Added a manual page

[xsv.git] / xsv.c

diff --git a/xsv.c b/xsv.c
index 8fd72779d23e5bd11ff80711e2453adc44e31853..545a39efe3e4a98421e201a5d719902bf89ca5e6 100644 (file)
--- a/xsv.c
+++ b/xsv.c
@@ -1,5 +1,5 @@
 /*
- *     A Swiss-Army Knife for CSV-like Files
+ *     The Swiss-Army Knife for CSV-like Files
  *
  *     (c) 2012 Martin Mares <mj@ucw.cz>
  */
@@ -24,6 +24,9 @@
 #define UNUSED
 #endif
 
+static void select_fields(void);
+static void select_all_fields(void);
+
 /*** General functions ***/
 
 static void NONRET die(char *msg, ...)
@@ -99,6 +102,7 @@ struct format {
        int fs;
        int quote;
        int quiet;
+       int sloppy;
        int (*read_line)(struct format *fmt);
        void (*write_line)(struct format *fmt);
        void (*write_grid)(struct format *fmt, int pos);        // -1=above, 1=below, 0=after header
@@ -112,9 +116,6 @@ struct format {
        // CSV backend:
        int always_quote;
 
-       // WS backend:
-       int strict_ws;
-
        // regex backend:
        pcre *pcre;
        pcre_extra *pcre_extra;
@@ -359,7 +360,7 @@ static int ws_read(struct format *fmt)
                        if (ws) {
                                if (!in_field->start_pos &&
                                    !in_field->len &&
-                                   !fmt->strict_ws)
+                                   fmt->sloppy)
                                        in_field->start_pos = i;
                                else
                                        new_field(i);
@@ -369,7 +370,7 @@ static int ws_read(struct format *fmt)
                }
        }
 
-       if (ws && fmt->strict_ws)
+       if (ws && !fmt->sloppy)
                new_field(n);
        return 1;
 }
@@ -404,17 +405,21 @@ static int regex_read(struct format *fmt)
        int i = 0;
        for (;;) {
                int ovec[3];
-               int sep = pcre_exec(fmt->pcre, fmt->pcre_extra, (char *) c, n, i, 0, ovec, 3);
-               if (sep < 0) {
-                       if (sep != PCRE_ERROR_NOMATCH)
-                               warn(fmt, "PCRE matching error %d", sep);
+               int err = pcre_exec(fmt->pcre, fmt->pcre_extra, (char *) c, n, i, 0, ovec, 3);
+               if (err < 0) {
+                       if (err != PCRE_ERROR_NOMATCH)
+                               warn(fmt, "PCRE matching error %d", err);
                        // No further occurrence of the separator: the rest is a single field
-                       new_field(i);
-                       in_field->len = n - i;
+                       if (!fmt->sloppy || i < n) {
+                               new_field(i);
+                               in_field->len = n - i;
+                       }
                        return 1;
                }
-               new_field(i);
-               in_field->len = ovec[0] - i;
+               if (!fmt->sloppy || ovec[0]) {
+                       new_field(i);
+                       in_field->len = ovec[0] - i;
+               }
                i = ovec[1];
        }
 }
@@ -609,20 +614,22 @@ static void write_header(void)
                return;
        }
 
+       int want_select_fields = 0;
        if (out_format->set_field_names) {
                struct field_names *fn = xmalloc_zero(sizeof(*fn));
                out_format->field_names = fn;
                add_field_names(fn, out_format->set_field_names);
-       } else if (in_format->field_names)
+       } else if (in_format->field_names) {
                out_format->field_names = in_format->field_names;
-       else
+               want_select_fields = 1;
+       } else
                die("Output header requested, but no field names specified");
 
        line_reset(&in_line);
-       fields_reset(&out_fields);
+       fields_reset(&in_fields);
        struct field_names *fn = out_format->field_names;
        for (int i = 0; i < stringarray_count(&fn->names); i++) {
-               struct field *f = fields_push(&out_fields);
+               struct field *f = fields_push(&in_fields);
                f->start_pos = line_count(&in_line);
                f->len = 0;
                char *s = *stringarray_nth(&fn->names, i);
@@ -632,6 +639,12 @@ static void write_header(void)
                }
        }
 
+       fields_reset(&out_fields);
+       if (want_select_fields)
+               select_fields();
+       else
+               select_all_fields();
+
        // 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.
@@ -694,7 +707,7 @@ static int parse_field(char *str)
        if (in_format->field_names && (f = find_field_by_name(in_format->field_names, str)) > 0)
                return f;
 
-       die("Unknown field %s", str);
+       die("Unknown field `%s'", str);
 }
 
 static char *parse_selector(char *str)
@@ -811,10 +824,9 @@ static void NONRET usage(void)
 Usage: xsv <in-format> [<out-format>] <options> [<fields>]\n\
 \n\
 Formats:\n\
--t, --tsv              TAB-separated values (default)\n\
+-t, --tsv              Tab-separated values (default)\n\
 -c, --csv              Comma-separated values\n\
 -w, --ws               Values separated by arbitrary whitespace\n\
--W, --strict-ws                Like --ws, but recognize empty columns at start/end\n\
 -r, --regex=<rx>       Separator given by Perl regular expression (input only)\n\
     --table            Format a table (output only)\n\
 \n\
@@ -826,6 +838,7 @@ Format parameters:\n\
     --always-quote     Put quotes around all fields (CSV output only)\n\
     --table-sep=<n>    Separate table columns by <n> spaces (default: 2)\n\
     --grid             Separate table columns by grid lines\n\
+-s, --sloppy           Ignore separators at the start/end of line (ws/regex only)\n\
 \n\
 Other options:\n\
     --trim             Trim leading and trailing whitespaces in fields\n\
@@ -870,7 +883,7 @@ static const struct option long_options[] = {
        { "header",             0,      NULL,   'h' },
        { "quiet",              0,      NULL,   'q' },
        { "regex",              1,      NULL,   'r' },
-       { "strict-ws",          0,      NULL,   'W' },
+       { "sloppy",             0,      NULL,   's' },
        { "table",              0,      NULL,   OPT_TABLE },
        { "table-sep",          1,      NULL,   OPT_TABLE_SEP },
        { "trim",               0,      NULL,   OPT_TRIM },
@@ -964,19 +977,20 @@ int main(int argc, char **argv)
                                if (err)
                                        bad_args("Error compiling regex: %s", err);
                                break;
+                       case 's':
+                               if (current_format()->id != FORM_WS && current_format()->id != FORM_REGEX)
+                                       bad_args("--sloppy makes sense only for --ws or --regex.");
+                               current_format()->sloppy = 1;
+                               break;
                        case 't':
                                set_format(FORM_TSV);
                                break;
                        case 'w':
                                set_format(FORM_WS);
                                break;
-                       case 'W':
-                               set_format(FORM_WS);
-                               current_format()->strict_ws = 1;
-                               break;
                        case OPT_ALWAYS_QUOTE:
                                if (current_format()->id != FORM_CSV)
-                                       bad_args("--always-quote makes sense only for CSV.");
+                                       bad_args("--always-quote makes sense only for --csv.");
                                current_format()->always_quote = 1;
                                break;
                        case OPT_HELP: