* Re: [9front] merge3: a first draft
2023-02-12 23:17 [9front] merge3: a first draft ori
@ 2023-02-25 16:56 ` ori
2023-03-04 20:06 ` ori
0 siblings, 1 reply; 3+ messages in thread
From: ori @ 2023-02-25 16:56 UTC (permalink / raw)
To: 9front
Quoth ori@eigenstate.org:
> I've been annoyed that we have to use ape/diff3 to merge
> files for a while, so I finally sat down this weekend and
> hacked 3-way merge capabilities into diff(1).
>
> This is a rough draft, but here's the current approach.
>
> 3 way merging operates as follows:
>
Now bugfixed, and with tests.
diff e6c6217b35c319127f0200fdb28ec86e1b774a4f uncommitted
--- /dev/null
+++ b/sys/src/cmd/diff/diff.c
@@ -1,0 +1,84 @@
+#include <u.h>
+#include <libc.h>
+#include <bio.h>
+#include "diff.h"
+
+void
+done(int status)
+{
+ switch(status)
+ {
+ case 0:
+ exits("");
+ case 1:
+ exits("some");
+ default:
+ exits("error");
+ }
+}
+
+void
+usage(void)
+{
+ fprint(2, "usage: %s [-abcefmnrw] file1 ... file2\n", argv0);
+ exits("usage");
+}
+
+void
+main(int argc, char *argv[])
+{
+ int i;
+ Dir *fsb, *tsb;
+
+ Binit(&stdout, 1, OWRITE);
+ ARGBEGIN{
+ case 'e':
+ case 'f':
+ case 'n':
+ case 'c':
+ case 'a':
+ case 'u':
+ mode = ARGC();
+ break;
+ case 'w':
+ bflag = 2;
+ break;
+
+ case 'b':
+ bflag = 1;
+ break;
+
+ case 'r':
+ rflag = 1;
+ break;
+
+ case 'm':
+ mflag = 1;
+ break;
+
+ case 'h':
+ default:
+ usage();
+ }ARGEND;
+ if (argc < 2)
+ usage();
+ if ((tsb = dirstat(argv[argc-1])) == nil)
+ sysfatal("can't stat %s", argv[argc-1]);
+ if (argc > 2) {
+ if (!DIRECTORY(tsb))
+ sysfatal("not directory: %s", argv[argc-1]);
+ mflag = 1;
+ } else {
+ if ((fsb = dirstat(argv[0])) == nil)
+ sysfatal("can't stat %s", argv[0]);
+ if (DIRECTORY(fsb) && DIRECTORY(tsb))
+ mflag = 1;
+ free(fsb);
+ }
+ free(tsb);
+ for (i = 0; i < argc-1; i++)
+ diff(argv[i], argv[argc-1], 0);
+
+ done(anychange);
+ /*NOTREACHED*/
+}
--- a/sys/src/cmd/diff/diff.h
+++ b/sys/src/cmd/diff/diff.h
@@ -1,13 +1,53 @@
-typedef struct Line Line;
+typedef struct Line Line;
+typedef struct Cand Cand;
+typedef struct Diff Diff;
+typedef struct Change Change;
struct Line {
int serial;
int value;
};
-extern Line *file[2];
-extern int len[2];
-extern long *ixold, *ixnew;
-extern int *J;
+
+struct Cand {
+ int x;
+ int y;
+ int pred;
+};
+
+struct Change
+{
+ int a;
+ int b;
+ int c;
+ int d;
+};
+
+struct Diff {
+ Cand cand;
+ Line *file[2], line;
+ int len[2];
+ int binary;
+ Line *sfile[2]; /*shortened by pruning common prefix and suffix*/
+ int slen[2];
+ int pref, suff; /*length of prefix and suffix*/
+ int *class; /*will be overlaid on file[0]*/
+ int *member; /*will be overlaid on file[1]*/
+ int *klist; /*will be overlaid on file[0] after class*/
+ Cand *clist; /* merely a free storage pot for candidates */
+ int clen;
+ int *J; /*will be overlaid on class*/
+ long *ixold; /*will be overlaid on klist*/
+ long *ixnew; /*will be overlaid on file[1]*/
+ char *file1;
+ char *file2;
+ Biobuf *input[2];
+ Biobuf *b0;
+ Biobuf *b1;
+ int firstchange;
+ Change *changes;
+ int nchanges;
+};
+
extern char mode;
extern char bflag;
extern char rflag;
@@ -14,19 +54,24 @@
extern char mflag;
extern int anychange;
extern Biobuf stdout;
-extern int binary;
#define MAXPATHLEN 1024
+#define DIRECTORY(s) ((s)->qid.type&QTDIR)
+#define REGULAR_FILE(s) ((s)->type == 'M' && !DIRECTORY(s))
+
int mkpathname(char *, char *, char *);
+char *mktmpfile(int, Dir **);
+char *statfile(char *, Dir **);
void *emalloc(unsigned);
void *erealloc(void *, unsigned);
void diff(char *, char *, int);
+void diffreg(char*, char*, char*, char*);
void diffdir(char *, char *, int);
-void diffreg(char *, char *, char *, char *);
-Biobuf *prepare(int, char *, char *);
-void panic(int, char *, ...);
-void check(Biobuf *, Biobuf *);
-void change(int, int, int, int);
-void flushchanges(void);
-
+void calcdiff(Diff *, char *, char *, char *, char *);
+Biobuf *prepare(Diff*, int, char *, char *);
+void check(Diff *, Biobuf *, Biobuf *);
+void change(Diff *, int, int, int, int);
+void freediff(Diff *);
+void flushchanges(Diff *);
+void fetch(Diff *d, long *f, int a, int b, Biobuf *bp, char *s);
--- a/sys/src/cmd/diff/diffdir.c
+++ b/sys/src/cmd/diff/diffdir.c
@@ -111,3 +111,45 @@
free(dirf);
free(dirt);
}
+
+void
+diff(char *f, char *t, int level)
+{
+ char *fp, *tp, *p, fb[MAXPATHLEN+1], tb[MAXPATHLEN+1];
+ Dir *fsb, *tsb;
+
+ fsb = nil;
+ tsb = nil;
+ if ((fp = statfile(f, &fsb)) == 0)
+ goto Return;
+ if ((tp = statfile(t, &tsb)) == 0)
+ goto Return;
+ if (DIRECTORY(fsb) && DIRECTORY(tsb)) {
+ if (rflag || level == 0)
+ diffdir(fp, tp, level);
+ else
+ Bprint(&stdout, "Common subdirectories: %s and %s\n", fp, tp);
+ }
+ else if (REGULAR_FILE(fsb) && REGULAR_FILE(tsb))
+ diffreg(fp, f, tp, t);
+ else {
+ if (REGULAR_FILE(fsb)) {
+ if ((p = utfrrune(f, '/')) == 0)
+ p = f;
+ else
+ p++;
+ if (mkpathname(tb, tp, p) == 0)
+ diffreg(fp, f, tb, t);
+ } else {
+ if ((p = utfrrune(t, '/')) == 0)
+ p = t;
+ else
+ p++;
+ if (mkpathname(fb, fp, p) == 0)
+ diffreg(fb, f, tp, t);
+ }
+ }
+Return:
+ free(fsb);
+ free(tsb);
+}
--- a/sys/src/cmd/diff/diffio.c
+++ b/sys/src/cmd/diff/diffio.c
@@ -4,10 +4,6 @@
#include <ctype.h>
#include "diff.h"
-static Biobuf *input[2];
-static char *file1, *file2;
-static int firstchange;
-
#define MAXLINELEN 4096
#define MIN(x, y) ((x) < (y) ? (x): (y))
@@ -104,7 +100,7 @@
}
Biobuf *
-prepare(int i, char *arg, char *orig)
+prepare(Diff *d, int i, char *arg, char *orig)
{
Line *p;
int j, h;
@@ -115,10 +111,10 @@
bp = Bopen(arg, OREAD);
if (!bp) {
- panic(mflag ? 0: 2, "cannot open %s: %r\n", arg);
+ sysfatal("cannot open %s: %r", arg);
return 0;
}
- if (binary)
+ if (d->binary)
return bp;
nbytes = Bread(bp, buf, MIN(1024, MAXLINELEN));
if (nbytes > 0) {
@@ -130,7 +126,7 @@
*/
cp += chartorune(&r, cp);
if (r == 0 || (r > 0x7f && r <= 0xa0)) {
- binary++;
+ d->binary++;
return bp;
}
}
@@ -139,14 +135,14 @@
p = emalloc(3*sizeof(Line));
for (j = 0; h = readhash(bp, buf); p[j].value = h)
p = erealloc(p, (++j+3)*sizeof(Line));
- len[i] = j;
- file[i] = p;
- input[i] = bp;
+ d->len[i] = j;
+ d->file[i] = p;
+ d->input[i] = bp;
if (i == 0) {
- file1 = orig;
- firstchange = 0;
+ d->file1 = orig;
+ d->firstchange = 0;
} else
- file2 = orig;
+ d->file2 = orig;
return bp;
}
@@ -175,31 +171,32 @@
* need to fix up for unexpected EOF's
*/
void
-check(Biobuf *bf, Biobuf *bt)
+check(Diff *d, Biobuf *bf, Biobuf *bt)
{
int f, t, flen, tlen;
char fbuf[MAXLINELEN], tbuf[MAXLINELEN];
- ixold[0] = ixnew[0] = 0;
- for (f = t = 1; f < len[0]; f++) {
+ d->ixold[0] = 0;
+ d->ixnew[0] = 0;
+ for (f = t = 1; f < d->len[0]; f++) {
flen = readline(bf, fbuf);
- ixold[f] = ixold[f-1] + flen + 1; /* ftell(bf) */
- if (J[f] == 0)
+ d->ixold[f] = d->ixold[f-1] + flen + 1; /* ftell(bf) */
+ if (d->J[f] == 0)
continue;
do {
tlen = readline(bt, tbuf);
- ixnew[t] = ixnew[t-1] + tlen + 1; /* ftell(bt) */
- } while (t++ < J[f]);
+ d->ixnew[t] = d->ixnew[t-1] + tlen + 1; /* ftell(bt) */
+ } while (t++ < d->J[f]);
if (bflag) {
flen = squishspace(fbuf);
tlen = squishspace(tbuf);
}
if (flen != tlen || strcmp(fbuf, tbuf))
- J[f] = 0;
+ d->J[f] = 0;
}
- while (t < len[1]) {
+ while (t < d->len[1]) {
tlen = readline(bt, tbuf);
- ixnew[t] = ixnew[t-1] + tlen + 1; /* fseek(bt) */
+ d->ixnew[t] = d->ixnew[t-1] + tlen + 1; /* fseek(bt) */
t++;
}
}
@@ -212,8 +209,8 @@
Bprint(&stdout, "%s%d", separator, b);
}
-static void
-fetch(long *f, int a, int b, Biobuf *bp, char *s)
+void
+fetch(Diff *d, long *f, int a, int b, Biobuf *bp, char *s)
{
char buf[MAXLINELEN];
int maxb;
@@ -220,10 +217,10 @@
if(a <= 1)
a = 1;
- if(bp == input[0])
- maxb = len[0];
+ if(bp == d->input[0])
+ maxb = d->len[0];
else
- maxb = len[1];
+ maxb = d->len[1];
if(b > maxb)
b = maxb;
if(a > maxb)
@@ -232,23 +229,12 @@
while (a++ <= b) {
readline(bp, buf);
Bprint(&stdout, "%s%s\n", s, buf);
+ Bflush(&stdout);
}
}
-typedef struct Change Change;
-struct Change
-{
- int a;
- int b;
- int c;
- int d;
-};
-
-Change *changes;
-int nchanges;
-
void
-change(int a, int b, int c, int d)
+change(Diff *df, int a, int b, int c, int d)
{
char verb;
char buf[4];
@@ -257,7 +243,7 @@
if (a > b && c > d)
return;
anychange = 1;
- if (mflag && firstchange == 0) {
+ if (mflag && df->firstchange == 0) {
if(mode) {
buf[0] = '-';
buf[1] = mode;
@@ -266,8 +252,8 @@
} else {
buf[0] = '\0';
}
- Bprint(&stdout, "diff %s%s %s\n", buf, file1, file2);
- firstchange = 1;
+ Bprint(&stdout, "diff %s%s %s\n", buf, df->file1, df->file2);
+ df->firstchange = 1;
}
verb = a > b ? 'a': c > d ? 'd': 'c';
switch(mode) {
@@ -281,10 +267,10 @@
range(c, d, ",");
break;
case 'n':
- Bprint(&stdout, "%s:", file1);
+ Bprint(&stdout, "%s:", df->file1);
range(a, b, ",");
Bprint(&stdout, " %c ", verb);
- Bprint(&stdout, "%s:", file2);
+ Bprint(&stdout, "%s:", df->file2);
range(c, d, ",");
break;
case 'f':
@@ -294,9 +280,9 @@
case 'c':
case 'a':
case 'u':
- if(nchanges%1024 == 0)
- changes = erealloc(changes, (nchanges+1024)*sizeof(changes[0]));
- ch = &changes[nchanges++];
+ if(df->nchanges%1024 == 0)
+ df->changes = erealloc(df->changes, (df->nchanges+1024)*sizeof(df->changes[0]));
+ ch = &df->changes[df->nchanges++];
ch->a = a;
ch->b = b;
ch->c = c;
@@ -305,11 +291,11 @@
}
Bputc(&stdout, '\n');
if (mode == 0 || mode == 'n') {
- fetch(ixold, a, b, input[0], "< ");
+ fetch(df, df->ixold, a, b, df->input[0], "< ");
if (a <= b && c <= d)
Bprint(&stdout, "---\n");
}
- fetch(ixnew, c, d, input[1], mode == 0 || mode == 'n' ? "> ": "");
+ fetch(df, df->ixnew, c, d, df->input[1], mode == 0 || mode == 'n' ? "> ": "");
if (mode != 0 && mode != 'n' && c <= d)
Bprint(&stdout, ".\n");
}
@@ -320,69 +306,69 @@
};
int
-changeset(int i)
+changeset(Diff *d, int i)
{
- while(i<nchanges && changes[i].b+1+2*Lines > changes[i+1].a)
+ while(i < d->nchanges && d->changes[i].b + 1 + 2*Lines > d->changes[i+1].a)
i++;
- if(i<nchanges)
+ if(i < d->nchanges)
return i+1;
- return nchanges;
+ return d->nchanges;
}
void
-flushchanges(void)
+flushchanges(Diff *df)
{
- int a, b, c, d, at, hdr;
- int i, j;
+ vlong a, b, c, d, at, hdr;
+ vlong i, j;
- if(nchanges == 0)
+ if(df->nchanges == 0)
return;
hdr = 0;
- for(i=0; i<nchanges; ){
- j = changeset(i);
- a = changes[i].a-Lines;
- b = changes[j-1].b+Lines;
- c = changes[i].c-Lines;
- d = changes[j-1].d+Lines;
+ for(i=0; i < df->nchanges; ){
+ j = changeset(df, i);
+ a = df->changes[i].a - Lines;
+ b = df->changes[j-1].b + Lines;
+ c = df->changes[i].c - Lines;
+ d = df->changes[j-1].d + Lines;
if(a < 1)
a = 1;
if(c < 1)
c = 1;
- if(b > len[0])
- b = len[0];
- if(d > len[1])
- d = len[1];
+ if(b > df->len[0])
+ b = df->len[0];
+ if(d > df->len[1])
+ d = df->len[1];
if(mode == 'a'){
a = 1;
- b = len[0];
+ b = df->len[0];
c = 1;
- d = len[1];
- j = nchanges;
+ d = df->len[1];
+ j = df->nchanges;
}
if(mode == 'u'){
if(!hdr){
- Bprint(&stdout, "--- %s\n", file1);
- Bprint(&stdout, "+++ %s\n", file2);
+ Bprint(&stdout, "--- %s\n", df->file1);
+ Bprint(&stdout, "+++ %s\n", df->file2);
hdr = 1;
}
- Bprint(&stdout, "@@ -%d,%d +%d,%d @@\n", a, b-a+1, c, d-c+1);
+ Bprint(&stdout, "@@ -%lld,%lld +%lld,%lld @@\n", a, b-a+1, c, d-c+1);
}else{
- Bprint(&stdout, "%s:", file1);
+ Bprint(&stdout, "%s:", df->file1);
range(a, b, ",");
Bprint(&stdout, " - ");
- Bprint(&stdout, "%s:", file2);
+ Bprint(&stdout, "%s:", df->file2);
range(c, d, ",");
Bputc(&stdout, '\n');
}
at = a;
for(; i<j; i++){
- fetch(ixold, at, changes[i].a-1, input[0], mode == 'u' ? " " : " ");
- fetch(ixold, changes[i].a, changes[i].b, input[0], mode == 'u' ? "-" : "- ");
- fetch(ixnew, changes[i].c, changes[i].d, input[1], mode == 'u' ? "+" : "+ ");
- at = changes[i].b+1;
+ fetch(df, df->ixold, at, df->changes[i].a-1, df->input[0], mode == 'u' ? " " : " ");
+ fetch(df, df->ixold, df->changes[i].a, df->changes[i].b, df->input[0], mode == 'u' ? "-" : "- ");
+ fetch(df, df->ixnew, df->changes[i].c, df->changes[i].d, df->input[1], mode == 'u' ? "+" : "+ ");
+ at = df->changes[i].b+1;
}
- fetch(ixold, at, b, input[0], mode == 'u' ? " " : " ");
+ fetch(df, df->ixold, at, b, df->input[0], mode == 'u' ? " " : " ");
}
- nchanges = 0;
+ df->nchanges = 0;
}
--- a/sys/src/cmd/diff/diffreg.c
+++ b/sys/src/cmd/diff/diffreg.c
@@ -66,30 +66,7 @@
* 3*(number of k-candidates installed), typically about
* 6n words for files of length n.
