table

чување 87a587014e2ac983134f36fde443c965723d74da
родитељ 66eea0abf2b8886d89ce550a87d083bd000a8188
Аутор: Страхиња Радић <contact@strahinja.org>
Датум:   Wed, 10 May 2023 21:27:56 +0200

Add -q

Signed-off-by: Страхиња Радић <contact@strahinja.org>

Diffstat:
M
TODO
 | 
2
+-
M
defs.h
 | 
62
+++++++++++++++++++++++++++++++-------------------------------
M
table.1.in
 | 
24
+++++++++++++++++++++++-
M
table.c
 | 
210
++++++++++++++++++++++++++++++++++++++++++++++++-------------------------------

измењених датотека: 4, додавања: 184(+), брисања: 114(-)
diff --git a/TODO b/TODO
@@ -3,7 +3,7 @@ TODO
 
 [ ] Add column alignment
 
-< > Add switch to turn off processing of quotes; have tsvtable use it
+<x> Add switch to turn off processing of quotes; have tsvtable use it
     automatically
 
 
diff --git a/defs.h b/defs.h
@@ -4,49 +4,49 @@
 
 #include "utf8.h"
 
-#define BUFSIZE       4096
+#define BUFSIZE	      4096
 #define SMALL_BUFSIZE 256
 
-//#define ANSI_SGR_RESET "\e[0m\e[?25h"
+// #define ANSI_SGR_RESET "\e[0m\e[?25h"
 #define ANSI_SGR_BOLD_ON  "\e[1m"
 #define ANSI_SGR_BOLD_OFF "\e[0m"
 
 typedef unsigned char UBYTE;
-typedef unsigned int  UINT;
+typedef unsigned int UINT;
 typedef unsigned long ULONG;
 
-typedef enum
-{
-    CMD_NONE,
-    CMD_COLUMNS,
-    CMD_DELIMITER,
-    CMD_FORMAT,
-    CMD_SYMBOLS,
-    CMD_VERSION
+typedef enum {
+	CMD_NONE,
+	CMD_COLUMNS,
+	CMD_DELIMITER,
+	CMD_FORMAT,
+	CMD_QUOTES,
+	CMD_SYMBOLS,
+	CMD_VERSION
 } Command;
 
-enum
-{
-    TABLE_SYMBOLS_ASCII,
-    TABLE_SYMBOLS_SINGLE,
-    TABLE_SYMBOLS_DOUBLE,
-    TABLE_SYMBOLS_EMPTY
+enum { Q_PROCESS, Q_IGNORE };
+
+enum {
+	TABLE_SYMBOLS_ASCII,
+	TABLE_SYMBOLS_SINGLE,
+	TABLE_SYMBOLS_DOUBLE,
+	TABLE_SYMBOLS_EMPTY
 };
 
-enum
-{
-    TABLE_INNER_ASCII_ASCII,
-    TABLE_INNER_SINGLE_SINGLE,
-    TABLE_INNER_SINGLE_DOUBLE,
-    TABLE_INNER_DOUBLE_SINGLE,
-    TABLE_INNER_DOUBLE_DOUBLE,
-    TABLE_INNER_EMPTY_EMPTY,
-    TABLE_INNER_ASCII_EMPTY,
-    TABLE_INNER_SINGLE_EMPTY,
-    TABLE_INNER_DOUBLE_EMPTY,
-    TABLE_INNER_EMPTY_ASCII,
-    TABLE_INNER_EMPTY_SINGLE,
-    TABLE_INNER_EMPTY_DOUBLE,
+enum {
+	TABLE_INNER_ASCII_ASCII,
+	TABLE_INNER_SINGLE_SINGLE,
+	TABLE_INNER_SINGLE_DOUBLE,
+	TABLE_INNER_DOUBLE_SINGLE,
+	TABLE_INNER_DOUBLE_DOUBLE,
+	TABLE_INNER_EMPTY_EMPTY,
+	TABLE_INNER_ASCII_EMPTY,
+	TABLE_INNER_SINGLE_EMPTY,
+	TABLE_INNER_DOUBLE_EMPTY,
+	TABLE_INNER_EMPTY_ASCII,
+	TABLE_INNER_EMPTY_SINGLE,
+	TABLE_INNER_EMPTY_DOUBLE,
 };
 
