otherwise, the first parameter is the source filename.
  74
  75 if "${cmd_TEST}" "${1}" = '--'
  76 then :
  77   shift
  78 fi
  79
  80 ## If no input file is provided, it defaults to `-´ (standard input).
  81
  82 if "${cmd_TEST}" -n "${1}" && "${cmd_TEST}" "${1}" != '-'
  83 then :
  84   inputfile="$(
  85     "${cmd_REALPATH}" -- "${1}"
  86   )"
  87 else :
  88   inputfile='-'
  89 fi
  90
  91 ## § Implementation
  92 ##
  93 ## ※ This implementation is derived from
  94 ##   {🔗⛩📰 书社<https://git.ladys.computer/Shushe/>} code written
  95 ##   2023–2024.
  96 ##
  97 ## The `sanitizeprog´ variable contains a Sed program for sanitizing
  98 ##   an arbitrary string into characters which are valid in X·M·L.
  99 ## Ideally this would be written using dollar‐single‐quote literals,
 100 ##   but Mac·O·S Dash is outdated and does not support them.
 101
 102 sanitizeprog="$(
 103   "${cmd_PRINTF}" '%b' \
 104     's/]]>/]]]]><!\\[CDATA\\[>/g\n' \
 105     's/\0357\0277\0276/�/g\n' \
 106     's/\0357\0277\0277/�/g\n' \
 107     '$!s/\r$//g\n' \
 108     's/\r/\\n/g\n' \
 109     '$!s/\0302\0205$//g\n' \
 110     's/\0302\0205/\\n/g\n' \
 111     's/\0342\0200\0250/\\n/g\n' \
 112     's/[\0001\0002\0003\0004\0005\0006\0007\0010]/�/g\n' \
 113     's/[\0016\0017\0020\0021\0022\0023\0024\0025\0026\0027]/�/g\n' \
 114     's/[\0031\0032\0033\0034\0035\0036\0037]/�/g'
 115 )"
 116
 117 ## A compound statement is used to group together the wrapping X·M·L
 118 ##   and the sanitized input into one output which can be piped to
 119 ##   `xsltproc´.
 120
 121 {
 122   "${cmd_PRINTF}" '%s\n%s%s%s' \
 123     '<?xml version="1.0"?>' \
 124     '<script xmlns="http://www.w3.org/1999/xhtml" type="' \
 125     "${name_MEDIATYPE}" \
 126     '"><![CDATA['
 127   if "${cmd_TEST}" "${inputfile}" = '-'
 128   then :
 129     "${cmd_TR}" '\000\013\014' '\032\011\012'
 130   else :
 131     "${cmd_TR}" '\000\013\014' '\032\011\012' <"${inputfile}"
 132   fi |
 133   "${cmd_SED}" "${sanitizeprog}"
 134   "${cmd_PRINTF}" '%s\n' ']]></script>'
 135 } |
 136 "${cmd_XSLTPROC}" --nonet --nowrite --novalid \
 137   "${path_TRANSFORM}" \
 138   -