void get_args(void)
{
int i = 0;
int arg_ptr = 0;
int flag;
num_args = 0;
do {
if (num_args == MAX_ARGS)
cuterror(MAX_ARGS_EXCEEDED_ERROR);
if (!isdigit(line[i]) && line[i] != '-')
cuterror(SYNTAX_ERROR);
args[arg_ptr] = 1;
args[arg_ptr + 1] = BUFSIZ;
flag = 1;
while (line[i] != ',' && line[i] != 0) {
if (isdigit(line[i])) {
args[arg_ptr] = 0;
while (isdigit(line[i]))
args[arg_ptr] = 10 * args[arg_ptr] + line[i++] - '0';
if (!args[arg_ptr])
cuterror(POSITION_ERROR);
arg_ptr++;
}
if (line[i] == '-') {
arg_ptr |= 1;
i++;
flag = 0;
}
}
if (flag && arg_ptr & 1)
args[arg_ptr] = args[arg_ptr - 1];
if (args[num_args * 2] > args[num_args * 2 + 1])
cuterror(RANGE_ERROR);
num_args++;
arg_ptr = num_args * 2;
}
while (line[i++]);
}
int main(int argc, char *argv[])
{
int i = 1;
int numberFilenames = 0;
name = argv[0];
if (argc == 1)
cuterror(USAGE);
while (i < argc) {
if (argv[i][0] == '-') {
switch (argv[i++][1]) {
case 'd':
if (mode == OPTIONC || mode == OPTIONB)
warn(DELIMITER_NOT_APPLICABLE, "d");
delim = argv[i++][0];
break;
case 'f':
sprintf(line, "%s", argv[i++]);
if (mode == OPTIONC || mode == OPTIONB)
warn(OVERRIDING_PREVIOUS_MODE, "f");
mode = OPTIONF;
break;
case 'b':
sprintf(line, "%s", argv[i++]);
if (mode == OPTIONF || mode == OPTIONC)
warn(OVERRIDING_PREVIOUS_MODE, "b");
mode = OPTIONB;
break;
case 'c':
sprintf(line, "%s", argv[i++]);
if (mode == OPTIONF || mode == OPTIONB)
warn(OVERRIDING_PREVIOUS_MODE, "c");
mode = OPTIONC;
break;
case 'i':
flag_i = SET;
break;
case 's':
flag_s = SET;
break;
case '\0': /* - means: read from stdin */
numberFilenames++;
break;
case 'n': /* needed for Posix, but no effect here */
if (mode != OPTIONB)
warn(OPTION_NOT_APPLICABLE, "n");
break;
default:
warn(UNKNOWN_OPTION, &(argv[i - 1][1]));
}
} else {
i++;
numberFilenames++;
}
}
/* Here follow the checks, if the selected options are reasonable. */
if (mode == OPTIONB) /* since in Minix char == byte */
mode = OPTIONC;
/* Flag -s is only allowed with -f, otherwise warn and reset flag_s */
if (flag_s == SET && (mode == OPTIONB || mode == OPTIONC)) {
warn(OPTION_NOT_APPLICABLE, "s");
flag_s = NOTSET;
}
/* Flag -i is only allowed with -f, otherwise warn and reset flag_i */
if (flag_i == SET && mode == OPTIONF) {
warn(OPTION_NOT_APPLICABLE, "s");
flag_i = NOTSET;
}
get_args();
if (numberFilenames != 0) {
i = 1;
while (i < argc) {
if (argv[i][0] == '-') {
switch (argv[i][1]) {
case 'f':
case 'c':
case 'b':
case 'd':
i += 2;
break;
case 'n':
case 'i':
case 's':
i++;
break;
case '\0':
fd = stdin;
i++;
cut();
break;
default:
i++;
}
} else {
if ((fd = fopen(argv[i++], "r")) == NULL) {
warn(FILE_NOT_READABLE, argv[i - 1]);