*/
-typedef struct Cand Cand;
-struct Cand {
- int x;
- int y;
- int pred;
-};
-
-Cand cand;
-Line *file[2], line;
-int len[2];
-int binary;
-Line *sfile[2]; /*shortened by pruning common prefix and suffix*/
-int slen[2];
-int pref, suff; /*length of prefix and suffix*/
-int *class; /*will be overlaid on file[0]*/
-int *member; /*will be overlaid on file[1]*/
-int *klist; /*will be overlaid on file[0] after class*/
-Cand *clist; /* merely a free storage pot for candidates */
-int clen;
-int *J; /*will be overlaid on class*/
-long *ixold; /*will be overlaid on klist*/
-long *ixnew; /*will be overlaid on file[1]*/
-
static void
sort(Line *a, int n) /*shellsort CACM #201*/
{
@@ -137,21 +114,21 @@
}
static void
-prune(void)
+prune(Diff *d)
{
int i,j;
- for(pref=0;pref<len[0]&&pref<len[1]&&
- file[0][pref+1].value==file[1][pref+1].value;
- pref++ ) ;
- for(suff=0;suff<len[0]-pref&&suff<len[1]-pref&&
- file[0][len[0]-suff].value==file[1][len[1]-suff].value;
- suff++) ;
+ for(d->pref = 0; d->pref < d->len[0] && d->pref < d->len[1] &&
+ d->file[0][d->pref+1].value == d->file[1][d->pref+1].value;
+ d->pref++) ;
+ for(d->suff=0; d->suff < d->len[0] - d->pref && d->suff < d->len[1] - d->pref &&
+ d->file[0][d->len[0] - d->suff].value == d->file[1][d->len[1] - d->suff].value;
+ d->suff++) ;
for(j=0;j<2;j++) {
- sfile[j] = file[j]+pref;
- slen[j] = len[j]-pref-suff;
- for(i=0;i<=slen[j];i++)
- sfile[j][i].serial = i;
+ d->sfile[j] = d->file[j]+d->pref;
+ d->slen[j] = d->len[j]-d->pref-d->suff;
+ for(i=0;i<=d->slen[j];i++)
+ d->sfile[j][i].serial = i;
}
}
@@ -184,30 +161,30 @@
}
static int
-newcand(int x, int y, int pred)
+newcand(Diff *d, int x, int y, int pred)
{
Cand *q;
- clist = erealloc(clist, (clen+1)*sizeof(Cand));
- q = clist + clen;
+ d->clist = erealloc(d->clist, (d->clen+1)*sizeof(Cand));
+ q = d->clist + d->clen;
q->x = x;
q->y = y;
q->pred = pred;
- return clen++;
+ return d->clen++;
}
static int
-search(int *c, int k, int y)
+search(Diff *d, int *c, int k, int y)
{
int i, j, l;
int t;
- if(clist[c[k]].y < y) /*quick look for typical case*/
+ if(d->clist[c[k]].y < y) /*quick look for typical case*/
return k+1;
i = 0;
j = k+1;
while((l=(i+j)/2) > i) {
- t = clist[c[l]].y;
+ t = d->clist[c[l]].y;
if(t > y)
j = l;
else if(t < y)
@@ -219,7 +196,7 @@
}
static int
-stone(int *a, int n, int *b, int *c)
+stone(Diff *d, int *a, int n, int *b, int *c)
{
int i, k,y;
int j, l;
@@ -227,7 +204,7 @@
int oldl;
k = 0;
- c[0] = newcand(0,0,0);
+ c[0] = newcand(d, 0, 0, 0);
for(i=1; i<=n; i++) {
j = a[i];
if(j==0)
@@ -236,20 +213,20 @@
oldl = 0;
oldc = c[0];
do {
- if(y <= clist[oldc].y)
+ if(y <= d->clist[oldc].y)
continue;
- l = search(c, k, y);
+ l = search(d, c, k, y);
if(l!=oldl+1)
oldc = c[l-1];
if(l<=k) {
- if(clist[c[l]].y <= y)
+ if(d->clist[c[l]].y <= y)
continue;
tc = c[l];
- c[l] = newcand(i,y,oldc);
+ c[l] = newcand(d, i, y, oldc);
oldc = tc;
oldl = l;
} else {
- c[l] = newcand(i,y,oldc);
+ c[l] = newcand(d, i,y,oldc);
k++;
break;
}
@@ -259,61 +236,23 @@
}
static void
-unravel(int p)
+unravel(Diff *d, int p)
{
int i;
Cand *q;
- for(i=0; i<=len[0]; i++) {
- if (i <= pref)
- J[i] = i;
- else if (i > len[0]-suff)
- J[i] = i+len[1]-len[0];
+ for(i=0; i<=d->len[0]; i++) {
+ if (i <= d->pref)
+ d->J[i] = i;
+ else if (i > d->len[0]-d->suff)
+ d->J[i] = i+d->len[1] - d->len[0];
else
- J[i] = 0;
+ d->J[i] = 0;
}
- for(q=clist+p;q->y!=0;q=clist+q->pred)
- J[q->x+pref] = q->y+pref;
+ for(q=d->clist+p; q->y != 0; q= d->clist + q->pred)
+ d->J[q->x+d->pref] = q->y+d->pref;
}
-static void
-output(void)
-{
- int m, i0, i1, j0, j1;
-
- m = len[0];
- J[0] = 0;
- J[m+1] = len[1]+1;
- if (mode != 'e') {
- for (i0 = 1; i0 <= m; i0 = i1+1) {
- while (i0 <= m && J[i0] == J[i0-1]+1)
- i0++;
- j0 = J[i0-1]+1;
- i1 = i0-1;
- while (i1 < m && J[i1+1] == 0)
- i1++;
- j1 = J[i1+1]-1;
- J[i1] = j1;
- change(i0, i1, j0, j1);
- }
- } else {
- for (i0 = m; i0 >= 1; i0 = i1-1) {
- while (i0 >= 1 && J[i0] == J[i0+1]-1 && J[i0])
- i0--;
- j0 = J[i0+1]-1;
- i1 = i0+1;
- while (i1 > 1 && J[i1-1] == 0)
- i1--;
- j1 = J[i1-1]+1;
- J[i1] = j1;
- change(i1 , i0, j1, j0);
- }
- }
- if (m == 0)
- change(1, 0, 1, len[1]);
- flushchanges();
-}
-
#define BUF 4096
static int
cmp(Biobuf* b1, Biobuf* b2)
@@ -361,21 +300,20 @@
}
void
-diffreg(char *f, char *fo, char *t, char *to)
+calcdiff(Diff *d, char *f, char *fo, char *t, char *to)
{
Biobuf *b0, *b1;
int k;
- binary = 0;
- b0 = prepare(0, f, fo);
+ b0 = prepare(d, 0, f, fo);
if (!b0)
return;
- b1 = prepare(1, t, to);
+ b1 = prepare(d, 1, t, to);
if (!b1) {
Bterm(b0);
return;
}
- if (binary){
+ if (d->binary){
// could use b0 and b1 but this is simpler.
if (cmp(b0, b1))
print("binary files %s %s differ\n", f, t);
@@ -383,38 +321,91 @@
Bterm(b1);
return;
}
- clen = 0;
- prune();
- sort(sfile[0], slen[0]);
- sort(sfile[1], slen[1]);
+ d->clen = 0;
+ prune(d);
+ sort(d->sfile[0], d->slen[0]);
+ sort(d->sfile[1], d->slen[1]);
- member = (int *)file[1];
- equiv(sfile[0], slen[0], sfile[1], slen[1], member);
- member = erealloc(member, (slen[1]+2)*sizeof(int));
+ d->member = (int *)d->file[1];
+ equiv(d->sfile[0], d->slen[0], d->sfile[1], d->slen[1], d->member);
+ d->member = erealloc(d->member, (d->slen[1]+2)*sizeof(int));
- class = (int *)file[0];
- unsort(sfile[0], slen[0], class);
- class = erealloc(class, (slen[0]+2)*sizeof(int));
+ d->class = (int *)d->file[0];
+ unsort(d->sfile[0], d->slen[0], d->class);
+ d->class = erealloc(d->class, (d->slen[0]+2)*sizeof(int));
- klist = emalloc((slen[0]+2)*sizeof(int));
- clist = emalloc(sizeof(Cand));
- k = stone(class, slen[0], member, klist);
- free(member);
- free(class);
+ d->klist = emalloc((d->slen[0]+2)*sizeof(int));
+ d->clist = emalloc(sizeof(Cand));
+ k = stone(d, d->class, d->slen[0], d->member, d->klist);
+ free(d->member);
+ free(d->class);
- J = emalloc((len[0]+2)*sizeof(int));
- unravel(klist[k]);
- free(clist);
- free(klist);
+ d->J = emalloc((d->len[0]+2)*sizeof(int));
+ unravel(d, d->klist[k]);
+ free(d->clist);
+ free(d->klist);
- ixold = emalloc((len[0]+2)*sizeof(long));
- ixnew = emalloc((len[1]+2)*sizeof(long));
+ d->ixold = emalloc((d->len[0]+2)*sizeof(long));
+ d->ixnew = emalloc((d->len[1]+2)*sizeof(long));
Bseek(b0, 0, 0); Bseek(b1, 0, 0);
- check(b0, b1);
- output();
- free(J);
- free(ixold);
- free(ixnew);
- Bterm(b0);
- Bterm(b1);
+ check(d, b0, b1);
+}
+
+static void
+output(Diff *d)
+{
+ int m, i0, i1, j0, j1;
+
+ m = d->len[0];
+ d->J[0] = 0;
+ d->J[m+1] = d->len[1]+1;
+ if (mode != 'e') {
+ for (i0 = 1; i0 <= m; i0 = i1+1) {
+ while (i0 <= m && d->J[i0] == d->J[i0-1]+1)
+ i0++;
+ j0 = d->J[i0-1]+1;
+ i1 = i0-1;
+ while (i1 < m && d->J[i1+1] == 0)
+ i1++;
+ j1 = d->J[i1+1]-1;
+ d->J[i1] = j1;
+ change(d, i0, i1, j0, j1);
+ }
+ } else {
+ for (i0 = m; i0 >= 1; i0 = i1-1) {
+ while (i0 >= 1 && d->J[i0] == d->J[i0+1]-1 && d->J[i0])
+ i0--;
+ j0 = d->J[i0+1]-1;
+ i1 = i0+1;
+ while (i1 > 1 && d->J[i1-1] == 0)
+ i1--;
+ j1 = d->J[i1-1]+1;
+ d->J[i1] = j1;
+ change(d, i1 , i0, j1, j0);
+ }
+ }
+ if (m == 0)
+ change(d, 1, 0, 1, d->len[1]);
+ flushchanges(d);
+}
+
+void
+diffreg(char *f, char *fo, char *t, char *to)
+{
+ Diff d;
+
+ memset(&d, 0, sizeof(d));
+ calcdiff(&d, f, fo, t, to);
+ output(&d);
+ freediff(&d);
+}
+
+void
+freediff(Diff *d)
+{
+ Bterm(d->input[0]);
+ Bterm(d->input[1]);
+ free(d->J);
+ free(d->ixold);
+ free(d->ixnew);
}
--- /dev/null
+++ b/sys/src/cmd/diff/merge3.c
@@ -1,0 +1,169 @@
+#include <u.h>
+#include <libc.h>
+#include <bio.h>
+#include "diff.h"
+
+static int
+changecmp(void *a, void *b)
+{
+ return ((Change*)a)->a - ((Change*)b)->a;
+}
+
+static void
+addchange(Diff *df, int a, int b, int c, int d)
+{
+ Change *ch;
+
+ if (a > b && c > d)
+ return;
+ if(df->nchanges%1024 == 0)
+ df->changes = erealloc(df->changes, (df->nchanges+1024)*sizeof(df->changes[0]));
+ ch = &df->changes[df->nchanges++];
+ ch->a = a;
+ ch->b = b;
+ ch->c = c;
+ ch->d = d;
+}
+
+static void
+collect(Diff *d)
+{
+ int m, i0, i1, j0, j1;
+
+ m = d->len[0];
+ d->J[0] = 0;
+ d->J[m+1] = d->len[1]+1;
+ for (i0 = m; i0 >= 1; i0 = i1-1) {
+ while (i0 >= 1 && d->J[i0] == d->J[i0+1]-1 && d->J[i0])
+ i0--;
+ j0 = d->J[i0+1]-1;
+ i1 = i0+1;
+ while (i1 > 1 && d->J[i1-1] == 0)
+ i1--;
+ j1 = d->J[i1-1]+1;
+ d->J[i1] = j1;
+ addchange(d, i1 , i0, j1, j0);
+ }
+ if (m == 0)
+ change(d, 1, 0, 1, d->len[1]);
+ qsort(d->changes, d->nchanges, sizeof(Change), changecmp);
+}
+
+static int
+overlaps(Change *l, Change *r)
+{
+ if(l == nil || r == nil)
+ return 0;
+ if(l->a <= r->a)
+ return l->b >= r->a;
+ else
+ return r->b >= l->a;
+}
+
+char*
+merge(Diff *l, Diff *r)
+{
+ int il, ir, x, y, δ;
+ Change *lc, *rc;
+ char *status;
+ vlong ln;
+
+ il = 0;
+ ir = 0;
+ ln = 0;
+ status = nil;
+ collect(l);
+ collect(r);
+ while(il < l->nchanges || ir < r->nchanges){
+ lc = nil;
+ rc = nil;
+ if(il < l->nchanges)
+ lc = &l->changes[il];
+ if(ir < r->nchanges)
+ rc = &r->changes[ir];
+ if(overlaps(lc, rc)){
+ /*
+ * align the edges of the chunks
+ */
+ if(lc->a < rc->a){
+ x = lc->c;
+ δ = rc->a - lc->a;
+ rc->a -= δ;
+ rc->c -= δ;
+ }else{
+ x = rc->c;
+ δ = lc->a - rc->a;
+ lc->a -= δ;
+ lc->c -= δ;
+ }
+ if(lc->b > rc->b){
+ y = lc->d;
+ δ = lc->b - rc->b;
+ rc->b += δ;
+ rc->d += δ;
+ }else{
+ y = rc->d;
+ δ = rc->b - lc->b;
+ lc->b += δ;
+ lc->d += δ;
+ }
+ fetch(l, l->ixold, ln, x-1, l->input[0], "");
+ Bprint(&stdout, "<<<<<<<<<< %s\n", l->file2);
+ fetch(l, l->ixnew, lc->c, lc->d, l->input[1], "");
+ Bprint(&stdout, "========== original\n");
+ fetch(l, l->ixold, x, y, l->input[0], "");
+ Bprint(&stdout, "========== %s\n", r->file2);
+ fetch(r, r->ixnew, rc->c, rc->d, r->input[1], "");
+ Bprint(&stdout, ">>>>>>>>>>\n");
+ ln = y+1;
+ il++;
+ ir++;
+ status = "conflict";
+ }else if(rc == nil || (lc != nil && lc->a < rc->a)){
+ fetch(l, l->ixold, ln, lc->a-1, l->input[0], "");
+ fetch(l, l->ixnew, lc->c, lc->d, l->input[1], "");
+ ln = lc->b+1;
+ il++;
+ }else if(lc == nil || (rc != nil && rc->a < lc->a)){
+ fetch(l, l->ixold, ln, rc->a-1, l->input[0], "");
+ fetch(r, r->ixnew, rc->c, rc->d, r->input[1], "");
