From c8bf6529829949d87ec87e5f4ee256fc0de08cde Mon Sep 17 00:00:00 2001
From: Joshua Bell <inexorabletash@gmail.com>
Date: Sat, 5 Jan 2019 11:29:54 -0800
Subject: [PATCH] ca65: Add string_escapes feature. Resolves #535

---
 doc/ca65.sgml      | 18 ++++++++++++++++++
 src/ca65/feature.c |  2 ++
 src/ca65/feature.h |  1 +
 src/ca65/global.c  |  2 +-
 src/ca65/global.h  |  1 +
 src/ca65/scanner.c | 37 +++++++++++++++++++++++++++++++++++++
 6 files changed, 60 insertions(+), 1 deletion(-)

diff --git a/doc/ca65.sgml b/doc/ca65.sgml
index 52e9634ae..bc22ba354 100644
--- a/doc/ca65.sgml
+++ b/doc/ca65.sgml
@@ -2831,6 +2831,24 @@ Here's a list of all control commands and a description, what they do:
     removing the lines with the assignments may also be an option when porting
     code written for older assemblers).
 
+  <tag><tt>string_escapes</tt><label id="string_escapes"></tag>
+
+    Allow C-style backslash escapes within string constants to embed
+    special characters. The following escapes are accepted:
+
+<itemize>
+<item><tt>\\</tt> backslash (<tt>$5C</tt>)
+<item><tt>\'</tt> single quote (<tt>$27</tt>)
+<item><tt>\&quot;</tt> double quote (<tt>$22</tt>)
+<item><tt>\t</tt> tab (<tt>$09</tt>)
+<item><tt>\r</tt> carriage return (<tt>$0D</tt>)
+<item><tt>\n</tt> newline (<tt>$0A</tt>)
+<item><tt>\xNN</tt>  (<tt>$NN</tt>)
+</itemize>
+
+    Note that <tt>\n</tt> maps to ASCII <tt>$0A</tt>, not a platform specific
+    line ending character.
+
   <tag><tt>ubiquitous_idents</tt><label id="ubiquitous_idents"></tag>
 
     Allow the use of instructions names as names for macros and symbols. This
diff --git a/src/ca65/feature.c b/src/ca65/feature.c
index 0fb766b6f..b11345338 100644
--- a/src/ca65/feature.c
+++ b/src/ca65/feature.c
@@ -65,6 +65,7 @@ static const char* const FeatureKeys[FEAT_COUNT] = {
     "underline_in_numbers",
     "addrsize",
     "bracket_as_indirect",
+    "string_escapes",
 };
 
 
@@ -123,6 +124,7 @@ feature_t SetFeature (const StrBuf* Key)
         case FEAT_UNDERLINE_IN_NUMBERS:       UnderlineInNumbers= 1;    break;
         case FEAT_ADDRSIZE:                   AddrSize          = 1;    break;
         case FEAT_BRACKET_AS_INDIRECT:        BracketAsIndirect = 1;    break;
+        case FEAT_STRING_ESCAPES:             StringEscapes     = 1;    break;
         default:                         /* Keep gcc silent */          break;
     }
 
diff --git a/src/ca65/feature.h b/src/ca65/feature.h
index 050c197f0..876f3c4a8 100644
--- a/src/ca65/feature.h
+++ b/src/ca65/feature.h
@@ -67,6 +67,7 @@ typedef enum {
     FEAT_UNDERLINE_IN_NUMBERS,
     FEAT_ADDRSIZE,
     FEAT_BRACKET_AS_INDIRECT,
+    FEAT_STRING_ESCAPES,
 
     /* Special value: Number of features available */
     FEAT_COUNT
diff --git a/src/ca65/global.c b/src/ca65/global.c
index 31e599f00..a216fc406 100644
--- a/src/ca65/global.c
+++ b/src/ca65/global.c
@@ -66,6 +66,7 @@ unsigned char DbgSyms            = 0;   /* Add debug symbols */
 unsigned char LineCont           = 0;   /* Allow line continuation */
 unsigned char LargeAlignment     = 0;   /* Don't warn about large alignments */
 unsigned char RelaxChecks        = 0;   /* Relax a few assembler checks */
+unsigned char StringEscapes      = 0;   /* Allow C-style escapes in strings */
 
 /* Emulation features */
 unsigned char DollarIsPC         = 0;   /* Allow the $ symbol as current PC */
@@ -84,4 +85,3 @@ unsigned char ForceRange         = 0;   /* Force values into expected range */
 unsigned char UnderlineInNumbers = 0;   /* Allow underlines in numbers */
 unsigned char AddrSize           = 0;   /* Allow .ADDRSIZE function */
 unsigned char BracketAsIndirect  = 0;   /* Use '[]' not '()' for indirection */
-
diff --git a/src/ca65/global.h b/src/ca65/global.h
index 397d9221b..77d90beca 100644
--- a/src/ca65/global.h
+++ b/src/ca65/global.h
@@ -68,6 +68,7 @@ extern unsigned char    DbgSyms;            /* Add debug symbols */
 extern unsigned char    LineCont;           /* Allow line continuation */
 extern unsigned char    LargeAlignment;     /* Don't warn about large alignments */
 extern unsigned char    RelaxChecks;        /* Relax a few assembler checks */
+extern unsigned char    StringEscapes;      /* Allow C-style escapes in strings */
 
 /* Emulation features */
 extern unsigned char    DollarIsPC;         /* Allow the $ symbol as current PC */
diff --git a/src/ca65/scanner.c b/src/ca65/scanner.c
index 7c291666c..965fa3581 100644
--- a/src/ca65/scanner.c
+++ b/src/ca65/scanner.c
@@ -794,6 +794,43 @@ static void ReadStringConst (int StringTerm)
             break;
         }
 
+        if (C == '\\' && StringEscapes) {
+            NextChar ();
+
+            switch (C) {
+                case EOF:
+                    Error ("Unterminated escape sequence in string constant");
+                    break;
+                case '\\':
+                case '\'':
+                case '"':
+                    break;
+                case 't':
+                    C = '\x09';
+                    break;
+                case 'r':
+                    C = '\x0D';
+                    break;
+                case 'n':
+                    C = '\x0A';
+                    break;
+                case 'x':
+                    NextChar ();
+                    if (IsXDigit (C)) {
+                        char high_nibble = DigitVal (C) << 4;
+                        NextChar ();
+                        if (IsXDigit (C)) {
+                            C = high_nibble | DigitVal (C);
+                            break;
+                        }
+                    }
+                    /* otherwise, fall through */
+                default:
+                    Error ("Unsupported escape sequence in string constant");
+                    break;
+            }
+        }
+
         /* Append the char to the string */
         SB_AppendChar (&CurTok.SVal, C);
 
-- 
2.39.5