From 796add0ee2999f7434bf10f0275ed89ce254dd49 Mon Sep 17 00:00:00 2001
From: Irmen de Jong <irmen@razorvine.net>
Date: Wed, 13 Dec 2023 00:28:34 +0100
Subject: [PATCH] add string.isspace and string.isprint

---
 compiler/res/prog8lib/string.p8         | 46 ++++++++++++++++++++++---
 compiler/res/prog8lib/virtual/string.p8 | 10 +++++-
 docs/source/libraries.rst               |  8 ++++-
 docs/source/todo.rst                    |  5 +++
 4 files changed, 63 insertions(+), 6 deletions(-)

diff --git a/compiler/res/prog8lib/string.p8 b/compiler/res/prog8lib/string.p8
index ad9d0e2d4..6bc4cdded 100644
--- a/compiler/res/prog8lib/string.p8
+++ b/compiler/res/prog8lib/string.p8
@@ -384,7 +384,7 @@ fail    clc             ; yes, no match found, return with c=0
         }}
     }
 
-    asmsub isdigit(ubyte character @A) -> bool @Pc {
+    asmsub isdigit(ubyte petsciichar @A) -> bool @Pc {
         %asm {{
             cmp  #'0'
             bcs  +
@@ -398,7 +398,7 @@ fail    clc             ; yes, no match found, return with c=0
         }}
     }
 
-    asmsub isupper(ubyte character @A) -> bool @Pc {
+    asmsub isupper(ubyte petsciichar @A) -> bool @Pc {
         ; shifted petscii has 2 ranges that contain the upper case letters...
         %asm {{
             cmp  #97
@@ -418,7 +418,7 @@ _yes        sec
         }}
     }
 
-    asmsub islower(ubyte character @A) -> bool @Pc {
+    asmsub islower(ubyte petsciichar @A) -> bool @Pc {
         %asm {{
             cmp  #'a'
             bcs  +
@@ -432,7 +432,7 @@ _yes        sec
         }}
     }
 
-    asmsub isletter(ubyte character @A) -> bool @Pc {
+    asmsub isletter(ubyte petsciichar @A) -> bool @Pc {
         %asm {{
             jsr  islower
             bcs  +
@@ -440,4 +440,42 @@ _yes        sec
 +           rts
         }}
     }
+
+    asmsub isspace(ubyte petsciichar @A) -> bool @Pc {
+        %asm {{
+            cmp  #32
+            beq  +
+            cmp  #13
+            beq  +
+            cmp  #9
+            beq  +
+            cmp  #10
+            beq  +
+            cmp  #141
+            beq  +
+            cmp  #160
+            beq  +
+            clc
+            rts
++           sec
+            rts
+        }}
+    }
+
+    asmsub isprint(ubyte petsciichar @A) -> bool @Pc {
+        %asm {{
+            cmp  #160
+            bcc  +
+            rts
++           cmp  #32
+            bcs  +
+            rts
++           cmp  #128
+            bcc  +
+            clc
+            rts
++           sec
+            rts
+        }}
+    }
 }
diff --git a/compiler/res/prog8lib/virtual/string.p8 b/compiler/res/prog8lib/virtual/string.p8
index d01155434..487df01ce 100644
--- a/compiler/res/prog8lib/virtual/string.p8
+++ b/compiler/res/prog8lib/virtual/string.p8
@@ -190,4 +190,12 @@ string {
     sub isletter(ubyte character) -> bool {
         return islower(character) or isupper(character)
     }
- }
+
+    sub isspace(ubyte character) -> bool {
+        return character in [32, 13, 9, 10, 141, 160]
+    }
+
+    sub isprint(ubyte character) -> bool {
+        return character>=32 and character<=127 or character>=160
+    }
+}
diff --git a/docs/source/libraries.rst b/docs/source/libraries.rst
index f67310fce..187db6543 100644
--- a/docs/source/libraries.rst
+++ b/docs/source/libraries.rst
@@ -275,7 +275,13 @@ Provides string manipulation routines.
     Returns boolean if the character is a numerical digit 0-9
 
 ``islower (char)``, ``isupper (char)``, ``isletter (char)``
-    Returns boolean if the character is a shifted-PETSCII lowercase letter, uppercase letter, or any letter.
+    Returns true if the character is a shifted-PETSCII lowercase letter, uppercase letter, or any letter, respectively.
+
+``isspace (char)``
+    Returns true if the PETSCII character is a whitespace (tab, space, return, and shifted versions)
+
+``isprint (char)``
+    Returns true if the PETSCII character is a "printable" character (space or any visible symbol)
 
 ``startswith (string, prefix) -> bool``
     Returns true if string starts with prefix, otherwise false
diff --git a/docs/source/todo.rst b/docs/source/todo.rst
index 1e3191a92..b6a57bf39 100644
--- a/docs/source/todo.rst
+++ b/docs/source/todo.rst
@@ -2,6 +2,11 @@
 TODO
 ====
 
+- optimize if-else expressions whose condition returns the boolean status in a status register to use a branch opcode instead of a comparison against 0
+- fix "can't use Z or N flags as return 'values'" in 6502 codegen?
+
+- merge branch optimize-st  for some optimizations regardign SymbolTable use
+
 - [on branch: call-pointers] allow calling a subroutine via a pointer variable (indirect JSR, optimized form of callfar())
    modify programs (shell, paint) that now use callfar