implemented a way for a console input handler to indicate medium switching

lib/Hawk.hpp

@@ -621,8 +621,8 @@ public:
 		const void* getHandle () const { return this->riod->handle; }
 		void setHandle (void* handle) { this->riod->handle = handle; }
 
-		int getUflags () const { return this->riod->uflags; }
-		void setUflags (int uflags) { this->riod->uflags = uflags; }
+		hawk_uint16_t getUflags () const { return this->riod->uflags; }
+		void setUflags (hawk_uint16_t uflags) { this->riod->uflags = uflags; }
 
 		operator Hawk* () const { return this->run->hawk; }
 		operator hawk_t* () const
@@ -775,7 +775,7 @@ public:
 			/// The information set here is available in subsequent
 			/// calls to other methods and are accessible with
 			/// getHandle() and getUflags().
-			virtual int open  (Console& io) = 0;
+			virtual int open (Console& io) = 0;
 
 			/// The close() function is called when the console
 			/// is not needed any more. It must return 0 for success
@@ -811,6 +811,9 @@ public:
 			virtual int next  (Console& io) = 0;
 		};
 
+		bool getSwitched() const { return this->riod->console_switched; }
+		void setSwitched(bool v) { this->riod->console_switched = v; }
+
 	protected:
 		Console (Run* run, hawk_rio_arg_t* riod);
 		~Console ();

lib/Std.cpp

@@ -258,8 +258,8 @@ int HawkStd::build_argcv (Run* run)
 		if (argv.setIndexedStr(Value::IntIndex(i), this->runarg.ptr[i].ptr, this->runarg.ptr[i].len, true) <= -1) return -1;
 	}
 
-	run->setGlobal (this->gbl_argc, (hawk_int_t)this->runarg.len);
-	run->setGlobal (this->gbl_argv, argv);
+	run->setGlobal(this->gbl_argc, (hawk_int_t)this->runarg.len);
+	run->setGlobal(this->gbl_argv, argv);
 	return 0;
 }
 
@@ -423,7 +423,7 @@ int HawkStd::setioattr (
 	    hawk_find_oochar_in_oochars(ptr[1], l[1], '\0') ||
 	    hawk_find_oochar_in_oochars(ptr[2], l[2], '\0'))
 	{
-		return ret.setInt ((hawk_int_t)-1);
+		return ret.setInt((hawk_int_t)-1);
 	}
 
 	int tmout;
@@ -451,7 +451,7 @@ int HawkStd::setioattr (
 			nsec = fv - ioattr->tmout[tmout].sec;
 			ioattr->tmout[tmout].nsec = HAWK_SEC_TO_NSEC(nsec);
 		}
-		return ret.setInt ((hawk_int_t)0);
+		return ret.setInt((hawk_int_t)0);
 	}
 #if defined(HAWK_OOCH_IS_UCH)
 	else if (hawk_comp_oocstr(ptr[1], HAWK_T("codepage"), 1) == 0 ||
@@ -464,7 +464,7 @@ int HawkStd::setioattr (
 		else
 		{
 			cmgr = hawk_get_cmgr_by_name(ptr[2]);
-			if (cmgr == HAWK_NULL) return ret.setInt ((hawk_int_t)-1);
+			if (cmgr == HAWK_NULL) return ret.setInt((hawk_int_t)-1);
 		}
 
 		ioattr = find_or_make_ioattr(ptr[0], l[0]);
@@ -478,7 +478,7 @@ int HawkStd::setioattr (
 	else
 	{
 		// unknown attribute name
-		return ret.setInt ((hawk_int_t)-1);
+		return ret.setInt((hawk_int_t)-1);
 	}
 }
 
