fixed a bug in the builtin split() function

qse/cmd/awk/awk.c

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.c 182 2009-06-03 21:50:32Z hyunghwan.chung $
+ * $Id: awk.c 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -63,6 +63,7 @@ struct argout_t
 	qse_char_t*  osf;  /* output source file */
 
 
+	qse_char_t** arg;
 	qse_char_t** icf;  /* input console files */
 	qse_size_t   icfl; /* the number of input console files */
 	qse_map_t*   vm;   /* global variable map */
@@ -351,8 +352,9 @@ static int handle_args (int argc, qse_char_t* argv[], struct argout_t* ao)
 	qse_size_t isfc = 16; /* the capacity of isf */
 	qse_size_t isfl = 0; /* number of input source files */
 
+	qse_size_t argl = 0;
 	qse_size_t icfc = 0; /* the capacity of icf */
-	qse_size_t icfl = 0;  /* the number of input console files */
+	qse_size_t icfl = 0; /* the number of input console files */
 
 	qse_char_t** isf = QSE_NULL; /* input source files */
 	qse_char_t*  osf = QSE_NULL; /* output source file */
@@ -509,7 +511,7 @@ static int handle_args (int argc, qse_char_t* argv[], struct argout_t* ao)
 
 	/* the remaining arguments are input console file names */
 	icfc = (opt.ind >= argc)? 2: (argc - opt.ind + 1);
-	icf = (qse_char_t**) malloc (QSE_SIZEOF(*icf)*icfc);
+	icf = (qse_char_t**) malloc (QSE_SIZEOF(qse_char_t*)*icfc);
 	if (icf == QSE_NULL)
 	{
 		out_of_memory ();
@@ -644,7 +646,7 @@ static int awk_main (int argc, qse_char_t* argv[])
 	rcb.data = &ao;
 
 	rtx = qse_awk_rtx_openstd (
-		awk, 0, ao.icf, QSE_AWK_RTX_OPENSTD_STDIO);
+		awk, 0, QSE_T("qseawk"), ao.icf, QSE_AWK_RTX_OPENSTD_STDIO);
 	if (rtx == QSE_NULL) 
 	{
 		qse_printf (

qse/include/qse/awk/awk.h

@@ -1,5 +1,5 @@
 /*
- * $Id: awk.h 171 2009-06-01 09:34:34Z hyunghwan.chung $
+ * $Id: awk.h 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -413,7 +413,8 @@ enum qse_awk_option_t
 
 	/* option aggregtes */
 	QSE_AWK_CLASSIC  = QSE_AWK_IMPLICIT | QSE_AWK_RIO | 
-	                   QSE_AWK_NEWLINE | QSE_AWK_PABLOCK
+	                   QSE_AWK_NEWLINE | QSE_AWK_PABLOCK | 
+	                   QSE_AWK_STRIPSPACES
 };
 
 /**

qse/include/qse/awk/std.h

@@ -1,5 +1,5 @@
 /*
- * $Id: std.h 86 2009-02-26 12:55:05Z hyunghwan.chung $
+ * $Id: std.h 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -21,6 +21,10 @@
 
 #include <qse/awk/awk.h>
 
+/** @file
+ * Standard AWK Interpreter
+ */
+
 /****e* AWK/qse_awk_parsestd_type_t
  * NAME
  *  qse_awk_parsestd_type_t - define a source type
@@ -131,22 +135,19 @@ int qse_awk_parsestd (
 );
 /******/
 
-/****f* AWK/qse_awk_rtx_openstd
+/**
- * NAME
- *  qse_awk_rtx_openstd - create a runtime context
  * DESCRIPTION
- *  The caller should keep the contents of icf and ocf valid throughout
+ * The qse_awk_rtx_openstd() function creates a standard runtime context.
- *  the lifetime of the runtime context created. The runtime context 
+ * The caller should keep the contents of icf and ocf valid throughout
- *  remembers the pointers without copying in the contents.
+ * the lifetime of the runtime context created. The runtime context 
- * SYNOPSIS
  */
 qse_awk_rtx_t* qse_awk_rtx_openstd (
-	qse_awk_t*             awk,
+	qse_awk_t*              awk,
-	qse_size_t             xtnsize,
+	qse_size_t              xtn,
-	const qse_char_t*const icf[],
+	const qse_char_t*       id,
-	const qse_char_t*const ocf[]
+	const qse_char_t*const* icf,
+	const qse_char_t*const* ocf
 );
-/******/
 