} else {
cut();
fclose(fd);
}
}
}
} else {
fd = stdin;
cut();
}
return (exit_status);
}
void cut(void)
{
int i, j, length, maxcol;
char *columns[MAX_FIELD];
while (fgets(line, BUFSIZ, fd)) {
length = strlen(line) - 1;
*(line + length) = 0;
switch (mode) {
case DUMP_STDIN:
printf("%s", line);
break;
case OPTIONF:
maxcol = 0;
columns[maxcol++] = line;
for (i = 0; i < length; i++) {
if (*(line + i) == delim) {
*(line + i) = 0;
if (maxcol == MAX_FIELD)
cuterror(MAX_FIELDS_EXCEEDED_ERROR);
columns[maxcol] = line + i + 1;
while (*(line + i + 1) == delim && flag_i) {
columns[maxcol]++;
i++;
}
maxcol++;
}
}
if (maxcol == 1) {
if (flag_s != SET)
printf("%s", line);
} else {
for (i = 0; i < num_args; i++) {
for (j = args[i * 2]; j <= args[i * 2 + 1]; j++)
if (j <= maxcol) {
printf("%s", columns[j - 1]);
if (i != num_args - 1 || j != args[i * 2 + 1])
putchar(delim);
}
}
}
break;
case OPTIONC:
for (i = 0; i < num_args; i++) {
for (j = args[i * 2]; j <= (args[i * 2 + 1] > length ? length :
args[i * 2 + 1]); j++)
putchar(*(line + j - 1));
}
}
if (maxcol == 1 && flag_s == SET) {
;
} else {
putchar('\n');
}
}
}
void fcut_fields(FILE *fp, int PIDNO){
int i, j, k,length, maxcol=0;
char buff[256];
char *columns[MAX_FIELD];
char input[NBUFSIZ];
int lenfln;
char *tempfiles[MAX_ARGS];
char *tempflno[MAX_ARGS];
/*
char *dir[]={"/tmp/", "./"};
int dirno=1;
*/
char *fileparts[]={"t.",".",".tab"};
char *md[] ={"w","a"};
FILE *fw;
int counter=0;
int mode=0;
char pid[6];
int jj;
int frstobs=0;
int lastobs=0;
int ln =0;
sprintf(pid, "%d",PIDNO);
/*
printf("PIDNO= %d\n", PIDNO);
printf("process ID= %s\n", pid);
printf("num_args= %d\n", num_args);
printf("fileparts[0]= %s\n", fileparts[0]);
dirno=0;
*/
/* set up temp files*/
for (k=0; k < num_args; k++) {
tempflno[k]=malloc(4+1);
if (tempflno[k] == NULL) {
printf("Error: malloc failed in file no\n");
exit(EXIT_FAILURE);
}
sprintf(tempflno[k],"%d", (k+1));
/* lenfln =strlen(dir[dirno]) +2 + strlen(pid)+1 + strlen(tempflno[k])+4 + 1; */
/* lenfln =strlen(dir[dirno]) +2 + strlen(PIDNOS)+1 + strlen(tempflno[k])+4 + 1;*/
lenfln = strlen(PIDNOS)+ strlen(fileparts[1]) + strlen(tempflno[k])+ strlen(fileparts[2]) + 1;
/* printf("filename length=%d\n",lenfln);*/
tempfiles[k]= malloc(lenfln);
if (tempfiles[k] == NULL) {
printf("Error: malloc failed in file name\n");
exit(EXIT_FAILURE);
}
/*
strcpy(tempfiles[k], dir[dirno]);
strcat(tempfiles[k], fileparts[0]);
strcat(tempfiles[k], pid);
*/
strcat(tempfiles[k], PIDNOS);
strcat(tempfiles[k], fileparts[1]);
strcat(tempfiles[k], tempflno[k]);
strcat(tempfiles[k], fileparts[2]);
free(tempflno[k]);
}
jj=0;
while (fgets(input, NBUFSIZ, fp)) {
jj++;
if (DEBUG){
printf("\n\n++++++++++++++ %d-th iteration ++++++++++++++\n\n",jj);
}
if ( (num_sets) && (jj == 1) ) {
frstobs=rsse[ln];
lastobs=rsse[ln+1];
}
if (num_sets){
printf("1st=%d\tlast=%d\n", frstobs, lastobs);
if ( (jj < frstobs) || (jj >lastobs)) {
continue;
}
if (jj == lastobs){
printf("within:jj=%d\t1st=%d\tlast=%d\n", jj, frstobs, lastobs);
if ( ((ln+2)/2) <= num_sets){
ln=ln+2;
frstobs=rsse[ln];
lastobs=rsse[ln+1];
}
}
}
maxcol=0;
length = strlen(input) - 1;
*(input + length) = 0;
columns[maxcol++] = input ;
for (i = 0; i < length; i++) {
if (*(input + i) == DELIM) {
*(input+ i) = 0;
if (maxcol == MAX_FIELD)
cuterror(MAX_FIELDS_EXEEDED_ERROR);
columns[maxcol] = input + i + 1;
maxcol++;
}
}
if (maxcol != 1) {
if (counter > 0) {mode = 1;}
for (i = 0; i < num_args; i++) {
buff[0]='\0';
/* printf("%d-th tempfile name=%s\n",i,tempfiles[i]);*/
/* field set boundary */
if ( (fw = fopen(tempfiles[i], md[mode])) == NULL){
fprintf(stderr, "Error opening file.\n");
exit(1);
}
for (j = args[i * 2]; j <= args[i * 2 + 1]; j++){
if (j <= maxcol) {
fprintf(fw,"%s", columns[j-1]);
/* fprintf(fw, "%s", columns[j - 1]);*/
if (i != num_args - 1 || j != args[i * 2 + 1]) {fputc(ODELIM, fw);}
}
}
/* print newline character */
fputc('\n', fw);
fclose(fw);
/* field set boundary */
}
counter++;
} else if (OPTS==1) {
if ( (fw = fopen("temp.1554.z.tab", "w")) == NULL){
fprintf(stderr, "Error opening file.\n");
exit(1);
}
fprintf(fw, "%s",input);
fclose(fw);
}
columns[1]='\0';/* */
}
for (k=0;k<num_args;k++){
free(tempfiles[k]);
}
}
void cut_fields(FILE *fd)
{
int i, j, length, maxcol=0;
char *columns[MAX_FIELD];
char input[NBUFSIZ];
char buff[256];
int frstobs=0;
int lastobs=0;
int ln =0;
int k,kk,jj;
if(DEBUG){printf("/////////////// cut_fields ///////////////\n");}
jj=0;
while (fgets(input, NBUFSIZ, fd)) {
jj++;
if (DEBUG){
printf("\n++++++++++++++ %d-th iteration ++++++++++++++\n",jj);
}
if ( (num_sets) && (jj == 1) ) {
frstobs=rsse[ln];
lastobs=rsse[ln+1];
}
if (num_sets){
if (DEBUG){
printf("1st=%d\tlast=%d\n", frstobs, lastobs);
}
if ( (jj < frstobs) || (jj >lastobs)) {
continue;
}
if (jj == lastobs){
if (DEBUG){
printf("within:jj=%d\t1st=%d\tlast=%d\n", jj, frstobs, lastobs);
}
if ( ((ln+2)/2) <= num_sets){
ln=ln+2;
frstobs=rsse[ln];
lastobs=rsse[ln+1];
}
}
}
k=0;
kk=0;
maxcol=0;
length = strlen(input) - 1;
/* length = a line of data without a new line character*/
if (DEBUG){
printf("length=%d\n",length);
}
/*
for(k=0;columns[k]!='\0';k++){
printf("check k=%d\t",k);
if (jj == 0){
for (kk=0;columns[k][kk]!='\0';kk++){*/
/* printf("%d\n",strlen(columns[k]));
}
printf("- %d\n",kk);}
