From 042e43247f396a9000fead59d9bff87bf12806d6 Mon Sep 17 00:00:00 2001
From: Thomas Voss <mail@thomasvoss.com>
Date: Wed, 30 Oct 2024 01:51:14 +0100
Subject: Completely revamp the grab source code

Some of the (many) few changes are:
- Multithreading for significantly faster performance
- The -p/--predicate flag
- Byte offsets as the default
- No customizable colors (maybe this will come back later)
- Newer edition of mlib (formerly librune)
---
 vendor/librune/gen/gbrk           | 114 -------------------------------------
 vendor/librune/gen/mkfile         |  37 ------------
 vendor/librune/gen/rtype-cat      | 104 ----------------------------------
 vendor/librune/gen/rtype-prop     | 116 --------------------------------------
 vendor/librune/gen/rtype-prop.awk |  61 --------------------
 5 files changed, 432 deletions(-)
 delete mode 100755 vendor/librune/gen/gbrk
 delete mode 100755 vendor/librune/gen/mkfile
 delete mode 100755 vendor/librune/gen/rtype-cat
 delete mode 100755 vendor/librune/gen/rtype-prop
 delete mode 100644 vendor/librune/gen/rtype-prop.awk

(limited to 'vendor/librune/gen')

diff --git a/vendor/librune/gen/gbrk b/vendor/librune/gen/gbrk
deleted file mode 100755
index 1146327..0000000
--- a/vendor/librune/gen/gbrk
+++ /dev/null
@@ -1,114 +0,0 @@
-#!/bin/sh
-
-cache()
-{
-	name="/tmp/librune/gbrk/$(basename "$1")"
-	if test ! -f "$name"
-	then
-		mkdir -p /tmp/librune/gbrk
-		wget -q "$1" -O "$name"
-	fi
-}
-
-set -e
-cd "${0%/*}/.."
-exec >include/internal/gbrk_lookup.h
-
-readonly URL1='https://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakProperty.txt'
-readonly URL2='https://www.unicode.org/Public/UCD/latest/ucd/emoji/emoji-data.txt'
-readonly URL3='https://www.unicode.org/Public/UCD/latest/ucd/DerivedCoreProperties.txt'
-
-cache "$URL1" &
-cache "$URL2" &
-cache "$URL3" &
-wait
-
-cat <<C
-/* This file is autogenerated by gen/gbrk; DO NOT EDIT. */
-
-#ifndef RUNE_INTERNAL_GBRK_LOOKUP_H
-#define RUNE_INTERNAL_GBRK_LOOKUP_H
-
-/* IWYU pragma: private */
-/* clang-format off */
-
-#include "types.h"
-
-typedef enum {
-	GBP_OTHER = 0,
-
-	GBP_CTRL = 1 << 0, /* Control */
-	GBP_EXT  = 1 << 1, /* Extend */
-	GBP_PIC  = 1 << 2, /* Extended_Pictographic */
-	GBP_PREP = 1 << 3, /* Prepend */
-	GBP_RI   = 1 << 4, /* Regional_Indicator */
-	GBP_SM   = 1 << 5, /* SpacingMark */
-	GBP_ZWJ  = 1 << 6, /* ZWJ */
-
-	GBP_HNGL_L   = 1 <<  7, /* Hangul L */
-	GBP_HNGL_LV  = 1 <<  8, /* Hangul LV */
