xref: /csrg-svn/old/awk/lib.c (revision 6671)
1*6671Smckusick /*	lib.c	4.1	82/05/07	*/
2*6671Smckusick 
3*6671Smckusick #include "stdio.h"
4*6671Smckusick #include "awk.def"
5*6671Smckusick #include "awk.h"
6*6671Smckusick #include "ctype.h"
7*6671Smckusick 
8*6671Smckusick FILE	*infile	= NULL;
9*6671Smckusick char	*file;
10*6671Smckusick #define	RECSIZE	(5 * 512)
11*6671Smckusick char	record[RECSIZE];
12*6671Smckusick char	fields[RECSIZE];
13*6671Smckusick 
14*6671Smckusick #define	MAXFLD	100
15*6671Smckusick int	donefld;	/* 1 = implies rec broken into fields */
16*6671Smckusick int	donerec;	/* 1 = record is valid (no flds have changed) */
17*6671Smckusick int	mustfld;	/* 1 = NF seen, so always break*/
18*6671Smckusick 
19*6671Smckusick #define	FINIT	{0, NULL, 0.0, FLD|STR}
20*6671Smckusick cell fldtab[MAXFLD] = {	/*room for fields */
21*6671Smckusick 	{ "$record", record, 0.0, STR|FLD},
22*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
23*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
24*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
25*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
26*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
27*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
28*6671Smckusick 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT
29*6671Smckusick };
30*6671Smckusick int	maxfld	= 0;	/* last used field */
31*6671Smckusick 
32*6671Smckusick 
33*6671Smckusick getrec()
34*6671Smckusick {
35*6671Smckusick 	register char *rr;
36*6671Smckusick 	extern int svargc;
37*6671Smckusick 	extern char **svargv;
38*6671Smckusick 	register c, sep;
39*6671Smckusick 
40*6671Smckusick 	dprintf("**RS=%o, **FS=%o\n", **RS, **FS, NULL);
41*6671Smckusick 	donefld = 0;
42*6671Smckusick 	donerec = 1;
43*6671Smckusick 	record[0] = 0;
44*6671Smckusick 	while (svargc > 0) {
45*6671Smckusick 		dprintf("svargc=%d, *svargv=%s\n", svargc, *svargv, NULL);
46*6671Smckusick 		if (infile == NULL) {	/* have to open a new file */
47*6671Smckusick 			if (member('=', *svargv)) {	/* it's a var=value argument */
48*6671Smckusick 				setclvar(*svargv);
49*6671Smckusick 				svargv++;
50*6671Smckusick 				svargc--;
51*6671Smckusick 				continue;
52*6671Smckusick 			}
53*6671Smckusick 			*FILENAME = file = *svargv;
54*6671Smckusick 			dprintf("opening file %s\n", file, NULL, NULL);
55*6671Smckusick 			if (*file == '-')
56*6671Smckusick 				infile = stdin;
57*6671Smckusick 			else if ((infile = fopen(file, "r")) == NULL)
58*6671Smckusick 				error(FATAL, "can't open %s", file);
59*6671Smckusick 		}
60*6671Smckusick 		if ((sep = **RS) == 0)
61*6671Smckusick 			sep = '\n';
62*6671Smckusick 		for (rr = record; ; ) {
63*6671Smckusick 			for (; (c=getc(infile)) != sep && c != EOF; *rr++ = c)
64*6671Smckusick 				;
65*6671Smckusick 			if (**RS == sep || c == EOF)
66*6671Smckusick 				break;
67*6671Smckusick 			if ((c = getc(infile)) == '\n' || c == EOF)	/* 2 in a row */
68*6671Smckusick 				break;
69*6671Smckusick 			*rr++ = '\n';
70*6671Smckusick 			*rr++ = c;
71*6671Smckusick 		}
72*6671Smckusick 		if (rr > record+RECSIZE)
73*6671Smckusick 			error(FATAL, "record `%.20s...' too long", record);
74*6671Smckusick 		*rr = 0;
75*6671Smckusick 		if (mustfld)
76*6671Smckusick 			fldbld();
77*6671Smckusick 		if (c != EOF || rr > record) {	/* normal record */
78*6671Smckusick 			recloc->tval &= ~NUM;
79*6671Smckusick 			recloc->tval |= STR;
80*6671Smckusick 			++nrloc->fval;
81*6671Smckusick 			nrloc->tval &= ~STR;
82*6671Smckusick 			nrloc->tval |= NUM;
83*6671Smckusick 			return(1);
84*6671Smckusick 		}
85*6671Smckusick 		/* EOF arrived on this file; set up next */
86*6671Smckusick 		if (infile != stdin)
87*6671Smckusick 			fclose(infile);