}
if(length >= MAX_FIELD){
cuterror(MAX_FIELDS_EXEEDED_ERROR);
}
*/
*(input + length) = '\0';/* input[length] = */
columns[maxcol++] = input ;/*if maxcol=0, columns[0] = input then maxcol=1 */
/*
printf("maxcol=%d\n",maxcol);
kk++;
printf("kk=%d\n",kk);
*/
/*
if (DEBUG){
printf("1st iteration\n");
for(k=0;columns[k]!='\0';k++){
printf("k=%d\t",k);
for (kk=0;columns[k][kk]!='\0';kk++){
printf("[%c]",columns[k][kk]);
}
printf("- %d\n",kk);
}
}
*/
for (i = 0; i < length; i++) {
if (*(input + i) == DELIM) {
*(input+ i) = '\0';
if (maxcol == MAX_FIELD){
cuterror(MAX_FIELDS_EXEEDED_ERROR);
}
columns[maxcol] = input + i + 1;
maxcol++;
}
}
/* after the above block maxcol = no of vars in a line of data*/
/*
if (DEBUG){
printf("2nd iteration\n");
printf("maxcol=%d\n\n",maxcol);
for(k=0;columns[k]!='\0';k++){
printf("k=%d\t",k);
for (kk=0;columns[k][kk]!='\0';kk++){
printf("[%c]",columns[k][kk]);
}
printf("- %d\n",kk);
}
}
*/
if (maxcol != 1) {
for (i = 0; i < num_args; i++) {
buff[0]='\0';
for (j = args[i * 2]; j <= args[i * 2 + 1]; j++){
if (j <= maxcol) {
/* print a data point */
printf("%s", columns[j - 1]);
if (i != num_args - 1 || j != args[i * 2 + 1]){
/* print a delimiter */
putchar(ODELIM);
}
}
}
}
} else if (OPTS==1) {
printf("%s",input);
}
if (maxcol != 1){
/* print a new line character */
putchar('\n');
}
/*
if (DEBUG){
printf("3rd iteration\n");
for(k=0;columns[k]!='\0';k++){
printf("k=%d\t%s\n", k, columns[k]);
printf("k=%d\t",k);
for (kk=0;columns[k][kk]!='\0';kk++){
printf("[%c]",columns[k][kk]);
}
printf("- %d\n",kk);
}
}
*/
columns[1]='\0';/* */
}
}
static void parse_list_g(char *lists){
/*int EOL,BOL;*/
char *ltok = NULL;
char *ntok = NULL;
char *junk;
int s = 0, e = 0;
int l;
/* take apart the lists, one by one (they are seperated with commas */
num_sets=0;
#if DBG
if(DEBUG){ printf("entering parse_list_g\n");}
#endif
while ((ltok = strsep(&lists, ",")) != NULL) {
#if DBG
if(DEBUG){printf("token split by comma(ltok)=%s\n", ltok);}
#endif
/* it's actually legal to pass an empty list */
if (strlen(ltok) == 0){
continue;
}
/* get the start pos */
ntok = strsep(<ok, "-");
#if DBG
if(DEBUG){printf("start_string(ntok)=%s\n", ntok);}
#endif
if (ntok == NULL) {
#if DBG
if(DEBUG){printf("\tstart_string(ntok) is NULL\n");}
#endif
cuterror(SYNTAX_ERROR);
} else {
s = strtoul(ntok, &junk, 10);
#if DBG
if(DEBUG){printf("\tstart_string: after ulong =%d\n",s);}
#endif
if(e < 0){
cuterror(SYNTAX_ERROR);
} else if (*junk != '\0' ) {
cuterror(SYNTAX_ERROR);
}
}
/* get the end pos */
ntok = strsep(<ok, "-");
#if DBG
if(DEBUG){
if (ntok != NULL){
printf("\tend_string=%s\n", ntok);
}
}
#endif