+ ln = rc->b+1;
+ ir++;
+ }else
+ abort();
+ }
+ if(ln < l->len[0])
+ fetch(l, l->ixold, ln, l->len[0], l->input[0], "");
+ return status;
+}
+
+void
+usage(void)
+{
+ fprint(2, "usage: %s theirs base ours\n", argv0);
+ exits("usage");
+}
+
+void
+main(int argc, char **argv)
+{
+ Diff l, r;
+ char *x;
+
+ ARGBEGIN{
+ default:
+ usage();
+ }ARGEND;
+
+ if(argc != 3)
+ usage();
+ Binit(&stdout, 1, OWRITE);
+ memset(&l, 0, sizeof(l));
+ memset(&r, 0, sizeof(r));
+ calcdiff(&l, argv[1], argv[1], argv[0], argv[0]);
+ calcdiff(&r, argv[1], argv[1], argv[2], argv[2]);
+ x = merge(&l, &r);
+ freediff(&l);
+ freediff(&r);
+ exits(x);
+}
--- a/sys/src/cmd/diff/mkfile
+++ b/sys/src/cmd/diff/mkfile
@@ -1,13 +1,13 @@
< /$objtype/mkfile
-TARG=diff
+TARG=diff merge3
OFILES=\
diffdir.$O\
diffio.$O\
diffreg.$O\
- main.$O\
+ util.$O
HFILES=diff.h
BIN=/$objtype/bin
-</sys/src/cmd/mkone
+</sys/src/cmd/mkmany
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t10.1
@@ -1,0 +1,1 @@
+a line of text
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t10.2
@@ -1,0 +1,1 @@
+Another line of text
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t10.expected
@@ -1,0 +1,5 @@
+--- diff-t10.1
++++ diff-t10.2
+@@ -1 +1 @@
+-a line of text
++Another line of text
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t11.1
@@ -1,0 +1,1003 @@
+.\" $OpenBSD: t11.1,v 1.2 2007/11/27 16:22:12 martynas Exp $
+.\" $NetBSD: ed.1,v 1.13 1995/03/21 09:04:38 cgd Exp $
+.\"
+.TH ED 1 "21 May 1993"
+.SH NAME
+.\" ed, red \- text editor
+ed \- text editor
+.SH SYNOPSIS
+ed [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+.\" .LP
+.\" red [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+.SH DESCRIPTION
+.B ed
+is a line-oriented text editor.
+It is used to create, display, modify and otherwise manipulate text
+files.
+.\" .B red
+.\" is a restricted
+.\" .BR ed :
+.\" it can only edit files in the current
+.\" directory and cannot execute shell commands.
+
+If invoked with a
+.I file
+argument, then a copy of
+.I file
+is read into the editor's buffer.
+Changes are made to this copy and not directly to
+.I file
+itself.
+Upon quitting
+.BR ed ,
+any changes not explicitly saved with a
+.I `w'
+command are lost.
+
+Editing is done in two distinct modes:
+.I command
+and
+.IR input .
+When first invoked,
+.B ed
+is in command mode.
+In this mode commands are read from the standard input and
+executed to manipulate the contents of the editor buffer.
+A typical command might look like:
+.sp
+.RS
+,s/\fIold\fR/\fInew\fR/g
+.RE
+.sp
+which replaces all occurrences of the string
+.I old
+with
+.IR new .
+
+When an input command, such as
+.I `a'
+(append),
+.I `i'
+(insert) or
+.I `c'
+(change), is given,
+.B ed
+enters input mode. This is the primary means
+of adding text to a file.
+In this mode, no commands are available;
+instead, the standard input is written
+directly to the editor buffer. Lines consist of text up to and
+including a
+.IR newline
+character.
+Input mode is terminated by
+entering a single period (\fI.\fR) on a line.
+
+All
+.B ed
+commands operate on whole lines or ranges of lines; e.g.,
+the
+.I `d'
+command deletes lines; the
+.I `m'
+command moves lines, and so on.
+It is possible to modify only a portion of a line by means of replacement,
+as in the example above. However even here, the
+.I `s'
+command is applied to whole lines at a time.
+
+In general,
+.B ed
+commands consist of zero or more line addresses, followed by a single
+character command and possibly additional parameters; i.e.,
+commands have the structure:
+.sp
+.RS
+.I [address [,address]]command[parameters]
+.RE
+.sp
+The address(es) indicate the line or range of lines to be affected by the
+command. If fewer addresses are given than the command accepts, then
+default addresses are supplied.
+
+.SS OPTIONS
+.TP 8
+-s
+Suppresses diagnostics. This should be used if
+.BR ed 's
+standard input is from a script.
+
+.TP 8
+-x
+Prompts for an encryption key to be used in subsequent reads and writes
+(see the
+.I `x'
+command).
+
+.TP 8
+.RI \-p \ string
+Specifies a command prompt. This may be toggled on and off with the
+.I `P'
+command.
+
+.TP 8
+.I file
+Specifies the name of a file to read. If
+.I file
+is prefixed with a
+bang (!), then it is interpreted as a shell command. In this case,
+what is read is
+the standard output of
+.I file
+executed via
+.IR sh (1).
+To read a file whose name begins with a bang, prefix the
+name with a backslash (\\).
+The default filename is set to
+.I file
+only if it is not prefixed with a bang.
+
+.SS LINE ADDRESSING
+An address represents the number of a line in the buffer.
+.B ed
+maintains a
+.I current address
+which is
+typically supplied to commands as the default address when none is specified.
+When a file is first read, the current address is set to the last line
+of the file. In general, the current address is set to the last line
+affected by a command.
+
+A line address is
+constructed from one of the bases in the list below, optionally followed
+by a numeric offset. The offset may include any combination
+of digits, operators (i.e.,
+.IR + ,
+.I -
+and
+.IR ^ )
+and whitespace.
+Addresses are read from left to right, and their values are computed
+relative to the current address.
+
+One exception to the rule that addresses represent line numbers is the
+address
+.I 0
+(zero).
+This means "before the first line,"
+and is legal wherever it makes sense.
+
+An address range is two addresses separated either by a comma or
+semi-colon. The value of the first address in a range cannot exceed the
+value of the second. If only one address is given in a range, then
+the second address is set to the given address. If an
+.IR n- tuple
+of addresses is given where
+.I n > 2,
+then the corresponding range is determined by the last two addresses in
+the
+.IR n- tuple.
+If only one address is expected, then the last address is used.
+
+Each address in a comma-delimited range is interpreted relative to the
+current address. In a semi-colon-delimited range, the first address is
+used to set the current address, and the second address is interpreted
+relative to the first.
+
+
+The following address symbols are recognized.
+
+.TP 8
+\&.
+The current line (address) in the buffer.
+
+.TP 8
+$
+The last line in the buffer.
+
+.TP 8
+n
+The
+.IR n th,
+line in the buffer
+where
+.I n
+is a number in the range
+.I [0,$].
+
+.TP 8
+- or ^
+The previous line.
+This is equivalent to
+.I -1
+and may be repeated with cumulative effect.
+
+.TP 8
+-\fIn\fR or ^\fIn\fR
+The
+.IR n th
+previous line, where
+.I n
+is a non-negative number.
+
+.TP 8
++
+The
+next line.
+This is equivalent to
+.I +1
+and may be repeated with cumulative effect.
+
+.TP 8
++\fIn\fR or whitespace\fIn\fR
+The
+.IR n th
+next line, where
+.I n
+is a non-negative number.
+.I whitespace
+followed by a number
+.I n
+is interpreted as
+.IR +n .
+
+.TP 8
+, \fRor\fB %
+The first through last lines in the buffer. This is equivalent to
+the address range
+.I 1,$.
+
+.TP 8
+;
+The
+current through last lines in the buffer. This is equivalent to
+the address range
+.I .,$.
+
+.TP 8
+.RI / re/
+The
+next line containing the regular expression
+.IR re .
+The search wraps to the beginning of the buffer and continues down to the
+current line, if necessary.
+// repeats the last search.
+
+.TP 8
+.RI ? re?
+The
+previous line containing the regular expression
+.IR re .
+The search wraps to the end of the buffer and continues up to the
+current line, if necessary.
+?? repeats the last search.
+
+.TP 8
+.RI \' lc
+The
+line previously marked by a
+.I `k'
+(mark) command, where
+.I lc
+is a lower case letter.
+
+.SS REGULAR EXPRESSIONS
+Regular expressions are patterns used in selecting text.
+For example, the
+.B ed
+command
+.sp
+.RS
+g/\fIstring\fR/
+.RE
+.sp
+prints all lines containing
+.IR string .
+Regular expressions are also
+used by the
+.I `s'
+command for selecting old text to be replaced with new.
+
+In addition to a specifying string literals, regular expressions can
+represent
+classes of strings. Strings thus represented are said to be matched
+by the corresponding regular expression.
+If it is possible for a regular expression
+to match several strings in a line, then the left-most longest match is
+the one selected.
+
+The following symbols are used in constructing regular expressions:
+
+.TP 8
+c
+Any character
+.I c
+not listed below, including `{', '}', `(', `)', `<' and `>',
+matches itself.
+
+.TP 8
+\fR\e\fIc\fR
+Any backslash-escaped character
+.IR c ,
+except for `{', '}', `(', `)', `<' and `>',
+matches itself.
+
+.TP 8
+\fR.\fR
+Matches any single character.
+
+.TP 8
+.I [char-class]
+Matches any single character in
+.IR char-class .
+To include a `]'
+in
+.IR char-class ,
+it must be the first character.
+A range of characters may be specified by separating the end characters
+of the range with a `-', e.g., `a-z' specifies the lower case characters.
+The following literal expressions can also be used in
+.I char-class
+to specify sets of characters:
+.sp
+\ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
+.PD 0
+\ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
+.PD 0
+\ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
+.sp
+If `-' appears as the first or last
+character of
+.IR char-class ,
+then it matches itself.
+All other characters in
+.I char-class
+match themselves.
+.sp
+Patterns in
+.I char-class
+of the form:
+.sp
+\ \ [.\fIcol-elm\fR.] or,
+.PD 0
+\ \ [=\fIcol-elm\fR=]
+.sp
+where
+.I col-elm
+is a
+.I collating element
+are interpreted according to
+.IR locale (5)
+(not currently supported).
+See
+.IR regex (3)
+for an explanation of these constructs.
+
+.TP 8
+[^\fIchar-class\fR]
+Matches any single character, other than newline, not in
+.IR char-class .
+.IR char-class
+is defined
+as above.
+
+.TP 8
+^
+If `^' is the first character of a regular expression, then it
+anchors the regular expression to the beginning of a line.
+Otherwise, it matches itself.
