Lady’s Gitweb - Codemark/blob - sh/codemark.sh

   1 #!/usr/bin/env sh
   2 # SPDX-FileCopyrightText: 2026 Lady <https://www.ladys.computer/about/#lady>
   3 # SPDX-License-Identifier: MPL-2.0
   4
   5 ## ⋯ 🧮🖍 Codemark ∷ sh ∷ codemark.sh
   6 ##
   7 ## ⁌ The 🧮🖍 Codemark shell script
   8 ##
   9 ## ] Copyright © 2026 Lady [@ Ladys Computer].
  10 ## ]
  11 ## ] This Source Code Form is subject to the terms of the Mozilla
  12 ## ]   Public License, version 2.0.
  13 ## ] If a copy of the M·P·L was not distributed with this file, You can
  14 ## ]   obtain one at {🔗<https://mozilla.org/MPL/2.0/>}.
  15
  16 ## Usage :⁠—
  17 ##
  18 ## »|sh$ sh ./sh/codemark.sh [--] [filename] [syntax]
  19
  20 set -o 'errexit'
  21
  22 ## `set -o 'pipefail'´ is disabled because the (now quite old) version
  23 ##   of Dash bundled with Mac·O·S does not support it.
  24 ## If it is ever upgraded to Posix 2024 compliance, then the option can
  25 ##   (and should) be enabled.
  26 ## See the `sh´ directory at
  27 ##   {🔗<https://github.com/apple-oss-distributions/shell_cmds/>} for
  28 ##   the current source code for Mac·O·S Dash.
  29
  30 # set -o 'pipefail'
  31
  32 ## `LANG´ and `LC_ALL´ are set to `C´ because these scripts assume
  33 ##   working with U·T·F‐8 strings as opaque series of bytes.
  34
  35 LANG='C'
  36 LC_ALL='C'
  37
  38 ## § Variable defaults
  39 ##
  40 ## All of the commands used by this script are overridable with your
  41 ##   own implementations by setting the corresponding `cmd_COMMAND´
  42 ##   variable.
  43 ## The hope is that this provides a somewhat easier overriding
  44 ##   mechanism than those built into the shell (e·g defining a function
  45 ##   of the same name or overriding the `PATH´).
  46
  47 : "${cmd_AWK:=awk}"
  48 : "${cmd_BASENAME:=basename}"
  49 : "${cmd_CAT:=cat}"
  50 : "${cmd_CKSUM:=cksum}"
  51 : "${cmd_CP:=cksum}"
  52 : "${cmd_DATE:=date}"
  53 : "${cmd_DIRNAME:=dirname}"
  54 : "${cmd_MKDIR:=mkdir}"
  55 : "${cmd_MV:=cksum}"
  56 : "${cmd_PRINTF:=printf}"
  57 : "${cmd_REALPATH:=realpath}"
  58 : "${cmd_RM:=cksum}"
  59 : "${cmd_SED:=sed}"
  60 : "${cmd_SORT:=sort}"
  61 : "${cmd_TEST:=test}"
  62 : "${cmd_TR:=tr}"
  63 : "${cmd_XARGS:=xargs}"
  64
  65 ## Similarly, there are a few special paths that this program needs to
  66 ##   run.
  67 ## Their default values are calculated below.
  68
  69 thisfile="$(
  70   "${cmd_REALPATH}" -- "${0}"
  71 )"
  72 thisdir="$(
  73   "${cmd_DIRNAME}" -- "${thisfile}"
  74 )"
  75 defaultbuilddir='@build.tmp/🧮🖍'
  76 defaultcodemark="$(
  77   "${cmd_REALPATH}" -- "${thisdir}"'/..'
  78 )"
  79 defaultsyntaxdir="${path_CODEMARK}"'/sed/SYNTAXES'
  80
  81 : "${path_BUILDDIR:=${defaultbuilddir}}"
  82 : "${path_CODEMARK:=${defaultcodemark}}"
  83 : "${path_SYNTAXDIR:=${defaultsyntaxdir}}"
  84
  85 ## § Parameter processing
  86 ##
  87 ## A leading `--´ is allowed (and ignored).
  88 ## Otherwise, the first parameter is the source filename, and the
  89 ##   second the syntax id.
  90
  91 if "${cmd_TEST}" "${1}" = '--'
  92 then :
  93   shift
  94 fi
  95
  96 ## If no input file is provided, it defaults to `-´ (standard input).
  97
  98 if "${cmd_TEST}" -n "${1}" && "${cmd_TEST}" "${1}" != '-'
  99 then :
 100   inputfile="$(
 101     "${cmd_REALPATH}" -- "${1}"
 102   )"
 103 else :
 104   inputfile='-'
 105 fi
 106
 107 ## If no syntax is provided, it is taken from the extension of the
 108 ##   input file (the characters following the final `.´).
 109 ## If the input file does not have an extension, the syntax defaults to
 110 ##   `lesml´.
 111
 112 mainsyntax="${2}"
 113 if "${cmd_TEST}" -z "${mainsyntax}"
 114 then :
 115   mainsyntax="$(
 116     "${cmd_PRINTF}" '%s\n' "${inputfile}" |
 117     "${cmd_SED}" '/[.][^/][^/]*$/!s/$/.lesml/;s/.*[.]\([^/]*\)$/\1/'
 118   )"
 119 fi
 120
 121 ## The `sedprog´ variable gives the filename of the Sed program which
 122 ##   describes the current syntax.