 /****f* AWK/qse_awk_rtx_getxtnstd
  * NAME

qse/lib/awk/fnc.c

@@ -1,5 +1,5 @@
 /*
- * $Id: fnc.c 171 2009-06-01 09:34:34Z hyunghwan.chung $
+ * $Id: fnc.c 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -584,7 +584,7 @@ static int fnc_split (
 	qse_awk_val_t* a0, * a1, * a2, * t1, * t2, ** a1_ref;
 	qse_char_t* str, * str_free, * p, * tok;
 	qse_size_t str_len, str_left, tok_len, org_len;
-	qse_long_t num;
+	qse_long_t nflds;
 	qse_char_t key[QSE_SIZEOF(qse_long_t)*8+2];
 	qse_size_t key_len;
 	qse_char_t* fs_ptr, * fs_free;
@@ -727,7 +727,7 @@ static int fnc_split (
 	qse_awk_rtx_refupval (run, *a1_ref);
 
 	p = str; str_left = str_len; org_len = str_len;
-	num = 1;
+	nflds = 0;
 
 	while (p != QSE_NULL)
 	{
@@ -755,14 +755,13 @@ static int fnc_split (
 			}
 		}
 
-		if (num == 0 && p == QSE_NULL && tok_len == 0) 
+		if (nflds == 0 && p == QSE_NULL && tok_len == 0) 
 		{
 			/* no field at all*/
 			break; 
 		}	
 
-		QSE_ASSERT (
+		QSE_ASSERT ((tok != QSE_NULL && tok_len > 0) || tok_len == 0);
-			(tok != QSE_NULL && tok_len > 0) || tok_len == 0);
 
 		/* create the field string */
 		t2 = qse_awk_rtx_makestrval (run, tok, tok_len);
@@ -780,7 +779,7 @@ static int fnc_split (
 
 		/* put it into the map */
 		key_len = qse_awk_longtostr (
-			num, 10, QSE_NULL, key, QSE_COUNTOF(key));
+			++nflds, 10, QSE_NULL, key, QSE_COUNTOF(key));
 		QSE_ASSERT (key_len != (qse_size_t)-1);
 
 		/* don't forget to update the reference count when you 
@@ -810,7 +809,6 @@ static int fnc_split (
 			return -1;
 		}
 
-		num++;
 		str_len = str_left - (p - str);
 	}
 
@@ -818,9 +816,9 @@ static int fnc_split (
 	if (fs_free != QSE_NULL) QSE_AWK_FREE (run->awk, fs_free);
 	if (fs_rex_free != QSE_NULL) QSE_AWK_FREEREX (run->awk, fs_rex_free);
 
-	num--;
+	/*nflds--;*/
 
-	t1 = qse_awk_rtx_makeintval (run, num);
+	t1 = qse_awk_rtx_makeintval (run, nflds);
 	if (t1 == QSE_NULL)
 	{
 		/*qse_awk_rtx_seterrnum (run, QSE_AWK_ENOMEM);*/

qse/lib/awk/parse.c

@@ -1,5 +1,5 @@
 /*
- * $Id: parse.c 171 2009-06-01 09:34:34Z hyunghwan.chung $
+ * $Id: parse.c 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -3760,6 +3760,17 @@ static qse_awk_nde_t* parse_if (qse_awk_t* awk, qse_size_t line)
 		return QSE_NULL;
 	}
 
+	/* skip any new lines before the else block */
+	while (MATCH(awk,TOKEN_NEWLINE))
+	{
+		if (get_token(awk) == -1) 
+		{
+			qse_awk_clrpt (awk, then_part);
+			qse_awk_clrpt (awk, test);
+			return QSE_NULL;
+		}
+	} 
+
 	if (MATCH(awk,TOKEN_ELSE)) 
 	{
 		if (get_token(awk) == -1) 

qse/lib/awk/run.c

@@ -1,5 +1,5 @@
 /*
- * $Id: run.c 182 2009-06-03 21:50:32Z hyunghwan.chung $
+ * $Id: run.c 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -1002,8 +1002,8 @@ static int build_runarg (
 				return -1;
 			}
 