+
+.TP 8
+$
+If `$' is the last character of a regular expression, it
+anchors the regular expression to the end of a line.
+Otherwise, it matches itself.
+
+.TP 8
+\fR\e<\fR
+Anchors the single character regular expression or subexpression
+immediately following it to the beginning of a word.
+(This may not be available)
+
+.TP 8
+\fR\e>\fR
+Anchors the single character regular expression or subexpression
+immediately following it to the end of a word.
+(This may not be available)
+
+.TP 8
+\fR\e(\fIre\fR\e)\fR
+Defines a subexpression
+.IR re .
+Subexpressions may be nested.
+A subsequent backreference of the form \fI`\en'\fR, where
+.I n
+is a number in the range [1,9], expands to the text matched by the
+.IR n th
+subexpression.
+For example, the regular expression `\e(.*\e)\e1' matches any string
+consisting of identical adjacent substrings.
+Subexpressions are ordered relative to
+their left delimiter.
+
+.TP 8
+*
+Matches the single character regular expression or subexpression
+immediately preceding it zero or more times. If '*' is the first
+character of a regular expression or subexpression, then it matches
+itself. The `*' operator sometimes yields unexpected results.
+For example, the regular expression `b*' matches the beginning of
+the string `abbb' (as opposed to the substring `bbb'), since a null match
+is the only left-most match.
+
+.TP 8
+\fR\e{\fIn,m\fR\e}\fR or \fR\e{\fIn,\fR\e}\fR or \fR\e{\fIn\fR\e}\fR
+Matches the single character regular expression or subexpression
+immediately preceding it at least
+.I n
+and at most
+.I m
+times.
+If
+.I m
+is omitted, then it matches at least
+.I n
+times.
+If the comma is also omitted, then it matches exactly
+.I n
+times.
+
+.LP
+Additional regular expression operators may be defined depending on the
+particular
+.IR regex (3)
+implementation.
+
+.SS COMMANDS
+All
+.B ed
+commands are single characters, though some require additonal parameters.
+If a command's parameters extend over several lines, then
+each line except for the last
+must be terminated with a backslash (\\).
+
+In general, at most one command is allowed per line.
+However, most commands accept a print suffix, which is any of
+.I `p'
+(print),
+.I `l'
+(list) ,
+or
+.I `n'
+(enumerate),
+to print the last line affected by the command.
+
+An interrupt (typically ^C) has the effect of aborting the current command
+and returning the editor to command mode.
+
+.B ed
+recognizes the following commands. The commands are shown together with
+the default address or address range supplied if none is
+specified (in parenthesis).
+
+.TP 8
+(.)a
+Appends text to the buffer after the addressed line.
+Text is entered in input mode.
+The current address is set to last line entered.
+
+.TP 8
+(.,.)c
+Changes lines in the buffer. The addressed lines are deleted
+from the buffer, and text is appended in their place.
+Text is entered in input mode.
+The current address is set to last line entered.
+
+.TP 8
+(.,.)d
+Deletes the addressed lines from the buffer.
+If there is a line after the deleted range, then the current address is set
+to this line. Otherwise the current address is set to the line
+before the deleted range.
+
+.TP 8
+.RI e \ file
+Edits
+.IR file ,
+and sets the default filename.
+If
+.I file
+is not specified, then the default filename is used.
+Any lines in the buffer are deleted before
+the new file is read.
+The current address is set to the last line read.
+
+.TP 8
+.RI e \ !command
+Edits the standard output of
+.IR `!command' ,
+(see
+.RI ! command
+below).
+The default filename is unchanged.
+Any lines in the buffer are deleted before the output of
+.I command
+is read.
+The current address is set to the last line read.
+
+.TP 8
+.RI E \ file
+Edits
+.I file
+unconditionally.
+This is similar to the
+.I e
+command,
+except that unwritten changes are discarded without warning.
+The current address is set to the last line read.
+
+.TP 8
+.RI f \ file
+Sets the default filename to
+.IR file .
+If
+.I file
+is not specified, then the default unescaped filename is printed.
+
+.TP 8
+.RI (1,$)g /re/command-list
+Applies
+.I command-list
+to each of the addressed lines matching a regular expression
+.IR re .
+The current address is set to the
+line currently matched before
+.I command-list
+is executed.
+At the end of the
+.I `g'
+command, the current address is set to the last line affected by
+.IR command-list .
+
+Each command in
+.I command-list
+must be on a separate line,
+and every line except for the last must be terminated by a backslash
+(\\).
+Any commands are allowed, except for
+.IR `g' ,
+.IR `G' ,
+.IR `v' ,
+and
+.IR `V' .
+A newline alone in
+.I command-list
+is equivalent to a
+.I `p'
+command.
+
+.TP 8
+.RI (1,$)G /re/
+Interactively edits the addressed lines matching a regular expression
+.IR re.
+For each matching line,
+the line is printed,
+the current address is set,
+and the user is prompted to enter a
+.IR command-list .
+At the end of the
+.I `G'
+command, the current address
+is set to the last line affected by (the last)
+.IR command-list .
+
+The format of
+.I command-list
+is the same as that of the
+.I `g'
+command. A newline alone acts as a null command list.
+A single `&' repeats the last non-null command list.
+
+.TP 8
+H
+Toggles the printing of error explanations.
+By default, explanations are not printed.
+It is recommended that ed scripts begin with this command to
+aid in debugging.
+
+.TP 8
+h
+Prints an explanation of the last error.
+
+.TP 8
+(.)i
+Inserts text in the buffer before the current line.
+Text is entered in input mode.
+The current address is set to the last line entered.
+
+.TP 8
+(.,.+1)j
+Joins the addressed lines. The addressed lines are
+deleted from the buffer and replaced by a single
+line containing their joined text.
+The current address is set to the resultant line.
+
+.TP 8
+.RI (.)k lc
+Marks a line with a lower case letter
+.IR lc .
+The line can then be addressed as
+.I 'lc
+(i.e., a single quote followed by
+.I lc
+) in subsequent commands. The mark is not cleared until the line is
+deleted or otherwise modified.
+
+.TP 8
+(.,.)l
+Prints the addressed lines unambiguously.
+If a single line fills for than one screen (as might be the case
+when viewing a binary file, for instance), a `--More--'
+prompt is printed on the last line.
+.B ed
+waits until the RETURN key is pressed
+before displaying the next screen.
+The current address is set to the last line
+printed.
+
+.TP 8
+(.,.)m(.)
+Moves lines in the buffer. The addressed lines are moved to after the
+right-hand destination address, which may be the address
+.IR 0
+(zero).
+The current address is set to the
+last line moved.
+
+.TP 8
+(.,.)n
+Prints the addressed lines along with
+their line numbers. The current address is set to the last line
+printed.
+
+.TP 8
+(.,.)p
+Prints the addressed lines. The current address is set to the last line
+printed.
+
+.TP 8
+P
+Toggles the command prompt on and off.
+Unless a prompt was specified by with command-line option
+\fI-p string\fR, the command prompt is by default turned off.
+
+.TP 8
+q
+Quits ed.
+
+.TP 8
+Q
+Quits ed unconditionally.
+This is similar to the
+.I q
+command,
+except that unwritten changes are discarded without warning.
+
+.TP 8
+.RI ($)r \ file
+Reads
+.I file
+to after the addressed line. If
+.I file
+is not specified, then the default
+filename is used. If there was no default filename prior to the command,
+then the default filename is set to
+.IR file .
+Otherwise, the default filename is unchanged.
+The current address is set to the last line read.
+
+.TP 8
+.RI ($)r \ !command
+Reads
+to after the addressed line
+the standard output of
+.IR `!command' ,
+(see the
+.RI ! command
+below).
+The default filename is unchanged.
+The current address is set to the last line read.
+
+.HP
+.RI (.,.)s /re/replacement/
+.PD 0
+.HP
+.RI (.,.)s /re/replacement/\fRg\fR
+.HP
+.RI (.,.)s /re/replacement/n
+.br
+Replaces text in the addressed lines
+matching a regular expression
+.I re
+with
+.IR replacement .
+By default, only the first match in each line is replaced.
+If the
+.I `g'
+(global) suffix is given, then every match to be replaced.
+The
+.I `n'
+suffix, where
+.I n
+is a postive number, causes only the
+.IR n th
+match to be replaced.
+It is an error if no substitutions are performed on any of the addressed
+lines.
+The current address is set the last line affected.
+
+.I re
+and
+.I replacement
+may be delimited by any character other than space and newline
+(see the
+.I `s'
+command below).
+If one or two of the last delimiters is omitted, then the last line
+affected is printed as though the print suffix
+.I `p'
+were specified.
+
+
+An unescaped `&' in
+.I replacement
+is replaced by the currently matched text.
+The character sequence
+\fI`\em'\fR,
+where
+.I m
+is a number in the range [1,9], is replaced by the
+.IR m th
+backreference expression of the matched text.
+If
+.I replacement
+consists of a single `%', then
+.I replacement
+from the last substitution is used.
+Newlines may be embedded in
+.I replacement
+if they are escaped with a backslash (\\).
+
+.TP 8
+(.,.)s
+Repeats the last substitution.
+This form of the
+.I `s'
+command accepts a count suffix
+.IR `n' ,
+or any combination of the characters
+.IR `r' ,
+.IR `g' ,
+and
+.IR `p' .
+If a count suffix
+.I `n'
+is given, then only the
+.IR n th
+match is replaced.
+The
+.I `r'
+suffix causes
+the regular expression of the last search to be used instead of the
+that of the last substitution.
+The
+.I `g'
+suffix toggles the global suffix of the last substitution.
+The
+.I `p'
+suffix toggles the print suffix of the last substitution
+The current address is set to the last line affected.
+
+.TP 8
+(.,.)t(.)
+Copies (i.e., transfers) the addressed lines to after the right-hand
+destination address, which may be the address
+.IR 0
+(zero).
+The current address is set to the last line
+copied.
+
+.TP 8
+u
+Undoes the last command and restores the current address
+to what it was before the command.
+The global commands
+.IR `g' ,
+.IR `G' ,
+.IR `v' ,
+and
+.IR `V' .
+are treated as a single command by undo.
+.I `u'
+is its own inverse.
+
+.TP 8
+.RI (1,$)v /pat/command-list
+Applies
+.I command-list
+to each of the addressed lines not matching a regular expression
+.IR re .
+This is similar to the
+.I `g'
+command.
+
+.TP 8
+.RI (1,$)V /re/
+Interactively edits the addressed lines not matching a regular expression
+.IR re.
+This is similar to the
+.I `G'
+command.
+
+.TP 8
+.RI (1,$)w \ file
+Writes the addressed lines to
+.IR file .
+Any previous contents of
+.I file
+is lost without warning.
+If there is no default filename, then the default filename is set to
+.IR file,
+otherwise it is unchanged. If no filename is specified, then the default
+filename is used.
+The current address is unchanged.
+
+.TP 8
+.RI (1,$)wq \ file
+Writes the addressed lines to
+.IR file ,
+and then executes a
+.I `q'
+command.
+
+.TP 8
+.RI (1,$)w \ !command
+Writes the addressed lines to the standard input of
+.IR `!command' ,
+(see the
+.RI ! command
+below).
+The default filename and current address are unchanged.
+
+.TP 8
+.RI (1,$)W \ file
+Appends the addressed lines to the end of
+.IR file .
+This is similar to the
+.I `w'
+command, expect that the previous contents of file is not clobbered.
+The current address is unchanged.
+
+.TP 8
+x
+Prompts for an encryption key which is used in subsequent reads and
+writes. If a newline alone is entered as the key, then encryption is
+turned off. Otherwise, echoing is disabled while a key is read.
+Encryption/decryption is done using the bdes(1) algorithm.
+
+.TP 8
+.RI (.+1)z n
+Scrolls
+.I n
+lines at a time starting at addressed line. If
+.I n
+is not specified, then the current window size is used.
+The current address is set to the last line printed.
+
+.TP 8
+.RI ! command
+Executes
+.I command
+via
+.IR sh (1).
+If the first character of
+.I command
+is `!', then it is replaced by text of the
+previous
+.IR `!command' .
+.B ed
+does not process
+.I command
+for backslash (\\) escapes.
+However, an unescaped
+.I `%'
+is replaced by the default filename.
+When the shell returns from execution, a `!'
+is printed to the standard output.
+The current line is unchanged.
+
+.TP 8
+($)=
+Prints the line number of the addressed line.
+
+.TP 8
+(.+1)newline
+Prints the addressed line, and sets the current address to
+that line.
+
+.SH FILES
+.TP 20
+/tmp/ed.*
+Buffer file
+.PD 0
+.TP 20
+ed.hup
+The file to which
+.B ed
+attempts to write the buffer if the terminal hangs up.
+
+.SH SEE ALSO
+
+.IR vi (1),
+.IR sed (1),
+.IR regex (3),
+.IR bdes (1),
+.IR sh (1).
+
+USD:12-13
+
+B. W. Kernighan and P. J. Plauger,
+.I Software Tools in Pascal ,
+Addison-Wesley, 1981.
+
+.SH LIMITATIONS
+.B ed
+processes
+.I file
+arguments for backslash escapes, i.e., in a filename,
+any characters preceded by a backslash (\\) are
+interpreted literally.
+
+If a text (non-binary) file is not terminated by a newline character,
+then
+.B ed
+appends one on reading/writing it. In the case of a binary file,
+.B ed
+does not append a newline on reading/writing.
+
+per line overhead: 4 ints
+
+.SH DIAGNOSTICS
+When an error occurs,
+.B ed
+prints a `?' and either returns to command mode
+or exits if its input is from a script.
+An explanation of the last error can be
+printed with the
+.I `h'
+(help) command.
+
+Since the
+.I `g'
+(global) command masks any errors from failed searches and substitutions,
+it can be used to perform conditional operations in scripts; e.g.,
+.sp
+.RS
+g/\fIold\fR/s//\fInew\fR/
+.RE
+.sp
+replaces any occurrences of
+.I old
+with
+.IR new .
+If the
+.I `u'
+(undo) command occurs in a global command list, then
+the command list is executed only once.
+
+If diagnostics are not disabled, attempting to quit
+.B ed
+or edit another file before writing a modified buffer
+results in an error.
+If the command is entered a second time, it succeeds,
+but any changes to the buffer are lost.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t11.2
@@ -1,0 +1,908 @@
+.\" $OpenBSD: t11.2,v 1.1 2003/07/21 20:16:21 otto Exp $
+.\"
+.Dd May 2, 1993
+.Dt ED 1
+.Os
+.Sh NAME
+.Nm ed
+.Nd text editor
+.Sh SYNOPSIS
+.Nm ed
+.Op Fl
+.Op Fl sx
+.Op Fl p Ar string
+.Op Ar file
+.Sh DESCRIPTION
+.Nm
+is a line-oriented text editor.