@@ -506,21 +506,21 @@ int HawkStd::getioattr (
 		if ((tmout = timeout_code(ptr[1])) >= 0)
 		{
 			if (ioattr->tmout[tmout].nsec == 0)
-				xx = args[2].setInt ((hawk_int_t)ioattr->tmout[tmout].sec);
+				xx = args[2].setInt((hawk_int_t)ioattr->tmout[tmout].sec);
 			else
-				xx = args[2].setFlt ((hawk_flt_t)ioattr->tmout[tmout].sec + HAWK_NSEC_TO_SEC((hawk_flt_t)ioattr->tmout[tmout].nsec));
+				xx = args[2].setFlt((hawk_flt_t)ioattr->tmout[tmout].sec + HAWK_NSEC_TO_SEC((hawk_flt_t)ioattr->tmout[tmout].nsec));
 		}
 	#if defined(HAWK_OOCH_IS_UCH)
 		else if (hawk_comp_oocstr(ptr[1], HAWK_T("codepage"), 1) == 0 ||
 		         hawk_comp_oocstr(ptr[1], HAWK_T("encoding"), 1) == 0)
 		{
-			xx = args[2].setStr (ioattr->cmgr_name);
+			xx = args[2].setStr(ioattr->cmgr_name);
 		}
 	#endif
 	}
 
 	// unknown attribute name or errors
-	return ret.setInt ((hawk_int_t)xx);
+	return ret.setInt((hawk_int_t)xx);
 }
 
 #if defined(ENABLE_NWIO)
@@ -782,7 +782,7 @@ int HawkStd::openFile (File& io)
 
 int HawkStd::closeFile (File& io)
 {
-	hawk_sio_close ((hawk_sio_t*)io.getHandle());
+	hawk_sio_close((hawk_sio_t*)io.getHandle());
 	return 0;
 }
 
@@ -946,6 +946,11 @@ nextfile:
 	pair = hawk_map_search(map, ibuf, ibuflen);
 	if (!pair)
 	{
+		if (this->runarg_count < i_argc)
+		{
+			this->runarg_index++;
+			goto nextfile;
+		}
 		if (this->runarg_count <= 0) goto console_open_stdin;
 		return 0;
 	}
@@ -1004,7 +1009,7 @@ nextfile:
 
 	if (hawk_rtx_setfilenamewithoochars(rtx, file, hawk_count_oocstr(file)) <= -1)
 	{
-		hawk_sio_close (sio);
+		hawk_sio_close(sio);
 		hawk_rtx_freevaloocstr (rtx, v_pair, as.ptr);
 		return -1;
 	}
@@ -1075,7 +1080,7 @@ int HawkStd::open_console_out (Console& io)
 
 		if (hawk_rtx_setofilenamewithoochars(rtx, file, hawk_count_oocstr(file)) == -1)
 		{
-			hawk_sio_close (sio);
+			hawk_sio_close(sio);
 			return -1;
 		}
 
@@ -1110,7 +1115,7 @@ int HawkStd::openConsole (Console& io)
 
 int HawkStd::closeConsole (Console& io)
 {
-	hawk_sio_close ((hawk_sio_t*)io.getHandle());
+	hawk_sio_close((hawk_sio_t*)io.getHandle());
 	return 0;
 }
 
@@ -1132,8 +1137,9 @@ hawk_ooi_t HawkStd::readConsole (Console& io, hawk_ooch_t* data, hawk_oow_t size
 			return 0;
 		}
 
-		if (sio) hawk_sio_close (sio);
-		((Run*)io)->setGlobal (HAWK_GBL_FNR, (hawk_int_t)0);
+		if (sio) hawk_sio_close(sio);
+		((Run*)io)->setGlobal(HAWK_GBL_FNR, (hawk_int_t)0);
+		io.setSwitched(true); // indicates that the console medium switched
 	}
 
 	return nn;
@@ -1157,8 +1163,9 @@ hawk_ooi_t HawkStd::readConsoleBytes (Console& io, hawk_bch_t* data, hawk_oow_t
 			return 0;
 		}
 
