Recovered from cvs revision 2007-05-19 04:43:00

ase/awk/StdAwk.cpp

@@ -1,5 +1,5 @@
 /*
- * $Id: StdAwk.cpp,v 1.14 2007/05/16 14:44:13 bacon Exp $
+ * $Id: StdAwk.cpp,v 1.15 2007/05/18 15:36:02 bacon Exp $
  */
 
 #include <ase/awk/StdAwk.hpp>
@@ -26,44 +26,49 @@ namespace ASE
 		::srand (seed);
 	}
 
+	#define ADD_FUNC(name,min,max,impl) \
+		do { \
+			if (addFunction (name, min, max, \
+				(FunctionHandler)impl) == -1) return -1; \
+		} while (0)
+
 	int StdAwk::open ()
 	{
 		int n = Awk::open ();
-		if (n == 0)
+		if (n == -1) return n;
-		{
+
-			int opt = 
+		int opt = 
-				ASE_AWK_IMPLICIT |
+			ASE_AWK_IMPLICIT |
-				ASE_AWK_EXPLICIT | 
+			ASE_AWK_EXPLICIT | 
-				ASE_AWK_UNIQUEFN | 
+			ASE_AWK_UNIQUEFN | 
-				ASE_AWK_IDIV |
+			ASE_AWK_IDIV |
-				ASE_AWK_SHADING | 
+			ASE_AWK_SHADING | 
-				ASE_AWK_SHIFT | 
+			ASE_AWK_SHIFT | 
-				ASE_AWK_EXTIO | 
+			ASE_AWK_EXTIO | 
-				ASE_AWK_BLOCKLESS | 
+			ASE_AWK_BLOCKLESS | 
-				ASE_AWK_STRBASEONE | 
+			ASE_AWK_STRBASEONE | 
-				ASE_AWK_STRIPSPACES | 
+			ASE_AWK_STRIPSPACES | 
-				ASE_AWK_NEXTOFILE |
+			ASE_AWK_NEXTOFILE |
-				ASE_AWK_ARGSTOMAIN;
+			ASE_AWK_ARGSTOMAIN;
-			ase_awk_setoption (awk, opt);
+		ase_awk_setoption (awk, opt);
 
 
-			addFunction (ASE_T("sin"), 1, 1, (FunctionHandler)&StdAwk::sin);
+		ADD_FUNC (ASE_T("sin"),        1, 1, &StdAwk::sin);
-			addFunction (ASE_T("cos"), 1, 1, (FunctionHandler)&StdAwk::cos);
+		ADD_FUNC (ASE_T("cos"),        1, 1, &StdAwk::cos);
-			addFunction (ASE_T("tan"), 1, 1, (FunctionHandler)&StdAwk::tan);
+		ADD_FUNC (ASE_T("tan"),        1, 1, &StdAwk::tan);
-			addFunction (ASE_T("atan2"), 2, 2, (FunctionHandler)&StdAwk::atan2);
+		ADD_FUNC (ASE_T("atan2"),      2, 2, &StdAwk::atan2);
-			addFunction (ASE_T("log"), 1, 1, (FunctionHandler)&StdAwk::log);
+		ADD_FUNC (ASE_T("log"),        1, 1, &StdAwk::log);
-			addFunction (ASE_T("exp"), 1, 1, (FunctionHandler)&StdAwk::exp);
+		ADD_FUNC (ASE_T("exp"),        1, 1, &StdAwk::exp);
-			addFunction (ASE_T("sqrt"), 1, 1, (FunctionHandler)&StdAwk::sqrt);
+		ADD_FUNC (ASE_T("sqrt"),       1, 1, &StdAwk::sqrt);
-			addFunction (ASE_T("int"), 1, 1, (FunctionHandler)&StdAwk::fnint);
+		ADD_FUNC (ASE_T("int"),        1, 1, &StdAwk::fnint);
-			addFunction (ASE_T("rand"), 0, 0, (FunctionHandler)&StdAwk::rand);
+		ADD_FUNC (ASE_T("rand"),       0, 0, &StdAwk::rand);
-			addFunction (ASE_T("srand"), 1, 1, (FunctionHandler)&StdAwk::srand);
+		ADD_FUNC (ASE_T("srand"),      1, 1, &StdAwk::srand);
-			addFunction (ASE_T("systime"), 0, 0, (FunctionHandler)&StdAwk::systime);
+		ADD_FUNC (ASE_T("systime"),    0, 0, &StdAwk::systime);
-			addFunction (ASE_T("strftime"), 0, 2, (FunctionHandler)&StdAwk::strftime);
+		ADD_FUNC (ASE_T("strftime"),   0, 2, &StdAwk::strftime);
-			addFunction (ASE_T("strfgmtime"), 0, 2, (FunctionHandler)&StdAwk::strfgmtime);
+		ADD_FUNC (ASE_T("strfgmtime"), 0, 2, &StdAwk::strfgmtime);
-			addFunction (ASE_T("system"), 1, 1, (FunctionHandler)&StdAwk::system);
+		ADD_FUNC (ASE_T("system"),     1, 1, &StdAwk::system);
-		}
 