-			key_len = qse_awk_longtostr (argc+1, 
+			key_len = qse_awk_longtostr (
-				10, QSE_NULL, key, QSE_COUNTOF(key));
+				argc, 10, QSE_NULL, key, QSE_COUNTOF(key));
 			QSE_ASSERT (key_len != (qse_size_t)-1);
 
 			/* increment reference count of v_tmp in advance as if 

qse/lib/awk/std.c

@@ -1,5 +1,5 @@
 /*
- * $Id: std.c 182 2009-06-03 21:50:32Z hyunghwan.chung $
+ * $Id: std.c 194 2009-06-09 13:07:42Z hyunghwan.chung $
  *
    Copyright 2006-2009 Chung, Hyung-Hwan.
 
@@ -440,8 +440,6 @@ static qse_ssize_t awk_rio_pipe (
 			}
 			else return -1; /* TODO: any way to set the error number? */
 
-			/*dprint (QSE_T("opening %s of type %d (pipe)\n"),  riod->name, riod->type);*/
-
 			handle = qse_pio_open (
 				rtx->awk->mmgr,
 				0, 
@@ -456,7 +454,6 @@ static qse_ssize_t awk_rio_pipe (
 
 		case QSE_AWK_RIO_CLOSE:
 		{
-			/*dprint (QSE_T("closing %s of type (pipe) %d\n"),  riod->name, riod->type);*/
 			qse_pio_close ((qse_pio_t*)riod->handle);
 			riod->handle = QSE_NULL;
 			return 0;
@@ -525,7 +522,6 @@ static qse_ssize_t awk_rio_file (
 			}
 			else return -1; /* TODO: any way to set the error number? */
 
-			/*dprint (QSE_T("opening %s of type %d (file)\n"), riod->name, riod->type);*/
 			handle = qse_fio_open (
 				rtx->awk->mmgr,
 				0,
@@ -551,7 +547,6 @@ static qse_ssize_t awk_rio_file (
 
 		case QSE_AWK_RIO_CLOSE:
 		{
-			/*dprint (QSE_T("closing %s of type %d (file)\n"), riod->name, riod->type);*/
 			qse_fio_close ((qse_fio_t*)riod->handle);
 			riod->handle = QSE_NULL;
 			return 0;
@@ -594,8 +589,6 @@ static int open_rio_console (qse_awk_rtx_t* rtx, qse_awk_riod_t* riod)
 {
 	rxtn_t* rxtn = (rxtn_t*) QSE_XTN (rtx);
 
-	/*dprint (QSE_T("opening console[%s] of type %x\n"), riod->name, riod->type);*/
-
 	if (riod->mode == QSE_AWK_RIO_CONSOLE_READ)
 	{
 		if (rxtn->c.in.files == QSE_NULL)
@@ -607,13 +600,11 @@ static int open_rio_console (qse_awk_rtx_t* rtx, qse_awk_riod_t* riod)
 		if (rxtn->c.in.files[rxtn->c.in.index] == QSE_NULL)
 		{
 			/* no more input file */
-			/*dprint (QSE_T("console - no more file\n"));*/
 			return 0;
 		}
 
 		if (rxtn->c.in.files[rxtn->c.in.index][0] == QSE_T('\0'))
 		{
-			/*dprint (QSE_T("    console(r) - <standard input>\n"));*/
 			riod->handle = qse_sio_in;
 		}
 		else
@@ -639,7 +630,6 @@ static int open_rio_console (qse_awk_rtx_t* rtx, qse_awk_riod_t* riod)
 				return -1;
 			}
 
-			/*dprint (QSE_T("    console(r) - %s\n"), rxtn->c.in.files[rxtn->c.in.index]);*/
 			if (qse_awk_rtx_setfilename (
 				rtx, rxtn->c.in.files[rxtn->c.in.index], 
 				qse_strlen(rxtn->c.in.files[rxtn->c.in.index])) == -1)
@@ -665,13 +655,11 @@ static int open_rio_console (qse_awk_rtx_t* rtx, qse_awk_riod_t* riod)
 		if (rxtn->c.out.files[rxtn->c.out.index] == QSE_NULL)
 		{
 			/* no more input file */
-			/*dprint (QSE_T("console - no more file\n"));*/
 			return 0;
 		}
 
 		if (rxtn->c.out.files[rxtn->c.out.index][0] == QSE_T('\0'))
 		{
-			/*dprint (QSE_T("    console(w) - <standard output>\n"));*/
 			riod->handle = qse_sio_out;
 		}
 		else
@@ -697,7 +685,6 @@ static int open_rio_console (qse_awk_rtx_t* rtx, qse_awk_riod_t* riod)
 				return -1;
 			}
 
