~sircmpwn/ctools: Implement comm

5 files changed, 350 insertions(+), 1 deletions(-)

M STATUS
M meson.build
A src/comm.c
A test/comm
M test/meson.build

M STATUS => STATUS +1 -1

@@ 25,7 25,7 @@ T       cal
   D     chown
   D     cksum
   D     cmp
-T       comm
+  D     comm
       N command
     W   compress
 T       cp

M meson.build => meson.build +1 -0

@@ 18,6 18,7 @@ oneshots = [
 	'chown',
 	'cksum',
 	'cmp',
+	'comm',
 	'false',
 	'logname',
 	'nice',

A src/comm.c => src/comm.c +188 -0

@@ 0,0 1,188 @@
+#include <assert.h>
+#include <getopt.h>
+#include <stdbool.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <sys/stat.h>
+
+enum display_mode {
+	LINES_LEFT = 1 << 0,
+	LINES_RIGHT = 1 << 1,
+	LINES_BOTH = 1 << 2,
+};
+
+static void
+usage(void)
+{
+	fprintf(stderr, "usage: comm [-123] file1 file2\n");
+}
+
+static int
+clamp(int n, int min, int max)
+{
+	return n < min ? min : n > max ? max : n;
+}
+
+static const char *
+leadfor(int diff, unsigned int display_mode)
+{
+	const unsigned int leftright = LINES_LEFT | LINES_RIGHT;
+	switch (diff) {
+	case 0:
+		if ((display_mode & leftright) == leftright) {
+			return "\t\t";
+		} else if ((display_mode & leftright)) {
+			return "\t";
+		}
+		return "";
+	case 1:
+		return (display_mode & LINES_LEFT) ?  "\t" : "";
+	case -1:
+		/* fallthrough */
+	default:
+		assert(false);
+	}
+}
+
+int
+main(int argc, char *argv[])
+{
+	unsigned int display_mode =
+		LINES_LEFT | LINES_RIGHT | LINES_BOTH;
+
+	char c;
+	while ((c = getopt(argc, argv, "123")) != -1) {
+		switch (c) {
+		case '1':
+			display_mode &= ~LINES_LEFT;
+			break;
+		case '2':
+			display_mode &= ~LINES_RIGHT;
+			break;
+		case '3':
+			display_mode &= ~LINES_BOTH;
+			break;
+		}
+	}
+
+	if (argc - optind != 2) {
+		usage();
+		return 1;
+	}
+
+	FILE *f[2] = { 0 };
+	char *names[2];
+	names[0] = argv[optind], names[1] = argv[optind + 1];
+	if (names[0][0] == '-' && names[0][1] == '\0') {
+		f[0] = stdin;
+	}
+	if (names[1][0] == '-' && names[1][1] == '\0') {
+		f[1] = stdin;
+	}
+
+	if (f[0] == stdin && f[1] == stdin) {
+		fprintf(stderr, "comm: behavior is unspecified when "
+				"both inputs are stdin\n");
+		return 1;
+	}
+
+	if (strcmp(names[0], names[1]) == 0) {
+		struct stat st;
+		if (stat(names[0], &st) != 0) {
+			perror(names[0]);
+		}
+		if (S_ISBLK(st.st_mode) || S_ISCHR(st.st_mode)
+				|| S_ISFIFO(st.st_mode)) {
+			fprintf(stderr, "comm: behavior is unspecified when "
+					"both inputs are the same special "
+					"file\n");
+			return 1;
+		}
+	}
+
+	if (!f[0] && (f[0] = fopen(names[0], "r")) == NULL) {
+		perror(names[0]);
+		return 1;
+	}
+	if (!f[1] && (f[1] = fopen(names[1], "r")) == NULL) {
+		perror(names[1]);
+		return 1;
+	}
+
+	int ret = 0, diff = 0;
+	char *lp[2] = { 0 }, *_lp[2] = { 0 };
+	size_t sz[2] = { 0 };
+
+	while (true) {
+		for (int i = 0; i < 2; ++i) {
+			if ((diff != 0 && i == (diff < 0)) || feof(f[i])) {
+				/* Processing unique lines */
+				continue;
+			}
+			if (lp[i]) {
+				free(_lp[i]);
+				_lp[i] = strdup(lp[i]);
+			}
+			if (getline(&lp[i], &sz[i], f[i]) > 0) {
+				if (_lp[i] && strcmp(_lp[i], lp[i]) > 0) {
+					fprintf(stderr, "Error: the behavior "
+							"of this utility is "
+							"unspecified for "
+							"unsorted files.\n");
+					ret = 1;
+					goto cleanup;
+				}
+				continue;
+			}
+			/* Error occured */
+			if (ferror(f[i])) {
+				ret = 1;
+				perror(names[i]);
+				goto cleanup;
+			} else if (!feof(f[!i])) {
+				lp[i] = NULL;
+				continue;
+			} else if (feof(f[i])) {
+				goto cleanup;
+			}
+		}
+
+		if (lp[0] == NULL) {
+			diff = 1;
+		} else if (lp[1] == NULL) {
+			diff = -1;
+		} else {
+			diff = strcmp(lp[0], lp[1]);
+			diff = clamp(diff, -1, 1);
+		}
+
+		switch (diff) {
+		case -1:
+			if ((display_mode & LINES_LEFT)) {
+				printf("%s", lp[0]);
+			}
+			break;
+		case 0:
+			if ((display_mode & LINES_BOTH)) {
+				printf("%s%s", leadfor(
+					diff, display_mode), lp[0]);
+			}
+			break;
+		case 1:
+			if ((display_mode & LINES_RIGHT)) {
+				printf("%s%s", leadfor(
+					diff, display_mode), lp[1]);
+			}
+			break;
+		}
+	}
+
+cleanup:
+	for (int i = 0; i < 2; ++i) {
+		fclose(f[i]);
+		free(lp[i]);
+		free(_lp[i]);
+	}
+	return ret;
+}

A test/comm => test/comm +159 -0

@@ 0,0 1,159 @@
+#!/bin/sh
+. "$HARNESS" comm
+
+compare_with_ref() (
+	ref=$(cat "$TMPDIR"/"$1")
+	ref_sum=$(printf '%s' "$ref" | cksum)
+	shift
+	(
+		cd "$TMPDIR"
+		actual=$(comm "$@")
+		actual_sum=$(printf '%s' "$actual" | cksum)
+		if [ "$ref_sum" != "$actual_sum" ]
+		then
+			printf 'Args: %s\nExpected:\n%s\nActual:\n%s\n' \
+				"$*" "$ref" "$actual"
+			return 1
+		fi
+	)
+)
+
+cat >"$TMPDIR"/case_a <<EOF
+1 same line
+2 same line
+3 different line A
+4 different line A
+5 same line
+EOF
+
+cat >"$TMPDIR"/case_b <<EOF
+1 same line
+2 same line
+3 different line B
+4 different line B
+5 same line
+EOF
+
+cat >"$TMPDIR"/ref_ab_noflags <<EOF
+		1 same line
+		2 same line
+3 different line A
+	3 different line B
+4 different line A
+	4 different line B
+		5 same line
+EOF
+
+should_handle_samelength() (
+	compare_with_ref ref_ab_noflags case_a case_b
+)
+
+cat >"$TMPDIR"/ref_ab_1 <<EOF
+	1 same line
+	2 same line
+3 different line B
+4 different line B
+	5 same line
+EOF
+
+cat >"$TMPDIR"/ref_ab_2 <<EOF
+	1 same line
+	2 same line
+3 different line A
+4 different line A
+	5 same line
+EOF
+
+cat >"$TMPDIR"/ref_ab_3 <<EOF
+3 different line A
+	3 different line B
+4 different line A
+	4 different line B
+EOF
+
+cat >"$TMPDIR"/ref_ab_12 <<EOF
+1 same line
+2 same line
+5 same line
+EOF
+
+cat >"$TMPDIR"/ref_ab_13 <<EOF
+3 different line B
+4 different line B
+EOF
+
+cat >"$TMPDIR"/ref_ab_23 <<EOF
+3 different line A
+4 different line A
+EOF
+
+should_handle_flags() (
+	compare_with_ref ref_ab_1 -1 -- case_a case_b || return 1
+	compare_with_ref ref_ab_2 -2 case_a case_b || return 1
+	compare_with_ref ref_ab_3 -3 -- case_a case_b || return 1
+	compare_with_ref ref_ab_12 -12 case_a case_b || return 1
+	compare_with_ref ref_ab_13 -13 -- case_a case_b || return 1
+	compare_with_ref ref_ab_23 -23 case_a case_b || return 1
+)
+
+cat >"$TMPDIR"/case_c <<EOF
+1 same line
+2 same line
+3 different line A
+4 different line A
+5 same line
+6 extra line 1
+7 extra line 2
+EOF
+
+cat >"$TMPDIR"/case_d <<EOF
+1 same line
+2 same line
+3 different line B
+4 different line B
+5 same line
+EOF
+
+cat >"$TMPDIR"/ref_cd <<EOF
+		1 same line
+		2 same line
+3 different line A
+	3 different line B
+4 different line A
+	4 different line B
+		5 same line
+6 extra line 1
+7 extra line 2
+EOF
+
+cat >"$TMPDIR"/ref_cd_reverse <<EOF
+		1 same line
+		2 same line
+	3 different line A
+3 different line B
+	4 different line A
+4 different line B
+		5 same line
+	6 extra line 1
+	7 extra line 2
+EOF
+
+should_handle_difflen() (
+	compare_with_ref ref_cd case_c case_d || return 1
+	compare_with_ref ref_cd_reverse case_d case_c || return 1
+)
+
+should_handle_stdin() (
+	ref=$(cksum <"$TMPDIR"/ref_ab_noflags)
+	actual=$(comm "$TMPDIR"/case_a - <"$TMPDIR"/case_b | cksum)
+	[ "$ref" = "$actual" ] || return 1
+	ref=$(cksum <"$TMPDIR"/ref_ab_noflags)
+	actual=$(comm - "$TMPDIR"/case_b <"$TMPDIR"/case_a | cksum)
+	[ "$ref" = "$actual" ] || return 1
+)
+
+runtests \
+	should_handle_samelength \
+	should_handle_flags \
+	should_handle_difflen \
+	should_handle_stdin

M test/meson.build => test/meson.build +1 -0

@@ 8,6 8,7 @@ test_files = [
 	'chmod',
 	'cksum',
 	'cmp',
+	'comm',
 	'false',
 	'link',
 	'logname',