sumsrc

#!/bin/bash
_is_yielded=false;
yell() { echo "$@" >&2; }
die() { yell "$@"; exit 111; }
flag() { _is_yielded=true; yell "ERROR: $@"; }
yield() { if [ "$_is_yielded" == "true" ]; then die "Aborting because of above errors."; fi; }

SCRIPT_NAME="${0##*/}";
DEFAULT_DIGEST_LENGTH=32;
DEFAULT_OUTPUT_FORMAT='query';
DEFAULT_ALGORITHM_NAME_EXE='b2sum';
DEFAULT_ALGORITHM_NAME_SHORT='b2';
showHelp() {
	echo "USAGE (STDIN or PATH modes):";
	echo "[…] | ${SCRIPT_NAME} [OPTIONS]";
	echo "${SCRIPT_NAME} [OPTIONS] PATH";
	echo "";
	echo "OPTIONS:";
	echo " -h    Show this help";
  echo " -a A  Set algorithm to A; any name that has a *sum command, e.g. b2, sha256.";
	#echo " -f FORMAT";
	#echo "       Specify source FORMAT, so that comments can be escaped correctly (only sh and vim are used right now)";
	echo " -j, --json    Output JSON";
	echo " -y, --yaml    Output YAML";
	echo " -t, --text    Output plain text";
	echo " -q, --query   Output query string (default)";
	echo " -l N  Set digest length in bits, default ${DEFAULT_DIGEST_LENGTH} (see ${DEFAULT_ALGORITHM_NAME_SHORT} -l)";
	echo "";
	echo "EXAMPLE:";
	echo "  cat myfile.txt | ${SCRIPT_NAME}";
}

algorithmNameExe="$DEFAULT_ALGORITHM_NAME_EXE";
algorithmNameShort="$DEFAULT_ALGORITHM_NAME_SHORT";
targetFile=
opt_sumLength=$DEFAULT_DIGEST_LENGTH;
opt_outputFormat=$DEFAULT_OUTPUT_FORMAT;
while [ $# -gt 0 ]; do
	case $1 in
		-h) showHelp; exit;;
		#-f) opt_inputFormat="$2"; shift;;
    -a)
      a="${2}";
      ax="${a}sum";
      if command -v "$ax" &>/dev/null; then
        algorithmNameShort="$a";
        algorithmNameExe="$ax";
        shift;
      fi;;
		-j|--json) opt_outputFormat='json';;
		-y|--yaml) opt_outputFormat='yaml';;
		-t|--text) opt_outputFormat='plain';;
		-q|--query) opt_outputFormat='query';;
		-l) opt_sumLength=$2; shift;;
		*) if [[ -z $targetFile ]]; then targetFile="$1"; else flag "Unknown arg '$1'"; fi;;
	esac
	shift;
done

if [[ -t 0 && ( -z "$targetFile" || ! -s "$targetFile" ) ]]; then
	echo "Please provide input data, by STDIN (e.g. 'cat ~/myfile.txt | ${SCRIPT_NAME}') or as first non-option argument (e.g. '${SCRIPT_NAME} myfile.txt')" >&2;
	echo '' >&2;
	showHelp >&2;
	exit 1;
fi

[[ $opt_sumLength =~ ^[0-9]+$ ]] || flag "Please specify only integer numbers for the length value of option '-l'.";

yield;

getOmitLinesPattern() {
	local commentLineStart='#';
	#local blockCommentPattern=''; # WISH, LATER: omit block comments. Maybe use some other program to do the comment identification (Vim)?
	local filename=$1;
	#echo "DEBUG(filename): ${filename}"; #XXX
	if [[ -n $filename ]]; then
		local ftype="${filename##*.}";
		ftype="${ftype##*/}";
		case $ftype in
			vim|vimrc) commentLineStart='"';;
			js|javascript|ts|typescript|c|cpp) commentLineStart='//';;
			sh) commentLineStart='##';;
			# LATER: add more.
		esac
	fi
	local OMIT_KEYWORD='nosum';
	# NOTE: grep is using -E extended RegExp option, so pipes are not escaped.
	local pattern='^\s*$'; # empty line
	pattern+='|^\s*'"${commentLineStart}"'[^!]'; # commented line, besides execution hint  FIXME (B) 2024-06-05 only applies when commentLineStart='#'
	pattern+='|'"${commentLineStart}${OMIT_KEYWORD}"'$'; # lines ENDING like '#nosum'
	echo "$pattern";
}

if [[ -n $targetFile ]]; then
	omitPattern="$(getOmitLinesPattern "$targetFile")";
	data="$(grep -E -v "$omitPattern" "$targetFile")";
else
	omitPattern="$(getOmitLinesPattern)";
	data="$(grep -E -v "$omitPattern")";
fi
#>&2 echo -n "[DEBUG(omitPattern):\n${omitPattern}]";
#>&2 echo -n "[DEBUG(DATA):\n${data}]" | head -20;exit 130;
sumCommand=("$algorithmNameExe");
if [[ $algorithmNameShort == 'b2' && $opt_sumLength -gt 0 ]]; then
  sumCommand+=(-l ${opt_sumLength});
fi
sum="$("${sumCommand[@]}" <(echo "$data") | awk '{ print $1 }')";
size="$(echo "$data" | wc -c| awk '{ print $1 }')";
case $opt_outputFormat in
  json) echo "{\"sum\":\"${sum}\",\"size\":${size},\"date\":\"$(date '+%Y-%m-%dT%H:%M:%S%z')\",\"provider\":{\"name\":\"${SCRIPT_NAME}\",\"method\":\"${algorithmNameExe}\",\"length\":${opt_sumLength}}}";;
  yaml) echo -e "---\nsum: ${sum}\nsize: ${size}\nprovider:\n  name: ${SCRIPT_NAME}\n  method: ${algorithmNameExe}\n  length: ${opt_sumLength}\ndate: $(date '+%Y-%m-%dT%H:%M:%S%z')\n...";;
	plain) echo "sum:${sum}#a=${algorithmNameShort},l=${opt_sumLength};size:${size};";;
	query) echo "${algorithmNameExe}=${sum}&size=${size}";;
	*) die "Unknown output form at '${opt_outputFormat}'";;
esac

## HISTORY:
# v0.1.0 2022-12-13  Optional JSON output. Adding -h documentation.   b2sum=0c305a3b&size=1592
# v0.2.0 2022-12-16  2nd mode, PATH arg instead of STDIN; Renamed 'sumsrc' from 'sum-stream'.   b2sum=bf10f132&size=2041
# v0.3.0 2022-12-19  Add query output format. Include '#!...' execution comment in sum. Change 'sum' key to '*sum' where '*' is algorithm short name.   b2sum=b64e99f0&size=2364
# v0.4.0 2022-12-20  Adapt comment pattern to input source language (stub: only Vim by file name).   b2sum=beaa596f&size=2691
# v0.5.0 2024-07-07  Fix omit pattern for grep -E. Adding to shell-utilities.   b2sum=d23acaeb&size=2787
# v0.5.1 2025-04-25  Add full flags for --text, --query, for parameterized scripting.   b2sum=c26fba89&size=3871
# STUB  2024-06-05  Support more formats, and allow specifying with input option arg.