-		return n;
+		return 0;
 	}
 
 	int StdAwk::sin (Return* ret, const Argument* args, size_t nargs)
@@ -228,8 +233,20 @@ namespace ASE
 
 	StdAwk::ssize_t StdAwk::readPipe (Pipe& io, char_t* buf, size_t len) 
 	{ 
-		// TODO: change this implementation...
 		FILE* fp = (FILE*)io.getHandle();
+		ssize_t n = 0;
+
+		while (n < len)
+		{
+			ase_cint_t c = ase_fgetc (fp);
+			if (c == ASE_CHAR_EOF) break;
+
+			buf[n++] = c;
+			if (c == ASE_T('\n')) break;
+		}
+
+		return n;
+		/*
 		if (ase_fgets (buf, len, fp) == ASE_NULL)
 		{
 			if (ferror(fp)) return -1;
@@ -237,6 +254,7 @@ namespace ASE
 		}
 
 		return ase_strlen(buf);
+		*/
 	}
 
 	
@@ -326,8 +344,21 @@ namespace ASE
 
 	StdAwk::ssize_t StdAwk::readFile (File& io, char_t* buf, size_t len) 
 	{
-		// TODO: replace ase_fgets by something else
 		FILE* fp = (FILE*)io.getHandle();
+		ssize_t n = 0;
+
+		while (n < len)
+		{
+			ase_cint_t c = ase_fgetc (fp);
+			if (c == ASE_CHAR_EOF) break;
+
+			buf[n++] = c;
+			if (c == ASE_T('\n')) break;
+		}
+
+		return n;
+
+		/*
 		if (ase_fgets (buf, len, fp) == ASE_NULL)
 		{
 			if (ferror(fp)) return -1;
@@ -335,6 +366,7 @@ namespace ASE
 		}
 
 		return ase_strlen(buf);
+		*/
 	}
 
 	StdAwk::ssize_t StdAwk::writeFile (File& io, char_t* buf, size_t len)

