String manipulation improvements

std/std.cn

@@ -298,26 +298,7 @@ nproc str_to_nstr @str -> ptr:
   buf
 end
 
-proc streq @str @str -> int:
-  let len data len2 data2;
-  len2 len == if
-    0 1
-    len 0 >
-    while
-      drop
-      let curlen;
-      data curlen ptr+ @8
-      data2 curlen ptr+ @8
-      == let res;
-      curlen 1 + res
-      curlen 1 + len <
-      res and
-    end
-    swap drop
-  else
-    0
-  end
-end
+include str.cn
 
 #if platform Platform.fasm_x86_64_linux ==;
   nproc exec_cmd ptr argv ptr exec:
@@ -341,113 +322,3 @@ end
   end
 #endif
 
-proc str_find @str @str -> int:
-  let len1 data1 len2 data2;
-  if len2 0 > len1 len2 >= and do
-    0
-    1 while
-      if len2 over data1 +ptr len2 data2 streq do
-        0
-      else
-        1 +
-        dup len1 len2 - <=
-      end
-    end
-    if dup len1 len2 - 1 + == do
-      drop -1
-    end
-  else
-    -1
-  end
-end
-
-proc str_rfind @str @str -> int:
-  let len1 data1 len2 data2;
-  if len2 0 > len1 len2 >= and do
-    len1 len2 -
-    1 while
-      if len2 over data1 +ptr len2 data2 streq do
-        0
-      else
-        1 -
-        dup 0 >=
-      end
-    end
-  else
-    -1
-  end
-end
-
-proc str_count @str @str -> int:
-  var counter int
-  let len1 data1 len2 data2;
-  0 
-  len1 0 != len1 len2 >= and while
-    if len2 over data1 +ptr len2 data2 streq do
-      *counter inc
-      len2 +
-    else
-      1 +
-    end
-    dup len1 len2 - <=
-  end drop
-  counter
-end
-
-nproc str_slice1 int start @str -> @str:
-  if start len >= start 0 < or do
-    "Index out of range for the start of the slice: " puts
-    start print
-    1 exit
-  end
-  len start - data start ptr+
-end
-
-nproc str_slice2 int start int _end @str -> @str:
-  if start len >= start 0 < or do
-    "Index out of range for the start of the string slice: " puts
-    start print
-    1 exit
-  end
-  if _end len > _end 0 < or do
-    "Index out of range for the end of the string slice: " puts
-    _end print
-    1 exit
-  end
-  _end start - data start ptr+
-end
-
-proc str_split @str @str -> [DYNAMIC_ARRAY_SIZE] str:
-  let len data sep_len sep_data;
-  len data sep_len sep_data str_count
-  sizeoftype *str * malloc let res;
-  sizeoftype str malloc drop
-  res
-end
-
-struct CursoredString
-  str data
-  int cursor
-
-  sproc __init__ @str:
-    str !self.data
-    0 !self.cursor
-  end
-
-  nproc rest -> @str:
-    self.data.len self.cursor - self.data.data self.cursor ptr+
-  end
-
-  // Jumps to the CHARACTER AFTER the sub-string
-  // Returns a bool
-  // true - jumped
-  // false - no next line found
-  nproc jump_to @str -> int:
-    self.rest len data str_find
-    let offset;
-    offset -1 != offset len + self.data.len != and
-    dup if
-      *self.cursor offset len + incby
-    end
-  end
-end

