tdevdraw: generate Latin-1 table from lib/keyboard - plan9port - [fork] Plan 9 from user space
(HTM) git clone git://src.adamsgaard.dk/plan9port
(DIR) Log
(DIR) Files
(DIR) Refs
(DIR) README
(DIR) LICENSE
---
(DIR) commit bb7ff349fb314edffabb01c418e146d563548058
(DIR) parent 342866f89cff5d35bc5e8fc7990767ad5d080a3c
(HTM) Author: Michael Teichgräber <mt@ib.wmipf.de>
Date: Wed, 8 Jul 2009 08:34:42 -0700
devdraw: generate Latin-1 table from lib/keyboard
Diffstat:
M lib/keyboard | 13 +++++++++----
M src/cmd/devdraw/latin1.c | 101 +------------------------------
M src/cmd/devdraw/mkfile | 10 ++++++++++
A src/cmd/devdraw/mklatinkbd.c | 232 ++++++++++++++++++++++++++++++
4 files changed, 252 insertions(+), 104 deletions(-)
---
(DIR) diff --git a/lib/keyboard b/lib/keyboard
t@@ -392,11 +392,14 @@
2020 dg † dagger
2021 dd ‡ double dagger
2022 bu • bullet
+2024 1. ․ one dot leader
+2025 2. ‥ two dot leader
+2026 3. … horizontal ellipsis
203D !? ?! ‽ interrobang
2070 s0 ⁰ superscript digit zero
-00B9 s1 ⁱ superscript digit one
-00B2 s2 superscript digit two
-00B3 s3 superscript digit three
+00B9 s1 ¹ superscript digit one
+00B2 s2 ² superscript digit two
+00B3 s3 ³ superscript digit three
2074 s4 ⁴ superscript digit four
2075 s5 ⁵ superscript digit five
2076 s6 ⁶ superscript digit six
t@@ -408,9 +411,10 @@
207C s= ⁼ superscript equals sign
207D s( ⁽ superscript opening parenthesis
207E s) ⁾ superscript closing parenthesis
+2071 si ⁱ superscript latin small letter i
207F sn ⁿ superscript latin small letter n
2013 en – en dash
-2014 em – em dash
+2014 em — em dash
2080 b0 ₀ subscript digit zero
2081 b1 ₁ subscript digit one
2082 b2 ₂ subscript digit two
t@@ -541,6 +545,7 @@
22E7 >!~ ⋧ greater than but not equivalent to
22EF el ⋯ midline horizontal ellipsis
2423 ␣ open box, visible space
+2639 :( ☹ sad face
263A :) ☺ smiley face
2654 wk ♔ white chess king
2655 wq ♕ white chess queen
(DIR) diff --git a/src/cmd/devdraw/latin1.c b/src/cmd/devdraw/latin1.c
t@@ -12,106 +12,7 @@ static struct cvlist
char *si; /* options for last input characters */
Rune so[60]; /* the corresponding Rune for each si entry */
} latintab[] = {
- " ", " i", { 0x2423, 0x0131 },
- "!~", "-=~", { 0x2244, 0x2247, 0x2249 },
- "!", "!<=>?bmp", { 0x00a1, 0x226e, 0x2260, 0x226f, 0x203d, 0x2284, 0x2209, 0x2285 },