ase/awk/extio.c

@@ -1,5 +1,5 @@
 /*
- * $Id: extio.c,v 1.3 2007/04/30 05:47:33 bacon Exp $
+ * $Id: extio.c,v 1.4 2007/05/18 16:19:20 bacon Exp $
  *
  * {License}
  */
@@ -247,7 +247,44 @@ int ase_awk_readextio (
 			if (n == 0) 
 			{
 				p->in.eof = ase_true;
+
 				if (ASE_STR_LEN(buf) == 0) ret = 0;
+				else if (rs_len >= 2)
+				{
+					/* when RS is multiple characters, it needs to check
+					 * for the match at the end of the input stream as
+					 * the buffer has been appened with the last character
+					 * after the previous matchrex has failed */
+
+					const ase_char_t* match_ptr;
+					ase_size_t match_len;
+
+					ASE_ASSERT (run->global.rs != ASE_NULL);
+
+					n = ase_awk_matchrex (
+						run->awk, run->global.rs, 
+						((run->global.ignorecase)? ASE_AWK_REX_IGNORECASE: 0),
+						ASE_STR_BUF(buf), ASE_STR_LEN(buf), 
+						&match_ptr, &match_len, &run->errnum);
+					if (n == -1)
+					{
+						ret = -1;
+						break;
+					}
+
+					if (n == 1)
+					{
+						/* the match should be found at the end of
+						 * the current buffer */
+						ASE_ASSERT (
+							ASE_STR_BUF(buf) + ASE_STR_LEN(buf) ==
+							match_ptr + match_len);
+
+						ASE_STR_LEN(buf) -= match_len;
+						break;
+					}
+				}
+
 				break;
 			}
 
@@ -320,6 +357,7 @@ int ase_awk_readextio (
 			if (n == -1)
 			{
 				ret = -1;
+				p->in.pos--; /* unread the character in c */
 				break;
 			}
 
@@ -332,6 +370,7 @@ int ase_awk_readextio (
 					match_ptr + match_len);
 
 				ASE_STR_LEN(buf) -= match_len;
+				p->in.pos--; /* unread the character in c */
 				break;
 			}
 		}

ase/change.log

@@ -4,6 +4,10 @@
 * fixed bug (nextofile not allowed in the BEGIN and END block)
 * fixed bug (ASE_MALLOC/ASE_FREE defined wrongly causing test programs
              to crash in the debug mode when compiled with MSVC/C++.
+* fixed bug (the first character of the next record lost when RS are
+             multiple characters)
+* fixed bug (the last field of the last record lost when RS are 
+             multiple characters)
 
 * added awk c++ classes (awk/Awk.hpp awk/Awk.cpp awk/StdAwk.hpp awk/StdAwk.cpp)
 * added c++ test program (test/awk/Awk.cpp)

ase/test/awk/adr-001.awk

@@ -0,0 +1,9 @@
+#!/bin/awk
+
+BEGIN { 
+        RS = "\n\n"; 
+        FS = "\n"; 
+}
+{ 
+	print $1, $NF;         
+}

ase/test/awk/adr-001.out

@@ -0,0 +1,12 @@
+BEGIN {
+	RS = "\n\n";
+	FS = "\n";
+}
+
+{
+	print $1,$NF;
+}
+
+James Brown 012-345-678
+Richie Ren 02-3473-9192
+Toh WeeKung 9102-1203

ase/test/awk/adr-en.data

@@ -0,0 +1,15 @@
+James Brown
+IBM
+Somewhere over the rainbow
+012-345-678
+
+Richie Ren
+Ezsystem
+Taipei Taiwan
+02-3473-9192
+
+Toh WeeKung
+Topaz
+Singapore
+9102-1203
+

ase/test/awk/regress.sh

@@ -21,6 +21,11 @@ run_init()
 		run_script_for_init "$script" "cou-en.data"
 	done
 
+	for script in adr-???.awk
+	do
+		run_script_for_init "$script" "adr-en.data"
+	done
+
 	for script in err-???.awk
 	do
 		run_script_for_init "$script" "err-en.data"
@@ -88,6 +93,24 @@ run_test()
 		fi
 	done
 
+	for script in adr-???.awk
+	do
+		run_script_for_test "$script" "adr-en.data"
+		if [ $? -ne 0 ]
+		then
+			echo "###################################"
+			echo "PROBLEM(S) DETECTED IN $script.".
+			echo "###################################"
+
+			echo "Do you want to abort? [y/n]"
+			read ans
+			if [ "$ans" = "y" -o "$ans" = "Y" ]
+			then
+				return 1
+			fi
+		fi
+	done
+
 	for script in err-???.awk
 	do
 		run_script_for_test "$script" "err-en.data"