if (ntok == NULL) {
#if DBG
if(DEBUG){printf("\tend_string is NULL: start_no = end_no\n");}
#endif
e = s;
} else {
e = strtoul(ntok, &junk, 10);
#if DBG
if (DEBUG){
printf("\tend_string: after ulong=%d\n",e);
printf("check the leftover string=%s\n", junk );
}
#endif
if(e < 0){
cuterror(SYNTAX_ERROR);
} else if (*junk != '\0' ) {
cuterror(SYNTAX_ERROR);
}
}
#if DBG
if (DEBUG){
printf("entering the leftover processing\n");
}
#endif
/* if there's something left to tokenize, the user past an invalid list */
if (ltok){
cuterror(SYNTAX_ERROR);
}
#if DBG
if (DEBUG){
printf("start_string=%d\tend_string=%d\n",s,e);
}
#endif
/* add the new list */
rsse[num_sets*2]=s;
rsse[num_sets*2+1]=e;
#if DBG
if (DEBUG){
printf("no of argument set(num_sets)=%d\n", num_sets);
printf("contents of argment_set array\n");
for (l=0;rsse[l] != 0 ;l++){
printf("%d:%d\n",l, rsse[l]);
}
}
#endif
num_sets++;
if ((num_sets*2+1)>= MAX_ARGS){
cuterror(MAX_ARGS_EXEEDED_ERROR);
}
}
/* make sure we got some cut positions out of all that */
/* printf("entering last of parse_lists\n");*/
if (num_sets <= 0){
cuterror(SYNTAX_ERROR);
} else {
if(DEBUG){
printf("no of case-subsetting num_sets=%d\n", num_sets);
for (l=0;rsse[l] != 0 ;l++){
printf("%d:%d\n",l, rsse[l]);
}
printf("\n");
}
}
}
static void parse_lists(char *lists){
int EOL,BOL;
char *ltok = NULL;
char *ntok = NULL;
char *junk;
int s = 0, e = 0,recid;
int l;
int fmt;
/* take apart the lists, one by one (they are seperated with commas */
num_args=0;
#if DBG
if(DEBUG){printf("entering parse_lists\n");}
#endif
while ((ltok = strsep(&lists, ",")) != NULL) {
#if DBG
if(DEBUG){printf("token split by comma(ltok)=%s\n", ltok);}
#endif
fmt =0;
/* it's actually legal to pass an empty list */
if (strlen(ltok) == 0){
continue;
}
/* get the record pos */
ntok = strsep(<ok, ":");
#if DBG
if(DEBUG){printf("\trecord number(ntok)=%s\n", ntok);}
#endif
if (ltok) {
recid= atoi(ntok);
#if DBG
if(DEBUG){
printf("\tcheck the remaining part\n");
printf("\trecord number(ntok)=%s\n", ntok);
printf("\trecord number after atoi=%d\n", recid);
}
#endif
if ((recid>RECNUM) || (recid<1)){
cuterror(SYNTAX_ERROR);
}
BOL = (recid-1)*(RECLEN+1);
EOL = recid*(RECLEN+1)-1;
#if DBG
if(DEBUG){printf("BOL=%d\tEOL=%d\n",BOL,EOL);}
#endif
} else {
recid=0;
BOL = 0;
EOL = INT_MAX;
ltok=ntok;
}
/* get the start pos */
ntok = strsep(<ok, "-");
if (ntok == NULL) {
cuterror(SYNTAX_ERROR);
} else if (strlen(ntok) == 0) {
s = BOL;
#if DBG
if(DEBUG){
printf("\tstart_string number(ntok) is empty\n");
printf("\tstart_string number(s)=BOL(%d)\n",s);
}
#endif
} else {
s = strtoul(ntok, &junk, 10) + BOL;
#if DBG
if(DEBUG){