 123
 124 if "${cmd_TEST}" "${mainsyntax}" = 'lesml'
 125 then :
 126   sedprog=
 127 else :
 128   sedprog="${path_SYNTAXDIR}"'/'"${mainsyntax}"'.sed'
 129 fi
 130
 131 ## It is an error if the main syntax is not `lesml´ and has no syntax
 132 ##   definition.
 133
 134 if "${cmd_TEST}" -n "${sedprog}" && "${cmd_TEST}" ! -f "${sedprog}"
 135 then :
 136   "${cmd_PRINTF}" '%s\n' 'Unrecognized syntax: '"${mainsyntax}"'.' >&2
 137   exit 1
 138 fi
 139
 140 ## § Processing the file
 141 ##
 142 ## The input file is processed and saved to a temporary file whose
 143 ##   filename is generated from the current time and a hash of the
 144 ##   input filename.
 145 ## The hope is that this is enough to prevent collisions (assuming two
 146 ##   versions of this script do not attempt to process the same file at
 147 ##   the same second).
 148
 149 temphash="$(
 150   "${cmd_PRINTF}" '%s\n' "${inputfile}" | "${cmd_CKSUM}"
 151 )"
 152 temptime="$(
 153   "${cmd_DATE}" -u '+%Y%m%d%H%M%S'
 154 )"
 155 tempfile="${path_BUILDDIR}"'/'"${temphash}"'@'"${temptime}"'.lesml'
 156
 157 ## Assuming the syntax is not `lesml´, the input file is processed to
 158 ##   produce Les·M·L output.
 159 ## This is then saved to the tempfile.
 160 ## If the syntax ⹐is⹑ `lesml´, the input file is simply copied.
 161
 162 if "${cmd_TEST}" "${mainsyntax}" = 'lesml'
 163 then :
 164   if "${cmd_TEST}" "${inputfile}" = '-'
 165   then :
 166     "${cmd_CAT}" >|"${tempfile}"
 167   else :
 168     "${cmd_CP}" -- "${inputfile}" "${tempfile}"
 169   fi
 170 else :
 171   if "${cmd_TEST}" "${inputfile}" = '-'
 172   then :
 173     "${cmd_SED}" -f "${sedprog}" |
 174     "${cmd_SED}" -f "${path_CODEMARK}"'/sed/syntax-comment.sed' |
 175     "${cmd_SED}" 's//'"${mainsyntax}"'/g' >|"${tempfile}"
 176   else :
 177     "${cmd_SED}" -f "${sedprog}" <"${inputfile}" |
 178     "${cmd_SED}" -f "${path_CODEMARK}"'/sed/syntax-comment.sed' |
 179     "${cmd_SED}" 's//'"${mainsyntax}"'/g' >|"${tempfile}"
 180   fi
 181 fi
 182
 183 ## Next, tagged code blocks in the Les·M·L need to be processed.
 184 ## First, a list of syntaxes must be collected.
 185
 186 syntaxes=$(
 187   "${cmd_SED}" '/^»|[^ $]*$ /!d;s/»|\([^ $]*\).*/\1/' <"${tempfile}" |
 188   "${cmd_SORT}" -u |
 189   "${cmd_TR}" -s '\n' ' '
 190 )
 191
 192 ## Each syntax can then be processed.
 193 ## Processing blanks out all lines not in the syntax, removes the
 194 ##   syntax prefix, processes the resulting file, and then restores the
 195 ##   blanked out lines.
 196 ## Unlike with the main syntax, missing syntaxes at this stage are
 197 ##   simply ignored.
 198
 199 for s in ${syntaxes}
 200 do :
 201   syntaxprog="${path_SYNTAXDIR}/${s}.sed"
 202   if "${cmd_TEST}" -f "${syntaxprog}"
 203   then :
 204     "${cmd_MV}" -- "${tempfile}" "${tempfile}"'~'
 205     "${cmd_SED}" \
 206       -e '/^»|'${s}'$ /!s/.*/\7f/;s/^»|'${s}'$ //' \
 207       -f "${syntaxprog}" \
 208       -e '/\7f/s/.*//' \
 209       <"${tempfile}"'~' |
 210     "${cmd_SED}" -f "${path_CODEMARK}"'/sed/syntax-comment.sed' |
 211     "${cmd_SED}" 's//'${s}'/g;/./s/^/»/' |
 212     "${cmd_AWK}" \
 213       -F '\037' \
 214       'FNR==NR{a[NR]=$0;next};{r=/^»\|'${s}'\$ /};r{print a[FNR]};!r' \
 215       - "${tempfile}"'~' \
 216       >|"${tempfile}"
 217     "${cmd_RM}" -- "${tempfile}"'~'
 218   fi
 219 done
 220
 221 ## Finally, the resulting file is written to standard output and
 222 ##   deleted.
 223
 224 if "${cmd_TEST}" -z "$(
 225   "${cmd_SED}" -n '/#!lesml/p;q' <"${tempfile}"
 226 )"
 227 then :
 228   defaultheader='#!lesml@'"${lesml_LANG}"'$ profile='"${lesml_PROFILE}"
 229   : "${lesml_LANG:=en}"
 230   : "${lesml_PROFILE:=urn:fdc:ladys.computer:20260226:codemark:pf:out}"
 231   : "${lesml_HEADER:=${defaultheader}}"
 232   "${cmd_PRINTF}" '%s\n' "${lesml_HEADER}"
 233 fi
 234 "${cmd_CAT}" -- "${tempfile}"
 235 "${cmd_RM}" -- "${tempfile}"