Mercurial > hg > toybox

comparison toys/posix/cut.c @ 706:35a9f9c5b53f

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Commit 698 adding cut should ahve included the actual cut.c file. (Oops.)
author Rob Landley <rob@landley.net>
date Tue, 20 Nov 2012 01:00:17 -0600
parents
children 6cc69be43c42
comparison
equal deleted inserted replaced
705:3e81cd0bad4b 706:35a9f9c5b53f
1 /* cut.c - Cut from a file.
2 *
3 * Copyright 2012 Ranjan Kumar <ranjankumar.bth@gmail.com>, Kyungwan Han <asura321@gamil.com>
4 *
5 * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/cut.html
6 *
7 USE_CUT(NEWTOY(cut, "b:c:f:d:sn", TOYFLAG_BIN))
8
9 config CUT
10 bool "cut"
11 default y
12 help
13 usage: cut OPTION... [FILE]...
14 Print selected parts of lines from each FILE to standard output.
15 -b LIST select only these bytes from LIST.
16 -c LIST select only these characters from LIST.
17 -f LIST select only these fields.
18 -d DELIM use DELIM instead of TAB for field delimiter.
19 -s do not print lines not containing delimiters.
20 -n don't split multibyte characters (Ignored).
21 */
22 #define FOR_cut
23 #include "toys.h"
24
25 typedef struct _slist {
26 int start_position;
27 int end_position;
28 struct _slist *next;
29 }SLIST;
30
31 GLOBALS(
32 char *delim;
33 char *flist;
34 char *clist;
35 char *blist;
36 struct _slist *slist_head;
37 unsigned nelem;
38 )
39
40 #define BEGINOFLINE 0
41 #define ENDOFLINE INT_MAX
42 #define NORANGE -1
43
44 static int get_user_options(void);
45 void (*do_cut)(int);
46 static void do_fcut(int fd);
47 static void do_bccut(int fd);
48 static void free_list(void);
49
50 /*
51 * add items in the slist.
52 */
53 static void add_to_list(int start, int end)
54 {
55 SLIST *current, *head_ref, *temp1_node;
56
57 head_ref = TT.slist_head;
58 temp1_node = (SLIST *)xzalloc(sizeof(SLIST));
59 temp1_node->start_position = start;
60 temp1_node->end_position = end;
61 temp1_node->next = NULL;
62
63 /* Special case for the head end */
64 if (head_ref == NULL || (head_ref)->start_position >= start) {
65 temp1_node->next = head_ref;
66 head_ref = temp1_node;
67 }
68 else {
69 /* Locate the node before the point of insertion */
70 current = head_ref;
71 while (current->next!=NULL && current->next->start_position < temp1_node->start_position)
72 current = current->next;
73 temp1_node->next = current->next;
74 current->next = temp1_node;
75 }
76 TT.slist_head = head_ref;
77 return;
78 }
79
80 /*
81 * parse list and add to slist.
82 */
83 static void parse_list(char *list)
84 {
85 char *ctoken, *dtoken;
86 int start = 0, end = 0;
87 while((ctoken = strsep(&list, ",")) != NULL) {
88 if(!ctoken[0]) continue;
89
90 //Get start position.
91 dtoken = strsep(&ctoken, "-");
92 if(!dtoken[0]) start = BEGINOFLINE;
93 else {
94 start = get_int_value(dtoken, 0, INT_MAX);
95 start = (start?(start-1):start);
96 }
97 //Get end position.
98 if(ctoken == NULL) end = NORANGE; //case e.g. 1,2,3
99 else if(!ctoken[0]) end = ENDOFLINE; //case e.g. N-
100 else {//case e.g. N-M
101 end = get_int_value(ctoken, 0, INT_MAX);
102 end = (end?end:ENDOFLINE);
103 end--;
104 if(end == start) end = NORANGE;
105 }
106 add_to_list(start, end);
107 TT.nelem++;
108 }
109 //if list is missing in command line.
110 if(TT.nelem == 0) error_exit("missing positions list:");
111 return;
112 }
113
114 /*
115 * retrive data from the file/s.
116 */
117 static void get_data(void)
118 {
119 char **argv = toys.optargs; //file name.
120 toys.exitval = EXIT_SUCCESS;
121
122 if(!*argv) do_cut(0); //for stdin
123 else {
124 for(; *argv; ++argv) {
125 if(strcmp(*argv, "-") == 0) do_cut(0); //for stdin
126 else {
127 int fd = open(*argv, O_RDONLY, 0);
128 if(fd < 0) {//if file not present then continue with other files.
129 perror_msg(*argv);
130 toys.exitval = EXIT_FAILURE;
131 continue;
132 }
133 do_cut(fd);
134 xclose(fd);
135 }
136 }
137 }
138 return;
139 }
140
141 /*
142 * cut main function.
143 */
144 void cut_main(void)
145 {
146 char delimiter = '\t'; //default delimiter.
147 int num_of_options = 0;
148 char *list;
149
150 TT.nelem = 0;
151 TT.slist_head = NULL;
152 //verify the number of options provided by user.
153 num_of_options = get_user_options();
154 if(num_of_options == 0) error_exit("specify a list of bytes, characters, or fields");
155 else if(num_of_options > 1) error_exit("only one type of list may be specified");
156
157 //Get list and assign the function.
