1/***********************************************************************
2*                                                                      *
3*               This software is part of the ast package               *
4*          Copyright (c) 1992-2010 AT&T Intellectual Property          *
5*                      and is licensed under the                       *
6*                  Common Public License, Version 1.0                  *
7*                    by AT&T Intellectual Property                     *
8*                                                                      *
9*                A copy of the License is available at                 *
10*            http://www.opensource.org/licenses/cpl1.0.txt             *
11*         (with md5 checksum 059e8cd6165cb4c31e351f2b69388fd9)         *
12*                                                                      *
13*              Information and Software Systems Research               *
14*                            AT&T Research                             *
15*                           Florham Park NJ                            *
16*                                                                      *
17*                 Glenn Fowler <gsf@research.att.com>                  *
18*                  David Korn <dgk@research.att.com>                   *
19*                                                                      *
20***********************************************************************/
21#pragma prototyped
22/*
23 * David Korn
24 * AT&T Bell Laboratories
25 *
26 * fold
27 */
28
29static const char usage[] =
30"[-?\n@(#)$Id: fold (AT&T Research) 2004-11-18 $\n]"
31USAGE_LICENSE
32"[+NAME?fold - fold lines]"
33"[+DESCRIPTION?\bfold\b is a filter that folds lines from its input, "
34	"breaking the lines to have a maximum of \awidth\a column "
35	"positions (or bytes if the \b-b\b option is specified).  Lines "
36	"are broken by the insertion of a newline character such that "
37	"each output line is the maximum width possible that does not "
38	"exceed the specified number of column positions, (or bytes).  A line "
39	"will not be broken in the middle of a character.] "
40"[+?Unless the \b-b\b option is specified, the following will be treated "
41	"specially:]{"
42	"[+carriage-return?The current count of line width will be set "
43		"to zero.  \bfold\b will not insert a newline immediately "
44		"before or after a carriage-return.]"
45	"[+backspace?If positive, the current count of line width will be "
46		"decremented by  one.  \bfold\b will not insert a newline "
47		"immediately before or after a backspace.]"
48	"[+tab?Each tab character encountered will advance the column "
49		"position to the next tab stop.  Tab stops are at each "
50		"column position \an\a, where \an\a modulo 8 equals 1.]"
51	"}"
52"[+?If no \afile\a is given, or if the \afile\a is \b-\b, \bfold\b "
53        "reads from standard input.   The start of the file is defined "
54        "as the current offset.]"
55
56"[b:bytes?Count bytes rather than columns so that each carriage-return, "
57	"backspace, and tab counts as 1.]"
58"[c:continue?Emit \atext\a at line splits.]:[text:='\\n']"
59"[d:delimiter?Break at \adelim\a boundaries.]:[delim]"
60"[s:spaces?Break at word boundaries.  If the line contains any blanks, "
61	"(spaces or tabs), within the first \awidth\a column positions or "
62	"bytes, the line is broken after the last blank meeting the "
63	"\awidth\a constraint.]"
64"[w:width]#[width:=80?Use a maximum line length of \awidth\a columns "
65	"instead of the default.]"
66"\n"
67"\n[file ...]\n"
68"\n"
69"[+EXIT STATUS?]{"
70	"[+0?All files processed successfully.]"
71	"[+>0?An error occurred.]"