- "\"*", "IUiu", { 0x03aa, 0x03ab, 0x03ca, 0x03cb },
- "\"", "\"AEIOUYaeiouy", { 0x00a8, 0x00c4, 0x00cb, 0x00cf, 0x00d6, 0x00dc, 0x0178, 0x00e4, 0x00eb, 0x00ef, 0x00f6, 0x00fc, 0x00ff },
- "$*", "fhk", { 0x03d5, 0x03d1, 0x03f0 },
- "$", "BEFHILMRVaefglopv", { 0x212c, 0x2130, 0x2131, 0x210b, 0x2110, 0x2112, 0x2133, 0x211b, 0x01b2, 0x0251, 0x212f, 0x0192, 0x210a, 0x2113, 0x2134, 0x2118, 0x028b },
- "\'\"", "Uu", { 0x01d7, 0x01d8 },
- "\'", "\'ACEILNORSUYZacegilnorsuyz", { 0x00b4, 0x00c1, 0x0106, 0x00c9, 0x00cd, 0x0139, 0x0143, 0x00d3, 0x0154, 0x015a, 0x00da, 0x00dd, 0x0179, 0x00e1, 0x0107, 0x00e9, 0x0123, 0x00ed, 0x013a, 0x0144, 0x00f3, 0x0155, 0x015b, 0x00fa, 0x00fd, 0x017a },
- "*", "*ABCDEFGHIKLMNOPQRSTUWXYZabcdefghiklmnopqrstuwxyz", { 0x2217, 0x0391, 0x0392, 0x039e, 0x0394, 0x0395, 0x03a6, 0x0393, 0x0398, 0x0399, 0x039a, 0x039b, 0x039c, 0x039d, 0x039f, 0x03a0, 0x03a8, 0x03a1, 0x03a3, 0x03a4, 0x03a5, 0x03a9, 0x03a7, 0x0397, 0x0396, 0x03b1, 0x03b2, 0x03be, 0x03b4, 0x03b5, 0x03c6, 0x03b3, 0x03b8, 0x03b9, 0x03ba, 0x03bb, 0x03bc, 0x03bd, 0x03bf, 0x03c0, 0x03c8, 0x03c1, 0x03c3, 0x03c4, 0x03c5, 0x03c9, 0x03c7, 0x03b7, 0x03b6 },
- "+", "-O", { 0x00b1, 0x2295 },
- ",", ",ACEGIKLNORSTUacegiklnorstu", { 0x00b8, 0x0104, 0x00c7, 0x0118, 0x0122, 0x012e, 0x0136, 0x013b, 0x0145, 0x01ea, 0x0156, 0x015e, 0x0162, 0x0172, 0x0105, 0x00e7, 0x0119, 0x0123, 0x012f, 0x0137, 0x013c, 0x0146, 0x01eb, 0x0157, 0x015f, 0x0163, 0x0173 },
- "-*", "l", { 0x019b },
- "-", "+-2:>DGHILOTZbdghiltuz~", { 0x2213, 0x00ad, 0x01bb, 0x00f7, 0x2192, 0x00d0, 0x01e4, 0x0126, 0x0197, 0x0141, 0x2296, 0x0166, 0x01b5, 0x0180, 0x00f0, 0x01e5, 0x210f, 0x0268, 0x0142, 0x0167, 0x0289, 0x01b6, 0x2242 },
- ".", ".CEGILOZceglz", { 0x00b7, 0x010a, 0x0116, 0x0120, 0x0130, 0x013f, 0x2299, 0x017b, 0x010b, 0x0117, 0x0121, 0x0140, 0x017c },
- "/", "Oo", { 0x00d8, 0x00f8 },
- "1", "234568", { 0x00bd, 0x2153, 0x00bc, 0x2155, 0x2159, 0x215b },
- "2", "-35", { 0x01bb, 0x2154, 0x2156 },
- "3", "458", { 0x00be, 0x2157, 0x215c },
- "4", "5", { 0x2158 },
- "5", "68", { 0x215a, 0x215d },
- "7", "8", { 0x215e },
- ":", "()-=", { 0x2639, 0x263a, 0x00f7, 0x2254 },
- "<!", "=~", { 0x2268, 0x22e6 },
- "<", "-<=>~", { 0x2190, 0x00ab, 0x2264, 0x2276, 0x2272 },
- "=", ":<=>OV", { 0x2255, 0x22dc, 0x2261, 0x22dd, 0x229c, 0x21d2 },
