xref: /csrg-svn/old/awk/lib.c (revision 10794)
1 /*	lib.c	4.2	83/02/09	*/
2 
3 #include "stdio.h"
4 #include "awk.def"
5 #include "awk.h"
6 #include "ctype.h"
7 
8 FILE	*infile	= NULL;
9 char	*file;
10 #define	RECSIZE	(5 * 512)
11 char	record[RECSIZE];
12 char	fields[RECSIZE];
13 char	EMPTY[] = "";
14 
15 #define	MAXFLD	100
16 int	donefld;	/* 1 = implies rec broken into fields */
17 int	donerec;	/* 1 = record is valid (no flds have changed) */
18 int	mustfld;	/* 1 = NF seen, so always break*/
19 
20 #define	FINIT	{EMPTY, EMPTY, 0.0, FLD|STR}
21 cell fldtab[MAXFLD] = {	/*room for fields */
22 	{ "$record", record, 0.0, STR|FLD},
23 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
24 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
25 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
26 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
27 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
28 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT,
29 	FINIT, FINIT, FINIT, FINIT, FINIT, FINIT, FINIT
30 };
31 int	maxfld	= 0;	/* last used field */
32 
33 
34 getrec()
35 {
36 	register char *rr;
37 	extern int svargc;
38 	extern char **svargv;
39 	register c, sep;
40 
41 	dprintf("**RS=%o, **FS=%o\n", **RS, **FS, NULL);
42 	donefld = 0;
43 	donerec = 1;
44 	record[0] = 0;
45 	while (svargc > 0) {
46 		dprintf("svargc=%d, *svargv=%s\n", svargc, *svargv, NULL);
47 		if (infile == NULL) {	/* have to open a new file */
48 			if (member('=', *svargv)) {	/* it's a var=value argument */
49 				setclvar(*svargv);
50 				svargv++;
51 				svargc--;
52 				continue;
53 			}
54 			*FILENAME = file = *svargv;
55 			dprintf("opening file %s\n", file, NULL, NULL);
56 			if (*file == '-')
57 				infile = stdin;
58 			else if ((infile = fopen(file, "r")) == NULL)
59 				error(FATAL, "can't open %s", file);
60 		}
61 		if ((sep = **RS) == 0)
62 			sep = '\n';
63 		for (rr = record; ; ) {
64 			for (; (c=getc(infile)) != sep && c != EOF; *rr++ = c)
65 				;
66 			if (**RS == sep || c == EOF)
67 				break;
68 			if ((c = getc(infile)) == '\n' || c == EOF)	/* 2 in a row */
69 				break;
70 			*rr++ = '\n';
71 			*rr++ = c;
72 		}
73 		if (rr > record+RECSIZE)
74 			error(FATAL, "record `%.20s...' too long", record);
75 		*rr = 0;
76 		if (mustfld)
77 			fldbld();
78 		if (c != EOF || rr > record) {	/* normal record */
79 			recloc->tval &= ~NUM;
80 			recloc->tval |= STR;
81 			++nrloc->fval;
82 			nrloc->tval &= ~STR;
83 			nrloc->tval |= NUM;
84 			return(1);
85 		}
86 		/* EOF arrived on this file; set up next */
87 		if (infile != stdin)
88 			fclose(infile);
89 		infile = NULL;
90 		svargc--;
91 		svargv++;
92 	}
93 	return(0);	/* true end of file */
94 }
95 
96 setclvar(s)	/* set var=value from s */
97 char *s;
98 {
99 	char *p;
100 	cell *q;
101 
102 	for (p=s; *p != '='; p++)
103 		;
104 	*p++ = 0;
105 	q = setsymtab(s, tostring(p), 0.0, STR, symtab);
106 	setsval(q, p);
107 	dprintf("command line set %s to |%s|\n", s, p, NULL);
108 }
109 
110 fldbld()
111 {
112 	register char *r, *fr, sep;
113 	int i, j;
114 
115 	r = record;
116 	fr = fields;
117 	i = 0;	/* number of fields accumulated here */
118 	if ((sep = **FS) == ' ')
119 		for (i = 0; ; ) {
120 			while (*r == ' ' || *r == '\t' || *r == '\n')
121 				r++;
122 			if (*r == 0)
123 				break;
124 			i++;
125 			if (i >= MAXFLD)
126 				error(FATAL, "record `%.20s...' has too many fields", record);
127 			if (!(fldtab[i].tval&FLD))
128 				strfree(fldtab[i].sval);
129 			fldtab[i].sval = fr;
130 			fldtab[i].tval = FLD | STR;
131 			do
132 				*fr++ = *r++;
133 			while (*r != ' ' && *r != '\t' && *r != '\n' && *r != '\0');
