1 #include "rc.h"
2 #include "exec.h"
3 #include "io.h"
4 #include "getflags.h"
5 #include "fns.h"
6 int getnext(void);
7 
8 int
9 wordchr(int c)
10 {
11 	return !strchr("\n \t#;&|^$=`'{}()<>", c) && c!=EOF;
12 }
13 
14 int
15 idchr(int c)
16 {
17 	/*
18 	 * Formerly:
19 	 * return 'a'<=c && c<='z' || 'A'<=c && c<='Z' || '0'<=c && c<='9'
20 	 *	|| c=='_' || c=='*';
21 	 */
22 	return c>' ' && !strchr("!\"#$%&'()+,-./:;<=>?@[\\]^`{|}~", c);
23 }
24 int future = EOF;
25 int doprompt = 1;
26 int inquote;
27 int incomm;
28 /*
29  * Look ahead in the input stream
30  */
31 
32 int
33 nextc(void)
34 {
35 	if(future==EOF)
36 		future = getnext();
37 	return future;
38 }
39 /*
40  * Consume the lookahead character.
41  */
42 
43 int
44 advance(void)
45 {
46 	int c = nextc();
47 	lastc = future;
48 	future = EOF;
49 	return c;
50 }
51 /*
52  * read a character from the input stream
53  */
54 
55 int
56 getnext(void)
57 {
58 	int c;
59 	static int peekc = EOF;
60 	if(peekc!=EOF){
61 		c = peekc;
62 		peekc = EOF;
63 		return c;
64 	}
65 	if(runq->eof)
66 		return EOF;
67 	if(doprompt)
68 		pprompt();
69 	c = rchr(runq->cmdfd);
70 	if(!inquote && c=='\\'){
71 		c = rchr(runq->cmdfd);
72 		if(c=='\n' && !incomm){		/* don't continue a comment */
73 			doprompt = 1;
74 			c=' ';
75 		}
76 		else{
77 			peekc = c;
78 			c='\\';
79 		}
80 	}
81 	doprompt = doprompt || c=='\n' || c==EOF;
82 	if(c==EOF)
83 		runq->eof++;
84 	else if(flag['V'] || ndot>=2 && flag['v']) pchr(err, c);
85 	return c;
86 }
87 
88 void
89 pprompt(void)
90 {
91 	var *prompt;
92 	if(runq->iflag){
93 		pstr(err, promptstr);
94 		flush(err);
95 		prompt = vlook("prompt");
96 		if(prompt->val && prompt->val->next)
97 			promptstr = prompt->val->next->word;
98 		else
99 			promptstr="\t";
100 	}
101 	runq->lineno++;
102 	doprompt = 0;
103 }
104 
105 int
106 skipwhite(void)
107 {
108 	int c, skipped;
109 	skipped = 0;
110 	for(;;){
111 		c = nextc();
112 		/* Why did this used to be  if(!inquote && c=='#') ?? */
113 		if(c=='#'){
114 			incomm = 1;
115 			skipped = 1;
116 			for(;;){
117 				c = nextc();
118 				if(c=='\n' || c==EOF) {
119 					incomm = 0;
120 					break;
121 				}
122 				advance();
123 			}
124 		}
125 		if(c==' ' || c=='\t') {
126 			skipped = 1;
127 			advance();
128 		}
129 		else
130 			return skipped;
131 	}
132 }
133 
134 void
135 skipnl(void)
136 {
137 	int c;
138 	for(;;){
139 		skipwhite();
140 		c = nextc();
141 		if(c!='\n')
142 			return;
143 		advance();
144 	}
145 }
146 
147 int
148 nextis(int c)
149 {
150 	if(nextc()==c){
151 		advance();
152 		return 1;
153 	}
154 	return 0;
155 }
156 
157 char*
158 addtok(char *p, int val)
159 {
160 	if(p==0)
161 		return 0;
162 	if(p==&tok[NTOK-1]){
163 		*p = 0;
164 		yyerror("token buffer too short");
165 		return 0;
166 	}
167 	*p++=val;
168 	return p;
169 }
170 
171 char*
172 addutf(char *p, int c)
173 {
174 	p = addtok(p, c);
175 	if(twobyte(c))	 /* 2-byte escape */
176 		return addtok(p, advance());
177 	if(threebyte(c)){	/* 3-byte escape */
178 		p = addtok(p, advance());
179 		return addtok(p, advance());
180 	}
181 	if(fourbyte(c)){	/* 4-byte escape */
182 		p = addtok(p, advance());
183 		p = addtok(p, advance());
184 		return addtok(p, advance());
185 	}
186 	return p;
187 }
188 int lastdol;	/* was the last token read '$' or '$#' or '"'? */
189 int lastword;	/* was the last token read a word or compound word terminator? */
190 
191 int
192 yylex(void)
193 {
194 	int c, d = nextc();
195 	char *w = tok;
196 	tree *t;
197 	yylval.tree = 0;
198 	/*
199 	 * Embarassing sneakiness:  if the last token read was a quoted or unquoted
200 	 * WORD then we alter the meaning of what follows.  If the next character
201 	 * is `(', we return SUB (a subscript paren) and consume the `('.  Otherwise,
202 	 * if the next character is the first character of a simple or compound word,
203 	 * we insert a `^' before it.