- ">!", "=~", { 0x2269, 0x22e7 },
- ">", "<=>~", { 0x2277, 0x2265, 0x00bb, 0x2273 },
- "?", "!?", { 0x203d, 0x00bf },
- "@\'", "\'", { 0x044a },
- "@@", "\'EKSTYZekstyz", { 0x044c, 0x0415, 0x041a, 0x0421, 0x0422, 0x042b, 0x0417, 0x0435, 0x043a, 0x0441, 0x0442, 0x044b, 0x0437 },
- "@C", "Hh", { 0x0427, 0x0427 },
- "@E", "Hh", { 0x042d, 0x042d },
- "@K", "Hh", { 0x0425, 0x0425 },
- "@S", "CHch", { 0x0429, 0x0428, 0x0429, 0x0428 },
- "@T", "Ss", { 0x0426, 0x0426 },
- "@Y", "AEOUaeou", { 0x042f, 0x0415, 0x0401, 0x042e, 0x042f, 0x0415, 0x0401, 0x042e },
- "@Z", "Hh", { 0x0416, 0x0416 },
- "@c", "h", { 0x0447 },
- "@e", "h", { 0x044d },
- "@k", "h", { 0x0445 },
- "@s", "ch", { 0x0449, 0x0448 },
- "@t", "s", { 0x0446 },
- "@y", "aeou", { 0x044f, 0x0435, 0x0451, 0x044e },
- "@z", "h", { 0x0436 },
- "@", "ABDFGIJLMNOPRUVXabdfgijlmnopruvx", { 0x0410, 0x0411, 0x0414, 0x0424, 0x0413, 0x0418, 0x0419, 0x041b, 0x041c, 0x041d, 0x041e, 0x041f, 0x0420, 0x0423, 0x0412, 0x0425, 0x0430, 0x0431, 0x0434, 0x0444, 0x0433, 0x0438, 0x0439, 0x043b, 0x043c, 0x043d, 0x043e, 0x043f, 0x0440, 0x0443, 0x0432, 0x0445 },
- "A", "E", { 0x00c6 },
- "C", "ACU", { 0x22c2, 0x2102, 0x22c3 },
- "Dv", "Zz", { 0x01c4, 0x01c5 },
- "D", "-e", { 0x00d0, 0x2206 },
- "G", "-", { 0x01e4 },
- "H", "-H", { 0x0126, 0x210d },
- "I", "-J", { 0x0197, 0x0132 },
- "L", "&-Jj|", { 0x22c0, 0x0141, 0x01c7, 0x01c8, 0x22c1 },
- "M", "#48bs", { 0x266e, 0x2669, 0x266a, 0x266d, 0x266f },
- "N", "JNj", { 0x01ca, 0x2115, 0x01cb },
- "O", "*+-./=EIcoprx", { 0x229b, 0x2295, 0x2296, 0x2299, 0x2298, 0x229c, 0x0152, 0x01a2, 0x00a9, 0x229a, 0x2117, 0x00ae, 0x2297 },
- "P", "P", { 0x2119 },
- "Q", "Q", { 0x211a },
- "R", "R", { 0x211d },
- "S", "123S", { 0x00b9, 0x00b2, 0x00b3, 0x00a7 },
- "T", "-u", { 0x0166, 0x22a8 },
- "V", "=", { 0x21d0 },
- "Y", "R", { 0x01a6 },
- "Z", "-ACSZ", { 0x01b5, 0xf015, 0xf017, 0xf016, 0x2124 },
- "^", "ACEGHIJOSUWYaceghijosuwy", { 0x00c2, 0x0108, 0x00ca, 0x011c, 0x0124, 0x00ce, 0x0134, 0x00d4, 0x015c, 0x00db, 0x0174, 0x0176, 0x00e2, 0x0109, 0x00ea, 0x011d, 0x0125, 0x00ee, 0x0135, 0x00f4, 0x015d, 0x00fb, 0x0175, 0x0177 },
- "_\"", "AUau", { 0x01de, 0x01d5, 0x01df, 0x01d6 },
- "_,", "Oo", { 0x01ec, 0x01ed },
- "_.", "Aa", { 0x01e0, 0x01e1 },
- "_", "AEIOU_aeiou", { 0x0100, 0x0112, 0x012a, 0x014c, 0x016a, 0x00af, 0x0101, 0x0113, 0x012b, 0x014d, 0x016b },
