[unix-history] / usr / src / usr.bin / diff / diff3 / diff3.c

#ifndef lint
static char sccsid[] = "@(#)diff3.c	4.2 (Berkeley) %G%";
#endif

#include <stdio.h>
#

/* diff3 - 3-way differential file comparison*/

/* diff3 [-e] d13 d23 f1 f2 f3 
 *
 * d13 = diff report on f1 vs f3
 * d23 = diff report on f2 vs f3
 * f1, f2, f3 the 3 files
*/

struct  range {int from,to; };
	/* from is first in range of changed lines
	 * to is last+1
	 * from=to=line after point of insertion
	* for added lines
	*/
struct diff {struct range old, new;};

#define NC 200
/* de is used to gather editing scripts,
 * that are later spewed out in reverse order.
 * its first element must be all zero
 * the "new" component of de contains line positions
 * or byte positions depending on when you look(!?)
*/
struct diff d13[NC];
struct diff d23[NC];
struct diff de[NC];
char line[256];
FILE *fp[3];
int linct[3] = {0,0,0};
/*	the number of the last-read line in each file
 *	is kept in cline[0-2]
*/
int cline[3];
/*	the latest known correspondence between line
 *	numbers of the 3 files is stored in last[1-3]
*/
int last[4];
int eflag;
int debug  = 0;

main(argc,argv)
char **argv;
{
	register i,m,n;
	if(*argv[1]=='-') {
		switch(argv[1][1]) {
		default:
			eflag = 3;
			break;
		case '3':
			eflag = 2;
			break;
		case 'x':
			eflag = 1;
		}
		argv++;
		argc--;
	}
	if(argc<6) {
		fprintf(stderr,"diff3: arg count\n");
		exit(1);
	}
	m = readin(argv[1],d13);
	n = readin(argv[2],d23);
	for(i=0;i<=2;i++)
		if((fp[i] = fopen(argv[i+3],"r")) == NULL) {
			printf("diff3: can't open %s\n",argv[i+3]);
			exit(1);
		}
	merge(m,n);
}

/*pick up the line numbers of allcahnges from
 * one change file
 * (this puts the numbers in a vector, which is not
 * strictly necessary, since the vector is processed
 * in one sequential pass. The vector could be optimized
 * out of existence)
*/

readin(name,dd)
char *name;
struct diff *dd;
{
	register i;
	int a,b,c,d;
	char kind;
	char *p;
	fp[0] = fopen(name,"r");
	for(i=0;getchange(fp[0]);i++) {
		if(i>=NC) {
			fprintf(stderr,"diff3: too many changes\n");
			exit(0);
		}
		p = line;
		a = b = number(&p);
		if(*p==',') {
			p++;
			b = number(&p);
		}
		kind = *p++;
		c = d = number(&p);
		if(*p==',') {
			p++;
			d = number(&p);
		}
		if(kind=='a')
			a++;
		if(kind=='d')
			c++;
		b++;
		d++;
		dd[i].old.from = a;
		dd[i].old.to = b;
		dd[i].new.from = c;
		dd[i].new.to = d;
	}
	dd[i].old.from = dd[i-1].old.to;
	dd[i].new.from = dd[i-1].new.to;
	fclose(fp[0]);
	return(i);
}

number(lc)
char **lc;
{
	register nn;
	nn = 0;
	while(digit(**lc))
		nn = nn*10 + *(*lc)++ - '0';
	return(nn);
}

digit(c)
{
	return(c>='0'&&c<='9');
}

getchange(b)
FILE *b;
{
	while(getline(b))
		if(digit(line[0]))
			return(1);
	return(0);
}

getline(b)
FILE *b;
{
	register i, c;
	for(i=0;i<sizeof(line)-1;i++) {
		c = getc(b);
		if(c==EOF)
			break;
		line[i] = c;
		if(c=='\n') {
			line[++i] = 0;
			return(i);
		}
	}
	return(0);
}