+It is used to create, display, modify, and otherwise manipulate text files.
+If invoked with a
+.Ar file
+argument, then a copy of
+.Ar file
+is read into the editor's buffer.
+Changes are made to this copy and not directly to
+.Ar file
+itself.
+Upon quitting
+.Nm ed ,
+any changes not explicitly saved with a
+.Em w
+command are lost.
+.Pp
+Editing is done in two distinct modes:
+.Em command
+and
+.Em input .
+When first invoked,
+.Nm
+is in command mode.
+In this mode, commands are read from the standard input and
+executed to manipulate the contents of the editor buffer.
+.Pp
+A typical command might look like:
+.Bd -literal -offset indent
+,s/old/new/g
+.Ed
+.Pp
+which replaces all occurrences of the string
+.Pa old
+with
+.Pa new .
+.Pp
+When an input command, such as
+.Em a
+(append),
+.Em i
+(insert),
+or
+.Em c
+(change) is given,
+.Nm
+enters input mode.
+This is the primary means of adding text to a file.
+In this mode, no commands are available;
+instead, the standard input is written directory to the editor buffer.
+Lines consist of text up to and including a newline character.
+Input mode is terminated by entering a single period
+.Pq Ql \&.
+on a line.
+.Pp
+All
+.Nm
+commands operate on whole lines or ranges of lines; e.g.,
+the
+.Em d
+command deletes lines; the
+.Em m
+command moves lines, and so on.
+It is possible to modify only a portion of a line by means of replacement,
+as in the example above.
+However, even here, the
+.Em s
+command is applied to whole lines at a time.
+.Pp
+In general,
+.Nm
+commands consist of zero or more line addresses, followed by a single
+character command and possibly additional parameters; i.e.,
+commands have the structure:
+.Bd -literal -offset indent
+[address [,address]]command[parameters]
+.Ed
+.Pp
+The address(es) indicate the line or range of lines to be affected by the
+command.
+If fewer addresses are given than the command accepts, then
+default addresses are supplied.
+.Pp
+The options are as follows:
+.Bl -tag -width Ds
+.It Fl
+Same as the
+.Fl s
+option (deprecated).
+.It Fl s
+Suppress diagnostics.
+This should be used if
+.Nm
+standard input is from a script.
+.Fl s
+flag.
+.It Fl x
+Prompt for an encryption key to be used in subsequent reads and writes
+(see the
+.Em x
+command).
+.It Fl p Ar string
+Specifies a command prompt.
+This may be toggled on and off with the
+.Em P
+command.
+.It Ar file
+Specifies the name of a file to read.
+If
+.Ar file
+is prefixed with a
+bang
+.Pq Ql \&! ,
+then it is interpreted as a shell command.
+In this case, what is read is the standard output of
+.Ar file
+executed via
+.Xr sh 1 .
+To read a file whose name begins with a bang, prefix the
+name with a backslash
+.Pq Ql \e .
+The default filename is set to
+.Ar file
+only if it is not prefixed with a bang.
+.El
+.Ss LINE ADDRESSING
+An address represents the number of a line in the buffer.
+.Nm
+maintains a
+.Em current address
+which is typically supplied to commands as the default address
+when none is specified.
+When a file is first read, the current address is set to the last line
+of the file.
+In general, the current address is set to the last line affected by a command.
+.Pp
+A line address is
+constructed from one of the bases in the list below, optionally followed
+by a numeric offset.
+The offset may include any combination of digits, operators (i.e.,
+.Em + ,
+.Em - ,
+and
+.Em ^ ) ,
+and whitespace.
+Addresses are read from left to right, and their values are computed
+relative to the current address.
+.Pp
+One exception to the rule that addresses represent line numbers is the
+address
+.Em 0
+(zero).
+This means
+.Dq before the first line ,
+and is legal wherever it makes sense.
+.Pp
+An address range is two addresses separated either by a comma or semi-colon.
+The value of the first address in a range cannot exceed the
+value of the second.
+If only one address is given in a range,
+then the second address is set to the given address.
+If an
+.Em n Ns No -tuple
+of addresses is given where
+.Em n > 2 ,
+then the corresponding range is determined by the last two addresses in the
+.Em n Ns No -tuple.
+If only one address is expected, then the last address is used.
+.Pp
+Each address in a comma-delimited range is interpreted relative to the
+current address.
+In a semi-colon-delimited range, the first address is
+used to set the current address, and the second address is interpreted
+relative to the first.
+.Pp
+The following address symbols are recognized:
+.Bl -tag -width Ds
+.It Em \&.
+The current line (address) in the buffer.
+.It Em $
+The last line in the buffer.
+.It Em n
+The
+.Em n Ns No th
+line in the buffer where
+.Em n
+is a number in the range
+.Em [0,$] .
+.It Em - No or Em ^
+The previous line.
+This is equivalent to
+.Em -1
+and may be repeated with cumulative effect.
+.It Em -n No or Em ^n
+The
+.Em n Ns No th
+previous line, where
+.Em n
+is a non-negative number.
+.It Em +
+The next line.
+This is equivalent to
+.Em +1
+and may be repeated with cumulative effect.
+.It Em +n
+The
+.Em n Ns No th
+next line, where
+.Em n
+is a non-negative number.
+.It Em \&, No or Em %
+The first through last lines in the buffer.
+This is equivalent to the address range
+.Em 1,$ .
+.It Em \&;
+The current through last lines in the buffer.
+This is equivalent to the address range
+.Em .,$ .
+.It Em / Ns No re Ns Em /
+The next line containing the regular expression
+.Em re .
+The search wraps to the beginning of the buffer and continues down to the
+current line, if necessary.
+.Em //
+repeats the last search.
+.It Em ? Ns No re Ns Em ?
+The previous line containing the regular expression
+.Em re .
+The search wraps to the end of the buffer and continues up to the
+current line, if necessary.
+.Em ??
+repeats the last search.
+.It Em \&\' Ns No lc
+The line previously marked by a
+.Em k
+(mark) command, where
+.Em lc
+is a lower case letter.
+.El
+.Ss REGULAR EXPRESSIONS
+Regular expressions are patterns used in selecting text.
+For example, the
+.Nm
+command
+.Bd -literal -offset indent
+g/string/
+.Ed
+.Pp
+prints all lines containing
+.Em string .
+Regular expressions are also used by the
+.Em s
+command for selecting old text to be replaced with new.
+.Pp
+In addition to a specifying string literals, regular expressions can
+represent classes of strings.
+Strings thus represented are said to be matched by the
+corresponding regular expression.
+If it is possible for a regular expression to match several strings in
+a line, then the leftmost longest match is the one selected.
+.Pp
+The following symbols are used in constructing regular expressions:
+.Bl -tag -width Dsasdfsd
+.It Em c
+Any character
+.Em c
+not listed below, including
+.Em { Ns No ,
+.Em } Ns No ,
+.Em \&( Ns No ,
+.Em \&) Ns No ,
+.Em < Ns No ,
+and
+.Em >
+matches itself.
+.It Em \ec
+Any backslash-escaped character
+.Em c Ns No ,
+except for
+.Em { Ns No ,
+.Em } Ns No ,
+.Em \&( Ns No ,
+.Em \&) Ns No ,
+.Em < Ns No , and
+.Em >
+matches itself.
+.It Em \&.
+Matches any single character.
+.It Em [char-class]
+Matches any single character in
+.Em char-class .
+To include a
+.Ql \&]
+in
+.Em char-class Ns No ,
+it must be the first character.
+A range of characters may be specified by separating the end characters
+of the range with a
+.Ql - ;
+e.g.,
+.Em a-z
+specifies the lower case characters.
+The following literal expressions can also be used in
+.Em char-class
+to specify sets of characters:
+.Pp
+.Em \ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
+.Em \ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
+.Em \ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
+.Pp
+If
+.Ql -
+appears as the first or last character of
+.Em char-class Ns No ,
+then it matches itself.
+All other characters in
+.Em char-class
+match themselves.
+.Pp
+Patterns in
+.Em char-class
+of the form
+.Em [.col-elm.] No or Em [=col-elm=]
+where
+.Em col-elm
+is a collating element are interpreted according to
+.Xr locale 5
+(not currently supported).
+See
+.Xr regex 3
+for an explanation of these constructs.
+.It Em [^char-class]
+Matches any single character, other than newline, not in
+.Em char-class Ns No .
+.Em char-class
+is defined as above.
+.It Em ^
+If
+.Em ^
+is the first character of a regular expression, then it
+anchors the regular expression to the beginning of a line.
+Otherwise, it matches itself.
+.It Em $
+If
+.Em $
+is the last character of a regular expression,
+it anchors the regular expression to the end of a line.
+Otherwise, it matches itself.
+.It Em \e<
+Anchors the single character regular expression or subexpression
+immediately following it to the beginning of a word.
+(This may not be available.)
+.It Em \e>
+Anchors the single character regular expression or subexpression
+immediately following it to the end of a word.
+(This may not be available.)
+.It Em \e( Ns No re Ns Em \e)
+Defines a subexpression
+.Em re .
+Subexpressions may be nested.
+A subsequent backreference of the form
+.Em \en Ns No ,
+where
+.Em n
+is a number in the range [1,9], expands to the text matched by the
+.Em n Ns No th
+subexpression.
+For example, the regular expression
+.Em \e(.*\e)\e1
+matches any string consisting of identical adjacent substrings.
+Subexpressions are ordered relative to their left delimiter.
+.It Em *
+Matches the single character regular expression or subexpression
+immediately preceding it zero or more times.
+If
+.Em *
+is the first character of a regular expression or subexpression,
+then it matches itself.
+The
+.Em *
+operator sometimes yields unexpected results.
+For example, the regular expression
+.Em b*
+matches the beginning of the string
+.Em abbb
+(as opposed to the substring
+.Em bbb Ns No ),
+since a null match is the only leftmost match.
+.Sm off
+.It Xo Em \e{ No n,m
+.Em \e}\ \e{ No n, Em \e}\
+.Em \e{ No n Em \e}
+.Xc
+.Sm on
+Matches the single character regular expression or subexpression
+immediately preceding it at least
+.Em n
+and at most
+.Em m
+times.
+If
+.Em m
+is omitted, then it matches at least
+.Em n
+times.
+If the comma is also omitted, then it matches exactly
+.Em n
+times.
+.El
+.Pp
+Additional regular expression operators may be defined depending on the
+particular
+.Xr regex 3
+implementation.
+.Ss COMMANDS
+All
+.Nm
+commands are single characters, though some require additional parameters.
+If a command's parameters extend over several lines, then
+each line except for the last must be terminated with a backslash
+.Pq Ql \e .
+.Pp
+In general, at most one command is allowed per line.
+However, most commands accept a print suffix, which is any of
+.Em p No (print),
+.Em l No (list),
+or
+.Em n No (enumerate),
+to print the last line affected by the command.
+.Pp
+An interrupt (typically ^C) has the effect of aborting the current command
+and returning the editor to command mode.
+.Pp
+.Nm
+recognizes the following commands.
+The commands are shown together with
+the default address or address range supplied if none is
+specified (in parentheses), and other possible arguments on the right.
+.Bl -tag -width Dxxs
+.It (.) Ns Em a
+Appends text to the buffer after the addressed line.
+Text is entered in input mode.
+The current address is set to last line entered.
+.It (.,.) Ns Em c
+Changes lines in the buffer.
+The addressed lines are deleted from the buffer,
+and text is appended in their place.
+Text is entered in input mode.
+The current address is set to last line entered.
+.It (.,.) Ns Em d
+Deletes the addressed lines from the buffer.
+If there is a line after the deleted range, then the current address is set
+to this line.
+Otherwise the current address is set to the line before the deleted range.
+.It Em e No file
+Edits
+.Em file Ns No ,
+and sets the default filename.
+If
+.Em file
+is not specified, then the default filename is used.
+Any lines in the buffer are deleted before the new file is read.
+The current address is set to the last line read.
+.It Em e No !command
+Edits the standard output of
+.Em !command Ns No ,
+(see
+.Em ! No command
+below).
+The default filename is unchanged.
+Any lines in the buffer are deleted before the output of
+.Em command
+is read.
+The current address is set to the last line read.
+.It Em E No file
+Edits
+.Em file
+unconditionally.
+This is similar to the
+.Em e
+command, except that unwritten changes are discarded without warning.
+The current address is set to the last line read.
+.It Em f No file
+Sets the default filename to
+.Em file Ns No .
+If
+.Em file
+is not specified, then the default unescaped filename is printed.
+.It (1,$) Ns Em g Ns No /re/command-list
+Applies
+.Em command-list
+to each of the addressed lines matching a regular expression
+.Em re Ns No .
+The current address is set to the line currently matched before
+.Em command-list
+is executed.
+At the end of the
+.Em g
+command, the current address is set to the last line affected by
+.Em command-list Ns No .
+.Pp
+Each command in
+.Em command-list
+must be on a separate line,
+and every line except for the last must be terminated by
+.Em \e No (backslash).
+Any commands are allowed, except for
+.Em g Ns No ,
+.Em G Ns No ,
+.Em v Ns No ,
+and
+.Em V Ns No .
+A newline alone in
+.Em command-list
+is equivalent to a
+.Em p
+command.
+.It (1,$) Ns Em G Ns No /re/
+Interactively edits the addressed lines matching a regular expression
+.Em re Ns No .
+For each matching line, the line is printed, the current address is set,
+and the user is prompted to enter a
+.Em command-list Ns No .
+At the end of the
+.Em g
+command, the current address is set to the last line affected by (the last)
+.Em command-list Ns No .
+.Pp
+The format of
+.Em command-list
+is the same as that of the
+.Em g
+command.
+A newline alone acts as a null command list.
+A single
+.Em &
+repeats the last non-null command list.
+.It Em H
+Toggles the printing of error explanations.
+By default, explanations are not printed.
+It is recommended that
+.Nm
+scripts begin with this command to aid in debugging.
+.It Em h
+Prints an explanation of the last error.
+.It (.) Ns Em i
+Inserts text in the buffer before the current line.
+Text is entered in input mode.
+The current address is set to the last line entered.
+.It (.,.+1) Ns Em j
+Joins the addressed lines.
+The addressed lines are deleted from the buffer and replaced by a single
+line containing their joined text.
+The current address is set to the resultant line.
+.It (.) Ns Em klc
+Marks a line with a lower case letter
+.Em lc Ns No \&.
+The line can then be addressed as
+.Em \&'lc
+(i.e., a single quote followed by
+.Em lc Ns No )
+in subsequent commands.
+The mark is not cleared until the line is deleted or otherwise modified.
+.It (.,.) Ns Em l
+Prints the addressed lines unambiguously.