134 			*fr++ = 0;
135 		}
136 	else if (*r != 0)	/* if 0, it's a null field */
137 		for (;;) {
138 			i++;
139 			if (i >= MAXFLD)
140 				error(FATAL, "record `%.20s...' has too many fields", record);
141 			if (!(fldtab[i].tval&FLD))
142 				strfree(fldtab[i].sval);
143 			fldtab[i].sval = fr;
144 			fldtab[i].tval = FLD | STR;
145 			while (*r != sep && *r != '\n' && *r != '\0')	/* \n always a separator */
146 				*fr++ = *r++;
147 			*fr++ = 0;
148 			if (*r++ == 0)
149 				break;
150 		}
151 	*fr = 0;
152 	for (j=MAXFLD-1; j>i; j--) {	/* clean out junk from previous record */
153 		if (!(fldtab[j].tval&FLD))
154 			strfree(fldtab[j].sval);
155 		fldtab[j].tval = STR | FLD;
156 		fldtab[j].sval = EMPTY;
157 	}
158 	maxfld = i;
159 	donefld = 1;
160 	for(i=1; i<=maxfld; i++)
161 		if(isnumber(fldtab[i].sval)) {
162 			fldtab[i].fval = atof(fldtab[i].sval);
163 			fldtab[i].tval |= NUM;
164 		}
165 	setfval(lookup("NF", symtab, 0), (awkfloat) maxfld);
166 	if (dbg)
167 		for (i = 0; i <= maxfld; i++)
168 			printf("field %d: |%s|\n", i, fldtab[i].sval);
169 }
170 
171 recbld()
172 {
173 	int i;
174 	register char *r, *p;
175 
176 	if (donefld == 0 || donerec == 1)
177 		return;
178 	r = record;
179 	for (i = 1; i <= *NF; i++) {
180 		p = getsval(&fldtab[i]);
181 		while (*r++ = *p++)
182 			;
183 		*(r-1) = **OFS;
184 	}
185 	*(r-1) = '\0';
186 	dprintf("in recbld FS=%o, recloc=%o\n", **FS, recloc, NULL);
187 	recloc->tval = STR | FLD;
188 	dprintf("in recbld FS=%o, recloc=%o\n", **FS, recloc, NULL);
189 	if (r > record+RECSIZE)
190 		error(FATAL, "built giant record `%.20s...'", record);
191 	dprintf("recbld = |%s|\n", record, NULL, NULL);
192 }
193 
194 cell *fieldadr(n)
195 {
196 	if (n >= MAXFLD)
197 		error(FATAL, "trying to access field %d", n);
198 	return(&fldtab[n]);
199 }
200 
201 int	errorflag	= 0;
202 
203 yyerror(s) char *s; {
204 	fprintf(stderr, "awk: %s near line %d\n", s, lineno);
205 	errorflag = 2;
206 }
207 
208 error(f, s, a1, a2, a3, a4, a5, a6, a7) {
209 	fprintf(stderr, "awk: ");
210 	fprintf(stderr, s, a1, a2, a3, a4, a5, a6, a7);
211 	fprintf(stderr, "\n");
212 	if (*NR > 0)
213 		fprintf(stderr, " record number %g\n", *NR);
214 	if (f)
215 		exit(2);
216 }
217 
218 PUTS(s) char *s; {
219 	dprintf("%s\n", s, NULL, NULL);
220 }
221 
222 #define	MAXEXPON	38	/* maximum exponenet for fp number */
223 
224 isnumber(s)
225 register char *s;
226 {
227 	register d1, d2;
228 	int point;
229 	char *es;
230 
231 	if (s == NULL)
232 		return (0);
233 	d1 = d2 = point = 0;
234 	while (*s == ' ' || *s == '\t' || *s == '\n')
235 		s++;
236 	if (*s == '\0')
237 		return(0);	/* empty stuff isn't number */
238 	if (*s == '+' || *s == '-')
239 		s++;
240 	if (!isdigit(*s) && *s != '.')
241 		return(0);
242 	if (isdigit(*s)) {
243 		do {
244 			d1++;
245 			s++;
246 		} while (isdigit(*s));
247 	}
248 	if(d1 >= MAXEXPON)
249 		return(0);	/* too many digits to convert */
250 	if (*s == '.') {
251 		point++;
252 		s++;
253 	}
254 	if (isdigit(*s)) {
255 		d2++;
256 		do {
257 			s++;
258 		} while (isdigit(*s));
259 	}
260 	if (!(d1 || point && d2))
261 		return(0);
262 	if (*s == 'e' || *s == 'E') {
263 		s++;
264 		if (*s == '+' || *s == '-')
265 			s++;
266 		if (!isdigit(*s))
267 			return(0);
268 		es = s;
269 		do {
270 			s++;
271 		} while (isdigit(*s));
272 		if (s - es > 2)
273 			return(0);
274 		else if (s - es == 2 && 10 * (*es-'0') + *(es+1)-'0' >= MAXEXPON)
275 			return(0);
276 	}
277 	while (*s == ' ' || *s == '\t' || *s == '\n')
278 		s++;
279 	if (*s == '\0')
280 		return(1);
281 	else
282 		return(0);
283 }
284 /*
285 isnumber(s) char *s; {return(0);}
286 */
287