- "`\"", "Uu", { 0x01db, 0x01dc },
- "`", "AEIOUaeiou", { 0x00c0, 0x00c8, 0x00cc, 0x00d2, 0x00d9, 0x00e0, 0x00e8, 0x00ec, 0x00f2, 0x00f9 },
- "a", "ben", { 0x2194, 0x00e6, 0x2220 },
- "b", "()+-0123456789=bknpqru", { 0x208d, 0x208e, 0x208a, 0x208b, 0x2080, 0x2081, 0x2082, 0x2083, 0x2084, 0x2085, 0x2086, 0x2087, 0x2088, 0x2089, 0x208c, 0x265d, 0x265a, 0x265e, 0x265f, 0x265b, 0x265c, 0x2022 },
- "c", "$Oagu", { 0x00a2, 0x00a9, 0x2229, 0x2245, 0x222a },
- "dv", "z", { 0x01c6 },
- "d", "-adegz", { 0x00f0, 0x2193, 0x2021, 0x00b0, 0x2020, 0x02a3 },
- "e", "$lmns", { 0x20ac, 0x22ef, 0x2014, 0x2013, 0x2205 },
- "f", "a", { 0x2200 },
- "g", "$-r", { 0x00a4, 0x01e5, 0x2207 },
- "h", "-v", { 0x210f, 0x0195 },
- "i", "-bfjps", { 0x0268, 0x2286, 0x221e, 0x0133, 0x2287, 0x222b },
- "l", "\"$&\'-jz|", { 0x201c, 0x00a3, 0x2227, 0x2018, 0x0142, 0x01c9, 0x22c4, 0x2228 },
- "m", "iou", { 0x00b5, 0x2208, 0x00d7 },
- "n", "jo", { 0x01cc, 0x00ac },
- "o", "AOUaeiu", { 0x00c5, 0x229a, 0x016e, 0x00e5, 0x0153, 0x01a3, 0x016f },
- "p", "Odgrt", { 0x2117, 0x2202, 0x00b6, 0x220f, 0x221d },
- "r", "\"\'O", { 0x201d, 0x2019, 0x00ae },
- "s", "()+-0123456789=abinoprstu", { 0x207d, 0x207e, 0x207a, 0x207b, 0x2070, 0x00b9, 0x00b2, 0x00b3, 0x2074, 0x2075, 0x2076, 0x2077, 0x2078, 0x2079, 0x207c, 0x00aa, 0x2282, 0x2071, 0x207f, 0x00ba, 0x2283, 0x221a, 0x00df, 0x220d, 0x2211 },
- "t", "-efmsu", { 0x0167, 0x2203, 0x2234, 0x2122, 0x03c2, 0x22a2 },
- "u", "-AEGIOUaegiou", { 0x0289, 0x0102, 0x0114, 0x011e, 0x012c, 0x014e, 0x016c, 0x2191, 0x0115, 0x011f, 0x012d, 0x014f, 0x016d },
- "v\"", "Uu", { 0x01d9, 0x01da },
- "v", "ACDEGIKLNORSTUZacdegijklnorstuz", { 0x01cd, 0x010c, 0x010e, 0x011a, 0x01e6, 0x01cf, 0x01e8, 0x013d, 0x0147, 0x01d1, 0x0158, 0x0160, 0x0164, 0x01d3, 0x017d, 0x01ce, 0x010d, 0x010f, 0x011b, 0x01e7, 0x01d0, 0x01f0, 0x01e9, 0x013e, 0x0148, 0x01d2, 0x0159, 0x0161, 0x0165, 0x01d4, 0x017e },
- "w", "bknpqr", { 0x2657, 0x2654, 0x2658, 0x2659, 0x2655, 0x2656 },
- "x", "O", { 0x2297 },
- "y", "$", { 0x00a5 },
- "z", "-", { 0x01b6 },
- "|", "Pp|", { 0x00de, 0x00fe, 0x00a6 },
- "~!", "=", { 0x2246 },
- "~", "-=AINOUainou~", { 0x2243, 0x2245, 0x00c3, 0x0128, 0x00d1, 0x00d5, 0x0168, 0x00e3, 0x0129, 0x00f1, 0x00f5, 0x0169, 0x2248 },
+#include "latin1.h"
0, 0, { 0 }
};
(DIR) diff --git a/src/cmd/devdraw/mkfile b/src/cmd/devdraw/mkfile
t@@ -22,3 +22,13 @@ $O.drawclient: drawclient.$O drawfcall.$O
$O.snarf: x11-alloc.$O x11-cload.$O x11-draw.$O x11-fill.$O x11-get.$O x11-init.$O x11-itrans.$O x11-keysym2ucs.$O x11-load.$O x11-pixelbits.$O x11-unload.$O x11-wsys.$O snarf.$O latin1.$O devdraw.$O
$LD -o $target $prereq
+
+$O.mklatinkbd: mklatinkbd.$O
+ $LD -o $target $prereq
+
+latin1.$O: latin1.h
+
+latin1.h: $PLAN9/lib/keyboard $O.mklatinkbd
+ ./$O.mklatinkbd -r $PLAN9/lib/keyboard | sed 's/, }/ }/' >$target
+
+CLEANFILES=latin1.h $O.mklatinkbd
(DIR) diff --git a/src/cmd/devdraw/mklatinkbd.c b/src/cmd/devdraw/mklatinkbd.c
t@@ -0,0 +1,232 @@
+/*
+ * Parse /lib/keyboard to create latin1.h table for kernel.
+ * mklatinkbd -r prints an array of integers rather than a Rune string literal.
+ */
+
+#include <u.h>
+#include <libc.h>
+#include <bio.h>
+#include <ctype.h>
+
+int rflag;
+
+enum {
+ MAXLD = 2, /* latin1.c assumes this is 2 */
+};
+
+char *head = ""
+"/*\n"
+" * This is automatically generated by %s from /lib/keyboard\n"
+" * Edit /lib/keyboard instead.\n"
+" */\n";
+
+/*
+ * latin1.c assumes that strlen(ld) is at most 2.
+ * It also assumes that latintab[i].ld can be a prefix of latintab[j].ld
+ * only when j < i. We ensure this by sorting the output by prefix length.
+ * The so array is indexed by the character value.
+ */
+
+typedef struct Trie Trie;
+struct Trie {
+ int n; /* of characters r */
+ char seq[MAXLD+1];
+ Rune r[256];
+ Trie *link[256];
+};
+
+Trie *root;
+
+Trie*
+mktrie(char *seq)
+{
+ uchar *q;
+ Trie **tp;
+
+ if(root == nil) {
+ root = malloc(sizeof *root);
+ memset(root, 0, sizeof *root);
+ }
+
+ assert(seq[0] != '\0');
+
+ tp = &root;
+ for(q=(uchar*)seq; *(q+1) != '\0'; q++) {
+ tp = &(*tp)->link[*q];
+ if(*tp == nil) {
+ *tp = malloc(sizeof(**tp));
+ assert(*tp != nil);
+ memset(*tp, 0, sizeof(**tp));
+ strcpy((*tp)->seq, seq);
+ (*tp)->seq[q+1-(uchar*)seq] = '\0';
+ }
+ }
+
+ assert(*tp != nil);
+ return *tp;
+}
+
+/* add character sequence s meaning rune r */
+void
+insert(char *s, Rune r)
+{
+ uchar lastc;
+ int len;
+ Trie *t;
+
+ len = strlen(s);
+ lastc = (uchar)s[len-1];
+
+ t = mktrie(s);
+ if(t->r[lastc]) {
+ fprint(2, "warning: table duplicate: %s is %C and %C\n", s, t->r[lastc], r);