+If a single line fills more than one screen (as might be the case
+when viewing a binary file, for instance), a
+.Dq --More--
+prompt is printed on the last line.
+.Nm
+waits until the RETURN key is pressed before displaying the next screen.
+The current address is set to the last line printed.
+.It (.,.) Ns Em m Ns No (.)
+Moves lines in the buffer.
+The addressed lines are moved to after the
+right-hand destination address, which may be the address
+.Em 0
+(zero).
+The current address is set to the last line moved.
+.It (.,.) Ns Em n
+Prints the addressed lines along with their line numbers.
+The current address is set to the last line printed.
+.It (.,.) Ns Em p
+Prints the addressed lines.
+The current address is set to the last line printed.
+.It Em P
+Toggles the command prompt on and off.
+Unless a prompt was specified by with command-line option
+.Fl p Ar string Ns No ,
+the command prompt is by default turned off.
+.It Em q
+Quits
+.Nm ed .
+.It Em Q
+Quits
+.Nm
+unconditionally.
+This is similar to the
+.Em q
+command, except that unwritten changes are discarded without warning.
+.It ($) Ns Em r No file
+Reads
+.Em file
+to after the addressed line.
+If
+.Em file
+is not specified, then the default filename is used.
+If there was no default filename prior to the command,
+then the default filename is set to
+.Em file Ns No .
+Otherwise, the default filename is unchanged.
+The current address is set to the last line read.
+.It ($) Ns Em r No !command
+Reads to after the addressed line the standard output of
+.Em !command Ns No ,
+(see the
+.Em !
+command below).
+The default filename is unchanged.
+The current address is set to the last line read.
+.Sm off
+.It Xo (.,.) Em s No /re/replacement/ , \ (.,.)
+.Em s No /re/replacement/ Em g , No \ (.,.)
+.Em s No /re/replacement/ Em n
+.Xc
+.Sm on
+Replaces text in the addressed lines matching a regular expression
+.Em re
+with
+.Em replacement Ns No .
+By default, only the first match in each line is replaced.
+If the
+.Em g
+(global) suffix is given, then every match to be replaced.
+The
+.Em n
+suffix, where
+.Em n
+is a positive number, causes only the
+.Em n Ns No th
+match to be replaced.
+It is an error if no substitutions are performed on any of the addressed
+lines.
+The current address is set the last line affected.
+.Pp
+.Em re
+and
+.Em replacement
+may be delimited by any character other than space and newline
+(see the
+.Em s
+command below).
+If one or two of the last delimiters is omitted, then the last line
+affected is printed as though the print suffix
+.Em p
+were specified.
+.Pp
+An unescaped
+.Ql \e
+in
+.Em replacement
+is replaced by the currently matched text.
+The character sequence
+.Em \em Ns No ,
+where
+.Em m
+is a number in the range [1,9], is replaced by the
+.Em m Ns No th
+backreference expression of the matched text.
+If
+.Em replacement
+consists of a single
+.Ql % ,
+then
+.Em replacement
+from the last substitution is used.
+Newlines may be embedded in
+.Em replacement
+if they are escaped with a backslash
+.Pq Ql \e .
+.It (.,.) Ns Em s
+Repeats the last substitution.
+This form of the
+.Em s
+command accepts a count suffix
+.Em n Ns No ,
+or any combination of the characters
+.Em r Ns No ,
+.Em g Ns No ,
+and
+.Em p Ns No .
+If a count suffix
+.Em n
+is given, then only the
+.Em n Ns No th
+match is replaced.
+The
+.Em r
+suffix causes
+the regular expression of the last search to be used instead of the
+that of the last substitution.
+The
+.Em g
+suffix toggles the global suffix of the last substitution.
+The
+.Em p
+suffix toggles the print suffix of the last substitution
+The current address is set to the last line affected.
+.It (.,.) Ns Em t Ns No (.)
+Copies (i.e., transfers) the addressed lines to after the right-hand
+destination address, which may be the address
+.Em 0
+(zero).
+The current address is set to the last line copied.
+.It Em u
+Undoes the last command and restores the current address
+to what it was before the command.
+The global commands
+.Em g Ns No ,
+.Em G Ns No ,
+.Em v Ns No ,
+and
+.Em V Ns No .
+are treated as a single command by undo.
+.Em u
+is its own inverse.
+.It (1,$) Ns Em v Ns No /re/command-list
+Applies
+.Em command-list
+to each of the addressed lines not matching a regular expression
+.Em re Ns No .
+This is similar to the
+.Em g
+command.
+.It (1,$) Ns Em V Ns No /re/
+Interactively edits the addressed lines not matching a regular expression
+.Em re Ns No .
+This is similar to the
+.Em G
+command.
+.It (1,$) Ns Em w No file
+Writes the addressed lines to
+.Em file Ns No .
+Any previous contents of
+.Em file
+is lost without warning.
+If there is no default filename, then the default filename is set to
+.Em file Ns No ,
+otherwise it is unchanged.
+If no filename is specified, then the default filename is used.
+The current address is unchanged.
+.It (1,$) Ns Em wq No file
+Writes the addressed lines to
+.Em file Ns No ,
+and then executes a
+.Em q
+command.
+.It (1,$) Ns Em w No !command
+Writes the addressed lines to the standard input of
+.Em !command Ns No ,
+(see the
+.Em !
+command below).
+The default filename and current address are unchanged.
+.It (1,$) Ns Em W No file
+Appends the addressed lines to the end of
+.Em file Ns No .
+This is similar to the
+.Em w
+command, expect that the previous contents of file is not clobbered.
+The current address is unchanged.
+.It Em x
+Prompts for an encryption key which is used in subsequent reads and writes.
+If a newline alone is entered as the key, then encryption is turned off.
+Otherwise, echoing is disabled while a key is read.
+Encryption/decryption is done using the
+.Xr bdes 1
+algorithm.
+.It (.+1) Ns Em z Ns No n
+Scrolls
+.Em n
+lines at a time starting at addressed line.
+If
+.Em n
+is not specified, then the current window size is used.
+The current address is set to the last line printed.
+.It ($) Ns Em =
+Prints the line number of the addressed line.
+.It (.+1) Ns Em newline
+Prints the addressed line, and sets the current address to that line.
+.It Em ! Ns No command
+Executes
+.Em command
+via
+.Xr sh 1 .
+If the first character of
+.Em command
+is
+.Em ! Ns No ,
+then it is replaced by text of the previous
+.Em !command Ns No .
+.Nm
+does not process
+.Em command
+for
+.Em \e
+(backslash) escapes.
+However, an unescaped
+.Em %
+is replaced by the default filename.
+When the shell returns from execution, a
+.Em !
+is printed to the standard output.
+The current line is unchanged.
+.El
+.Sh LIMITATIONS
+.Nm
+processes
+.Em file
+arguments for backslash escapes, i.e., in a filename,
+any characters preceded by a backslash
+.Pq Ql \e
+are interpreted literally.
+.Pp
+If a text (non-binary) file is not terminated by a newline character,
+then
+.Nm
+appends one on reading/writing it.
+In the case of a binary file,
+.Nm
+does not append a newline on reading/writing.
+.Sh DIAGNOSTICS
+When an error occurs,
+.Nm
+prints a
+.Dq ?
+and either returns to command mode or exits if its input is from a script.
+An explanation of the last error can be printed with the
+.Em h
+(help) command.
+.Pp
+Since the
+.Em g
+(global) command masks any errors from failed searches and substitutions,
+it can be used to perform conditional operations in scripts; e.g.,
+.Bd -literal -offset indent
+g/old/s//new/
+.Ed
+.Pp
+replaces any occurrences of
+.Em old
+with
+.Em new Ns No .
+.Pp
+If the
+.Em u
+(undo) command occurs in a global command list, then
+the command list is executed only once.
+.Pp
+If diagnostics are not disabled, attempting to quit
+.Nm
+or edit another file before writing a modified buffer results in an error.
+If the command is entered a second time, it succeeds,
+but any changes to the buffer are lost.
+.Sh FILES
+.Bl -tag -width /tmp/ed.* -compact
+.It Pa /tmp/ed.*
+buffer file
+.It Pa ed.hup
+where
+.Nm
+attempts to write the buffer if the terminal hangs up
+.El
+.Sh SEE ALSO
+.Xr bdes 1 ,
+.Xr sed 1 ,
+.Xr sh 1 ,
+.Xr vi 1 ,
+.Xr regex 3
+.Pp
+USD:12-13
+.Rs
+.%A B. W. Kernighan
+.%A P. J. Plauger
+.%B Software Tools in Pascal
+.%O Addison-Wesley
+.%D 1981
+.Re
+.Sh HISTORY
+An
+.Nm
+command appeared in
+.At v1 .
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t11.expected
@@ -1,0 +1,1612 @@
+--- diff-t11.1
++++ diff-t11.2
+@@ -1,1003 +1,908 @@
+-.\" $OpenBSD: t11.1,v 1.2 2007/11/27 16:22:12 martynas Exp $
+-.\" $NetBSD: ed.1,v 1.13 1995/03/21 09:04:38 cgd Exp $
++.\" $OpenBSD: t11.2,v 1.1 2003/07/21 20:16:21 otto Exp $
+ .\"
+-.TH ED 1 "21 May 1993"
+-.SH NAME
+-.\" ed, red \- text editor
+-ed \- text editor
+-.SH SYNOPSIS
+-ed [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+-.\" .LP
+-.\" red [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+-.SH DESCRIPTION
+-.B ed
++.Dd May 2, 1993
++.Dt ED 1
++.Os
++.Sh NAME
++.Nm ed
++.Nd text editor
++.Sh SYNOPSIS
++.Nm ed
++.Op Fl
++.Op Fl sx
++.Op Fl p Ar string
++.Op Ar file
++.Sh DESCRIPTION
++.Nm
+ is a line-oriented text editor.
+-It is used to create, display, modify and otherwise manipulate text
+-files.
+-.\" .B red
+-.\" is a restricted
+-.\" .BR ed :
+-.\" it can only edit files in the current
+-.\" directory and cannot execute shell commands.
+-
++It is used to create, display, modify, and otherwise manipulate text files.
+ If invoked with a
+-.I file
++.Ar file
+ argument, then a copy of
+-.I file
++.Ar file
+ is read into the editor's buffer.
+ Changes are made to this copy and not directly to
+-.I file
++.Ar file
+ itself.
+ Upon quitting
+-.BR ed ,
+-any changes not explicitly saved with a
+-.I `w'
++.Nm ed ,
++any changes not explicitly saved with a
++.Em w
+ command are lost.
+-
++.Pp
+ Editing is done in two distinct modes:
+-.I command
++.Em command
+ and
+-.IR input .
++.Em input .
+ When first invoked,
+-.B ed
++.Nm
+ is in command mode.
+-In this mode commands are read from the standard input and
++In this mode, commands are read from the standard input and
+ executed to manipulate the contents of the editor buffer.
++.Pp
+ A typical command might look like:
+-.sp
+-.RS
+-,s/\fIold\fR/\fInew\fR/g
+-.RE
+-.sp
++.Bd -literal -offset indent
++,s/old/new/g
++.Ed
++.Pp
+ which replaces all occurrences of the string
+-.I old
++.Pa old
+ with
+-.IR new .
+-
++.Pa new .
++.Pp
+ When an input command, such as
+-.I `a'
++.Em a
+ (append),
+-.I `i'
+-(insert) or
+-.I `c'
+-(change), is given,
+-.B ed
+-enters input mode. This is the primary means
+-of adding text to a file.
++.Em i
++(insert),
++or
++.Em c
++(change) is given,
++.Nm
++enters input mode.
++This is the primary means of adding text to a file.
+ In this mode, no commands are available;
+-instead, the standard input is written
+-directly to the editor buffer. Lines consist of text up to and
+-including a
+-.IR newline
+-character.
+-Input mode is terminated by
+-entering a single period (\fI.\fR) on a line.
+-
++instead, the standard input is written directory to the editor buffer.
++Lines consist of text up to and including a newline character.
++Input mode is terminated by entering a single period
++.Pq Ql \&.
++on a line.
++.Pp
+ All
+-.B ed
++.Nm
+ commands operate on whole lines or ranges of lines; e.g.,
+ the
+-.I `d'
++.Em d
+ command deletes lines; the
+-.I `m'
++.Em m
+ command moves lines, and so on.
+ It is possible to modify only a portion of a line by means of replacement,
+-as in the example above. However even here, the
+-.I `s'
++as in the example above.
++However, even here, the
++.Em s
+ command is applied to whole lines at a time.
+-
++.Pp
+ In general,
+-.B ed
++.Nm
+ commands consist of zero or more line addresses, followed by a single
+ character command and possibly additional parameters; i.e.,
+ commands have the structure:
+-.sp
+-.RS
+-.I [address [,address]]command[parameters]
+-.RE
+-.sp
++.Bd -literal -offset indent
++[address [,address]]command[parameters]
++.Ed
++.Pp
+ The address(es) indicate the line or range of lines to be affected by the
+-command. If fewer addresses are given than the command accepts, then
++command.
++If fewer addresses are given than the command accepts, then
+ default addresses are supplied.
+-
+-.SS OPTIONS
+-.TP 8
+--s
+-Suppresses diagnostics. This should be used if
+-.BR ed 's
++.Pp
++The options are as follows:
++.Bl -tag -width Ds
++.It Fl
++Same as the
++.Fl s
++option (deprecated).
++.It Fl s
++Suppress diagnostics.
++This should be used if
++.Nm
+ standard input is from a script.
+-
+-.TP 8
+--x
+-Prompts for an encryption key to be used in subsequent reads and writes
++.Fl s
++flag.
++.It Fl x
++Prompt for an encryption key to be used in subsequent reads and writes
+ (see the
+-.I `x'
++.Em x
+ command).
+-
+-.TP 8
+-.RI \-p \ string
+-Specifies a command prompt. This may be toggled on and off with the
+-.I `P'
++.It Fl p Ar string
++Specifies a command prompt.
++This may be toggled on and off with the
++.Em P
+ command.
+-
+-.TP 8
+-.I file
+-Specifies the name of a file to read. If
+-.I file
++.It Ar file
++Specifies the name of a file to read.
++If
++.Ar file
+ is prefixed with a
+-bang (!), then it is interpreted as a shell command. In this case,
+-what is read is
+-the standard output of
+-.I file
++bang
++.Pq Ql \&! ,
++then it is interpreted as a shell command.
++In this case, what is read is the standard output of
++.Ar file
+ executed via
+-.IR sh (1).
++.Xr sh 1 .
+ To read a file whose name begins with a bang, prefix the
+-name with a backslash (\\).
++name with a backslash
++.Pq Ql \e .