-		if (sio) hawk_sio_close (sio);
-		((Run*)io)->setGlobal (HAWK_GBL_FNR, (hawk_int_t)0);
+		if (sio) hawk_sio_close(sio);
+		((Run*)io)->setGlobal(HAWK_GBL_FNR, (hawk_int_t)0);
+		io.setSwitched(true); // indicates that the data comes from a new medium
 	}
 
 	return nn;
@@ -1194,7 +1201,7 @@ int HawkStd::nextConsole (Console& io)
 		return 0;
 	}
 
-	if (sio) hawk_sio_close (sio);
+	if (sio) hawk_sio_close(sio);
 	return n;
 }
 
@@ -1376,7 +1383,7 @@ int HawkStd::SourceFile::close (Data& io)
 {
 	hawk_sio_t* sio = (hawk_sio_t*)io.getHandle();
 	hawk_sio_flush (sio);
-	hawk_sio_close (sio);
+	hawk_sio_close(sio);
 	return 0;
 }
 
@@ -1520,7 +1527,7 @@ int HawkStd::SourceString::close (Data& io)
 	}
 	else
 	{
-		hawk_sio_close ((hawk_sio_t*)io.getHandle());
+		hawk_sio_close((hawk_sio_t*)io.getHandle());
 	}
 	return 0;
 }

lib/hawk.h

@@ -755,8 +755,9 @@ struct hawk_rio_arg_t
 	hawk_rio_rwcmode_t rwcmode;   /**< closing mode for rwpipe */
 
 	/* read-write. a user handler can do whatever it likes to do with these. */
-	void*              handle;    /**< I/O handle set by a handler */
-	int                uflags;    /**< user-flags set by a handler */
+	void*              handle;     /**< I/O handle set by a handler */
+	hawk_uint16_t      uflags; /**< user-flags set by a handler */
+	hawk_uint16_t      console_switched; /**< set by a console handler if it has opened a new underlying medium and continued reading */
 
 	/*--  from here down, internal use only --*/
 	int type;
@@ -771,15 +772,15 @@ struct hawk_rio_arg_t
 		} u;
 		hawk_oow_t pos;
 		hawk_oow_t len;
-		int        eof;
-		int        eos;
-		int        mbs;
+		unsigned int eof: 2;
+		unsigned int eos: 2;
+		unsigned int mbs: 2;
 	} in;
 
 	struct
 	{
-		int eof;
-		int eos;
+		unsigned int eof: 2;
+		unsigned int eos: 2;
 	} out;
 
 	struct hawk_rio_arg_t* next;

lib/rio.c

@@ -165,7 +165,7 @@ static int find_rio_in (
 		p->in.eof = 0;
 		p->in.eos = 0;
 		*/
-		p->in.mbs = mbs_if_new;
+		p->in.mbs = !!mbs_if_new;
 
 		/* request to open the stream */
 		x = handler(rtx, HAWK_RIO_CMD_OPEN, p, HAWK_NULL, 0);
@@ -453,14 +453,16 @@ int hawk_rtx_readio (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_ooch_t*
 				break;
 			}
 
+			/* used for console only. but reset it regardless of the stream type.
+			 * it must be faster than checking the type and resetting it */
+			p->console_switched = 0;
 			x = handler(rtx, HAWK_RIO_CMD_READ, p, p->in.u.buf, HAWK_COUNTOF(p->in.u.buf));
 			if (x <= -1)
 			{
 				ret = -1;
 				break;
 			}
-
-			if (x == 0)
+			else if (x == 0)
 			{
 				/* EOF reached */
 				p->in.eof = 1;
@@ -508,6 +510,18 @@ int hawk_rtx_readio (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_ooch_t*
 
 			p->in.len = x;
 			p->in.pos = 0;
+
+			/* the new data is a chunk from a new underlying stream.
+			 * let's end the record even if it didn't match RS yet.
+			 * for example, `hawk -v RS=separator -f script f1.txt f2.txt`
+			 * the console handler is invoked over all the specified files.
+			 * if `separator` is not found when the end of `f1.txt` is reached,
+			 * `f2.txt` is opened and read. When the handler returns the 
+			 * first part of the second file, it must indicate this special
+			 * condition. if it's not indicated, it goes on without breaking
+			 * the record at the end of the first file.
+			 */
+			if (p->console_switched) break;
 		}
 
 		if (rrs.ptr == HAWK_NULL)
@@ -820,6 +834,7 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 				break;
 			}
 
