From: Martin Mares <mj@ucw.cz>
Date: Tue, 24 Jul 2012 14:13:20 +0000 (+0200)
Subject: Added --equalize
X-Git-Tag: v1.0~25
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=e0be90cb432c569d2aac3bba0b9e9de460e39bf7;p=xsv.git

Added --equalize
---

diff --git a/xsv.c b/xsv.c
index 352771a..b0e9124 100644
--- a/xsv.c
+++ b/xsv.c
@@ -128,7 +128,7 @@ struct format {
 };
 
 static struct format *in_format, *out_format;
-static int want_trim;
+static int want_trim, want_equalize, want_stats;
 
 struct field {
 	int start_pos;
@@ -242,6 +242,9 @@ static intarray_t column_widths;
 
 static void update_stats(void)
 {
+	if (!want_stats)
+		return;
+
 	for (int i = 0; i < fields_count(&out_fields); i++) {
 		struct field *f = fields_nth(&out_fields, i);
 		intarray_t *w = &column_widths;
@@ -434,7 +437,7 @@ static void table_write(struct format *fmt)
 			unsigned char *p = get_field(&out_fields, i, &len);
 			fw = field_chars(fields_nth(&out_fields, i));
 			if (fw > cw) {
-				warn(fmt, "Internal error: Wrongly calculated column width (%d > %d)", fw, cw);
+				warn(fmt, "Internal error: Wrongly calculated width of column %d (%d > %d)", i, fw, cw);
 				cw = fw;
 			}
 			while (len--)
@@ -544,6 +547,14 @@ static void trim_fields(void)
 	}
 }
 
+static void equalize_fields(void)
+{
+	while (fields_count(&out_fields) < intarray_count(&column_widths)) {
+		struct field *f = fields_push(&out_fields);
+		f->start_pos = f->len = 0;
+	}
+}
+
 /*** Field names and headers ***/
 
 struct field_names {
@@ -626,7 +637,11 @@ static void write_header(void)
 	// This is tricky: when we are formatting a table, field names are normally
 	// calculated in pass 1, but the header is written in pass 2, so we have to
 	// update column statistics, because field name can be too wide to fit.
+	want_stats++;
 	update_stats();
+	want_stats--;
+	if (want_equalize)
+		equalize_fields();
 	write_grid(-1);
 	write_line();
 	write_grid(0);
@@ -754,9 +769,9 @@ static void one_pass(int pass)
 		else
 			select_all_fields();
 
-		if (out_format->needs_stats)
-			update_stats();
-
+		if (want_equalize && (pass & 2))
+			equalize_fields();
+		update_stats();
 		write_line();
 	}
 
@@ -777,7 +792,6 @@ static void two_pass(void)
 	out_format->read_line = tmp_read;
 	out_format->write_line = tmp_write;
 	out_format->tmp_file = tmpfile();
-	out_format->needs_stats = final_format->needs_stats;
 	out_format->field_names = in_format->field_names;
 	one_pass(1);
 
@@ -786,7 +800,7 @@ static void two_pass(void)
 	rewind(in_format->tmp_file);
 	line_number = 0;
 	out_format = final_format;
-	out_format->needs_stats = 0;
+	want_stats = 0;
 	one_pass(2);
 	fclose(in_format->tmp_file);
 }
@@ -817,6 +831,7 @@ Format parameters:\n\
 \n\
 Other options:\n\
     --trim		Trim leading and trailing whitespaces in fields\n\
+    --equalize		Pad all lines to the maximum number of fields\n\
 ");
 	exit(0);
 }
@@ -844,11 +859,13 @@ enum long_options {
 	OPT_TABLE,
 	OPT_TABLE_SEP,
 	OPT_GRID,
+	OPT_EQUALIZE,
 };
 
 static const struct option long_options[] = {
 	{ "always-quote",	0,	NULL,	OPT_ALWAYS_QUOTE },
 	{ "csv",		0,	NULL,	'c' },
+	{ "equalize",		0,	NULL,	OPT_EQUALIZE },
 	{ "fields",		1,	NULL,	'f' },
 	{ "fs",			1,	NULL,	'd' },
 	{ "grid",		0,	NULL,	OPT_GRID },
@@ -978,6 +995,9 @@ int main(int argc, char **argv)
 			case OPT_GRID:
 				current_format()->table_grid = 1;
 				break;
+			case OPT_EQUALIZE:
+				want_equalize = 1;
+				break;
 			default:
 				bad_args(NULL);
 		}
@@ -998,7 +1018,8 @@ int main(int argc, char **argv)
 	}
 	finish_parse_selectors();
 
-	if (out_format->needs_stats)
+	want_stats = out_format->needs_stats | want_equalize;
+	if (want_stats)
 		two_pass();
 	else
 		one_pass(3);