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 void
106 skipwhite(void)
107 {
108 	int c;
109 	for(;;){
110 		c = nextc();
111 		/* Why did this used to be  if(!inquote && c=='#') ?? */
112 		if(c=='#'){
113 			incomm = 1;
114 			for(;;){
115 				c = nextc();
116 				if(c=='\n' || c==EOF) {
117 					incomm = 0;
118 					break;
119 				}
120 				advance();
121 			}
122 		}
123 		if(c==' ' || c=='\t')
124 			advance();
125 		else return;
126 	}
127 }
128 
129 void
130 skipnl(void)
131 {
132 	int c;
133 	for(;;){
134 		skipwhite();
135 		c = nextc();
136 		if(c!='\n')
137 			return;
138 		advance();
139 	}
140 }
141 
142 int
143 nextis(int c)
144 {
145 	if(nextc()==c){
146 		advance();
147 		return 1;
148 	}
149 	return 0;
150 }
151 
152 char*
153 addtok(char *p, int val)
154 {
155 	if(p==0)
156 		return 0;
157 	if(p==&tok[NTOK-1]){
158 		*p = 0;
159 		yyerror("token buffer too short");
160 		return 0;
161 	}
162 	*p++=val;
163 	return p;
164 }
165 
166 char*
167 addutf(char *p, int c)
168 {
169 	p = addtok(p, c);
170 	if(twobyte(c))	 /* 2-byte escape */
171 		return addtok(p, advance());
172 	if(threebyte(c)){	/* 3-byte escape */
173 		p = addtok(p, advance());
174 		return addtok(p, advance());
175 	}
176 	if(fourbyte(c)){	/* 4-byte escape */
177 		p = addtok(p, advance());
178 		p = addtok(p, advance());
179 		return addtok(p, advance());
180 	}
181 	return p;
182 }
183 int lastdol;	/* was the last token read '$' or '$#' or '"'? */
184 int lastword;	/* was the last token read a word or compound word terminator? */
185 
186 int
187 yylex(void)
188 {
189 	int c, d = nextc();
190 	char *w = tok;
191 	struct tree *t;
192 	yylval.tree = 0;
193 	/*
194 	 * Embarassing sneakiness:  if the last token read was a quoted or unquoted
195 	 * WORD then we alter the meaning of what follows.  If the next character
196 	 * is `(', we return SUB (a subscript paren) and consume the `('.  Otherwise,
197 	 * if the next character is the first character of a simple or compound word,
198 	 * we insert a `^' before it.
199 	 */
200 	if(lastword){
201 		lastword = 0;
202 		if(d=='('){
203 			advance();
204 			strcpy(tok, "( [SUB]");
205 			return SUB;
206 		}
207 		if(wordchr(d) || d=='\'' || d=='`' || d=='$' || d=='"'){
208 			strcpy(tok, "^");
209 			return '^';
210 		}
211 	}
212 	inquote = 0;
213 	skipwhite();
214 	switch(c = advance()){
215 	case EOF:
216 		lastdol = 0;
217 		strcpy(tok, "EOF");
218 		return EOF;
219 	case '$':
220 		lastdol = 1;
221 		if(nextis('#')){
222 			strcpy(tok, "$#");
223 			return COUNT;
224 		}
225 		if(nextis('"')){
226 			strcpy(tok, "$\"");
227 			return '"';
228 		}
229 		strcpy(tok, "$");
230 		return '$';
231 	case '&':
232 		lastdol = 0;
233 		if(nextis('&')){
234 			skipnl();
235 			strcpy(tok, "&&");
236 			return ANDAND;
237 		}
238 		strcpy(tok, "&");
239 		return '&';
240 	case '|':
241 		lastdol = 0;
242 		if(nextis(c)){
243 			skipnl();
244 			strcpy(tok, "||");
245 			return OROR;
246 		}
247 	case '<':
248 	case '>':
249 		lastdol = 0;
250 		/*
251 		 * funny redirection tokens:
252 		 *	redir:	arrow | arrow '[' fd ']'
253 		 *	arrow:	'<' | '<<' | '>' | '>>' | '|'
254 		 *	fd:	digit | digit '=' | digit '=' digit
255 		 *	digit:	'0'|'1'|'2'|'3'|'4'|'5'|'6'|'7'|'8'|'9'
256 		 * some possibilities are nonsensical and get a message.
