enhanced a special form FS to affect record reading in bytes

lib/rio.c

@@ -123,7 +123,7 @@ static int find_rio_in (
 	if (handler == HAWK_NULL)
 	{
 		/* no I/O handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -388,9 +388,9 @@ int hawk_rtx_readio (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_ooch_t*
 	if (p->in.mbs)
 	{
 		if (name[0] == '\0')
-			hawk_rtx_seterrfmt (rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input"));
+			hawk_rtx_seterrfmt(rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input"));
 		else
-			hawk_rtx_seterrfmt (rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input on %js"), name);
+			hawk_rtx_seterrfmt(rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input on %js"), name);
 		return -1;
 	}
 
@@ -737,10 +737,16 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	hawk_rio_arg_t* p;
 	hawk_rio_impl_t handler;
 	int ret;
+	int esc_lq_rq;
+	int quoted;
+	hawk_bch_t esc, lq, rq;
 
 	hawk_val_t* brs;
 	hawk_bcs_t rrs;
 
+	hawk_val_t* bfs;
+	hawk_bcs_t ffs;
+
 	hawk_oow_t line_len = 0;
 	hawk_bch_t c = '\0', pc;
 
@@ -749,9 +755,9 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	if (!p->in.mbs)
 	{
 		if (name[0] == '\0')
-			hawk_rtx_seterrfmt (rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input"));
+			hawk_rtx_seterrfmt(rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input"));
 		else
-			hawk_rtx_seterrfmt (rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input on %js"), name);
+			hawk_rtx_seterrfmt(rtx, HAWK_NULL, HAWK_EPERM, HAWK_T("disallowed mixed mode input on %js"), name);
 		return -1;
 	}
 
@@ -762,12 +768,35 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	brs = hawk_rtx_getgbl(rtx, HAWK_GBL_RS);
 	hawk_rtx_refupval(rtx, brs);
 
+	bfs = hawk_rtx_getgbl(rtx, HAWK_GBL_FS);
+	hawk_rtx_refupval(rtx, bfs);
+
 	if (resolve_brs(rtx, brs, &rrs) <= -1)
 	{
 		hawk_rtx_refdownval(rtx, brs);
 		return -1;
 	}
 
+	if (resolve_brs(rtx, bfs, &ffs) <= -1)
+	{
+		hawk_rtx_refdownval(rtx, bfs);
+		hawk_rtx_refdownval(rtx, brs);
+		return -1;
+	}
+
+	/* RS set to @nil, FS set to a special string starting with ?, followed by esc lq rq */
+	esc_lq_rq = 0;
+	quoted = 0;
+	if (ffs.len == 5 && ffs.ptr[0] == '?' && !rrs.ptr)
+	{
+		esc = ffs.ptr[2];
+		lq = ffs.ptr[3];
+		rq = ffs.ptr[4];
+
+		esc_lq_rq = 1;
+		esc_lq_rq += (esc == lq && esc == rq);
+	}
+
 	ret = 1;
 
 	/* call the I/O handler */
@@ -859,6 +888,27 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 				c = p->in.u.bbuf[p->in.pos++];
 				end_pos = p->in.pos;
 
+				if (esc_lq_rq == 2)
+				{
+					/* if FS is something like [?,"""] and RS is @nil,
+					 * it supports multi-line quoted vlaues. */
+					if (quoted == 2)
+					{
+						quoted = (c == rq);
+						/* no continue here as c could be a new line */
+					}
+					else if (quoted == 1)
+					{
+						if (c == rq) quoted = 2;
+						continue;
+					}
+					else if (c == lq)
+					{
+						quoted = 1;
+						continue;
+					}
+				}
+
 				/* TODO: handle different line terminator */
 				/* separate by a new line */
 				if (c == '\n')
@@ -1044,6 +1094,7 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	}
 
 	if (rrs.ptr && HAWK_RTX_GETVALTYPE(rtx, brs) != HAWK_VAL_MBS) hawk_rtx_freemem(rtx, rrs.ptr);
+	hawk_rtx_refdownval(rtx, bfs);
 	hawk_rtx_refdownval(rtx, brs);
 
 	return ret;
@@ -1117,7 +1168,7 @@ static int prepare_for_write_io_data (hawk_rtx_t* rtx, hawk_out_type_t out_type,
 	if (HAWK_UNLIKELY(!handler))
 	{
 		/* no I/O handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -1261,7 +1312,7 @@ int hawk_rtx_flushio (hawk_rtx_t* rtx, hawk_out_type_t out_type, const hawk_ooch
 	if (!handler)
 	{
 		/* no I/O handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -1286,7 +1337,7 @@ int hawk_rtx_flushio (hawk_rtx_t* rtx, hawk_out_type_t out_type, const hawk_ooch
 	if (ok) return 0;
 