+			p->console_switched = 0;
 			x = handler(rtx, HAWK_RIO_CMD_READ_BYTES, p, p->in.u.bbuf, HAWK_COUNTOF(p->in.u.bbuf));
 			if (x <= -1)
 			{
@@ -875,6 +890,10 @@ int hawk_rtx_readiobytes (hawk_rtx_t* rtx, hawk_in_type_t in_type, const hawk_oo
 
 			p->in.len = x;
 			p->in.pos = 0;
+
+			/* the new data is a chunk from a new underlying medium.
+			 * let's end the record even if it didn't match RS yet. */
+			if (p->console_switched) break;
 		}
 
 		if (rrs.ptr == HAWK_NULL)

lib/std.c

@@ -2621,6 +2621,7 @@ static hawk_ooi_t hawk_rio_console (hawk_rtx_t* rtx, hawk_rio_cmd_t cmd, hawk_ri
 
 				/* reset FNR to 0 here since the caller doesn't know that the file has changed. */
 				hawk_rtx_setgbl(rtx, HAWK_GBL_FNR, hawk_rtx_makeintval(rtx, 0));
+				riod->console_switched = 1;
 			}
 
 			if (nn <= -1) set_rio_error(rtx, HAWK_EREAD, HAWK_T("unable to read"), hawk_sio_getpath((hawk_sio_t*)riod->handle));
@@ -2646,7 +2647,9 @@ static hawk_ooi_t hawk_rio_console (hawk_rtx_t* rtx, hawk_rio_cmd_t cmd, hawk_ri
 				}
 
 				if (sio) hawk_sio_close(sio);
+
 				hawk_rtx_setgbl(rtx, HAWK_GBL_FNR, hawk_rtx_makeintval(rtx, 0));
+				riod->console_switched = 1;
 			}
 
 			if (nn <= -1) set_rio_error(rtx, HAWK_EREAD, HAWK_T("unable to read"), hawk_sio_getpath((hawk_sio_t*)riod->handle));
@@ -2887,7 +2890,7 @@ static int build_environ (hawk_rtx_t* rtx, int gbl_id, env_char_t* envarr[])
 
 			kptr = hawk_rtx_dupbtoucstr(rtx, envarr[count], &klen, 1);
 			vptr = hawk_rtx_dupbtoucstr(rtx, eq + 1, &vlen, 1);
-			if (HAWK_UNLIKELY(!kptr || !vptr))
+			if (HAWK_UNLIKELY(!kptr) || HAWK_UNLIKELY(!vptr))
 			{
 				if (kptr) hawk_rtx_freemem(rtx, kptr);
 				if (vptr) hawk_rtx_freemem(rtx, vptr);
@@ -2904,7 +2907,7 @@ static int build_environ (hawk_rtx_t* rtx, int gbl_id, env_char_t* envarr[])
 
 			kptr = hawk_rtx_duputobcstr(rtx, envarr[count], &klen);
 			vptr = hawk_rtx_duputobcstr(rtx, eq + 1, &vlen);
-			if (HAWK_UNLIKELY(!kptr || !vptr))
+			if (HAWK_UNLIKELY(!kptr) || HAWK_UNLIKELY(!vptr))
 			{
 				if (kptr) hawk_rtx_freemem(rtx, kptr);
 				if (vptr) hawk_rtx_freeme(rtx, vptr):
@@ -2967,7 +2970,7 @@ static int build_environ (hawk_rtx_t* rtx, int gbl_id, env_char_t* envarr[])
 		}
 	}
 
-	if (hawk_rtx_setgbl(rtx, gbl_id, v_env) == -1)
+	if (hawk_rtx_setgbl(rtx, gbl_id, v_env) <= -1)
 	{
 		hawk_rtx_refdownval(rtx, v_env);
 		return -1;