freq

commit 1807bcc4ce93dfe03368f5a02ac8aaa81869f1ec
parent 7c83e66679339067eb36124cd613203d5c0e11f2
Author: phoebos <ben@bvnf.space>
Date:   Mon, 21 Mar 2022 19:58:43 +0000

allow other separator chars

Diffstat:
M
freq.1
 | 
14
++++++++++++--
M
freq.c
 | 
27
+++++++++++++++++++++++----

2 files changed, 35 insertions(+), 6 deletions(-)
diff --git a/freq.1 b/freq.1
@@ -6,6 +6,7 @@
 .Nd count frequencies of keys in text
 .Sh SYNOPSIS
 .Nm
+.Op Fl d Ar delim
 .Op Fl k Ar key
 .Op Ar file
 .Sh DESCRIPTION
@@ -26,14 +27,23 @@ character, then the value of the field.
 .Pp
 The following options are supported:
 .Bl -tag -width Ds
+.It Fl d Ar delim
+Set the field delimiter to the character
+.Ar delim .
+The default is any character of class
+.Vt space
+in the current locale (see
+.Xr isspace 3 ) .
 .It Fl k Ar key
 Count the frequencies of the field found at position
 .Ar key ,
-where fields are separated by any number of space characters in the current locale (see
-.Xr isspace 3 ) .
+where fields are separated by any number of
+.Ar delim
+characters.
 The default value of
 .Ar key
 is 1, meaning that the first fields on each line are counted.
+To use the whole line as a key, use the special value of 0.
 If
 .Ar key
 fields are not found on a line, a warning is printed and the whole line is ignored.
diff --git a/freq.c b/freq.c
@@ -4,6 +4,10 @@
 #include <string.h>
 #include <unistd.h>
 
+enum {
+	SEP_DEFAULT = -1,
+};
+
 void freq(FILE *f, int field);
 
 struct count {
@@ -13,6 +17,7 @@ struct count {
 
 struct count *cs;
 size_t cs_len = 0;
+char sep;
 
 size_t
 get_index(char *name) {
@@ -57,15 +62,27 @@ print_freqs(void) {
 }
 
 int
+is_sep(int c) {
+	if (sep == SEP_DEFAULT)
+		return isspace(c);
+	else
+		return (c == sep);
+}
+
+int
 main (int argc, char **argv) {
 	int c, field = 1;
-	while ((c = getopt(argc, argv, ":k:")) != -1) {
+	sep = SEP_DEFAULT;
+	while ((c = getopt(argc, argv, ":d:k:")) != -1) {
 		switch (c) {
+			case 'd':
+				sep = optarg[0];
+				break;
 			case 'k':
 				field = atoi(optarg);
 				break;
 			case '?':
-				fprintf(stderr, "usage: %s [-k fieldnum] [file]\n", argv[0]);
+				fprintf(stderr, "usage: %s [-d delim] [-k fieldnum] [file]\n", argv[0]);
 				return 1;
 		}
 	}
@@ -107,13 +124,15 @@ freq(FILE *f, int field) {
 		linenum++;
 		i = 0;
 		field_start = line;
+		if (field == 0)
+			goto field_found;
 		in_word = 0;
 		cur_field = 1;
 		/* eat up whitespace at front */
-		while (isspace(line[i]))
+		while (is_sep(line[i]))
 			i++;
 		for (; i < n; i++) {
-			if (isspace(line[i])) {
+			if (is_sep(line[i])) {
 				if (in_word) {
 					if (cur_field == field) {
 						line[i] = '\0';