 	/* there is no corresponding rio for name */
-	hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIONMNF);
+	hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIONMNF);
 	return -1;
 }
 
@@ -1310,7 +1361,7 @@ int hawk_rtx_nextio_read (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	if (!handler)
 	{
 		/* no I/O handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -1324,7 +1375,7 @@ int hawk_rtx_nextio_read (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	{
 		/* something is totally wrong */
 		HAWK_ASSERT(!"should never happen - cannot find the relevant rio entry");
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EINTERN);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EINTERN);
 		return -1;
 	}
 
@@ -1379,7 +1430,7 @@ int hawk_rtx_nextio_write (hawk_rtx_t* rtx, hawk_out_type_t out_type, const hawk
 	if (!handler)
 	{
 		/* no I/O handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -1394,7 +1445,7 @@ int hawk_rtx_nextio_write (hawk_rtx_t* rtx, hawk_out_type_t out_type, const hawk
 		/* something is totally wrong */
 		HAWK_ASSERT(!"should never happen - cannot find the relevant rio entry");
 
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EINTERN);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EINTERN);
 		return -1;
 	}
 
@@ -1443,7 +1494,7 @@ int hawk_rtx_closio_read (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	if (!handler)
 	{
 		/* no I/O handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -1459,7 +1510,7 @@ int hawk_rtx_closio_read (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 				if (handler (rtx, HAWK_RIO_CMD_CLOSE, p, HAWK_NULL, 0) <= -1)
 				{
 					/* this is not a rtx-time error.*/
-					hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOIMPL);
+					hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOIMPL);
 					return -1;
 				}
 			}
@@ -1477,7 +1528,7 @@ int hawk_rtx_closio_read (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 	}
 
 	/* the name given is not found */
-	hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIONMNF);
+	hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIONMNF);
 	return -1;
 }
 
@@ -1500,7 +1551,7 @@ int hawk_rtx_closio_write (hawk_rtx_t* rtx, hawk_out_type_t out_type, const hawk
 	if (HAWK_UNLIKELY(!handler))
 	{
 		/* no io handler provided */
-		hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIOUSER);
+		hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIOUSER);
 		return -1;
 	}
 
@@ -1525,7 +1576,7 @@ int hawk_rtx_closio_write (hawk_rtx_t* rtx, hawk_out_type_t out_type, const hawk
 		p = p->next;
 	}
 
-	hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIONMNF);
+	hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIONMNF);
 	return -1;
 }
 
@@ -1614,7 +1665,7 @@ int hawk_rtx_closeio (hawk_rtx_t* rtx, const hawk_ooch_t* name, const hawk_ooch_
 		p = p->next;
 	}
 
-	hawk_rtx_seterrnum (rtx, HAWK_NULL, HAWK_EIONMNF);
+	hawk_rtx_seterrnum(rtx, HAWK_NULL, HAWK_EIONMNF);
 	return -1;
 }
 

t/Makefile.am

@@ -24,7 +24,8 @@ check_ERRORS = e-001.err
 ##noinst_SCRIPTS = $(check_SCRIPTS)
 EXTRA_DIST = $(check_SCRIPTS) $(check_ERRORS) tap.inc err.sh \
 	journal-toc.hawk journal-toc.in journal-toc.out journal-toc-html.out \
-	bibtex-to-html.hawk bibtex-to-html.out
+	bibtex-to-html.hawk bibtex-to-html.out \
+	fs-test.hawk fs-test.in fs-test.out
 
 check_PROGRAMS = t-001 t-002 t-003 t-004 t-005 t-006 t-007 t-008 t-009
 

t/Makefile.in

@@ -648,7 +648,8 @@ check_SCRIPTS = $(am__append_1) h-003.hawk h-004.hawk h-009.hawk
 check_ERRORS = e-001.err
 EXTRA_DIST = $(check_SCRIPTS) $(check_ERRORS) tap.inc err.sh \
 	journal-toc.hawk journal-toc.in journal-toc.out journal-toc-html.out \
-	bibtex-to-html.hawk bibtex-to-html.out
+	bibtex-to-html.hawk bibtex-to-html.out \
+	fs-test.hawk fs-test.in fs-test.out
 
 t_001_SOURCES = t-001.c tap.h
 t_001_CPPFLAGS = $(CPPFLAGS_COMMON)

t/fs-test.hawk

@@ -0,0 +1,9 @@
+BEGIN {
+	## `FS` with a question mark(`?`) followed by four characters. the last three are equal.
+	FS="?,\"\"\"";
+}
+
+{
+	for (i = 0; i <= NF; i++) print i, "[" $i "]";
+}
+

t/fs-test.in

@@ -0,0 +1,8 @@
+"Product ID","Product Name","Description","Price","Available Sizes","Notes"
+101,"Deluxe Widget","A versatile widget for various applications. Features include:
+- Enhanced durability
+- Multi-functional design
+- Easy to use",29.99,"Small,Medium,Large","Customer feedback suggests excellent performance."
+102,"Super Gadget","The ultimate gadget for tech enthusiasts.",49.99,"One Size","Limited stock, order soon!"
+103,"Basic Tool Set","A comprehensive set of essential tools, including a hammer, screwdriver, and wrench.
+Perfect for home DIY projects.",35.50,"N/A","Tools are made from high-quality steel."