 static const u32* table_symbols[][9] =
diff --git a/table.1.in b/table.1.in
@@ -34,7 +34,7 @@ is a synonym for
 .br
 .B tsvtable
 is a synonym for
-.BR "table -d \(dq$(printf '\et')\(dq" .
+.BR "table -d \(dq$(printf '\et')\(dq -q ignore" .
 .
 .SH SYNOPSIS
 .
@@ -54,6 +54,7 @@ is a synonym for
 .OP "\-m \fR|\fP \-\-msdos"
 .OP "\-n \fR|\fP \-\-no\-ansi"
 .OP "\-p \fR|\fP \-\-pause"
+.OP "\-q \fR|\fP \-\-quotes=" behavior
 .OP "\-s \fR|\fP \-\-symbols=" set
 .OP "\-t \fR|\fP \-\-expand-tabs"
 .YS
@@ -191,6 +192,27 @@ from file managers which don't pause after executing programs or as a single
 command from terminal emulators.
 .
 .TP
+.BI \-q " behavior"
+.TQ
+.BI \-\-quotes= behavior
+.br
+.I behavior
+can be one of the following:
+.
+.RS
+.IP \fBprocess\fP
+Treats everything between the quotes as a single column.
+Treats consequtive quotes within the quoted text as literal quotes.
+This is the default.
+.IP \fBignore\fP
+Ignores the special meaning of quotes (like TSV).
+This is the default if 
+.B table
+is called as
+.BR tsvtable .
+.RE
+.
+.TP
 .BI \-s " set"
 .TQ
 .BI \-\-symbols= set
diff --git a/table.c b/table.c
@@ -81,6 +81,7 @@ int handle_ansi		     = 1;
 int msdos		     = 0;
 int pause		     = 0;
 int expand_tabs		     = 0;
+int quotes		     = Q_PROCESS;
 
 int
 version()
@@ -95,8 +96,9 @@ usage()
 	printf("Usage: %s [-b|--border-mode] [-c <cols>|--columns=<cols>] [-d"
 	       " <delim>|--delimiter=<delim>] [-f <format>|--format=<format>]"
 	       " [-h|--help] [-m|--msdos] [-n|--no-ansi]"
-	       " [-p|--pause] [-s <set>|--symbols=<set>] [-t|--expand-tabs] "
-	       "[-v|--version]\n",
+	       " [-p|--pause] [-q <behavior>|--quotes=<behavior>]"
+	       " [-s <set>|--symbols=<set>] [-t|--expand-tabs]"
+	       " [-v|--version]\n",
 		PROGRAMNAME);
 	return 0;
 }
@@ -311,6 +313,18 @@ set_format(const char* arg, ULONG** format, size_t* format_size)
 	return 0;
 }
 
+int
+set_quotes_behavior(const char* arg, int* quotes)
+{
+	if (!strcmp(arg, "process"))
+		*quotes = Q_PROCESS;
+	else if (!strcmp(arg, "ignore"))
+		*quotes = Q_IGNORE;
+	else
+		return 1;
+	return 0;
+}
+
 /* Return number of columns based on the input line */
 size_t
 number_of_columns(const u8* input, u32 delimiter)
@@ -365,7 +379,10 @@ main(int argc, char** argv)
 		progname = sep + 1;
 
 	if (!strcmp(progname, "tsvtable"))
+	{
 		delimiter = L'\t';
+		quotes	  = Q_IGNORE;
+	}
 	else if (!strcmp(progname, "tablep"))
 		pause = 1;
 
@@ -431,6 +448,15 @@ do_arg:
 				arg += strlen("pause");
 				pause = 1;
 			}
+			else if (startswith(arg, "quotes="))
+			{
+				arg += strlen("quotes=");
+				if (set_quotes_behavior(arg, &quotes))
+					return error(EINVAL,
+						"Invalid argument: "
+						"'%s'",
+						arg);
+			}
 			else if (startswith(arg, "symbols="))
 			{
 				arg += strlen("symbols=");
@@ -477,6 +503,9 @@ do_arg:
 			case 'p':
 				pause = 1;
 				break;
+			case 'q':
+				cmd = CMD_QUOTES;
+				break;
 			case 's':
 				cmd = CMD_SYMBOLS;
 				break;
@@ -516,6 +545,12 @@ do_arg:
 				return error(EINVAL, "Invalid argument: '%s'",
 					arg);
 		}
+		else if (cmd == CMD_QUOTES)
+		{
+			if (set_quotes_behavior(arg, &quotes))
+				return error(EINVAL, "Invalid argument: '%s'",
+					arg);
+		}
 		else if (cmd == CMD_SYMBOLS)
 		{
 			if (set_symbol_set(arg, &current_symbol_set,
@@ -543,15 +578,16 @@ done_arg:
 	else
 		input = stdin;
 
-	u8* line  = NULL;
-	u8* pline = NULL;
-	int quote = 0;
+	u8* line     = NULL;
+	u8* pline    = NULL;
+	int in_quote = 0;
 	u32 uch;
 	u8 u8ch[7];
 	size_t u8len	    = 0;
 	size_t ch_len	    = 0;
 	size_t output_lines = 0;
 	size_t column_start = 0;
+	u8* pch_end;
 
 	CALLOC(line, u8, BUFSIZE);
 
@@ -574,7 +610,7 @@ do_input:
 	if (!*line && lineno != 0)
 		goto empty_line;
 
