1From 6409a0c0efb552c6cbb8b8da17328bb65dd3473d Mon Sep 17 00:00:00 2001
2From: Michael Forney <mforney@mforney.org>
3Date: Tue, 10 Apr 2018 13:37:14 -0700
4Subject: [PATCH] m4: Use hand-written lexer to avoid cycle in bootstrap
5
6---
7 usr.bin/m4/tokenizer.c | 191 +++++++++++++++++++++++++++++++++++++++++
8 usr.bin/m4/tokenizer.l | 109 -----------------------
9 2 files changed, 191 insertions(+), 109 deletions(-)
10 create mode 100644 usr.bin/m4/tokenizer.c
11 delete mode 100644 usr.bin/m4/tokenizer.l
12
13diff --git a/usr.bin/m4/tokenizer.c b/usr.bin/m4/tokenizer.c
14new file mode 100644
15index 00000000000..fa19fc65035
16--- /dev/null
17+++ b/usr.bin/m4/tokenizer.c
18@@ -0,0 +1,191 @@
19+/* $OpenBSD: tokenizer.l,v 1.10 2017/06/17 01:55:16 bcallah Exp $ */
20+/*
21+ * Copyright (c) 2004 Marc Espie <espie@cvs.openbsd.org>
22+ *
23+ * Permission to use, copy, modify, and distribute this software for any
24+ * purpose with or without fee is hereby granted, provided that the above
25+ * copyright notice and this permission notice appear in all copies.
26+ *
27+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
28+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
29+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
30+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
31+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
32+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
33+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
34+ */
35+#include "parser.tab.h"
36+#include <assert.h>
37+#include <ctype.h>
38+#include <errno.h>
39+#include <limits.h>
40+#include <stdbool.h>
41+#include <stdio.h>
42+#include <stdlib.h>
43+#include <stdint.h>
44+
45+extern void m4_warnx(const char *, ...);
46+extern int mimic_gnu;
47+extern int32_t yylval;
48+static const char *yypos;
49+
50+void
51+yy_scan_string(const char *s)
52+{
53+ yypos = s;
54+}
55+
56+static int32_t
57+number(const char *yytext, size_t yylen)
58+{
59+ long l;
60+
61+ errno = 0;
62+ l = strtol(yytext, NULL, 0);
63+ if (((l == LONG_MAX || l == LONG_MIN) && errno == ERANGE) ||
64+ l > INT32_MAX || l < INT32_MIN)
65+ m4_warnx("numeric overflow in expr: %.*s", (int)yylen, yytext);
66+ return l;
67+}
68+
69+static int32_t
70+parse_radix(const char *yytext, size_t yylen)
71+{
72+ long base;
73+ char *next;
74+ long l;
75+ int d;
76+
77+ l = 0;
78+ base = strtol(yytext+2, &next, 0);
79+ if (base > 36 || next == NULL) {
80+ m4_warnx("error in number %.*s", (int)yylen, yytext);
81+ } else {
82+ next++;
83+ while (*next != 0) {
84+ if (*next >= '0' && *next <= '9')
85+ d = *next - '0';
86+ else if (*next >= 'a' && *next <= 'z')
87+ d = *next - 'a' + 10;
88+ else {
89+ assert(*next >= 'A' && *next <= 'Z');
90+ d = *next - 'A' + 10;
91+ }
92+ if (d >= base) {
93+ m4_warnx("error in number %.*s", (int)yylen, yytext);
94+ return 0;
95+ }
96+ l = base * l + d;
97+ next++;
98+ }
99+ }
100+ return l;
101+}
102+
103+static int
104+isodigit(int c)
105+{
106+ return c >= '0' && c <= '7';
107+}
108+
109+int yylex(void)
110+{
111+ const char *start;
112+
113+next:
114+ start = yypos;
115+ switch (*yypos) {
116+ case ' ':
117+ case '\t':
118+ case '\n':
119+ ++yypos;
120+ goto next;
121+ case '<':
122+ switch (yypos[1]) {
123+ case '=':
124+ yypos += 2;
125+ return LE;
126+ case '<':
127+ yypos += 2;
128+ return LSHIFT;
129+ }
130+ break;
131+ case '>':
132+ switch (yypos[1]) {
133+ case '=':
134+ yypos += 2;
135+ return GE;
136+ case '>':
137+ yypos += 2;
138+ return RSHIFT;
139+ }
140+ break;
141+ case '=':
142+ if (yypos[1] != '=')
143+ break;
144+ yypos += 2;
145+ return EQ;
146+ case '!':
147+ if (yypos[1] != '=')
148+ break;
149+ yypos += 2;
150+ return NE;
151+ case '&':
152+ if (yypos[1] != '&')
153+ break;
154+ yypos += 2;
155+ return LAND;
156+ case '|':
157+ if (yypos[1] != '|')
158+ break;
159+ yypos += 2;
160+ return LOR;
161+ case '*':
162+ if (!mimic_gnu || yypos[1] != '*')
163+ break;
164+ yypos += 2;
165+ return EXPONENT;
166+ case '0':
167+ switch (*++yypos) {
168+ case 'x':
169+ case 'X':
170+ if (!isxdigit(*++yypos))
171+ return ERROR;
172+ do ++yypos;
173+ while (isxdigit(*yypos));
174+ break;
175+ case 'r':
176+ case 'R':
177+ if (!mimic_gnu)
178+ break;
179+ if (!isdigit(*++yypos))
180+ return ERROR;
181+ do ++yypos;
182+ while (isdigit(*yypos));
183+ if (*yypos != ':')
184+ return ERROR;
185+ if (!isalnum(*++yypos))
186+ return ERROR;
187+ do ++yypos;
188+ while (isalnum(*yypos));
189+ yylval = parse_radix(start, yypos - start);
190+ return NUMBER;
191+ default:
192+ do ++yypos;
193+ while (isodigit(*yypos));
194+ break;
195+ }
196+ yylval = number(start, yypos - start);
197+ return NUMBER;
198+ case '\0':
199+ return '\0';
200+ }
201+ if (isdigit(*yypos)) {
202+ do ++yypos;
203+ while (isdigit(*yypos));
204+ yylval = number(start, yypos - start);
205+ return NUMBER;
206+ }
207+
208+ return *yypos++;
209+}
210diff --git a/usr.bin/m4/tokenizer.l b/usr.bin/m4/tokenizer.l
211deleted file mode 100644
212index 36f9b58853c..00000000000
213--- a/usr.bin/m4/tokenizer.l
214+++ /dev/null
215@@ -1,109 +0,0 @@
216-%{
217-/* $OpenBSD: tokenizer.l,v 1.11 2025/11/05 17:10:45 tb Exp $ */
218-/*
219- * Copyright (c) 2004 Marc Espie <espie@cvs.openbsd.org>
220- *
221- * Permission to use, copy, modify, and distribute this software for any
222- * purpose with or without fee is hereby granted, provided that the above
223- * copyright notice and this permission notice appear in all copies.
224- *
225- * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
226- * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
227- * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
228- * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
229- * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
230- * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
231- * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
232- */
233-#include "parser.h"
234-#include <assert.h>
235-#include <stdlib.h>
236-#include <errno.h>
237-#include <stdint.h>
238-#include <limits.h>
239-
240-extern void m4_warnx(const char *, ...);
241-extern int mimic_gnu;
242-extern int32_t yylval;
243-
244-int32_t number(void);
245-int32_t parse_radix(void);
246-%}
247-
248-delim [ \t\n]
249-ws {delim}+
250-hex 0[xX][0-9a-fA-F]+
251-oct 0[0-7]*
252-dec [1-9][0-9]*
253-radix 0[rR][0-9]+:[0-9a-zA-Z]+
254-
255-%option noyywrap
256-
257-%%
258-{ws} {/* just skip it */}
259-{hex}|{oct}|{dec} { yylval = number(); return(NUMBER); }
260-{radix} { if (mimic_gnu) {
261- yylval = parse_radix(); return(NUMBER);
262- } else {
263- return(ERROR);
264- }
265- }
266-"<=" { return(LE); }
267-">=" { return(GE); }
268-"<<" { return(LSHIFT); }
269-">>" { return(RSHIFT); }
270-"==" { return(EQ); }
271-"!=" { return(NE); }
272-"&&" { return(LAND); }
273-"||" { return(LOR); }
274-"**" { if (mimic_gnu) { return (EXPONENT); } }
275-. { return yytext[0]; }
276-%%
277-
278-int32_t
279-number(void)
280-{
281- long l;
282-
283- errno = 0;
284- l = strtol(yytext, NULL, 0);
285- if (((l == LONG_MAX || l == LONG_MIN) && errno == ERANGE) ||
286- l > INT32_MAX || l < INT32_MIN)
287- m4_warnx("numeric overflow in expr: %s", yytext);
288- return l;
289-}
290-
291-int32_t
292-parse_radix(void)
293-{
294- long base;
295- char *next;
296- long l;
297- int d;
298-
299- l = 0;
300- base = strtol(yytext+2, &next, 0);
301- if (base > 36 || next == NULL) {
302- m4_warnx("error in number %s", yytext);
303- } else {
304- next++;
305- while (*next != 0) {
306- if (*next >= '0' && *next <= '9')
307- d = *next - '0';
308- else if (*next >= 'a' && *next <= 'z')
309- d = *next - 'a' + 10;
310- else {
311- assert(*next >= 'A' && *next <= 'Z');
312- d = *next - 'A' + 10;
313- }
314- if (d >= base) {
315- m4_warnx("error in number %s", yytext);
316- return 0;
317- }
318- l = base * l + d;
319- next++;
320- }
321- }
322- return l;
323-}
324-
325--
3262.54.0
327
