More fixes in MH format parser

* mh/mh_fmtgram.y (codegen_node): Generate mhop_printlit instruction
for fmtnode_print node with a literal or number argument.
* mh/mh_format.c (mh_fvm_run): Handle mhop_printlit
* mh/mh_format.h (mhop_printlit): New opcode.
* mh/tests/fmtcomp.at: Update.
* mh/tests/fmtfunc.at: Add more tests.

5 files changed, 135 insertions, 44 deletions

diff --git a/mh/mh_fmtgram.y b/mh/mh_fmtgram.y
index 523bc5c7f..04a890e3d 100644
--- a/mh/mh_fmtgram.y
+++ b/mh/mh_fmtgram.y
@@ -703,13 +703,13 @@ yylex_initial (void)
 	  unput (c);
 	  return token_fmtspec (MH_FMT_DEFAULT);
 	default:
 	  return bogus ("component or function name expected");
       }
     }
-
+  
   c = peek ();
   
   if (c == 0)
     return 0;
 
   while ((c = input ()) != 0)
@@ -1389,22 +1389,38 @@ codegen_node (struct mh_format *fmt, struct node *node)
 {
   if (!node)
     return;
   switch (node->nodetype)
     {
     case fmtnode_print:
-      codegen_node (fmt, node->v.prt.arg);
-      if (node->v.prt.fmtspec)
+      if (node->v.prt.arg->nodetype == fmtnode_literal)
 	{
-	  emit_opcode (fmt, mhop_fmtspec);
-	  emit_instr (fmt, (mh_instr_t) (long) node->v.prt.fmtspec);
+	  emit_opcode (fmt, mhop_printlit);
+	  emit_string (fmt, node->v.prt.arg->v.str);
+	}
+      else if (node->v.prt.arg->nodetype == fmtnode_number)
+	{
+	  char *s;
+	  emit_opcode (fmt, mhop_printlit);
+	  mu_asprintf (&s, "%ld", node->v.prt.arg->v.num);
+	  emit_string (fmt, s);
+	  free (s);
+	}
+      else
+	{
+	  codegen_node (fmt, node->v.prt.arg);
+	  if (node->v.prt.fmtspec)
+	    {
+	      emit_opcode (fmt, mhop_fmtspec);
+	      emit_instr (fmt, (mh_instr_t) (long) node->v.prt.fmtspec);
+	    }
+	  
+	  if (node->v.prt.arg->datatype != mhtype_none)
+	    emit_opcode_typed (fmt, node->v.prt.arg->datatype,
+			       mhop_printn, mhop_prints);
 	}
-
-      if (node->v.prt.arg->datatype != mhtype_none)
-	emit_opcode_typed (fmt, node->v.prt.arg->datatype,
-			   mhop_printn, mhop_prints);
       break;
 
     case fmtnode_literal:
       emit_opcode (fmt, mhop_sets);
       emit_instr (fmt, (mh_instr_t) (long) R_REG);
       emit_string (fmt, node->v.str);
diff --git a/mh/mh_format.c b/mh/mh_format.c
index 9b19686e3..914a85b1e 100644
--- a/mh/mh_format.c
+++ b/mh/mh_format.c
@@ -630,12 +630,21 @@ mh_fvm_run (mh_fvm_t mach, mu_message_t msg, size_t msgno)
 	  break;
 
 	case mhop_prints:
 	  format_str (mach, mach->str[R_REG].ptr);
 	  break;
 
+	case mhop_printlit:
+	  {
+	    size_t skip = MHI_NUM (mach->prog[mach->pc++]);
+	    char const *str = MHI_STR (mach->prog[mach->pc]);
+	    format_str (mach, str);
+	    mach->pc += skip;
+	  }
+	  break;
+	  
 	case mhop_fmtspec:
 	  mach->fmtflags = MHI_NUM (mach->prog[mach->pc++]);
 	  break;
 
 	default:
 	  mu_error (_("INTERNAL ERROR: Unknown opcode: %x"), opcode);
@@ -731,13 +740,13 @@ builtin_size (struct mh_fvm *mach)
     mach->num[R_REG] = 0;
 }
 
 static void
 builtin_strlen (struct mh_fvm *mach)
 {
-  mach->num[R_REG] = mh_string_length (&mach->str[R_ARG]);
+  mach->num[R_REG] = mh_string_length (&mach->str[R_REG]);
 }
 
 static void
 builtin_width (struct mh_fvm *mach)
 {
   mach->num[R_REG] = mach->width;
@@ -2049,12 +2058,21 @@ mh_format_dump_disass (mh_format_t fmt)
 	  printf ("printn");
 	  break;
 	  
 	case mhop_prints:
 	  printf ("prints");
 	  break;
+
+	case mhop_printlit:
+	  {
+	    size_t skip = MHI_NUM (prog[pc++]);
+	    char const *str = MHI_STR (prog[pc]);
+	    pc += skip;
+	    printf ("printlit \"%s\"", str);
+	  }
+	  break;
 	  
 	case mhop_fmtspec:
 	  {
 	    int fmtspec = MHI_NUM (prog[pc++]);
 	    printf ("fmtspec ");
 	    mh_print_fmtspec (fmtspec);
diff --git a/mh/mh_format.h b/mh/mh_format.h
index afa94a2eb..8e2578683 100644
--- a/mh/mh_format.h
+++ b/mh/mh_format.h
@@ -75,12 +75,16 @@ enum mh_opcode
   /* Print num reg */
   mhop_printn,
 