88*6671Smckusick 		infile = NULL;
89*6671Smckusick 		svargc--;
90*6671Smckusick 		svargv++;
91*6671Smckusick 	}
92*6671Smckusick 	return(0);	/* true end of file */
93*6671Smckusick }
94*6671Smckusick 
95*6671Smckusick setclvar(s)	/* set var=value from s */
96*6671Smckusick char *s;
97*6671Smckusick {
98*6671Smckusick 	char *p;
99*6671Smckusick 	cell *q;
100*6671Smckusick 
101*6671Smckusick 	for (p=s; *p != '='; p++)
102*6671Smckusick 		;
103*6671Smckusick 	*p++ = 0;
104*6671Smckusick 	q = setsymtab(s, tostring(p), 0.0, STR, symtab);
105*6671Smckusick 	setsval(q, p);
106*6671Smckusick 	dprintf("command line set %s to |%s|\n", s, p, NULL);
107*6671Smckusick }
108*6671Smckusick 
109*6671Smckusick fldbld()
110*6671Smckusick {
111*6671Smckusick 	register char *r, *fr, sep;
112*6671Smckusick 	int i, j;
113*6671Smckusick 
114*6671Smckusick 	r = record;
115*6671Smckusick 	fr = fields;
116*6671Smckusick 	i = 0;	/* number of fields accumulated here */
117*6671Smckusick 	if ((sep = **FS) == ' ')
118*6671Smckusick 		for (i = 0; ; ) {
119*6671Smckusick 			while (*r == ' ' || *r == '\t' || *r == '\n')
120*6671Smckusick 				r++;
121*6671Smckusick 			if (*r == 0)
122*6671Smckusick 				break;
123*6671Smckusick 			i++;
124*6671Smckusick 			if (i >= MAXFLD)
125*6671Smckusick 				error(FATAL, "record `%.20s...' has too many fields", record);
126*6671Smckusick 			if (!(fldtab[i].tval&FLD))
127*6671Smckusick 				xfree(fldtab[i].sval);
128*6671Smckusick 			fldtab[i].sval = fr;
129*6671Smckusick 			fldtab[i].tval = FLD | STR;
130*6671Smckusick 			do
131*6671Smckusick 				*fr++ = *r++;
132*6671Smckusick 			while (*r != ' ' && *r != '\t' && *r != '\n' && *r != '\0');
133*6671Smckusick 			*fr++ = 0;
134*6671Smckusick 		}
135*6671Smckusick 	else if (*r != 0)	/* if 0, it's a null field */
136*6671Smckusick 		for (;;) {
137*6671Smckusick 			i++;
138*6671Smckusick 			if (i >= MAXFLD)
139*6671Smckusick 				error(FATAL, "record `%.20s...' has too many fields", record);
140*6671Smckusick 			if (!(fldtab[i].tval&FLD))
141*6671Smckusick 				xfree(fldtab[i].sval);
142*6671Smckusick 			fldtab[i].sval = fr;
143*6671Smckusick 			fldtab[i].tval = FLD | STR;
144*6671Smckusick 			while (*r != sep && *r != '\n' && *r != '\0')	/* \n always a separator */
145*6671Smckusick 				*fr++ = *r++;
146*6671Smckusick 			*fr++ = 0;
147*6671Smckusick 			if (*r++ == 0)
148*6671Smckusick 				break;
149*6671Smckusick 		}
150*6671Smckusick 	*fr = 0;
151*6671Smckusick 	for (j=MAXFLD-1; j>i; j--) {	/* clean out junk from previous record */
152*6671Smckusick 		if (!(fldtab[j].tval&FLD))
153*6671Smckusick 			xfree(fldtab[j].sval);
154*6671Smckusick 		fldtab[j].tval = STR | FLD;
155*6671Smckusick 		fldtab[j].sval = NULL;
156*6671Smckusick 	}
157*6671Smckusick 	maxfld = i;
158*6671Smckusick 	donefld = 1;
159*6671Smckusick 	for(i=1; i<=maxfld; i++)
160*6671Smckusick 		if(isnumber(fldtab[i].sval)) {
161*6671Smckusick 			fldtab[i].fval = atof(fldtab[i].sval);
162*6671Smckusick 			fldtab[i].tval |= NUM;
163*6671Smckusick 		}
164*6671Smckusick 	setfval(lookup("NF", symtab, 0), (awkfloat) maxfld);
165*6671Smckusick 	if (dbg)
166*6671Smckusick 		for (i = 0; i <= maxfld; i++)
167*6671Smckusick 			printf("field %d: |%s|\n", i, fldtab[i].sval);
168*6671Smckusick }
169*6671Smckusick 
170*6671Smckusick recbld()
171*6671Smckusick {
172*6671Smckusick 	int i;
173*6671Smckusick 	register char *r, *p;
174*6671Smckusick 
175*6671Smckusick 	if (donefld == 0 || donerec == 1)
176*6671Smckusick 		return;
177*6671Smckusick 	r = record;
178*6671Smckusick 	for (i = 1; i <= *NF; i++) {
179*6671Smckusick 		p = getsval(&fldtab[i]);
180*6671Smckusick 		while (*r++ = *p++)
181*6671Smckusick 			;
182*6671Smckusick 		*(r-1) = **OFS;