merge(m1,m2)
{
	register struct diff *d1, *d2, *d3;
	int dup;
	int j;
	int t1,t2;
	d1 = d13;
	d2 = d23;
	j = 0;
	for(;(t1 = d1<d13+m1) | (t2 = d2<d23+m2);) {
		if(debug) {
			printf("%d,%d=%d,%d %d,%d=%d,%d\n",
			d1->old.from,d1->old.to,
			d1->new.from,d1->new.to,
			d2->old.from,d2->old.to,
			d2->new.from,d2->new.to);
		}
/*			first file is different from others*/
		if(!t2||t1&&d1->new.to < d2->new.from) {
/*			stuff peculiar to 1st file */
			if(eflag==0) {
				separate("1");
				change(1,&d1->old,0);
				keep(2,&d1->old,&d1->new);
				change(3,&d1->new,0);
			}
			d1++;
			continue;
		}
/*			second file is different from others*/
		if(!t1||t2&&d2->new.to < d1->new.from) {
			if(eflag==0) {
				separate("2");
				keep(1,&d2->old,&d2->new);
				change(2,&d2->old,0);
				change(3,&d2->new,0);
			}
			d2++;
			continue;
		}
/*			merge overlapping changes in first file
 *			this happens after extension see below*/
		if(d1+1<d13+m1 &&
		   d1->new.to>=d1[1].new.from) {
			d1[1].old.from = d1->old.from;
			d1[1].new.from = d1->new.from;
			d1++;
			continue;
		}
/*			merge overlapping changes in second*/
		if(d2+1<d23+m2 &&
		   d2->new.to>=d2[1].new.from) {
			d2[1].old.from = d2->old.from;
			d2[1].new.from = d2->new.from;
			d2++;
			continue;
		}
/*			stuff peculiar to third file or different in all*/
		if(d1->new.from==d2->new.from&&
		   d1->new.to==d2->new.to) {
			dup = duplicate(&d1->old,&d2->old);
/*				dup=0 means all files differ
 *				dup =1 meands files 1&2 identical*/
			if(eflag==0) {
				separate(dup?"3":"");
				change(1,&d1->old,dup);
				change(2,&d2->old,0);
				d3 = d1->old.to>d1->old.from?d1:d2;
				change(3,&d3->new,0);
			} else
				j = edit(d1,dup,j);
			d1++;
			d2++;
			continue;
		}
/*			overlapping changes from file1 & 2
 *			extend changes appropriately to
 *			make them coincide*/
		 if(d1->new.from<d2->new.from) {
			d2->old.from -= d2->new.from-d1->new.from;
			d2->new.from = d1->new.from;
		}
		else if(d2->new.from<d1->new.from) {
			d1->old.from -= d1->new.from-d2->new.from;
			d1->new.from = d2->new.from;
		}
		if(d1->new.to >d2->new.to) {
			d2->old.to += d1->new.to - d2->new.to;
			d2->new.to = d1->new.to;
		}
		else if(d2->new.to >d1->new.to) {
			d1->old.to += d2->new.to - d1->new.to;
			d1->new.to = d2->new.to;
		}
	}
	if(eflag)
		edscript(j);
}

separate(s)
char *s;
{
	printf("====%s\n",s);
}

/*	the range of ines rold.from thru rold.to in file i
 *	is to be changed. it is to be printed only if
 *	it does not duplicate something to be printed later
*/
change(i,rold,dup)
struct range *rold;
{
	printf("%d:",i);
	last[i] = rold->to;
	prange(rold);
	if(dup)
		return;
	if(debug)
		return;
	i--;
	skip(i,rold->from,(char *)0);
	skip(i,rold->to,"  ");
}