158 if(toys.optflags & FLAG_f) {
159 list = TT.flist;
160 do_cut = do_fcut;
161 }
162 else if(toys.optflags & FLAG_c) {
163 list = TT.clist;
164 do_cut = do_bccut;
165 }
166 else {
167 list = TT.blist;
168 do_cut = do_bccut;
169 }
170
171 if(toys.optflags & FLAG_d) {
172 //delimiter must be 1 char.
173 if(TT.delim[0] && TT.delim[1]) perror_exit("the delimiter must be a single character");
174 delimiter = TT.delim[0];
175 }
176
177 if(!(toys.optflags & FLAG_d) && (toys.optflags & FLAG_f)) {
178 TT.delim = (char *)xzalloc(2);
179 TT.delim[0] = delimiter;
180 }
181
182 //when field is not specified, cutting has some special handling.
183 if(!(toys.optflags & FLAG_f)) {
184 if(toys.optflags & FLAG_s) perror_exit("suppressing non-delimited lines operating on fields");
185 if(delimiter != '\t') perror_exit("an input delimiter may be specified only when operating on fields");
186 }
187
188 parse_list(list);
189 get_data();
190 if(!(toys.optflags & FLAG_d) && (toys.optflags & FLAG_f)) {
191 free(TT.delim);
192 TT.delim = NULL;
193 }
194 free_list();
195 return;
196 }
197
198 /*
199 * perform cut operation on the given delimiter.
200 */
201 static void do_fcut(int fd)
202 {
203 char *buff;
204 char *delimiter = TT.delim;
205 while((buff = get_line(fd)) != NULL) {
206 //does line have any delimiter?.
207 if(strrchr(buff, (int)delimiter[0]) == NULL) {
208 //if not then print whole line and move to next line.
209 if(!(toys.optflags & FLAG_s)) xputs(buff);
210 continue;
211 }
212
213 unsigned cpos = 0;
214 int start, ndelimiters = -1;
215 int nprinted_fields = 0;
216 char *pfield = xzalloc(strlen(buff) + 1);
217 SLIST *temp_node = TT.slist_head;
218
219 if(temp_node != NULL) {
220 //process list on each line.
221 while(cpos < TT.nelem && buff) {
222 if(!temp_node) break;
223 start = temp_node->start_position;
224 do {
225 char *field = NULL;
226 //count number of delimeters per line.
227 while(buff) {
228 if(ndelimiters < start) {
229 ndelimiters++;
230 field = strsep(&buff, delimiter);
231 }
232 else break;
233 }
234 //print field (if not yet printed).
235 if(!pfield[ndelimiters]) {
236 if(ndelimiters == start) {
237 //put delimiter.
238 if(nprinted_fields++ > 0) xputc(delimiter[0]);
239 if(field) fputs(field, stdout);
240 //make sure this field won't print again.
241 pfield[ndelimiters] = (char) 0x23; //put some char at this position.
242 }
243 }
244 start++;
245 if((temp_node->end_position == NORANGE) || (!buff)) break;
246 }while(start <= temp_node->end_position);
247 temp_node = temp_node->next;
248 cpos++;
249 }
250 }
251 xputc('\n');
252 free(pfield);
253 pfield = NULL;
254 }//End of while loop.
255 return;
256 }
257
258 /*
259 * perform cut operation char or byte.
260 */
261 static void do_bccut(int fd)
262 {
263 char *buff;
264 while((buff = get_line(fd)) != NULL) {
265 unsigned cpos = 0;
266 int buffln = strlen(buff);
267 char *pfield = xzalloc(buffln + 1);
268 SLIST *temp_node = TT.slist_head;
269 if(temp_node != NULL) {
270 while(cpos < TT.nelem) {
271 int start;
272 if(!temp_node) break;
273 start = temp_node->start_position;
274 while(start < buffln) {
275 //to avoid duplicate field printing.
276 if(pfield[start]) {
277 if(++start <= temp_node->end_position) continue;
278 temp_node = temp_node->next;
279 break;
280 }
281 else {
282 //make sure this field won't print again.
283 pfield[start] = (char) 0x23; //put some char at this position.
284 xputc(buff[start]);
285 }
286 if(++start > temp_node->end_position) {
287 temp_node = temp_node->next;
288 break;
289 }
290 }
291 cpos++;
292 }
293 xputc('\n');
294 }
295 free(pfield);
296 pfield = NULL;
297 }
298 return;
299 }
300
301 /*
302 * used to verify "c", "b" and "f" options are present in command line.
303 * As cut command can only except any of them at a time.
304 */
305 static int get_user_options(void)
306 {
307 int i = 3, flag = 0;
308 for(; i<6; i++) {//verify 3rd, 4th and 5th bit is set.
309 int mask = 1;
310 mask = mask << i;
311 if(toys.optflags & mask) flag++;
312 }
313 return flag;
314 }
315
316 /*
317 * free the slist.
318 */
319 static void free_list(void)
320 {
321 SLIST *temp;
322 while(TT.slist_head != NULL) {
323 temp = TT.slist_head->next;
324 free(TT.slist_head);
325 TT.slist_head = temp;
326 }
327 return;
328 }