extended FS to support 5-character special form

qse/doc/page/awk.doc

@@ -128,6 +128,12 @@ BEGIN {
 }
 @endcode
 
+@subsection awk_ext_fs EXTENDED FS
+@code
+BEGIN { FS="?:\\\"\""; }
+@endcode
+
+
 @subsection awk_ext_binnum BINARY NUMBER
 Use 0b to begin a binary number sequence.
 

qse/lib/awk/misc.c

@@ -1,5 +1,5 @@
 /*
- * $Id: misc.c 291 2009-09-21 13:28:18Z hyunghwan.chung $
+ * $Id: misc.c 292 2009-09-23 10:19:30Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -940,10 +940,90 @@ exit_loop:
 
 qse_char_t* qse_awk_rtx_strxnfld (
 	qse_awk_rtx_t* rtx, qse_char_t* str, qse_size_t len,
-	qse_char_t fs, qse_char_t lq, qse_char_t rq, qse_char_t ec,
+	qse_char_t fs, qse_char_t ec, qse_char_t lq, qse_char_t rq,
 	qse_char_t** tok, qse_size_t* tok_len)
 {
-/* TODO: */
+	qse_char_t* p = str;
+	qse_char_t* end = str + len;
+	int escaped = 0, quoted = 0;
+	qse_char_t* ts; /* token start */
+	qse_char_t* tp; /* points to one char past the last token char */
+	qse_char_t* xp; /* points to one char past the last effective char */
+
+	/* skip leading spaces */
+	while (p < end && QSE_ISSPACE(*p)) p++;
+
+	/* initialize token pointers */
+	ts = tp = xp = p; 
+
+	while (p < end)
+	{
+		char c = *p;
+
+		if (escaped)
+		{
+			*tp++ = c; xp = tp; p++;
+			escaped = 0;
+		}
+		else
+		{
+			if (c == ec)
+			{
+				escaped = 1;
+				p++;
+			}
+			else if (quoted)
+			{
+				if (c == rq)
+				{
+					quoted = 0;
+					p++;
+				}
+				else
+				{
+					*tp++ = c; xp = tp; p++;
+				}
+			}
+			else 
+			{
+				if (c == fs)
+				{
+					*tok = ts;
+					*tok_len = xp - ts;
+					p++;
+
+					if (QSE_ISSPACE(fs))
+					{
+						while (p < end && *p == fs) p++;
+						if (p >= end) return QSE_NULL;
+					}
+
+					return p;
+				}
+		
+				if (c == lq)
+				{
+					quoted = 1;
+					p++;
+				}
+				else
+				{
+					*tp++ = c; p++;
+					if (!QSE_ISSPACE(c)) xp = tp; 
+				}
+			}
+		}
+	}
+
+	if (escaped) 
+	{
+		/* if it is still escaped, the last character must be 
+		 * the escaper itself. treat it as a normal character */
+		*xp++ = ec;
+	}
+	
+	*tok = ts;
+	*tok_len = xp - ts;
 	return QSE_NULL;
 }
 

qse/lib/awk/rec.c

@@ -1,5 +1,5 @@
 /*
- * $Id: rec.c 291 2009-09-21 13:28:18Z hyunghwan.chung $
+ * $Id: rec.c 292 2009-09-23 10:19:30Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -98,7 +98,7 @@ int qse_awk_rtx_setrec (
 
 static int split_record (qse_awk_rtx_t* rtx)
 {
-	qse_char_t* p, * tok;
+	qse_char_t* p, * px, * tok;
 	qse_size_t len, tok_len, nflds;
 	qse_awk_val_t* v, * fs;
 	qse_char_t* fs_ptr, * fs_free;
@@ -136,9 +136,6 @@ static int split_record (qse_awk_rtx_t* rtx)
 	}
 
 	/* scan the input record to count the fields */
-
-	len = QSE_STR_LEN(&rtx->inrec.line);
-
 	if (fs_len == 5 && fs_ptr[0] ==  QSE_T('?'))
 	{
 		if (qse_str_ncpy (
@@ -152,15 +149,18 @@ static int split_record (qse_awk_rtx_t* rtx)
 			return -1;
 		}
 
-		p = QSE_STR_PTR(&rtx->inrec.linew);
+		px = QSE_STR_PTR(&rtx->inrec.linew);
 		how = 1;
 	}
 	else
 	{
-		p = QSE_STR_PTR(&rtx->inrec.line);
+		px = QSE_STR_PTR(&rtx->inrec.line);
 		how = (fs_len <= 1)? 0: 2;
 	}
 
+	p = px; 
+	len = QSE_STR_LEN(&rtx->inrec.line);
+
 #if 0
 	nflds = 0;
 	while (p != QSE_NULL)
@@ -238,12 +238,14 @@ static int split_record (qse_awk_rtx_t* rtx)
 			qse_awk_rtx_seterrnum (rtx, QSE_AWK_ENOMEM, QSE_NULL);
 			return -1;
 		}
-		p = QSE_STR_PTR(&rtx->inrec.linew):
+		px = QSE_STR_PTR(&rtx->inrec.linew):
 	}
 	else
 	{
-		p = QSE_STR_PTR(&rtx->inrec.line);
+		px = QSE_STR_PTR(&rtx->inrec.line);
 	}
+
+	p = px; 
 	len = QSE_STR_LEN(&rtx->inrec.line);
 #endif
 
@@ -338,8 +340,7 @@ static int split_record (qse_awk_rtx_t* rtx)
 			rtx, rtx->inrec.flds[rtx->inrec.nflds].val);
 		rtx->inrec.nflds++;
 
-		len = QSE_STR_LEN(&rtx->inrec.line) - 
+		len = QSE_STR_LEN(&rtx->inrec.line) - (p - px);
-			(p - QSE_STR_PTR(&rtx->inrec.line));
 	}
 
 	if (fs_free != QSE_NULL) QSE_AWK_FREE (rtx->awk, fs_free);

qse/lib/awk/run.c

@@ -1,5 +1,5 @@
 /*
- * $Id: run.c 291 2009-09-21 13:28:18Z hyunghwan.chung $
+ * $Id: run.c 292 2009-09-23 10:19:30Z hyunghwan.chung $
  *
     Copyright 2006-2009 Chung, Hyung-Hwan.
     This file is part of QSE.
@@ -402,7 +402,7 @@ static int set_global (
 				fs_len = out.u.cpldup.len;
 			}
 
-			if (fs_len > 1)
+			if (fs_len > 1 && !(fs_len == 5 && fs_ptr[0] == QSE_T('?')))
 			{
 				void* rex;
 				qse_awk_errnum_t errnum;