-	GBP_HNGL_LVT = 1 <<  9, /* Hangul LVT */
-	GBP_HNGL_T   = 1 << 10, /* Hangul T */
-	GBP_HNGL_V   = 1 << 11, /* Hangul V */
-
-	GBP_INDC_CNSNT = 1 << 12, /* Indic Consonant */
-	GBP_INDC_EXT   = 1 << 13, /* Indic Extend */
-	GBP_INDC_LNK   = 1 << 14, /* Indic Linker */
-} gbrk_prop;
-
-static const struct {
-	rune lo, hi;
-	gbrk_prop prop;
-} gbrk_prop_tbl[] = {
-C
-
-gawk '
-BEGIN {
-	FS = "( *#.*| +; +)"
-	map["Control"]               = "CTRL"
-	map["Extend"]                = "EXT"
-	map["Extended_Pictographic"] = "PIC"
-	map["Prepend"]               = "PREP"
-	map["Regional_Indicator"]    = "RI"
-	map["SpacingMark"]           = "SM"
-	map["ZWJ"]                   = "ZWJ"
-
-	map["L"]   = "HNGL_L"
-	map["LV"]  = "HNGL_LV"
-	map["LVT"] = "HNGL_LVT"
-	map["T"]   = "HNGL_T"
-	map["V"]   = "HNGL_V"
-
-	map["InCB; Consonant"] = "INDC_CNSNT"
-	map["InCB; Extend"]    = "INDC_EXT"
-	map["InCB; Linker"]    = "INDC_LNK"
-}
-
-map[$2] {
-	n = split($1, a, /\.\./)
-	lo = strtonum("0X" a[1])
-	hi = strtonum("0X" a[n])
-
-	for (i = lo; i <= hi; i++) {
-		s = "GBP_" map[$2]
-		props[i] = props[i] ? props[i] " | " s : s
-	}
-}
-
-END {
-	for (i = 0; i <= 0x10FFFF; i++) {
-		if (!props[i])
-			continue
-		lo = i
-		while (props[lo] == props[i + 1])
-			i++
-		printf "\t{0x%06X, 0x%06X, %s},\n", lo, i, props[lo]
-	}
-}
-' /tmp/librune/gbrk/* | sort
-
-cat <<C
-};
-
-#endif /* !RUNE_INTERNAL_GBRK_LOOKUP_H */
-C
diff --git a/vendor/librune/gen/mkfile b/vendor/librune/gen/mkfile
deleted file mode 100755
index a55b08a..0000000
--- a/vendor/librune/gen/mkfile
+++ /dev/null
@@ -1,37 +0,0 @@
-#!/bin/sh
-
-set -e
-cd "${0%/*}/.."
-exec >Makefile
-
-cat <<make
-.POSIX:
-
-CC = cc
-CFLAGS = -Wall -Wextra -Wpedantic -g -ggdb3 -fanalyzer -Iinclude
-
-make
-
-printf 'objs = \\\n'
-find lib -name '*.c' | sort | sed '
-	s/c$/o/
-	s/.*/\t& \\/
-	$s/ \\$//
-'
-
-cat <<make
-
-all: \$(objs)
-
-make
-
-find lib -name '*.c' | sort | sed -E 's/(.*)c$/\1o: \1c/'
-
-cat <<make
-
-clean:
-	find lib -name '*.o' -delete
-
-ls:
-	@echo "alias ls='ls --color=auto -I \"*.o\" -I \"compile_commands.json\"'" >&2
-make
diff --git a/vendor/librune/gen/rtype-cat b/vendor/librune/gen/rtype-cat
deleted file mode 100755
index e35fb77..0000000
--- a/vendor/librune/gen/rtype-cat
+++ /dev/null
@@ -1,104 +0,0 @@
-#!/bin/sh
-
-cache()
-{
-	name="/tmp/librune/rtype/$(basename "$1")"
-	if test ! -f "$name"
-	then
-		mkdir -p /tmp/librune/rtype
-		wget -q "$1" -O "$name"
-	fi
-}
-
-set -e
-cd "${0%/*}/.."
-exec >include/internal/rtype/cat.h
-
-readonly URL='https://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt'
-cache "$URL"
-
-cat <<C
-/* This file is autogenerated by gen/rtype-cat; DO NOT EDIT. */
-
-#ifndef RUNE_INTERNAL_RTYPE_CAT_H
-#define RUNE_INTERNAL_RTYPE_CAT_H
-
-/* IWYU pragma: private */
-/* clang-format off */
-
-#include "../types.h"
-#include "../../rtype.h"
-
-static const enum unicat rtype_cat_lat1_tbl[] = {
-C
-
-gawk '
-BEGIN {
-	FS = ";"
-}
-
-{
-	s = "UC_" toupper($3)
-	lo = strtonum("0X" $1)
-
-	if ($2 ~ /First/) {
-		getline
-		hi = strtonum("0X" $1)
-	} else
-		hi = lo
-
-	for (i = lo; i <= hi; i++)
-		props[i] = s
-}
-
-END {
-	for (i = 0; i <= 0xFF; i++)
-		print props[i] ","
-}
-' /tmp/librune/rtype/* | paste -d' ' - - - - - - - - | sed 's/^/\t/'
-
-cat <<C
-};
-
-static const struct {
-	rune lo, hi;
-	enum unicat val;
-} rtype_cat_tbl[] = {
-C
-
-gawk '
-BEGIN {
-	FS = ";"
-}
-
-{
-	s = "UC_" toupper($3)
-	lo = strtonum("0X" $1)
-
-	if ($2 ~ /First/) {
-		getline
-		hi = strtonum("0X" $1)
-	} else
-		hi = lo
-
-	for (i = lo; i <= hi; i++)
-		props[i] = s
-}
-
-END {
-	for (i = 0x100; i <= 0x10FFFF; i++) {
-		if (!props[i])
-			continue
-		lo = i
-		while (props[lo] == props[i + 1])
-			i++
-		printf "\t{0x%06X, 0x%06X, %s},\n", lo, i, props[lo]
-	}
-}
-' /tmp/librune/rtype/* | sort
-
-cat <<C
-};
-
-#endif /* !RUNE_INTERNAL_RTYPE_CAT_H */
-C
diff --git a/vendor/librune/gen/rtype-prop b/vendor/librune/gen/rtype-prop
deleted file mode 100755
index 4c62884..0000000
--- a/vendor/librune/gen/rtype-prop
+++ /dev/null
@@ -1,116 +0,0 @@
-#!/bin/sh
-
-cache()
-{
-	name="/tmp/librune/rtype/$(basename "$1")"
-	if test ! -f "$name"
-	then
-		mkdir -p /tmp/librune/rtype
-		wget -q "$1" -O "$name"
-	fi
-}
-
-cd "${0%/*}/.."
-
-readonly URL1='https://www.unicode.org/Public/UCD/latest/ucd/PropList.txt'
-readonly URL2='https://www.unicode.org/Public/UCD/latest/ucd/DerivedCoreProperties.txt'
-
-cache "$URL1" &
-cache "$URL2" &
-wait
-
-props1='
-ASCII_Hex_Digit
-Bidi_Control
-Dash
-Deprecated
-Diacritic
-Extender
-Hex_Digit
-ID_Compat_Math_Continue
-ID_Compat_Math_Start
-Ideographic
-IDS_Binary_Operator
-IDS_Trinary_Operator
-IDS_Unary_Operator
-Join_Control
-Logical_Order_Exception
-Noncharacter_Code_Point
-Pattern_Syntax
-Pattern_White_Space
-Prepended_Concatenation_Mark
-Quotation_Mark
-Radical
-Regional_Indicator
-Sentence_Terminal
-Soft_Dotted
-Terminal_Punctuation
-Unified_Ideograph
-Variation_Selector
-White_Space
-'
-
-props2='
-Alphabetic
-Cased
-Case_Ignorable
-Changes_When_Casefolded
-Changes_When_Casemapped
-Changes_When_Lowercased
-Changes_When_Titlecased
-Changes_When_Uppercased
-Default_Ignorable_Code_Point