257 		 */
258 		*w++=c;
259 		t = newtree();
260 		switch(c){
261 		case '|':
262 			t->type = PIPE;
263 			t->fd0 = 1;
264 			t->fd1 = 0;
265 			break;
266 		case '>':
267 			t->type = REDIR;
268 			if(nextis(c)){
269 				t->rtype = APPEND;
270 				*w++=c;
271 			}
272 			else t->rtype = WRITE;
273 			t->fd0 = 1;
274 			break;
275 		case '<':
276 			t->type = REDIR;
277 			if(nextis(c)){
278 				t->rtype = HERE;
279 				*w++=c;
280 			} else if (nextis('>')){
281 				t->rtype = RDWR;
282 				*w++=c;
283 			} else t->rtype = READ;
284 			t->fd0 = 0;
285 			break;
286 		}
287 		if(nextis('[')){
288 			*w++='[';
289 			c = advance();
290 			*w++=c;
291 			if(c<'0' || '9'<c){
292 			RedirErr:
293 				*w = 0;
294 				yyerror(t->type==PIPE?"pipe syntax"
295 						:"redirection syntax");
296 				return EOF;
297 			}
298 			t->fd0 = 0;
299 			do{
300 				t->fd0 = t->fd0*10+c-'0';
301 				*w++=c;
302 				c = advance();
303 			}while('0'<=c && c<='9');
304 			if(c=='='){
305 				*w++='=';
306 				if(t->type==REDIR)
307 					t->type = DUP;
308 				c = advance();
309 				if('0'<=c && c<='9'){
310 					t->rtype = DUPFD;
311 					t->fd1 = t->fd0;
312 					t->fd0 = 0;
313 					do{
314 						t->fd0 = t->fd0*10+c-'0';
315 						*w++=c;
316 						c = advance();
317 					}while('0'<=c && c<='9');
318 				}
319 				else{
320 					if(t->type==PIPE)
321 						goto RedirErr;
322 					t->rtype = CLOSE;
323 				}
324 			}
325 			if(c!=']'
326 			|| t->type==DUP && (t->rtype==HERE || t->rtype==APPEND))
327 				goto RedirErr;
328 			*w++=']';
329 		}
330 		*w='\0';
331 		yylval.tree = t;
332 		if(t->type==PIPE)
333 			skipnl();
334 		return t->type;
335 	case '\'':
336 		lastdol = 0;
337 		lastword = 1;
338 		inquote = 1;
339 		for(;;){
340 			c = advance();
341 			if(c==EOF)
342 				break;
343 			if(c=='\''){
344 				if(nextc()!='\'')
345 					break;
346 				advance();
347 			}
348 			w = addutf(w, c);
349 		}
350 		if(w!=0)
351 			*w='\0';
352 		t = token(tok, WORD);
353 		t->quoted = 1;
354 		yylval.tree = t;
355 		return t->type;
356 	}
357 	if(!wordchr(c)){
358 		lastdol = 0;
359 		tok[0] = c;
360 		tok[1]='\0';
361 		return c;
362 	}
363 	for(;;){
364 		/* next line should have (char)c==GLOB, but ken's compiler is broken */
365 		if(c=='*' || c=='[' || c=='?' || c==(unsigned char)GLOB)
366 			w = addtok(w, GLOB);
367 		w = addutf(w, c);
368 		c = nextc();
369 		if(lastdol?!idchr(c):!wordchr(c)) break;
370 		advance();
371 	}
372 
373 	lastword = 1;
374 	lastdol = 0;
375 	if(w!=0)
376 		*w='\0';
377 	t = klook(tok);
378 	if(t->type!=WORD)
379 		lastword = 0;
380 	t->quoted = 0;
381 	yylval.tree = t;
382 	return t->type;
383 }