183*6671Smckusick 	}
184*6671Smckusick 	*(r-1) = '\0';
185*6671Smckusick 	dprintf("in recbld FS=%o, recloc=%o\n", **FS, recloc, NULL);
186*6671Smckusick 	recloc->tval = STR | FLD;
187*6671Smckusick 	dprintf("in recbld FS=%o, recloc=%o\n", **FS, recloc, NULL);
188*6671Smckusick 	if (r > record+RECSIZE)
189*6671Smckusick 		error(FATAL, "built giant record `%.20s...'", record);
190*6671Smckusick 	dprintf("recbld = |%s|\n", record, NULL, NULL);
191*6671Smckusick }
192*6671Smckusick 
193*6671Smckusick cell *fieldadr(n)
194*6671Smckusick {
195*6671Smckusick 	if (n >= MAXFLD)
196*6671Smckusick 		error(FATAL, "trying to access field %d", n);
197*6671Smckusick 	return(&fldtab[n]);
198*6671Smckusick }
199*6671Smckusick 
200*6671Smckusick int	errorflag	= 0;
201*6671Smckusick 
202*6671Smckusick yyerror(s) char *s; {
203*6671Smckusick 	fprintf(stderr, "awk: %s near line %d\n", s, lineno);
204*6671Smckusick 	errorflag = 2;
205*6671Smckusick }
206*6671Smckusick 
207*6671Smckusick error(f, s, a1, a2, a3, a4, a5, a6, a7) {
208*6671Smckusick 	fprintf(stderr, "awk: ");
209*6671Smckusick 	fprintf(stderr, s, a1, a2, a3, a4, a5, a6, a7);
210*6671Smckusick 	fprintf(stderr, "\n");
211*6671Smckusick 	if (*NR > 0)
212*6671Smckusick 		fprintf(stderr, " record number %g\n", *NR);
213*6671Smckusick 	if (f)
214*6671Smckusick 		exit(2);
215*6671Smckusick }
216*6671Smckusick 
217*6671Smckusick PUTS(s) char *s; {
218*6671Smckusick 	dprintf("%s\n", s, NULL, NULL);
219*6671Smckusick }
220*6671Smckusick 
221*6671Smckusick #define	MAXEXPON	38	/* maximum exponenet for fp number */
222*6671Smckusick 
223*6671Smckusick isnumber(s)
224*6671Smckusick register char *s;
225*6671Smckusick {
226*6671Smckusick 	register d1, d2;
227*6671Smckusick 	int point;
228*6671Smckusick 	char *es;
229*6671Smckusick 
230*6671Smckusick 	d1 = d2 = point = 0;
231*6671Smckusick 	while (*s == ' ' || *s == '\t' || *s == '\n')
232*6671Smckusick 		s++;
233*6671Smckusick 	if (*s == '\0')
234*6671Smckusick 		return(0);	/* empty stuff isn't number */
235*6671Smckusick 	if (*s == '+' || *s == '-')
236*6671Smckusick 		s++;
237*6671Smckusick 	if (!isdigit(*s) && *s != '.')
238*6671Smckusick 		return(0);
239*6671Smckusick 	if (isdigit(*s)) {
240*6671Smckusick 		do {
241*6671Smckusick 			d1++;
242*6671Smckusick 			s++;
243*6671Smckusick 		} while (isdigit(*s));
244*6671Smckusick 	}
245*6671Smckusick 	if(d1 >= MAXEXPON)
246*6671Smckusick 		return(0);	/* too many digits to convert */
247*6671Smckusick 	if (*s == '.') {
248*6671Smckusick 		point++;
249*6671Smckusick 		s++;
250*6671Smckusick 	}
251*6671Smckusick 	if (isdigit(*s)) {
252*6671Smckusick 		d2++;
253*6671Smckusick 		do {
254*6671Smckusick 			s++;
255*6671Smckusick 		} while (isdigit(*s));
256*6671Smckusick 	}
257*6671Smckusick 	if (!(d1 || point && d2))
258*6671Smckusick 		return(0);
259*6671Smckusick 	if (*s == 'e' || *s == 'E') {
260*6671Smckusick 		s++;
261*6671Smckusick 		if (*s == '+' || *s == '-')
262*6671Smckusick 			s++;
263*6671Smckusick 		if (!isdigit(*s))
264*6671Smckusick 			return(0);
265*6671Smckusick 		es = s;
266*6671Smckusick 		do {
267*6671Smckusick 			s++;
268*6671Smckusick 		} while (isdigit(*s));
269*6671Smckusick 		if (s - es > 2)
270*6671Smckusick 			return(0);
271*6671Smckusick 		else if (s - es == 2 && 10 * (*es-'0') + *(es+1)-'0' >= MAXEXPON)
272*6671Smckusick 			return(0);
273*6671Smckusick 	}
274*6671Smckusick 	while (*s == ' ' || *s == '\t' || *s == '\n')
275*6671Smckusick 		s++;
276*6671Smckusick 	if (*s == '\0')
277*6671Smckusick 		return(1);
278*6671Smckusick 	else
279*6671Smckusick 		return(0);
280*6671Smckusick }
281*6671Smckusick /*
282*6671Smckusick isnumber(s) char *s; {return(0);}
283*6671Smckusick */
284