204 	 */
205 	if(lastword && flag['Y']){
206 		lastword = 0;
207 		if(d=='('){
208 			advance();
209 			strcpy(tok, "(");
210 			return SUB;
211 		}
212 		if(wordchr(d) || d=='\'' || d=='`' || d=='$' || d=='"'){
213 			strcpy(tok, "^");
214 			return '^';
215 		}
216 	}
217 	inquote = 0;
218 	if(skipwhite() && !flag['Y'])
219 		return ' ';
220 	switch(c = advance()){
221 	case EOF:
222 		lastdol = 0;
223 		strcpy(tok, "EOF");
224 		return EOF;
225 	case '$':
226 		lastdol = 1;
227 		if(nextis('#')){
228 			strcpy(tok, "$#");
229 			return COUNT;
230 		}
231 		if(nextis('"')){
232 			strcpy(tok, "$\"");
233 			return '"';
234 		}
235 		strcpy(tok, "$");
236 		return '$';
237 	case '&':
238 		lastdol = 0;
239 		if(nextis('&')){
240 			if(flag['Y'])
241 				skipnl();
242 			strcpy(tok, "&&");
243 			return ANDAND;
244 		}
245 		strcpy(tok, "&");
246 		return '&';
247 	case '|':
248 		lastdol = 0;
249 		if(nextis(c)){
250 			if(flag['Y'])
251 				skipnl();
252 			strcpy(tok, "||");
253 			return OROR;
254 		}
255 	case '<':
256 	case '>':
257 		lastdol = 0;
258 		/*
259 		 * funny redirection tokens:
260 		 *	redir:	arrow | arrow '[' fd ']'
261 		 *	arrow:	'<' | '<<' | '>' | '>>' | '|'
262 		 *	fd:	digit | digit '=' | digit '=' digit
263 		 *	digit:	'0'|'1'|'2'|'3'|'4'|'5'|'6'|'7'|'8'|'9'
264 		 * some possibilities are nonsensical and get a message.
265 		 */
266 		*w++=c;
267 		t = newtree();
268 		switch(c){
269 		case '|':
270 			t->type = PIPE;
271 			t->fd0 = 1;
272 			t->fd1 = 0;
273 			break;
274 		case '>':
275 			t->type = REDIR;
276 			if(nextis(c)){
277 				t->rtype = APPEND;
278 				*w++=c;
279 			}
280 			else t->rtype = WRITE;
281 			t->fd0 = 1;
282 			break;
283 		case '<':
284 			t->type = REDIR;
285 			if(nextis(c)){
286 				t->rtype = HERE;
287 				*w++=c;
288 			} else if (nextis('>')){
289 				t->rtype = RDWR;
290 				*w++=c;
291 			} else t->rtype = READ;
292 			t->fd0 = 0;
293 			break;
294 		}
295 		if(nextis('[')){
296 			*w++='[';
297 			c = advance();
298 			*w++=c;
299 			if(c<'0' || '9'<c){
300 			RedirErr:
301 				*w = 0;
302 				yyerror(t->type==PIPE?"pipe syntax"
303 						:"redirection syntax");
304 				return EOF;
305 			}
306 			t->fd0 = 0;
307 			do{
308 				t->fd0 = t->fd0*10+c-'0';
309 				*w++=c;
310 				c = advance();
311 			}while('0'<=c && c<='9');
312 			if(c=='='){
313 				*w++='=';
314 				if(t->type==REDIR)
315 					t->type = DUP;
316 				c = advance();
317 				if('0'<=c && c<='9'){
318 					t->rtype = DUPFD;
319 					t->fd1 = t->fd0;
320 					t->fd0 = 0;
321 					do{
322 						t->fd0 = t->fd0*10+c-'0';
323 						*w++=c;
324 						c = advance();
325 					}while('0'<=c && c<='9');
326 				}
327 				else{
328 					if(t->type==PIPE)
329 						goto RedirErr;
330 					t->rtype = CLOSE;
331 				}
332 			}
333 			if(c!=']'
334 			|| t->type==DUP && (t->rtype==HERE || t->rtype==APPEND))
335 				goto RedirErr;
336 			*w++=']';
337 		}
338 		*w='\0';
339 		yylval.tree = t;
340 		if(t->type==PIPE && flag['Y'])
341 			skipnl();
342 		if(t->type==REDIR) {
343 			skipwhite();
344 			if(nextc() == '{')
345 				t->type = REDIRW;
346 		}
347 		return t->type;
348 	case '\'':
349 		lastdol = 0;
350 		lastword = 1;
351 		inquote = 1;
352 		for(;;){
353 			c = advance();
354 			if(c==EOF)
355 				break;
356 			if(c=='\''){
357 				if(nextc()!='\'')
358 					break;
359 				advance();
360 			}
361 			w = addutf(w, c);
362 		}
363 		if(w!=0)
364 			*w='\0';
365 		t = token(tok, WORD);
366 		t->quoted = 1;
367 		yylval.tree = t;
368 		return t->type;
369 	}
370 	if(!wordchr(c)){
371 		lastdol = 0;
372 		tok[0] = c;
373 		tok[1]='\0';
374 		return c;
375 	}
376 	for(;;){
377 		/* next line should have (char)c==GLOB, but ken's compiler is broken */
378 		if(c=='*' || c=='[' || c=='?' || c==(unsigned char)GLOB)
379 			w = addtok(w, GLOB);
380 		w = addutf(w, c);
381 		c = nextc();
382 		if(lastdol?!idchr(c):!wordchr(c)) break;
383 		advance();
384 	}
385 
386 	lastword = 1;
387 	lastdol = 0;
388 	if(w!=0)
389 		*w='\0';
390 	t = klook(tok);
391 	if(t->type!=WORD)
392 		lastword = 0;
393 	t->quoted = 0;
394 	yylval.tree = t;
395 	return t->type;
396 }