   /* Print str reg */
   mhop_prints,
 
+  /* Print literal
+     Format: mhop_printlit length string */
+  mhop_printlit,
+  
   /* Set format specification.
      Format: mhop_fmtspec number */
   mhop_fmtspec,
 };    
 
 enum regid { R_REG, R_ARG };
diff --git a/mh/tests/fmtcomp.at b/mh/tests/fmtcomp.at
index eb21a290c..253ecd9a5 100644
--- a/mh/tests/fmtcomp.at
+++ b/mh/tests/fmtcomp.at
@@ -28,15 +28,14 @@ AT_CHECK([fmtcheck -disass -dump -format '$2'
 [$3],[$4])
 AT_CLEANUP])
 
 FMTCOMP([literal],
 [text],
 [PRINT("text")
- 0001: sets reg, "text"
- 0007: prints
- 0008: stop
+ 0001: printlit "text"
+ 0006: stop
 ])
 
 FMTCOMP([component],
 [%{from}],
 [PRINT(COMPONENT.from)
  0001: ldcomp reg, "from"
@@ -136,36 +135,33 @@ FMTCOMP([nested function calls with typecast],
 FMTCOMP([simple conditional],
 [%<{replied}-%>],
 [IF (COMPONENT.replied) THEN
   PRINT("-")
 FI
  0001: ldcomp reg, "replied"
- 0008: brzs 18
- 0010: sets reg, "-"
- 0015: prints
- 0016: branch 26
- 0018: setn reg, 0
- 0021: sets reg, ""
- 0026: stop
+ 0008: brzs 16
+ 0010: printlit "-"
+ 0014: branch 24
+ 0016: setn reg, 0
+ 0019: sets reg, ""
+ 0024: stop
 ],[],[-format])
 
 FMTCOMP([if-else],
 [%<{replied}-%|+%>],
 [IF (COMPONENT.replied) THEN
   PRINT("-")
 ELSE
   PRINT("+")
 FI
  0001: ldcomp reg, "replied"
- 0008: brzs 18
- 0010: sets reg, "-"
- 0015: prints
- 0016: branch 24
- 0018: sets reg, "+"
- 0023: prints
- 0024: stop
+ 0008: brzs 16
+ 0010: printlit "-"
+ 0014: branch 20
+ 0016: printlit "+"
+ 0020: stop
 ])
 
 FMTCOMP([if-elsif-else],
 [%<{replied}-%?{encrypted}E%| %>],
 [IF (COMPONENT.replied) THEN
   PRINT("-")
@@ -174,24 +170,21 @@ ELSE
     PRINT("E")
   ELSE
     PRINT(" ")
   FI
 FI
  0001: ldcomp reg, "replied"
- 0008: brzs 18
- 0010: sets reg, "-"
- 0015: prints
- 0016: branch 41
- 0018: ldcomp reg, "encrypted"
- 0025: brzs 35
- 0027: sets reg, "E"
- 0032: prints
- 0033: branch 41
- 0035: sets reg, " "
- 0040: prints
- 0041: stop
+ 0008: brzs 16
+ 0010: printlit "-"
+ 0014: branch 35
+ 0016: ldcomp reg, "encrypted"
+ 0023: brzs 31
+ 0025: printlit "E"
+ 0029: branch 35
+ 0031: printlit " "
+ 0035: stop
 ])
 
 # The example below is taken from mh-format(1), subsection
 # "Other Hints and Tips".
 # It issues a spurious PRINT instruction, which is intended.  The following
 # testcase uses a fixed version. 
diff --git a/mh/tests/fmtfunc.at b/mh/tests/fmtfunc.at
index 65107f3ed..a7d72bce5 100644
--- a/mh/tests/fmtfunc.at
+++ b/mh/tests/fmtfunc.at
@@ -81,20 +81,18 @@ FMTFUNC([void],
 [
 
 ],
 [])
 
 FMTFUNC([strlen],
-[%(lit string)
-%(strlen)
+[%(lit string)=%(strlen)
 ],
 [
 
 ],
-[string
-6
+[string=6
 ])
 
 # FIXME: Not quite sure: perhaps it should report 80
 FMTFUNC([width],
 [%(width)
 ],
@@ -218,13 +216,13 @@ AT_CHECK([MH_TEST=defined fmtcheck -form input.fmt -width 80 message
 [0],
 [defined
 ])
 AT_CLEANUP
 
 AT_SETUP([profile])
-AT_KEYWORDS([fmtfunc format])
+AT_KEYWORDS([fmtfunc format])xo
 AT_DATA([input.fmt],[%(profile Local-Mailbox)
 %(profile undefined)
 %(profile moreproc)
 ])
 AT_DATA([message],[
 
@@ -305,9 +303,71 @@ FMTFUNC([compval],
 X-Level: 8
 
 ],
 [8
 ])
 
+# FIXME: This relies on LC_ALL=C
+FMTFUNC([decode],
+[%(decode{Subject})
+],
+[Subject: =?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?= ... =?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?=
+
+body
+],
+[If you can read this yo ... u understand the example.
+])
+
+# FIXME: unquote
+
+FMTFUNC([trim],
+[%(void(lit trailing    ))%(strlen)
+%(trim)%(strlen)
+],
+[
+
+],
+[12
+8
+])
+
+FMTFUNC([putstr],
+[%(void(lit string))%(putstr)
+%(putstr test)
+%(putstr{from})
+%(putstr(msg))
+],
+[From: foo@example.org
+
+],
+[string
+test
+foo@example.org
+1
+])
+
+# FIXME: putstrf
+
+FMTFUNC([putnum],
+[%(void(num 10))%(putnum)
+%(putnum 15)
+%(putnum{X-Number})
+%(putstr(msg))
+],
+[From: foo@example.org
+X-Number: 8
+
+body
+],
+[10
+15
+8
+1
+])
+
+# FIXME: putnumf
+
+# FIXME: putlit
+
 
 
 m4_popdef([FMTFUNC])
\ No newline at end of file