fixed quite a few bugs and enhanced error handling in sed

qse/include/qse/sed/sed.h

@@ -87,6 +87,7 @@ enum qse_sed_errnum_t
 	QSE_SED_EOCSDU,  /**< multiple occurrence specifiers */
 	QSE_SED_EOCSZE,  /**< occurrence specifier to s is zero */
 	QSE_SED_EOCSTL,  /**< occurrence specifier too large */
+	QSE_SED_EIOFIL,  /**< file io error */
 	QSE_SED_EIOUSR   /**< user io error */
 };
 typedef enum qse_sed_errnum_t qse_sed_errnum_t;
@@ -271,8 +272,8 @@ int qse_sed_comp (
  */
 int qse_sed_exec (
 	qse_sed_t*        sed,  /**< a stream editor */
-	qse_sed_io_fun_t in,  /**< stream reader */
+	qse_sed_io_fun_t  inf,  /**< stream reader */
-	qse_sed_io_fun_t out  /**< stream writer */
+	qse_sed_io_fun_t  outf  /**< stream writer */
 );
 
 #ifdef __cplusplus

qse/lib/sed/err.c

@@ -26,7 +26,7 @@ static const qse_char_t* geterrstr (int errnum)
 		QSE_T("out of memory"),
 		QSE_T("command '${0}' not recognized"),
 		QSE_T("command code missing"),
-		QSE_T("command '${0}' not terminated properly"),
+		QSE_T("command '${0}' incomplete"),
 		QSE_T("regular expression '${0}' incomplete"),
 		QSE_T("failed to compile regular expression '${0}'"),
 		QSE_T("failed to match regular expression"),
@@ -49,6 +49,7 @@ static const qse_char_t* geterrstr (int errnum)
 		QSE_T("multiple occurrence specifier"),
 		QSE_T("occurrence specifier is zero"),
 		QSE_T("occurrence specifier too large"),
+		QSE_T("io error with file '${0}'"),
 		QSE_T("error returned by user io handler")
 	};
 

qse/lib/sed/sed.h

@@ -49,8 +49,24 @@ struct qse_sed_t
 		const qse_char_t* cur; /**< current source text pointer */
 	} src;
 
-	/** temporary regular expression buffer */
+	/** temporary data for compiling */
-	qse_str_t rexbuf;
+	struct
+	{
+		qse_str_t rex; /**< regular expression buffer */
+		qse_str_t lab; /**< label name buffer */
+
+		/** data structure to compile command groups */
+		struct
+		{
+			/** current level of command group nesting */
+			int level;
+			/** keeps track of the begining of nested groups */
+			qse_sed_cmd_t* cmd[128];
+		} grp;
+
+		/** a table storing labels seen */
+		qse_map_t labs; 
+	} tmp;
 
 	/** compiled commands */
 	struct
@@ -60,18 +76,6 @@ struct qse_sed_t
 		qse_sed_cmd_t* cur; /**< points next to the last command */
 	} cmd;
 
-	/** a table storing labels seen */
-	qse_map_t labs; 
-
-	/** data structure to compile command groups */
-	struct
-	{
-		/** current level of command group nesting */
-		int level;
-		/** keeps track of the begining of nested command groups */
-		qse_sed_cmd_t* cmd[128];
-	} grp;
-
 	/** data for execution */
 	struct
 	{
@@ -174,6 +178,7 @@ struct qse_sed_adr_t
 struct qse_sed_cmd_t
 {
 	qse_char_t type;
+	qse_size_t lnum;
 
 	int negated;
 

qse/test/sed/sed01.c

@@ -22,6 +22,8 @@
 #include <qse/cmn/str.h>
 #include <qse/cmn/chr.h>
 
+const qse_char_t* instream = QSE_NULL;
+
 static qse_ssize_t in (
 	qse_sed_t* sed, qse_sed_io_cmd_t cmd, qse_sed_io_arg_t* arg)
 {
@@ -31,7 +33,19 @@ static qse_ssize_t in (
 			if (arg->open.path == QSE_NULL ||
 			    arg->open.path[0] == QSE_T('\0'))
 			{
-				arg->open.handle = QSE_STDIN;
+				if (instream)
+				{
+					arg->open.handle = qse_fopen (instream, QSE_T("r"));
+					if (arg->open.handle == QSE_NULL) 
+					{
+						qse_cstr_t errarg;
+						errarg.ptr = instream;
+						errarg.len = qse_strlen(instream);
+						qse_sed_seterror (sed, QSE_SED_EIOFIL, 0, &errarg);
+						return -1;
+					}
+				}
+				else arg->open.handle = QSE_STDIN;
 			}
 			else
 			{
@@ -102,10 +116,12 @@ int sed_main (int argc, qse_char_t* argv[])
 
 	if (argc != 2 && argc != 3)
 	{
-		qse_fprintf (QSE_STDERR, QSE_T("usage: %s string\n"), argv[0]);
+		qse_fprintf (QSE_STDERR, QSE_T("usage: %s string [file]\n"), argv[0]);
 		return -1;
 	}
 
+	if (argc == 3) instream = argv[2];
+
 	sed = qse_sed_open (QSE_NULL, 0);
 	if (sed == QSE_NULL)
 	{
@@ -116,21 +132,44 @@ int sed_main (int argc, qse_char_t* argv[])
 	if (argc == 3) qse_sed_setoption (sed, qse_strtoi(argv[2]));
 
 	if (qse_sed_comp (sed, argv[1], qse_strlen(argv[1])) == -1)
+	{
+		qse_size_t errlin = qse_sed_geterrlin(sed);
+		if (errlin > 0)
 		{
 			qse_fprintf (QSE_STDERR, 
 				QSE_T("cannot compile - %s at line %lu\n"),
 				qse_sed_geterrmsg(sed),
-			(unsigned long)qse_sed_geterrlin(sed)
+				(unsigned long)errlin
 			);
+		}
+		else
+		{
+			qse_fprintf (QSE_STDERR, 
+				QSE_T("cannot compile - %s\n"),
+				qse_sed_geterrmsg(sed)
+			);
+		}
 		goto oops;
 	}
 
 	if (qse_sed_exec (sed, in, out) == -1)
+	{
+		qse_size_t errlin = qse_sed_geterrlin(sed);
+		if (errlin > 0)
+		{
+			qse_fprintf (QSE_STDERR, 
+				QSE_T("cannot execute - %s at line %lu\n"),
+				qse_sed_geterrmsg(sed),
+				(unsigned long)errlin
+			);
+		}
+		else
 		{
 			qse_fprintf (QSE_STDERR, 
 				QSE_T("cannot execute - %s\n"),
 				qse_sed_geterrmsg(sed)
 			);
+		}
 		goto oops;
 	}