printf("\tset start_string number(s)=%d\n",s);
}
#endif
if(e < 0){
cuterror(SYNTAX_ERROR);
} else if (strlen(junk)) {
if (*junk == '@'){
junk++;
if (*junk == 'c') {
fmt = -1;
} else {
/* */
fmt = atoi(junk);
}
#if DBG
if(DEBUG){
printf("\thit the at mark\n");
printf("\tset start_string number(s)=%d\n",s);
printf("\tfrmt string=%s\n", junk);
printf("\tformat code=%d\n", fmt);
}
#endif
}
} else if (*junk != '\0' ) {
cuterror(SYNTAX_ERROR);
}
}
/* get the end pos */
ntok = strsep(<ok, "-");
if (ntok == NULL) {
e = s;
#if DBG
if(DEBUG){
printf("\tend string(e) is NULL\n");
printf("\tset end string(e)=%d[start_string(s)]\n", e);
}
#endif
} else if (strlen(ntok) == 0) {
e = EOL;
#if DBG
if(DEBUG){
printf("\tend string(e):length is 0\n");
printf("\tset end string(e)=%d[EOL]\n", e);
}
#endif
} else {
e = strtoul(ntok, &junk, 10) + BOL;
#if DBG
if(DEBUG){
printf("\tend_string after ul=%d\n",e);
printf("\tcheck remaining string=%s\n",junk);
}
#endif
if(e < 0){
cuterror(SYNTAX_ERROR);
} else if (e == 0) {
/* if the user specified and end position of 0, that means "til the
* end of the line */
e = EOL;
#if DBG
if(DEBUG){
printf("\tend_string is 0\n");
printf("\tset end_string(e)=%d[EOL]\n", e);
}
#endif
} else if (strlen(junk)) {
if (*junk == '@'){
junk++;
if (*junk == 'c') {
fmt = -1;
} else {
fmt = atoi(junk);
}
#if DBG
if(DEBUG){
printf("\thit the at mark\n");
printf("\tset start_string number(s)=%d\n",s);
printf("\tfrmt string=%s\n", junk);
printf("\tformat code=%d\n", fmt);
}
#endif
}
} else if (*junk != '\0' ) {
cuterror(SYNTAX_ERROR);
}
}
/* if there's something left to tokenize, the user past an invalid list */
if (ltok){
cuterror(SYNTAX_ERROR);
}
/*printf("s=%d\te=%d\n",s,e);*/
/* add the new list */
args[num_args*2]=s;
args[num_args*2+1]=e;
argsn[num_args]=fmt;
#if DBG
if (DEBUG){
printf("no of arguments set(num_args)=%d\n", num_args);
printf("contents of args array\n");
for (l=0;args[l] != 0 ;l++){
printf("%d:%d\n",l, args[l]);
}
printf("\n");
printf("contents of accompanying format set\n");
for (l=0;l<=(num_args) ;l++){
printf("%d:%d\n",l, argsn[l]);
}
printf("\n");
}
#endif
num_args++;
if ((num_args*2+1)>= MAX_ARGS){
cuterror(MAX_ARGS_EXEEDED_ERROR);
}
}
if(DEBUG){
printf("no of column no pairs(num_args)=%d\n", num_args);
printf("\ncolumn-subsetting array\n");
for (l=0;args[l] != 0 ;l++){
printf("%d:%d\n",l, args[l]);
}
printf("\ncontents of frmt array\n");
for (l=0;l<=(num_args) ;l++){
printf("%d:%d\n",l, argsn[l]);
}
printf("\n");
}
/* make sure we got some cut positions out of all that */
/* printf("entering last of parse_lists\n");*/
if (num_args == 0){
cuterror(SYNTAX_ERROR);
}
}