72"}"
73"[+SEE ALSO?\bpaste\b(1)]"
74;
75
76
77#include <cmd.h>
78
79#define WIDTH	80
80#define TABSIZE	8
81
82#define T_EOF	1
83#define T_NL	2
84#define T_BS	3
85#define T_TAB	4
86#define T_SP	5
87#define T_RET	6
88
89static void fold(Sfio_t *in, Sfio_t *out, register int width, const char *cont, size_t contsize, char *cols)
90{
91	register char *cp, *first;
92	register int n, col=0, x=0;
93	register char *last_space=0;
94	cols[0] = 0;
95	for (;;)
96	{
97		if (!(cp  = sfgetr(in,'\n',0)))
98		{
99			if (!(cp = sfgetr(in,'\n',-1)) || (n = sfvalue(in)) <= 0)
100				break;
101			x = cp[--n];
102			cp[n] = '\n';
103		}
104		/* special case -b since no column adjustment is needed */
105		if(cols['\b']==0 && (n=sfvalue(in))<=width)
106		{
107			sfwrite(out,cp,n);
108			continue;
109		}
110		first = cp;
111		col = 0;
112		last_space = 0;
113		for(;;)
114		{
115			while((n=cols[*(unsigned char*)cp++])==0);
116			while((cp-first) > (width-col))
117			{
118				if(last_space)
119					col = last_space - first;
120				else
121					col = width-col;
122				sfwrite(out,first,col);
123				first += col;
124				col = 0;
125				last_space = 0;
126				if(cp>first+1 || (n!=T_NL && n!=T_BS))
127					sfwrite(out, cont, contsize);
128			}
129			switch(n)
130			{
131			    case T_NL:
132				if(x)
133					*(cp-1) = x;
134				break;
135			    case T_RET:
136				col = 0;
137				continue;
138			    case T_BS:
139				if((cp+(--col)-first)>0)
140					col--;
141				continue;
142			    case T_TAB:
143				n = (TABSIZE-1) - (cp+col-1-first)&(TABSIZE-1);
144				col +=n;
145				if((cp-first) > (width-col))
146				{
147					sfwrite(out,first,(--cp)-first);
148					sfwrite(out, cont, contsize);
149					first = cp;
150					col =  TABSIZE-1;
151					last_space = 0;
152					continue;
153				}
154				if(cols[' '])
155					last_space = cp;
156				continue;
157			    case T_SP:
158				last_space = cp;
159				continue;
160			    default:
161				continue;
162			}
163			break;
164		}
165		sfwrite(out,first,cp-first);
166	}
167}
168
169int
170b_fold(int argc, char *argv[], void* context)
171{
172	register int n, width=WIDTH;
173	register Sfio_t *fp;
174	register char *cp;
175	char *cont="\n";
176	size_t contsize = 1;
177	char cols[1<<CHAR_BIT];
178
179	cmdinit(argc, argv, context, ERROR_CATALOG, 0);
180	memset(cols, 0, sizeof(cols));
181	cols['\t'] = T_TAB;
182	cols['\b'] = T_BS;
183	cols['\n'] = T_NL;
184	cols['\r'] = T_RET;
185	for (;;)
186	{
187		switch (optget(argv, usage))
188		{
189		case 0:
190			break;
191		case 'b':
192			cols['\r'] = cols['\b'] = 0;
193			cols['\t'] = cols[' '];
194			continue;
195		case 'c':
196			contsize = stresc(cont = strdup(opt_info.arg));
197			continue;
198		case 'd':
199			if (n = *opt_info.arg)
200				cols[n] = T_SP;
201			continue;
202		case 's':
203			cols[' '] = T_SP;
204			if(cols['\t']==0)
205				cols['\t'] = T_SP;
206			continue;
207		case 'w':
208			if ((width = opt_info.num) <= 0)
209				error(2, "%d: width must be positive", opt_info.num);
210			continue;
211		case ':':
212			error(2, "%s", opt_info.arg);
213			continue;
214		case '?':
215			error(ERROR_usage(2), "%s", opt_info.arg);
216			continue;
217		}
218		break;
219	}
220	argv += opt_info.index;
221	argc -= opt_info.index;
222	if(error_info.errors)
223		error(ERROR_usage(2),"%s", optusage(NiL));
224	if(cp = *argv)
225		argv++;
226	do
227	{
228		if(!cp || streq(cp,"-"))
229			fp = sfstdin;
230		else if(!(fp = sfopen(NiL,cp,"r")))
231		{
232			error(ERROR_system(0),"%s: cannot open",cp);
233			error_info.errors = 1;
234			continue;
235		}
236		fold(fp,sfstdout,width,cont,contsize,cols);
237		if(fp!=sfstdin)
238			sfclose(fp);
239	}
240	while(cp= *argv++);
241	return(error_info.errors);
242}
243