/*	print the range of line numbers, rold.from  thru rold.to
 *	as n1,n2 or n1
*/
prange(rold)
struct range *rold;
{
	if(rold->to<=rold->from)
		printf("%da\n",rold->from-1);
	else {
		printf("%d",rold->from);
		if(rold->to > rold->from+1)
			printf(",%d",rold->to-1);
		printf("c\n");
	}
}

/*	no difference was reported by diff between file 1(or 2)
 *	and file 3, and an artificial dummy difference (trange)
 *	must be ginned up to correspond to the change reported
 *	in the other file
*/
keep(i,rold,rnew)
struct range *rold, *rnew;
{
	register delta;
	struct range trange;
	delta = last[3] - last[i];
	trange.from = rnew->from - delta;
	trange.to = rnew->to - delta;
	change(i,&trange,1);
}

/*	skip to just befor line number from in file i
 *	if "pr" is nonzero, print all skipped stuff
 * w	with string pr as a prefix
*/
skip(i,from,pr)
char *pr;
{
	register j,n;
	for(n=0;cline[i]<from-1;n+=j) {
		if((j=getline(fp[i]))==0)
			trouble();
		if(pr)
			printf("%s%s",pr,line);
		cline[i]++;
	}
	return(n);
}

/*	return 1 or 0 according as the old range
 *	(in file 1) contains exactly the same data
 *	as the new range (in file 2)
*/
duplicate(r1,r2)
struct range *r1, *r2;
{
	register c,d;
	register nchar;
	int nline;
	if(r1->to-r1->from != r2->to-r2->from)
		return(0);
	skip(0,r1->from,(char *)0);
	skip(1,r2->from,(char *)0);
	nchar = 0;
	for(nline=0;nline<r1->to-r1->from;nline++) {
		do {
			c = getc(fp[0]);
			d = getc(fp[1]);
			if(c== -1||d== -1)
				trouble();
			nchar++;
			if(c!=d) {
				repos(nchar);
				return;
			}
		} while(c!= '\n');
	}
	repos(nchar);
	return(1);
}

repos(nchar)
{
	register i;
	for(i=0;i<2;i++) 
		fseek(fp[i], (long)-nchar, 1);
}

trouble()
{
	fprintf(stderr,"diff3: logic error\n");
	abort();
}

/*	collect an editing script for later regurgitation
*/
edit(diff,dup,j)
struct diff *diff;
{
	if(((dup+1)&eflag)==0)
		return(j);
	j++;
	de[j].old.from = diff->old.from;
	de[j].old.to = diff->old.to;
	de[j].new.from = de[j-1].new.to
	    +skip(2,diff->new.from,(char *)0);
	de[j].new.to = de[j].new.from
	    +skip(2,diff->new.to,(char *)0);
	return(j);
}