-			/*dprint (QSE_T("    console(w) - %s\n"), rxtn->c.out.files[rxtn->c.out.index]);*/
 			if (qse_awk_rtx_setofilename (
 				rtx, rxtn->c.out.files[rxtn->c.out.index], 
 				qse_strlen(rxtn->c.out.files[rxtn->c.out.index])) == -1)
@@ -728,8 +715,6 @@ static qse_ssize_t awk_rio_console (
 	}
 	else if (cmd == QSE_AWK_RIO_CLOSE)
 	{
-		/*dprint (QSE_T("closing console of type %x\n"), riod->type);*/
-
 		if (riod->handle != QSE_NULL &&
 		    riod->handle != qse_sio_in && 
 		    riod->handle != qse_sio_out && 
@@ -812,7 +797,6 @@ static qse_ssize_t awk_rio_console (
 					qse_sio_close ((qse_sio_t*)riod->handle);
 				}
 
-				/*dprint (QSE_T("open the next console [%s]\n"), rxtn->c.in.files[rxtn->c.in.index]);*/
 				riod->handle = sio;
 			}
 
@@ -838,8 +822,6 @@ static qse_ssize_t awk_rio_console (
 		int n;
 		qse_sio_t* sio = (qse_sio_t*)riod->handle;
 
-		/*dprint (QSE_T("switching console[%s] of type %x\n"), riod->name, riod->type);*/
-
 		n = open_rio_console (rtx, riod);
 		if (n == -1) return -1;
 
@@ -864,26 +846,72 @@ static qse_ssize_t awk_rio_console (
 }
 
 qse_awk_rtx_t* qse_awk_rtx_openstd (
-	qse_awk_t* awk,
+	qse_awk_t*              awk,
-	qse_size_t xtnsize,
+	qse_size_t              xtnsize,
-	const qse_char_t*const icf[],
+	const qse_char_t*       id,
-	const qse_char_t*const ocf[])
+	const qse_char_t*const* icf,
+	const qse_char_t*const* ocf)
 {
 	qse_awk_rtx_t* rtx;
 	qse_awk_rio_t rio;
 	rxtn_t* rxtn;
 	qse_ntime_t now;
 
+	const qse_char_t*const* p;
+	qse_size_t argc = 0;
+	qse_cstr_t argv[16];
+	qse_cstr_t* argvp = QSE_NULL, * p2;
+
 	rio.pipe    = awk_rio_pipe;
 	rio.file    = awk_rio_file;
 	rio.console = awk_rio_console;
 
+	if (icf != QSE_NULL)
+	{
+		for (p = icf; *p != QSE_NULL; p++);
+		argc = p - icf;
+	}
+
+	argc++; /* for id */
+
+	if (argc < QSE_COUNTOF(argv)) argvp = argv;
+	else
+	{
+		argvp = QSE_AWK_ALLOC (
+			awk, QSE_SIZEOF(*argvp) * (argc + 1));
+		if (argvp == QSE_NULL)
+		{
+			qse_awk_seterrnum (awk, QSE_AWK_ENOMEM);
+			return QSE_NULL;
+		}
+	}
+
+	p2 = argvp;
+
+	p2->ptr = id;
+	p2->len = qse_strlen(id);
+	p2++;
+	
+	if (icf != QSE_NULL)
+	{
+		for (p = icf; *p != QSE_NULL; p++, p2++) 
+		{
+			p2->ptr = *p;
+			p2->len = qse_strlen(*p);
+		}
+	}
+
+	p2->ptr = QSE_NULL;
+	p2->len = 0;
+
 	rtx = qse_awk_rtx_open (
 		awk, 
 		QSE_SIZEOF(rxtn_t) + xtnsize,
 		&rio,
-		QSE_NULL/*runarg*/
+		argvp
 	);
+
+	if (argvp != QSE_NULL && argvp != argv) QSE_AWK_FREE (awk, argvp);
 	if (rtx == QSE_NULL) return QSE_NULL;
 
 	rxtn = (rxtn_t*) QSE_XTN (rtx);