+ The default filename is set to
+-.I file
++.Ar file
+ only if it is not prefixed with a bang.
+-
+-.SS LINE ADDRESSING
++.El
++.Ss LINE ADDRESSING
+ An address represents the number of a line in the buffer.
+-.B ed
++.Nm
+ maintains a
+-.I current address
+-which is
+-typically supplied to commands as the default address when none is specified.
+-When a file is first read, the current address is set to the last line
+-of the file. In general, the current address is set to the last line
+-affected by a command.
+-
++.Em current address
++which is typically supplied to commands as the default address
++when none is specified.
++When a file is first read, the current address is set to the last line
++of the file.
++In general, the current address is set to the last line affected by a command.
++.Pp
+ A line address is
+ constructed from one of the bases in the list below, optionally followed
+-by a numeric offset. The offset may include any combination
+-of digits, operators (i.e.,
+-.IR + ,
+-.I -
++by a numeric offset.
++The offset may include any combination of digits, operators (i.e.,
++.Em + ,
++.Em - ,
+ and
+-.IR ^ )
++.Em ^ ) ,
+ and whitespace.
+ Addresses are read from left to right, and their values are computed
+ relative to the current address.
+-
++.Pp
+ One exception to the rule that addresses represent line numbers is the
+ address
+-.I 0
++.Em 0
+ (zero).
+-This means "before the first line,"
++This means
++.Dq before the first line ,
+ and is legal wherever it makes sense.
+-
+-An address range is two addresses separated either by a comma or
+-semi-colon. The value of the first address in a range cannot exceed the
+-value of the second. If only one address is given in a range, then
+-the second address is set to the given address. If an
+-.IR n- tuple
++.Pp
++An address range is two addresses separated either by a comma or semi-colon.
++The value of the first address in a range cannot exceed the
++value of the second.
++If only one address is given in a range,
++then the second address is set to the given address.
++If an
++.Em n Ns No -tuple
+ of addresses is given where
+-.I n > 2,
+-then the corresponding range is determined by the last two addresses in
+-the
+-.IR n- tuple.
++.Em n > 2 ,
++then the corresponding range is determined by the last two addresses in the
++.Em n Ns No -tuple.
+ If only one address is expected, then the last address is used.
+-
++.Pp
+ Each address in a comma-delimited range is interpreted relative to the
+-current address. In a semi-colon-delimited range, the first address is
++current address.
++In a semi-colon-delimited range, the first address is
+ used to set the current address, and the second address is interpreted
+ relative to the first.
+-
+-
+-The following address symbols are recognized.
+-
+-.TP 8
+-\&.
++.Pp
++The following address symbols are recognized:
++.Bl -tag -width Ds
++.It Em \&.
+ The current line (address) in the buffer.
+-
+-.TP 8
+-$
++.It Em $
+ The last line in the buffer.
+-
+-.TP 8
+-n
++.It Em n
+ The
+-.IR n th,
+-line in the buffer
+-where
+-.I n
++.Em n Ns No th
++line in the buffer where
++.Em n
+ is a number in the range
+-.I [0,$].
+-
+-.TP 8
+-- or ^
++.Em [0,$] .
++.It Em - No or Em ^
+ The previous line.
+ This is equivalent to
+-.I -1
++.Em -1
+ and may be repeated with cumulative effect.
+-
+-.TP 8
+--\fIn\fR or ^\fIn\fR
++.It Em -n No or Em ^n
+ The
+-.IR n th
++.Em n Ns No th
+ previous line, where
+-.I n
++.Em n
+ is a non-negative number.
+-
+-.TP 8
+-+
+-The
+-next line.
++.It Em +
++The next line.
+ This is equivalent to
+-.I +1
++.Em +1
+ and may be repeated with cumulative effect.
+-
+-.TP 8
+-+\fIn\fR or whitespace\fIn\fR
++.It Em +n
+ The
+-.IR n th
++.Em n Ns No th
+ next line, where
+-.I n
++.Em n
+ is a non-negative number.
+-.I whitespace
+-followed by a number
+-.I n
+-is interpreted as
+-.IR +n .
+-
+-.TP 8
+-, \fRor\fB %
+-The first through last lines in the buffer. This is equivalent to
+-the address range
+-.I 1,$.
+-
+-.TP 8
+-;
+-The
+-current through last lines in the buffer. This is equivalent to
+-the address range
+-.I .,$.
+-
+-.TP 8
+-.RI / re/
+-The
+-next line containing the regular expression
+-.IR re .
++.It Em \&, No or Em %
++The first through last lines in the buffer.
++This is equivalent to the address range
++.Em 1,$ .
++.It Em \&;
++The current through last lines in the buffer.
++This is equivalent to the address range
++.Em .,$ .
++.It Em / Ns No re Ns Em /
++The next line containing the regular expression
++.Em re .
+ The search wraps to the beginning of the buffer and continues down to the
+ current line, if necessary.
+-// repeats the last search.
+-
+-.TP 8
+-.RI ? re?
+-The
+-previous line containing the regular expression
+-.IR re .
++.Em //
++repeats the last search.
++.It Em ? Ns No re Ns Em ?
++The previous line containing the regular expression
++.Em re .
+ The search wraps to the end of the buffer and continues up to the
+ current line, if necessary.
+-?? repeats the last search.
+-
+-.TP 8
+-.RI \' lc
+-The
+-line previously marked by a
+-.I `k'
++.Em ??
++repeats the last search.
++.It Em \&\' Ns No lc
++The line previously marked by a
++.Em k
+ (mark) command, where
+-.I lc
++.Em lc
+ is a lower case letter.
+-
+-.SS REGULAR EXPRESSIONS
++.El
++.Ss REGULAR EXPRESSIONS
+ Regular expressions are patterns used in selecting text.
+ For example, the
+-.B ed
++.Nm
+ command
+-.sp
+-.RS
+-g/\fIstring\fR/
+-.RE
+-.sp
++.Bd -literal -offset indent
++g/string/
++.Ed
++.Pp
+ prints all lines containing
+-.IR string .
+-Regular expressions are also
+-used by the
+-.I `s'
++.Em string .
++Regular expressions are also used by the
++.Em s
+ command for selecting old text to be replaced with new.
+-
++.Pp
+ In addition to a specifying string literals, regular expressions can
+-represent
+-classes of strings. Strings thus represented are said to be matched
+-by the corresponding regular expression.
+-If it is possible for a regular expression
+-to match several strings in a line, then the left-most longest match is
+-the one selected.
+-
++represent classes of strings.
++Strings thus represented are said to be matched by the
++corresponding regular expression.
++If it is possible for a regular expression to match several strings in
++a line, then the leftmost longest match is the one selected.
++.Pp
+ The following symbols are used in constructing regular expressions:
+-
+-.TP 8
+-c
++.Bl -tag -width Dsasdfsd
++.It Em c
+ Any character
+-.I c
+-not listed below, including `{', '}', `(', `)', `<' and `>',
++.Em c
++not listed below, including
++.Em { Ns No ,
++.Em } Ns No ,
++.Em \&( Ns No ,
++.Em \&) Ns No ,
++.Em < Ns No ,
++and
++.Em >
+ matches itself.
+-
+-.TP 8
+-\fR\e\fIc\fR
++.It Em \ec
+ Any backslash-escaped character
+-.IR c ,
+-except for `{', '}', `(', `)', `<' and `>',
++.Em c Ns No ,
++except for
++.Em { Ns No ,
++.Em } Ns No ,
++.Em \&( Ns No ,
++.Em \&) Ns No ,
++.Em < Ns No , and
++.Em >
+ matches itself.
+-
+-.TP 8
+-\fR.\fR
++.It Em \&.
+ Matches any single character.
+-
+-.TP 8
+-.I [char-class]
++.It Em [char-class]
+ Matches any single character in
+-.IR char-class .
+-To include a `]'
++.Em char-class .
++To include a
++.Ql \&]
+ in
+-.IR char-class ,
++.Em char-class Ns No ,
+ it must be the first character.
+ A range of characters may be specified by separating the end characters
+-of the range with a `-', e.g., `a-z' specifies the lower case characters.
++of the range with a
++.Ql - ;
++e.g.,
++.Em a-z
++specifies the lower case characters.
+ The following literal expressions can also be used in
+-.I char-class
++.Em char-class
+ to specify sets of characters:
+-.sp
+-\ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
+-.PD 0
+-\ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
+-.PD 0
+-\ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
+-.sp
+-If `-' appears as the first or last
+-character of
+-.IR char-class ,
++.Pp
++.Em \ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
++.Em \ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
++.Em \ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
++.Pp
++If
++.Ql -
++appears as the first or last character of
++.Em char-class Ns No ,
+ then it matches itself.
+ All other characters in
+-.I char-class
++.Em char-class
+ match themselves.
+-.sp
++.Pp
+ Patterns in
+-.I char-class
+-of the form:
+-.sp
+-\ \ [.\fIcol-elm\fR.] or,
+-.PD 0
+-\ \ [=\fIcol-elm\fR=]
+-.sp
++.Em char-class
++of the form
++.Em [.col-elm.] No or Em [=col-elm=]
+ where
+-.I col-elm
+-is a
+-.I collating element
+-are interpreted according to
+-.IR locale (5)
++.Em col-elm
++is a collating element are interpreted according to
++.Xr locale 5
+ (not currently supported).
+ See
+-.IR regex (3)
++.Xr regex 3
+ for an explanation of these constructs.
+-
+-.TP 8
+-[^\fIchar-class\fR]
++.It Em [^char-class]
+ Matches any single character, other than newline, not in
+-.IR char-class .
+-.IR char-class
+-is defined
+-as above.
+-
+-.TP 8
+-^
+-If `^' is the first character of a regular expression, then it
++.Em char-class Ns No .
++.Em char-class
++is defined as above.
++.It Em ^
++If
++.Em ^
++is the first character of a regular expression, then it
+ anchors the regular expression to the beginning of a line.
+ Otherwise, it matches itself.
+-
+-.TP 8
+-$
+-If `$' is the last character of a regular expression, it
+-anchors the regular expression to the end of a line.
++.It Em $
++If
++.Em $
++is the last character of a regular expression,
++it anchors the regular expression to the end of a line.
+ Otherwise, it matches itself.
+-
+-.TP 8
+-\fR\e<\fR
++.It Em \e<
+ Anchors the single character regular expression or subexpression
+ immediately following it to the beginning of a word.
+-(This may not be available)
+-
+-.TP 8
+-\fR\e>\fR
++(This may not be available.)
++.It Em \e>
+ Anchors the single character regular expression or subexpression
+ immediately following it to the end of a word.
+-(This may not be available)
+-
+-.TP 8
+-\fR\e(\fIre\fR\e)\fR
++(This may not be available.)
++.It Em \e( Ns No re Ns Em \e)
+ Defines a subexpression
+-.IR re .
++.Em re .
+ Subexpressions may be nested.
+-A subsequent backreference of the form \fI`\en'\fR, where
+-.I n
++A subsequent backreference of the form
++.Em \en Ns No ,
++where
++.Em n
+ is a number in the range [1,9], expands to the text matched by the
+-.IR n th
++.Em n Ns No th
+ subexpression.
+-For example, the regular expression `\e(.*\e)\e1' matches any string
+-consisting of identical adjacent substrings.
+-Subexpressions are ordered relative to
+-their left delimiter.
+-
+-.TP 8
+-*
++For example, the regular expression
++.Em \e(.*\e)\e1
++matches any string consisting of identical adjacent substrings.
++Subexpressions are ordered relative to their left delimiter.
++.It Em *
+ Matches the single character regular expression or subexpression
+-immediately preceding it zero or more times. If '*' is the first
+-character of a regular expression or subexpression, then it matches
+-itself. The `*' operator sometimes yields unexpected results.
+-For example, the regular expression `b*' matches the beginning of
+-the string `abbb' (as opposed to the substring `bbb'), since a null match
+-is the only left-most match.
+-
+-.TP 8
+-\fR\e{\fIn,m\fR\e}\fR or \fR\e{\fIn,\fR\e}\fR or \fR\e{\fIn\fR\e}\fR
++immediately preceding it zero or more times.
++If
++.Em *
++is the first character of a regular expression or subexpression,
++then it matches itself.
++The
++.Em *
++operator sometimes yields unexpected results.
++For example, the regular expression
++.Em b*
++matches the beginning of the string
++.Em abbb
++(as opposed to the substring
++.Em bbb Ns No ),
++since a null match is the only leftmost match.
++.Sm off
++.It Xo Em \e{ No n,m
++.Em \e}\ \e{ No n, Em \e}\
++.Em \e{ No n Em \e}
++.Xc
++.Sm on
+ Matches the single character regular expression or subexpression
+ immediately preceding it at least
+-.I n
++.Em n
+ and at most
+-.I m
++.Em m
+ times.
+ If
+-.I m
++.Em m
+ is omitted, then it matches at least
+-.I n
++.Em n
+ times.
+ If the comma is also omitted, then it matches exactly
+-.I n
++.Em n
+ times.
+-
+-.LP
++.El
++.Pp
+ Additional regular expression operators may be defined depending on the
+ particular
+-.IR regex (3)
++.Xr regex 3
+ implementation.
+-
+-.SS COMMANDS
++.Ss COMMANDS
+ All
+-.B ed
+-commands are single characters, though some require additonal parameters.
++.Nm
++commands are single characters, though some require additional parameters.
+ If a command's parameters extend over several lines, then
+-each line except for the last
+-must be terminated with a backslash (\\).
+-
++each line except for the last must be terminated with a backslash
++.Pq Ql \e .
++.Pp
+ In general, at most one command is allowed per line.
+ However, most commands accept a print suffix, which is any of
+-.I `p'
+-(print),
+-.I `l'
+-(list) ,
++.Em p No (print),
++.Em l No (list),
+ or
+-.I `n'
+-(enumerate),
++.Em n No (enumerate),
+ to print the last line affected by the command.
+-
++.Pp
+ An interrupt (typically ^C) has the effect of aborting the current command
+ and returning the editor to command mode.
+-
+-.B ed
+-recognizes the following commands. The commands are shown together with
++.Pp
++.Nm
++recognizes the following commands.
++The commands are shown together with
+ the default address or address range supplied if none is
+-specified (in parenthesis).
+-
+-.TP 8
+-(.)a
++specified (in parentheses), and other possible arguments on the right.
++.Bl -tag -width Dxxs
++.It (.) Ns Em a
+ Appends text to the buffer after the addressed line.
+ Text is entered in input mode.
+ The current address is set to last line entered.
+-
+-.TP 8
+-(.,.)c
+-Changes lines in the buffer. The addressed lines are deleted
+-from the buffer, and text is appended in their place.