/*		regurgitate */
edscript(n)
{
	register j,k;
	char block[BUFSIZ];
	for(n=n;n>0;n--) {
		prange(&de[n].old);
		fseek(fp[2], (long)de[n].new.from, 0);
		for(k=de[n].new.to-de[n].new.from;k>0;k-= j) {
			j = k>BUFSIZ?BUFSIZ:k;
			if(fread(block,1,j,fp[2])!=j)
				trouble();
			fwrite(block, 1, j, stdout);
		}
		printf(".\n");
	}
}
Commit	Line	Data
f6227721 SL	1	#ifndef lint
	2	static char sccsid[] = "@(#)diff3.c 4.2 (Berkeley) %G%";
	3	#endif
953c3384 BS	4
	5	#include <stdio.h>
	6	#
	7
	8	/* diff3 - 3-way differential file comparison*/
	9
	10	/* diff3 [-e] d13 d23 f1 f2 f3
	11	*
	12	* d13 = diff report on f1 vs f3
	13	* d23 = diff report on f2 vs f3
	14	* f1, f2, f3 the 3 files
	15	*/
	16
	17	struct range {int from,to; };
	18	/* from is first in range of changed lines
	19	* to is last+1
	20	* from=to=line after point of insertion
	21	* for added lines
	22	*/
	23	struct diff {struct range old, new;};
	24
	25	#define NC 200
	26	/* de is used to gather editing scripts,
	27	* that are later spewed out in reverse order.
	28	* its first element must be all zero
	29	* the "new" component of de contains line positions
	30	* or byte positions depending on when you look(!?)
	31	*/
	32	struct diff d13[NC];
	33	struct diff d23[NC];
	34	struct diff de[NC];
	35	char line[256];
	36	FILE *fp[3];
	37	int linct[3] = {0,0,0};
	38	/* the number of the last-read line in each file
	39	* is kept in cline[0-2]
	40	*/
	41	int cline[3];
	42	/* the latest known correspondence between line
	43	* numbers of the 3 files is stored in last[1-3]
	44	*/
	45	int last[4];
	46	int eflag;
	47	int debug = 0;
	48
	49	main(argc,argv)
	50	char **argv;
	51	{
	52	register i,m,n;
	53	if(*argv[1]=='-') {
	54	switch(argv[1][1]) {
	55	default:
	56	eflag = 3;
	57	break;
	58	case '3':
	59	eflag = 2;
	60	break;
	61	case 'x':
	62	eflag = 1;
	63	}
	64	argv++;
	65	argc--;
	66	}
	67	if(argc<6) {
68	fprintf(stderr,"diff3: arg count\n");
69	exit(1);
70	}
71	m = readin(argv[1],d13);
72	n = readin(argv[2],d23);
73	for(i=0;i<=2;i++)
74	if((fp[i] = fopen(argv[i+3],"r")) == NULL) {
75	printf("diff3: can't open %s\n",argv[i+3]);
76	exit(1);
77	}
78	merge(m,n);
79	}
80
81	/*pick up the line numbers of allcahnges from
82	* one change file
83	* (this puts the numbers in a vector, which is not
84	* strictly necessary, since the vector is processed
85	* in one sequential pass. The vector could be optimized
86	* out of existence)
87	*/
88
89	readin(name,dd)
90	char *name;
91	struct diff *dd;
92	{
93	register i;
94	int a,b,c,d;
95	char kind;
96	char *p;
97	fp[0] = fopen(name,"r");
98	for(i=0;getchange(fp[0]);i++) {
99	if(i>=NC) {
100	fprintf(stderr,"diff3: too many changes\n");
101	exit(0);
102	}
103	p = line;
104	a = b = number(&p);
105	if(*p==',') {
106	p++;
107	b = number(&p);
108	}
109	kind = *p++;
110	c = d = number(&p);
111	if(*p==',') {
112	p++;
113	d = number(&p);
114	}
115	if(kind=='a')
116	a++;
117	if(kind=='d')
118	c++;
119	b++;
120	d++;
121	dd[i].old.from = a;
122	dd[i].old.to = b;
123	dd[i].new.from = c;
124	dd[i].new.to = d;
125	}
126	dd[i].old.from = dd[i-1].old.to;
127	dd[i].new.from = dd[i-1].new.to;
128	fclose(fp[0]);
129	return(i);
130	}
131