std/str.cn

@@ -0,0 +1,211 @@
+include core.cn
+
+proc streq @str @str -> int:
+  let len data len2 data2;
+  len2 len == if
+    var i int = 0;
+    var res int = 1;
+    while len i - 8 >=
+          res and do
+      data i ptr+ @ data2 i ptr+ @ == res and !res
+      *i 8 incby
+    end
+    while i len <
+          res and do
+      data i ptr+ @8 data2 i ptr+ @8 == res and !res
+      *i inc
+    end
+    res
+  else
+    0
+  end
+end
+
+nproc z_array @str -> [DYNAMIC_ARRAY_SIZE] int:
+  var l int = 0;
+  var r int = 0;
+  len sizeoftype int * malloc ([DYNAMIC_ARRAY_SIZE]) int let z;
+  len 0 z *[] !
+  var i int = 1;
+  while i len < do
+    var done int = 0;
+    if i r <= do
+      if i l - z [] i + r <= do
+        i l - z [] i z *[] ! 
+        1 !done
+      end
+    else
+      i 1 - !r
+    end
+    if done not do
+      i !l
+      if r len 1 - < do
+        data r 1 + ptr+ @8 data r 1 + i - ptr+ @8 ==
+      else 0 end
+      while
+        *r inc
+        if r len 1 - < do
+          data r 1 + ptr+ @8 data r 1 + i - ptr+ @8 ==
+        else 0 end
+      end
+      r i - 1 + i z *[] !
+    end
+    *i inc
+  end
+  z
+end
+
+proc str_find @str @str -> int:
+  let len1 data1 len2 data2;
+  if len2 0 > len1 len2 >= and do
+    if len2 16 < len1 len2 - 16 < or do
+      0
+      1 while
+        if len2 over data1 +ptr len2 data2 streq do
+          0
+        else
+          1 +
+          dup len1 len2 - <=
+        end
+      end
+      if dup len1 len2 - 1 + == do
+        drop -1
+      end
+    else
+      len1 len2 + malloc let data3;
+      len2 data2 data3 memcpy
+      len1 data1 data3 len2 ptr+ memcpy
+      len1 len2 + data3 z_array let z;
+      data3 free
+      var i int = 0;
+      var res int = 0 1 -;
+      while i len1 < res -1 == and do
+        if i len2 + z [] len2 >= do
+          i !res
+        end
+        *i inc
+      end
+      z free
+      res
+    end
+  else -1 end
+end
+
+proc str_rfind @str @str -> int:
+  let len1 data1 len2 data2;
+  if len2 0 > len1 len2 >= and do
+    if len2 16 < len1 len2 - 16 < or do
+      len1 len2 -
+      1 while
+        if len2 over data1 +ptr len2 data2 streq do
+          0
+        else
+          1 -
+          dup 0 >=
+        end
+      end
+    else
+      len1 len2 + malloc let data3;
+      len2 data2 data3 memcpy
+      len1 data1 data3 len2 ptr+ memcpy
+      len1 len2 + data3 z_array let z;
+      data3 free
+      var i int
+      len1 len2 - !i
+      var res int = 0 1 -;
+      while i len1 < res -1 == and do
+        if i len2 + z [] len2 >= do
+          i !res
+        end
+        *i dec
+      end
+      z free
+      res
+    end
+  else
+    -1
+  end
+end
+
+proc str_count @str @str -> int:
+  var counter int = 0;
+  let len1 data1 len2 data2;
+  if len2 16 < len1 len2 - 16 < or do
+    0 
+    len1 0 != len1 len2 >= and while
+      if len2 over data1 +ptr len2 data2 streq do
+        *counter inc
+        len2 +
+      else
+        1 +
+      end
+      dup len1 len2 - <=
+    end drop
+  counter print
+  else
+    len1 len2 + malloc let data3;
+    len2 data2 data3 memcpy
+    len1 data1 data3 len2 ptr+ memcpy
+    len1 len2 + data3 z_array let z;
+    data3 free
+    var i int = 0;
+    while i len1 len2 - <= do
+      if i len2 + z [] len2 >= do
+        *counter inc
+        *i len2 incby
+      else *i inc end
+    end
+    z free
+  end
+  counter
+end
+
+nproc str_slice1 int start @str -> @str:
+  if start len >= start 0 < or do
+    "Index out of range for the start of the slice: " puts
+    start print
+    1 exit
+  end
+  len start - data start ptr+
+end
+
+nproc str_slice2 int start int _end @str -> @str:
+  if start len >= start 0 < or do
+    "Index out of range for the start of the string slice: " puts
+    start print
+    1 exit
+  end
+  if _end len > _end 0 < or do
+    "Index out of range for the end of the string slice: " puts
+    _end print
+    1 exit
+  end
+  _end start - data start ptr+
+end
+
+struct CursoredString
+  str data
+  int cursor
+
+  sproc __init__ @str:
+    str !self.data
+    0 !self.cursor
+  end
+
+  nproc rest -> @str:
+    self.data.len self.cursor - self.data.data self.cursor ptr+
+  end
+
+  // Jumps to the CHARACTER AFTER the sub-string
+  // Returns a bool
+  // true - jumped
+  // false - no next line found
+  nproc jump_to @str -> int:
+    self.rest len data str_find
+    let offset;
+    offset -1 != offset len + self.data.len != and
+    dup if
+      *self.cursor offset len + incby
+    end
+  end
+end