t/fs-test.out

@@ -0,0 +1,36 @@
+0 ["Product ID","Product Name","Description","Price","Available Sizes","Notes"]
+1 [Product ID]
+2 [Product Name]
+3 [Description]
+4 [Price]
+5 [Available Sizes]
+6 [Notes]
+0 [101,"Deluxe Widget","A versatile widget for various applications. Features include:
+- Enhanced durability
+- Multi-functional design
+- Easy to use",29.99,"Small,Medium,Large","Customer feedback suggests excellent performance."]
+1 [101]
+2 [Deluxe Widget]
+3 [A versatile widget for various applications. Features include:
+- Enhanced durability
+- Multi-functional design
+- Easy to use]
+4 [29.99]
+5 [Small,Medium,Large]
+6 [Customer feedback suggests excellent performance.]
+0 [102,"Super Gadget","The ultimate gadget for tech enthusiasts.",49.99,"One Size","Limited stock, order soon!"]
+1 [102]
+2 [Super Gadget]
+3 [The ultimate gadget for tech enthusiasts.]
+4 [49.99]
+5 [One Size]
+6 [Limited stock, order soon!]
+0 [103,"Basic Tool Set","A comprehensive set of essential tools, including a hammer, screwdriver, and wrench.
+Perfect for home DIY projects.",35.50,"N/A","Tools are made from high-quality steel."]
+1 [103]
+2 [Basic Tool Set]
+3 [A comprehensive set of essential tools, including a hammer, screwdriver, and wrench.
+Perfect for home DIY projects.]
+4 [35.50]
+5 [N/A]
+6 [Tools are made from high-quality steel.]

t/h-003.hawk

@@ -10,7 +10,7 @@ function are_files_identical(a, b)
 	f1 = sys::open(a, sys::O_RDONLY);
 	if (f1 <= -1)
 	{
-		printf ("ERROR: unable to open %s\n", a);
+		printf("ERROR: unable to open %s\n", a);
 		return -1;
 	}
 
@@ -18,7 +18,7 @@ function are_files_identical(a, b)
 	if (f2 <= -1)
 	{
 		sys::close (a);
-		printf ("ERROR: unable to open %s\n", b);
+		printf("ERROR: unable to open %s\n", b);
 		return -1;
 	}
 
@@ -34,8 +34,8 @@ function are_files_identical(a, b)
 
 	if (sys::read(f2, y, 1) > 0) diff = 1;
 
-	sys::close (f2);
+	sys::close(f2);
-	sys::close (f1);
+	sys::close(f1);
 	return !diff;
 }
 
@@ -66,20 +66,21 @@ function run_test (x, more_opts, in_name, set_out_name, out_name)
 	if (same <= 0)
 	{
 		## don't delete the output file for review.
-		tap_fail (sprintf("%s[%d] %s - %s and %s differ", @SCRIPTNAME, @SCRIPTLINE, x, expf, outf));
+		tap_fail(sprintf("%s[%d] %s - %s and %s differ", @SCRIPTNAME, @SCRIPTLINE, x, expf, outf));
 	}
 	else
 	{
-		tap_ok (sprintf("%s[%d]", @SCRIPTNAME, @SCRIPTLINE));
+		tap_ok(sprintf("%s[%d]", @SCRIPTNAME, @SCRIPTLINE));
-		sys::unlink (outf);
+		sys::unlink(outf);
 	}
 }
 
 function main()
 {
-	run_test ("journal-toc", "", @nil, 0, @nil);
+	run_test("journal-toc", "", @nil, 0, @nil);
-	run_test ("journal-toc", "-vHTML=1", "journal-toc", 0, "journal-toc-html");
+	run_test("journal-toc", "-vHTML=1", "journal-toc", 0, "journal-toc-html");
-	run_test ("bibtex-to-html", "", "journal-toc", 1, "bibtex-to-html");
+	run_test("bibtex-to-html", "", "journal-toc", 1, "bibtex-to-html");
+	run_test("fs-test", "", "fs-test", 0, "fs-test");
 
 	tap_end ();
 }