+ return;
+ }
+ t->r[lastc] = r;
+ t->n++;
+}
+
+void
+cprintchar(Biobuf *b, int c)
+{
+ /* print a byte c safe for a C string. */
+ switch(c) {
+ case '\'':
+ case '\"':
+ case '\\':
+ Bprint(b, "\\%c", c);
+ break;
+ case '\t':
+ Bprint(b, "\\t");
+ break;
+ default:
+ if(isascii(c) && isprint(c))
+ Bprint(b, "%c", c);
+ else
+ Bprint(b, "\\x%.2x", c);
+ break;
+ }
+}
+
+void
+cprints(Biobuf *b, char *p)
+{
+ while(*p != '\0')
+ cprintchar(b, *p++);
+}
+
+
+void
+printtrie(Biobuf *b, Trie *t)
+{
+ int i;
+
+ for(i=0; i<256; i++)
+ if(t->link[i])
+ printtrie(b, t->link[i]);
+
+ if(t->n > 0) {
+ Bprint(b, "\t\"");
+ cprints(b, t->seq);
+ Bprint(b, "\", \"");
+ for(i=0; i<256; i++)
+ if(t->r[i])
+ cprintchar(b, i);
+ Bprint(b, "\",\t");
+ if(rflag) {
+ Bprint(b, "{");
+ for(i=0; i<256; i++)
+ if(t->r[i])
+ Bprint(b, " 0x%.4ux,", t->r[i]);
+ Bprint(b, " }");
+ } else {
+ Bprint(b, "L\"");
+ for(i=0; i<256; i++)
+ if(t->r[i])
+ Bprint(b, "%C", t->r[i]);
+ Bprint(b, "\"");
+ }
+ Bprint(b, ",\n");
+ }
+}
+
+void
+readfile(char *fname)
+{
+ Biobuf *b;
+ char *line, *p;
+ char *seq;
+ int inseq;
+ int lineno;
+ Rune r;
+
+ if((b = Bopen(fname, OREAD)) == 0) {
+ fprint(2, "cannot open \"%s\": %r\n", fname);
+ exits("open");
+ }
+
+ lineno = 0;
+ while((line = Brdline(b, '\n')) != 0) {
+ lineno++;
+ if(line[0] == '#')
+ continue;
+
+ r = strtol(line, nil, 16);
+ p = strchr(line, ' ');
+ if(r == 0 || p != line+4 || p[0] != ' ' || p[1] != ' ') {
+ fprint(2, "%s:%d: cannot parse line\n", fname, lineno);
+ continue;
+ }
+
+ p = line+6;
+/* 00AE Or rO ® registered trade mark sign */
+ for(inseq=1, seq=p; (uchar)*p < Runeself; p++) {
+ if(*p == '\0' || isspace(*p)) {
+ if(inseq && p-seq >= 2) {
+ *p = '\0';
+ inseq = 0;
+ insert(seq, r);
+ *p = ' ';
+ }
+ if(*p == '\0')
+ break;
+ } else {
+ if(!inseq) {
+ seq = p;
+ inseq = 1;
+ }
+ }
+ }
+ }
+}
+
+void
+usage(void)
+{
+ fprint(2, "usage: mklatinkbd [-r] [/lib/keyboard]\n");
+ exits("usage");
+}
+
+void
+main(int argc, char **argv)
+{
+ Biobuf bout;
+
+ ARGBEGIN{
+ case 'r': /* print rune values */
+ rflag = 1;
+ break;
+ default:
+ usage();
+ }ARGEND
+
+ if(argc > 1)
+ usage();
+
+ readfile(argc == 1 ? argv[0] : "/fd/0");
+
+ Binit(&bout, 1, OWRITE);
+ if(root)
+ printtrie(&bout, root);
+ exits(0);
+}