++.It (.,.) Ns Em c
++Changes lines in the buffer.
++The addressed lines are deleted from the buffer,
++and text is appended in their place.
+ Text is entered in input mode.
+ The current address is set to last line entered.
+-
+-.TP 8
+-(.,.)d
++.It (.,.) Ns Em d
+ Deletes the addressed lines from the buffer.
+ If there is a line after the deleted range, then the current address is set
+-to this line. Otherwise the current address is set to the line
+-before the deleted range.
+-
+-.TP 8
+-.RI e \ file
++to this line.
++Otherwise the current address is set to the line before the deleted range.
++.It Em e No file
+ Edits
+-.IR file ,
++.Em file Ns No ,
+ and sets the default filename.
+ If
+-.I file
+-is not specified, then the default filename is used.
+-Any lines in the buffer are deleted before
+-the new file is read.
++.Em file
++is not specified, then the default filename is used.
++Any lines in the buffer are deleted before the new file is read.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI e \ !command
++.It Em e No !command
+ Edits the standard output of
+-.IR `!command' ,
++.Em !command Ns No ,
+ (see
+-.RI ! command
++.Em ! No command
+ below).
+ The default filename is unchanged.
+ Any lines in the buffer are deleted before the output of
+-.I command
++.Em command
+ is read.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI E \ file
++.It Em E No file
+ Edits
+-.I file
++.Em file
+ unconditionally.
+ This is similar to the
+-.I e
+-command,
+-except that unwritten changes are discarded without warning.
++.Em e
++command, except that unwritten changes are discarded without warning.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI f \ file
++.It Em f No file
+ Sets the default filename to
+-.IR file .
++.Em file Ns No .
+ If
+-.I file
++.Em file
+ is not specified, then the default unescaped filename is printed.
+-
+-.TP 8
+-.RI (1,$)g /re/command-list
++.It (1,$) Ns Em g Ns No /re/command-list
+ Applies
+-.I command-list
++.Em command-list
+ to each of the addressed lines matching a regular expression
+-.IR re .
+-The current address is set to the
+-line currently matched before
+-.I command-list
++.Em re Ns No .
++The current address is set to the line currently matched before
++.Em command-list
+ is executed.
+ At the end of the
+-.I `g'
++.Em g
+ command, the current address is set to the last line affected by
+-.IR command-list .
+-
++.Em command-list Ns No .
++.Pp
+ Each command in
+-.I command-list
++.Em command-list
+ must be on a separate line,
+-and every line except for the last must be terminated by a backslash
+-(\\).
++and every line except for the last must be terminated by
++.Em \e No (backslash).
+ Any commands are allowed, except for
+-.IR `g' ,
+-.IR `G' ,
+-.IR `v' ,
++.Em g Ns No ,
++.Em G Ns No ,
++.Em v Ns No ,
+ and
+-.IR `V' .
++.Em V Ns No .
+ A newline alone in
+-.I command-list
+-is equivalent to a
+-.I `p'
++.Em command-list
++is equivalent to a
++.Em p
+ command.
+-
+-.TP 8
+-.RI (1,$)G /re/
++.It (1,$) Ns Em G Ns No /re/
+ Interactively edits the addressed lines matching a regular expression
+-.IR re.
+-For each matching line,
+-the line is printed,
+-the current address is set,
+-and the user is prompted to enter a
+-.IR command-list .
++.Em re Ns No .
++For each matching line, the line is printed, the current address is set,
++and the user is prompted to enter a
++.Em command-list Ns No .
+ At the end of the
+-.I `G'
+-command, the current address
+-is set to the last line affected by (the last)
+-.IR command-list .
+-
++.Em g
++command, the current address is set to the last line affected by (the last)
++.Em command-list Ns No .
++.Pp
+ The format of
+-.I command-list
++.Em command-list
+ is the same as that of the
+-.I `g'
+-command. A newline alone acts as a null command list.
+-A single `&' repeats the last non-null command list.
+-
+-.TP 8
+-H
++.Em g
++command.
++A newline alone acts as a null command list.
++A single
++.Em &
++repeats the last non-null command list.
++.It Em H
+ Toggles the printing of error explanations.
+ By default, explanations are not printed.
+-It is recommended that ed scripts begin with this command to
+-aid in debugging.
+-
+-.TP 8
+-h
++It is recommended that
++.Nm
++scripts begin with this command to aid in debugging.
++.It Em h
+ Prints an explanation of the last error.
+-
+-.TP 8
+-(.)i
++.It (.) Ns Em i
+ Inserts text in the buffer before the current line.
+ Text is entered in input mode.
+ The current address is set to the last line entered.
+-
+-.TP 8
+-(.,.+1)j
+-Joins the addressed lines. The addressed lines are
+-deleted from the buffer and replaced by a single
++.It (.,.+1) Ns Em j
++Joins the addressed lines.
++The addressed lines are deleted from the buffer and replaced by a single
+ line containing their joined text.
+ The current address is set to the resultant line.
+-
+-.TP 8
+-.RI (.)k lc
++.It (.) Ns Em klc
+ Marks a line with a lower case letter
+-.IR lc .
+-The line can then be addressed as
+-.I 'lc
++.Em lc Ns No \&.
++The line can then be addressed as
++.Em \&'lc
+ (i.e., a single quote followed by
+-.I lc
+-) in subsequent commands. The mark is not cleared until the line is
+-deleted or otherwise modified.
+-
+-.TP 8
+-(.,.)l
++.Em lc Ns No )
++in subsequent commands.
++The mark is not cleared until the line is deleted or otherwise modified.
++.It (.,.) Ns Em l
+ Prints the addressed lines unambiguously.
+-If a single line fills for than one screen (as might be the case
+-when viewing a binary file, for instance), a `--More--'
+-prompt is printed on the last line.
+-.B ed
+-waits until the RETURN key is pressed
+-before displaying the next screen.
+-The current address is set to the last line
+-printed.
+-
+-.TP 8
+-(.,.)m(.)
+-Moves lines in the buffer. The addressed lines are moved to after the
++If a single line fills more than one screen (as might be the case
++when viewing a binary file, for instance), a
++.Dq --More--
++prompt is printed on the last line.
++.Nm
++waits until the RETURN key is pressed before displaying the next screen.
++The current address is set to the last line printed.
++.It (.,.) Ns Em m Ns No (.)
++Moves lines in the buffer.
++The addressed lines are moved to after the
+ right-hand destination address, which may be the address
+-.IR 0
++.Em 0
+ (zero).
+-The current address is set to the
+-last line moved.
+-
+-.TP 8
+-(.,.)n
+-Prints the addressed lines along with
+-their line numbers. The current address is set to the last line
+-printed.
+-
+-.TP 8
+-(.,.)p
+-Prints the addressed lines. The current address is set to the last line
+-printed.
+-
+-.TP 8
+-P
++The current address is set to the last line moved.
++.It (.,.) Ns Em n
++Prints the addressed lines along with their line numbers.
++The current address is set to the last line printed.
++.It (.,.) Ns Em p
++Prints the addressed lines.
++The current address is set to the last line printed.
++.It Em P
+ Toggles the command prompt on and off.
+ Unless a prompt was specified by with command-line option
+-\fI-p string\fR, the command prompt is by default turned off.
+-
+-.TP 8
+-q
+-Quits ed.
+-
+-.TP 8
+-Q
+-Quits ed unconditionally.
++.Fl p Ar string Ns No ,
++the command prompt is by default turned off.
++.It Em q
++Quits
++.Nm ed .
++.It Em Q
++Quits
++.Nm
++unconditionally.
+ This is similar to the
+-.I q
+-command,
+-except that unwritten changes are discarded without warning.
+-
+-.TP 8
+-.RI ($)r \ file
++.Em q
++command, except that unwritten changes are discarded without warning.
++.It ($) Ns Em r No file
+ Reads
+-.I file
+-to after the addressed line. If
+-.I file
+-is not specified, then the default
+-filename is used. If there was no default filename prior to the command,
++.Em file
++to after the addressed line.
++If
++.Em file
++is not specified, then the default filename is used.
++If there was no default filename prior to the command,
+ then the default filename is set to
+-.IR file .
++.Em file Ns No .
+ Otherwise, the default filename is unchanged.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI ($)r \ !command
+-Reads
+-to after the addressed line
+-the standard output of
+-.IR `!command' ,
++.It ($) Ns Em r No !command
++Reads to after the addressed line the standard output of
++.Em !command Ns No ,
+ (see the
+-.RI ! command
+-below).
++.Em !
++command below).
+ The default filename is unchanged.
+ The current address is set to the last line read.
+-
+-.HP
+-.RI (.,.)s /re/replacement/
+-.PD 0
+-.HP
+-.RI (.,.)s /re/replacement/\fRg\fR
+-.HP
+-.RI (.,.)s /re/replacement/n
+-.br
+-Replaces text in the addressed lines
+-matching a regular expression
+-.I re
++.Sm off
++.It Xo (.,.) Em s No /re/replacement/ , \ (.,.)
++.Em s No /re/replacement/ Em g , No \ (.,.)
++.Em s No /re/replacement/ Em n
++.Xc
++.Sm on
++Replaces text in the addressed lines matching a regular expression
++.Em re
+ with
+-.IR replacement .
++.Em replacement Ns No .
+ By default, only the first match in each line is replaced.
+ If the
+-.I `g'
++.Em g
+ (global) suffix is given, then every match to be replaced.
+ The
+-.I `n'
++.Em n
+ suffix, where
+-.I n
+-is a postive number, causes only the
+-.IR n th
++.Em n
++is a positive number, causes only the
++.Em n Ns No th
+ match to be replaced.
+ It is an error if no substitutions are performed on any of the addressed
+ lines.
+ The current address is set the last line affected.
+-
+-.I re
++.Pp
++.Em re
+ and
+-.I replacement
++.Em replacement
+ may be delimited by any character other than space and newline
+ (see the
+-.I `s'
++.Em s
+ command below).
+ If one or two of the last delimiters is omitted, then the last line
+ affected is printed as though the print suffix
+-.I `p'
++.Em p
+ were specified.
+-
+-
+-An unescaped `&' in
+-.I replacement
++.Pp
++An unescaped
++.Ql \e
++in
++.Em replacement
+ is replaced by the currently matched text.
+ The character sequence
+-\fI`\em'\fR,
++.Em \em Ns No ,
+ where
+-.I m
++.Em m
+ is a number in the range [1,9], is replaced by the
+-.IR m th
++.Em m Ns No th
+ backreference expression of the matched text.
+ If
+-.I replacement
+-consists of a single `%', then
+-.I replacement
++.Em replacement
++consists of a single
++.Ql % ,
++then
++.Em replacement
+ from the last substitution is used.
+ Newlines may be embedded in
+-.I replacement
+-if they are escaped with a backslash (\\).
+-
+-.TP 8
+-(.,.)s
++.Em replacement
++if they are escaped with a backslash
++.Pq Ql \e .
++.It (.,.) Ns Em s
+ Repeats the last substitution.
+ This form of the
+-.I `s'
++.Em s
+ command accepts a count suffix
+-.IR `n' ,
++.Em n Ns No ,
+ or any combination of the characters
+-.IR `r' ,
+-.IR `g' ,
++.Em r Ns No ,
++.Em g Ns No ,
+ and
+-.IR `p' .
++.Em p Ns No .
+ If a count suffix
+-.I `n'
++.Em n
+ is given, then only the
+-.IR n th
++.Em n Ns No th
+ match is replaced.
+ The
+-.I `r'
++.Em r
+ suffix causes
+ the regular expression of the last search to be used instead of the
+ that of the last substitution.
+ The
+-.I `g'
++.Em g
+ suffix toggles the global suffix of the last substitution.
+ The
+-.I `p'
++.Em p
+ suffix toggles the print suffix of the last substitution
+ The current address is set to the last line affected.
+-
+-.TP 8
+-(.,.)t(.)
++.It (.,.) Ns Em t Ns No (.)
+ Copies (i.e., transfers) the addressed lines to after the right-hand
+ destination address, which may be the address
+-.IR 0
++.Em 0
+ (zero).
+-The current address is set to the last line
+-copied.
+-
+-.TP 8
+-u
++The current address is set to the last line copied.
++.It Em u
+ Undoes the last command and restores the current address
+ to what it was before the command.
+ The global commands
+-.IR `g' ,
+-.IR `G' ,
+-.IR `v' ,
++.Em g Ns No ,
++.Em G Ns No ,
++.Em v Ns No ,
+ and
+-.IR `V' .
++.Em V Ns No .
+ are treated as a single command by undo.
+-.I `u'
++.Em u
+ is its own inverse.
+-
+-.TP 8
+-.RI (1,$)v /pat/command-list
++.It (1,$) Ns Em v Ns No /re/command-list
+ Applies
+-.I command-list
++.Em command-list
+ to each of the addressed lines not matching a regular expression
+-.IR re .
++.Em re Ns No .
+ This is similar to the
+-.I `g'
++.Em g
+ command.
+-
+-.TP 8
+-.RI (1,$)V /re/
++.It (1,$) Ns Em V Ns No /re/
+ Interactively edits the addressed lines not matching a regular expression
+-.IR re.
++.Em re Ns No .
+ This is similar to the
+-.I `G'
++.Em G
+ command.
+-
+-.TP 8
+-.RI (1,$)w \ file
++.It (1,$) Ns Em w No file
+ Writes the addressed lines to
+-.IR file .
++.Em file Ns No .
+ Any previous contents of
+-.I file
++.Em file
+ is lost without warning.
+ If there is no default filename, then the default filename is set to
+-.IR file,
+-otherwise it is unchanged. If no filename is specified, then the default
+-filename is used.
++.Em file Ns No ,
++otherwise it is unchanged.
++If no filename is specified, then the default filename is used.
+ The current address is unchanged.
+-
+-.TP 8
+-.RI (1,$)wq \ file
++.It (1,$) Ns Em wq No file
+ Writes the addressed lines to
+-.IR file ,
++.Em file Ns No ,
+ and then executes a
+-.I `q'
++.Em q
+ command.
+-
+-.TP 8
+-.RI (1,$)w \ !command
++.It (1,$) Ns Em w No !command
+ Writes the addressed lines to the standard input of
+-.IR `!command' ,
++.Em !command Ns No ,
+ (see the
+-.RI ! command
+-below).
++.Em !
++command below).
+ The default filename and current address are unchanged.
+-
+-.TP 8
+-.RI (1,$)W \ file
++.It (1,$) Ns Em W No file
+ Appends the addressed lines to the end of
+-.IR file .
++.Em file Ns No .
+ This is similar to the
+-.I `w'
++.Em w
+ command, expect that the previous contents of file i
^ permalink raw reply [flat|nested] 3+ messages in thread