132	number(lc)
133	char **lc;
134	{
135	register nn;
136	nn = 0;
137	while(digit(**lc))
138	nn = nn10 + (*lc)++ - '0';
139	return(nn);
140	}
141
142	digit(c)
143	{
144	return(c>='0'&&c<='9');
145	}
146
147	getchange(b)
148	FILE *b;
149	{
150	while(getline(b))
151	if(digit(line[0]))
152	return(1);
153	return(0);
154	}
155
156	getline(b)
157	FILE *b;
158	{
159	register i, c;
160	for(i=0;i<sizeof(line)-1;i++) {
161	c = getc(b);
162	if(c==EOF)
163	break;
164	line[i] = c;
165	if(c=='\n') {
166	line[++i] = 0;
167	return(i);
168	}
169	}
170	return(0);
171	}
172
173	merge(m1,m2)
174	{
175	register struct diff d1, d2, *d3;
176	int dup;
177	int j;
178	int t1,t2;
179	d1 = d13;
180	d2 = d23;
181	j = 0;
182	for(;(t1 = d1<d13+m1) \| (t2 = d2<d23+m2);) {
183	if(debug) {
184	printf("%d,%d=%d,%d %d,%d=%d,%d\n",
185	d1->old.from,d1->old.to,
186	d1->new.from,d1->new.to,
187	d2->old.from,d2->old.to,
188	d2->new.from,d2->new.to);
189	}
190	/* first file is different from others*/
191	if(!t2\|\|t1&&d1->new.to < d2->new.from) {
192	/* stuff peculiar to 1st file */
193	if(eflag==0) {
194	separate("1");
195	change(1,&d1->old,0);
196	keep(2,&d1->old,&d1->new);
197	change(3,&d1->new,0);
198	}
199	d1++;
200	continue;
201	}
202	/* second file is different from others*/
203	if(!t1\|\|t2&&d2->new.to < d1->new.from) {
204	if(eflag==0) {
205	separate("2");
206	keep(1,&d2->old,&d2->new);
207	change(2,&d2->old,0);
208	change(3,&d2->new,0);
209	}
210	d2++;
211	continue;
212	}
213	/* merge overlapping changes in first file
214	* this happens after extension see below*/
215	if(d1+1<d13+m1 &&
216	d1->new.to>=d1[1].new.from) {
217	d1[1].old.from = d1->old.from;
218	d1[1].new.from = d1->new.from;
219	d1++;
220	continue;
221	}
222	/* merge overlapping changes in second*/
223	if(d2+1<d23+m2 &&
224	d2->new.to>=d2[1].new.from) {
225	d2[1].old.from = d2->old.from;
226	d2[1].new.from = d2->new.from;
227	d2++;
228	continue;
229	}
230	/* stuff peculiar to third file or different in all*/
231	if(d1->new.from==d2->new.from&&
232	d1->new.to==d2->new.to) {
233	dup = duplicate(&d1->old,&d2->old);
234	/* dup=0 means all files differ
235	* dup =1 meands files 1&2 identical*/
236	if(eflag==0) {
237	separate(dup?"3":"");
238	change(1,&d1->old,dup);
239	change(2,&d2->old,0);
240	d3 = d1->old.to>d1->old.from?d1:d2;
241	change(3,&d3->new,0);
242	} else
243	j = edit(d1,dup,j);
244	d1++;
245	d2++;
246	continue;
247	}
248	/* overlapping changes from file1 & 2
249	* extend changes appropriately to
250	* make them coincide*/
251	if(d1->new.from<d2->new.from) {
252	d2->old.from -= d2->new.from-d1->new.from;
253	d2->new.from = d1->new.from;
254	}
255	else if(d2->new.from<d1->new.from) {
256	d1->old.from -= d1->new.from-d2->new.from;
257	d1->new.from = d2->new.from;
258	}
259	if(d1->new.to >d2->new.to) {
260	d2->old.to += d1->new.to - d2->new.to;
261	d2->new.to = d1->new.to;
262	}
263	else if(d2->new.to >d1->new.to) {
264	d1->old.to += d2->new.to - d1->new.to;
265	d1->new.to = d2->new.to;
266	}
267	}
268	if(eflag)
269	edscript(j);
270	}
271
272	separate(s)
273	char *s;
274	{
275	printf("====%s\n",s);
276	}
277
278	/* the range of ines rold.from thru rold.to in file i