-	quote		    = 0;
+	in_quote	    = 0;
 	pline		    = line;
 	colno		    = 0;
 	current_rune_column = 0;
@@ -718,97 +754,109 @@ not_first:
 
 do_line:
 	if (!pline || !*pline)
-		goto done_line;
+	{
+		if (in_quote)
+		{
+			printf("\n");
+			return error(EINVAL, "Unmatched quote");
+		}
+		else
+			goto done_line;
+	}
 
-	if (*pline == '"')
+	if (in_quote && *pline == '"' && *(pline + 1) == '"')
 	{
-		quote = !quote;
 		pline++;
 		colno++;
+		goto continue_line;
 	}
-	else
+	else if (quotes == Q_PROCESS && *pline == '"')
+	{
+		in_quote = !in_quote;
+		pline++;
+		colno++;
+		goto do_line;
+	}
+
+continue_line:
+	u8_char_to_u32(&uch, pline, &ch_len);
+	if (uch == delimiter && !in_quote
+		&& current_table_column < table_columns - 1)
 	{
-		u8* pch_end;
-		u8_char_to_u32(&uch, pline, &ch_len);
-		if (ch_len <= 0)
+		if (handle_ansi && lineno == 0)
+			printf("%s", ANSI_SGR_BOLD_OFF);
+
+		while (within_column(column_start))
 		{
-			if (within_column(column_start))
-			{
-				printf("%c", *pline++);
-				current_rune_column++;
-			}
-			colno++;
+			printf(" ");
+			current_rune_column++;
 		}
-		else if (uch == (u32)L'\t' && expand_tabs)
+		column_start++;
+		if (format)
+			column_start += *(format + current_table_column);
+		else
+			column_start += format_value;
+		if (current_table_column != table_columns - 1)
 		{
-			while (within_column(column_start))
-			{
-				printf(" ");
-				current_rune_column++;
-				if (current_rune_column % tab_length == 0)
-					break;
-			}
-			pline++;
-			colno++;
+			u8len	    = u32_char_to_u8(u8ch,
+				      *table_inner_symbols[current_inner_symbol_set]
+							  [1]);
+			u8ch[u8len] = 0;
+			printf("%s", u8ch);
+			current_rune_column++;
+			current_table_column++;
 		}
-		else if (border_mode)
+
+		if (handle_ansi && lineno == 0)
+			printf("%s", ANSI_SGR_BOLD_ON);
+
+		pline += ch_len;
+		colno += ch_len;
+	}
+	else if (ch_len <= 0)
+	{
+		if (within_column(column_start))
 		{
-			pch_end = pline + ch_len;
-			if (within_column(column_start))
-			{
-				while (pline != pch_end)
-					printf("%c", *pline++);
-				current_rune_column++;
-			}
-			pline = pch_end;
-			colno += ch_len;
+			printf("%c", *pline++);
+			current_rune_column++;
 		}
-		else if (uch == delimiter && !quote
-			&& current_table_column < table_columns - 1)
+		colno++;
+	}
+	else if (uch == (u32)L'\t' && expand_tabs)
+	{
+		while (within_column(column_start))
 		{
-			if (handle_ansi && lineno == 0)
-				printf("%s", ANSI_SGR_BOLD_OFF);
-
-			while (within_column(column_start))
-			{
-				printf(" ");
-				current_rune_column++;
-			}
-			column_start++;
-			if (format)
-				column_start
-					+= *(format + current_table_column);
-			else
-				column_start += format_value;
-			if (current_table_column != table_columns - 1)
-			{
-				u8len	    = u32_char_to_u8(u8ch,
-					      *table_inner_symbols
-						      [current_inner_symbol_set][1]);
-				u8ch[u8len] = 0;
-				printf("%s", u8ch);
-				current_rune_column++;
-				current_table_column++;
-			}
-
-			if (handle_ansi && lineno == 0)
-				printf("%s", ANSI_SGR_BOLD_ON);
-
-			pline += ch_len;
-			colno += ch_len;
+			printf(" ");
+			current_rune_column++;
+			if (current_rune_column % tab_length == 0)
+				break;
 		}
-		else
+		pline++;
+		colno++;
+	}
+	else if (border_mode)
+	{
+		pch_end = pline + ch_len;
+		if (within_column(column_start))
 		{
-			pch_end = pline + ch_len;
-			if (within_column(column_start))
-			{
-				while (pline != pch_end)
-					printf("%c", *pline++);
-				current_rune_column++;
-			}
-			pline = pch_end;
-			colno += ch_len;
+			while (pline != pch_end)
+				printf("%c", *pline++);
+			current_rune_column++;
+		}
+		pline = pch_end;
+		colno += ch_len;
+	}
+	else
+	{
+		pch_end = pline + ch_len;
+		if (within_column(column_start))
+		{
+			while (pline != pch_end)
+				printf("%c", *pline++);
+			current_rune_column++;
 		}
+		pline = pch_end;
+		colno += ch_len;
 	}
 	goto do_line;