libtabular-c

commit af0a6916407b3f1549416e6bb4b7a9bec5ea2978
parent 46972dd2694b2b7ba4395861787c6a63d6dd4d53
Author: Louis Burda <quent.burda@gmail.com>
Date:   Mon, 20 Mar 2023 00:40:39 +0100

Fix wordaware column width calculation

Diffstat:
M
src/tabular.c
 | 
34
++++++++++++++++------------------
M
src/test.c
 | 
8
++++----

2 files changed, 20 insertions(+), 22 deletions(-)
diff --git a/src/tabular.c b/src/tabular.c
@@ -67,7 +67,7 @@ const char *
 calc_word_aware_line(const char *str, size_t maxwidth,
 	size_t *out_offset, size_t *out_width)
 {
-	const char *sep;
+	const char *sep, *nstr;
 	size_t width, nwidth;
 	size_t offset, wlen;
 
@@ -77,19 +77,22 @@ calc_word_aware_line(const char *str, size_t maxwidth,
 		return NULL;
 	}
 
-	offset = strspn(str, " ");
+	offset = strspn(str, " \v\t\r\n");
 	str += offset;
 
+	nstr = str;
 	nwidth = width = 0;
 	while (nwidth <= maxwidth) {
+		width = nwidth;
+		str = nstr;
 		if (!str) break;
 		sep = strchr(str, ' ');
 		if (!sep) {
 			wlen = u8strlen(str);
-			str = NULL;
+			nstr = NULL;
 		} else {
 			wlen = u8strnlen(str, (size_t) (sep - str));
-			str = sep + 1;
+			nstr = sep + 1;
 		}
 		nwidth = width + (width > 0) + wlen;
 	}
@@ -97,10 +100,12 @@ calc_word_aware_line(const char *str, size_t maxwidth,
 	if (out_offset) *out_offset = offset;
 	if (out_width) {
 		/* single word > maxwidth */
-		if (!width && nwidth)
+		if (!width && nwidth) {
 			*out_width = maxwidth;
-		else
-			*out_offset = offset;
+			str = nstr; /* skip single word */
+		} else {
+			*out_width = width;
+		}
 	}
 
 	return str;
@@ -116,12 +121,6 @@ calc_word_aware(struct tabular_entry *entry, size_t maxwidth,
 
 	if (out_offset) *out_offset = 0;
 
-	if (!entry->str) {
-		if (out_width) *out_width = 0;
-		if (out_lines) *out_lines = 1;
-		return;
-	}
-
 	lines = 0;
 	mwidth = 0;
 	str = entry->str;
@@ -147,13 +146,11 @@ recalc_col_word_aware(const struct tabular_cfg *cfg,
 	max_wwidth = cfg->columns[col].minwidth;
 
 	rowcnt = 0;
-	for (row = rows; row; row = row->next) {
+	for (row = rows; row && rowcnt < limit; row = row->next) {
 		calc_word_aware(&row->entries[col],
 			maxwidth, &off, &wwidth, NULL);
 		max_wwidth = MAX(max_wwidth, wwidth);
-
 		rowcnt += 1;
-		if (rowcnt == limit) break;
 	}
 
 	return max_wwidth;
@@ -519,7 +516,7 @@ output_row(FILE *file, const struct tabular_cfg *cfg,
 
 			/* check if anything other than whitespace left */
 			entry += u8rawlen(entry, outlen);
-			if (strspn(entry, " \t\v\n") != u8strlen(entry))
+			if (strspn(entry, " \t\v\r\n") != u8strlen(entry))
 				done = false;
 		}
 		fprintf(file, "\n");
@@ -577,7 +574,8 @@ tabular_format(FILE *file, const struct tabular_cfg *cfg,
 				>= cfg->columns[i].minwidth)
 			return 1;
 		if (cfg->columns[i].strategy != TABULAR_TRUNC
-				&& cfg->columns[i].strategy != TABULAR_SQUASH)
+				&& cfg->columns[i].strategy != TABULAR_SQUASH
+				&& cfg->columns[i].strategy != TABULAR_SQUASH_WORDAWARE)
 			return 1;
 	}
 
diff --git a/src/test.c b/src/test.c
@@ -44,13 +44,13 @@ static struct tabular_col columns[] = {
 		.is_hidden = NULL,
 
 		.minwidth = 5,
-		.maxwidth = 20,
+		.maxwidth = 80,
 		.lpad = 0,
 		.rpad = 0,
 
 		.align = TABULAR_ALIGN_LEFT,
 
-		.strategy = TABULAR_SQUASH,
+		.strategy = TABULAR_SQUASH_WORDAWARE,
 		.essential = true,
 	},
 };
@@ -140,11 +140,11 @@ main(int argc, const char **_argv)
 	for (i = 0; i < argc; i++) {
 		end = tabular_alloc_row(&cfg, &rc, end,
 			(struct tabular_user) { .id = (size_t) i });
-		if (!end) errx(1, "tabular_append_row");
+		if (!end) errx(1, "tabular_append_row %i", rc);
 	}
 
 	rc = tabular_format(stdout, &cfg, &stats, rows);
-	if (rc) errx(1, "tabular_format");
+	if (rc) errx(1, "tabular_format %i", rc);
 
 	printf("\n%lu lines, %lu rows\n",
 		stats.lines_used, stats.rows_displayed);