279	* is to be changed. it is to be printed only if
280	* it does not duplicate something to be printed later
281	*/
282	change(i,rold,dup)
283	struct range *rold;
284	{
285	printf("%d:",i);
286	last[i] = rold->to;
287	prange(rold);
288	if(dup)
289	return;
290	if(debug)
291	return;
292	i--;
293	skip(i,rold->from,(char *)0);
294	skip(i,rold->to," ");
295	}
296
297	/* print the range of line numbers, rold.from thru rold.to
298	* as n1,n2 or n1
299	*/
300	prange(rold)
301	struct range *rold;
302	{
303	if(rold->to<=rold->from)
304	printf("%da\n",rold->from-1);
305	else {
306	printf("%d",rold->from);
307	if(rold->to > rold->from+1)
308	printf(",%d",rold->to-1);
309	printf("c\n");
310	}
311	}
312
313	/* no difference was reported by diff between file 1(or 2)
314	* and file 3, and an artificial dummy difference (trange)
315	* must be ginned up to correspond to the change reported
316	* in the other file
317	*/
318	keep(i,rold,rnew)
319	struct range rold, rnew;
320	{
321	register delta;
322	struct range trange;
323	delta = last[3] - last[i];
324	trange.from = rnew->from - delta;
325	trange.to = rnew->to - delta;
326	change(i,&trange,1);
327	}
328
329	/* skip to just befor line number from in file i
330	* if "pr" is nonzero, print all skipped stuff
331	* w with string pr as a prefix
332	*/
333	skip(i,from,pr)
334	char *pr;
335	{
336	register j,n;
337	for(n=0;cline[i]<from-1;n+=j) {
338	if((j=getline(fp[i]))==0)
339	trouble();
340	if(pr)
341	printf("%s%s",pr,line);
342	cline[i]++;
343	}
344	return(n);
345	}
346
347	/* return 1 or 0 according as the old range
348	* (in file 1) contains exactly the same data
349	* as the new range (in file 2)
350	*/
351	duplicate(r1,r2)
352	struct range r1, r2;
353	{
354	register c,d;
355	register nchar;
356	int nline;
357	if(r1->to-r1->from != r2->to-r2->from)
358	return(0);
359	skip(0,r1->from,(char *)0);
360	skip(1,r2->from,(char *)0);
361	nchar = 0;
362	for(nline=0;nline<r1->to-r1->from;nline++) {
363	do {
364	c = getc(fp[0]);
365	d = getc(fp[1]);
366	if(c== -1\|\|d== -1)
367	trouble();
368	nchar++;
369	if(c!=d) {
370	repos(nchar);
371	return;
372	}
373	} while(c!= '\n');
374	}
375	repos(nchar);
376	return(1);
377	}
378
379	repos(nchar)
380	{
381	register i;
382	for(i=0;i<2;i++)
383	fseek(fp[i], (long)-nchar, 1);
384	}
385
386	trouble()
387	{
388	fprintf(stderr,"diff3: logic error\n");
389	abort();
390	}
391
392	/* collect an editing script for later regurgitation
393	*/
394	edit(diff,dup,j)
395	struct diff *diff;
396	{
397	if(((dup+1)&eflag)==0)
398	return(j);
399	j++;
400	de[j].old.from = diff->old.from;
401	de[j].old.to = diff->old.to;
402	de[j].new.from = de[j-1].new.to
403	+skip(2,diff->new.from,(char *)0);
404	de[j].new.to = de[j].new.from
405	+skip(2,diff->new.to,(char *)0);
406	return(j);
407	}
408
409	/* regurgitate */
410	edscript(n)
411	{
412	register j,k;
413	char block[BUFSIZ];
414	for(n=n;n>0;n--) {
415	prange(&de[n].old);
416	fseek(fp[2], (long)de[n].new.from, 0);
417	for(k=de[n].new.to-de[n].new.from;k>0;k-= j) {
418	j = k>BUFSIZ?BUFSIZ:k;
419	if(fread(block,1,j,fp[2])!=j)
420	trouble();
421	fwrite(block, 1, j, stdout);
422	}
423	printf(".\n");
424	}
425	}