-Grapheme_Base
-Grapheme_Extend
-ID_Continue
-ID_Start
-Indic_Conjunct_Break
-Lowercase
-Math
-Uppercase
-XID_Continue
-XID_Start
-'
-# InCB
-
-{
-	for prop in $props1
-	do
-		lprop=$(echo $prop | tr A-Z a-z)
-
-		printf 'rune_has_prop_%s…' $lprop >&2
-
-		gawk -M -v prop=$prop -f gen/rtype-prop.awk \
-			/tmp/librune/rtype/PropList.txt \
-			>lib/rtype/rune_has_prop_$lprop.c
-
-		echo "[[unsequenced]] bool rune_has_prop_$lprop(rune);"
-		echo ' DONE' >&2
-	done
-
-	for prop in $props2
-	do
-		lprop=$(echo $prop | tr A-Z a-z)
-
-		printf 'rune_has_prop_%s…' $lprop >&2
-
-		gawk -M -v prop=$prop -f gen/rtype-prop.awk \
-			/tmp/librune/rtype/DerivedCoreProperties.txt \
-			>lib/rtype/rune_has_prop_$lprop.c
-
-		echo "[[unsequenced]] bool rune_has_prop_$lprop(rune);"
-		echo ' DONE' >&2
-	done
-} | gawk '
-	/PROP PREDICATES END/ { no = 0 }
-	FILENAME != "-" && !no { print }
-	FILENAME == "-" { funcs[++i] = $0 }
-
-	/PROP PREDICATES START/ {
-		no = 1
-		asort(funcs)
-		for (i = 1; i <= length(funcs); i++)
-			print funcs[i]
-	}
-' - include/rtype.h \
-| sponge include/rtype.h
diff --git a/vendor/librune/gen/rtype-prop.awk b/vendor/librune/gen/rtype-prop.awk
deleted file mode 100644
index 59b4a99..0000000
--- a/vendor/librune/gen/rtype-prop.awk
+++ /dev/null
@@ -1,61 +0,0 @@
-BEGIN {
-	FS = "( *#.*| +; +)"
-
-	print "/* This file is autogenerated by gen/rtype-prop; DO NOT EDIT. */"
-	print ""
-	print "#include \"rtype.h\""
-	print ""
-	print "#include \"internal/common.h\""
-	print ""
-}
-
-$2 == prop || (prop == "Indic_Conjunct_Break" && $2 ~ /InCB;/) {
-	n = split($1, a, /\.\./)
-	lo = strtonum("0x" a[1])
-	hi = strtonum("0x" a[n])
-
-	for (i = lo; i <= hi; i++)
-		xs[i] = 1
-}
-
-END {
-	for (i = 0; i <= 0xFF; i++) {
-		if (xs[i])
-			mask = or(mask, lshift(1, i))
-	}
-	print  "#if BIT_LOOKUP"
-	printf "static const unsigned _BitInt(LATIN1_MAX + 1) mask = 0x%Xuwb;\n", \
-		mask
-	print  "#endif"
-	print  ""
-	print  "static const struct {"
-	print  "\trune lo, hi;"
-	print  "} lookup_tbl[] = {"
-
-	for (i = 0; i <= 0x10FFFF; i++) {
-		if (!xs[i])
-			continue
-		lo = i
-		while (xs[i + 1])
-			i++
-		printf "\t{0x%06X, 0x%06X},\n", lo, i
-	}
-
-	print  "};"
-	print  ""
-	print  "#define TYPE      bool"
-	print  "#define TABLE     lookup_tbl"
-	print  "#define DEFAULT   false"
-	print  "#define HAS_VALUE 0"
-	print  "#include \"internal/rtype/lookup-func.h\""
-	print  ""
-	print  "bool"
-	printf "rune_has_prop_%s(rune ch)\n", tolower(prop)
-	print  "{"
-	print  "\treturn"
-	print  "#if BIT_LOOKUP"
-	print  "\t\tch <= LATIN1_MAX ? (mask & ch) :"
-	print  "#endif"
-	print  "\t\tlookup(ch);"
-	print  "}"
-}
-- 
cgit v1.2.3