Changeset 416

Timestamp:

06/26/10 12:37:34 (3 years ago)

Author:

cameron

Message:

Prevalidation of XML name syntax

Location:

proto/parabix2

Files:

• byteclass.py (modified) (5 diffs)
• parabix2.py (modified) (1 diff)

proto/parabix2/byteclass.py

@@ -7 +7 @@
 #
 # Robert D. Cameron
-# August 14, 2009
-#
+# June 26, 2010 - Add definitions for lex.ASCII_name_char,
+# lex.ASCII_name_start, lex.colon
+# 
 
 class u8_streams ():
@@ -46 +47 @@
 class lexical_streams ():
   RefStart = 0
-  Semicolon = 0
+  Semicolon = 0  
+  Colon = 0
   LAngle = 0
   RAngle = 0
@@ -60 +62 @@
   Hash = 0
   x = 0
+  ASCII_name_start = 0
+  ASCII_name_char = 0
   NameScan = 0
   Digit = 0
@@ -120 +124 @@
   temp28 = (temp9 &~ temp13);
   lex.x = (temp27 & temp28);
-  temp29 = (temp18 & temp23);
-  temp30 = (temp29 | lex.Hyphen);
-  temp31 = (temp3 & temp15);
-  temp32 = (temp30 | temp31);
+  temp29 = (temp9 & temp5);
+  lex.Colon = (temp8 & temp29);
+  temp30 = (temp18 & temp23);
+  temp31 = (temp30 | lex.Colon);
+  temp32 = (temp16 &~ bit[2]);
   temp33 = (bit[5] | temp10);
   temp34 = (bit[4] & temp33);
-  temp35 = (temp8 &~ temp34);
-  temp36 = (temp32 | temp35);
-  temp37 = (temp16 &~ bit[2]);
-  temp38 = (~temp34);
-  temp39 = (temp21 | temp13);
-  temp40 = ((bit[3] & temp38)|(~(bit[3]) & temp39));
-  temp41 = (temp37 & temp40);
-  temp42 = (temp36 | temp41);
-  temp43 = (temp16 & bit[2]);
-  temp44 = (temp43 & temp40);
-  temp45 = (temp42 | temp44);
-  lex.NameScan = (temp45 | bit[0]);
-  temp46 = (temp1 | bit[2]);
-  control.x00_x1F = (~temp46);
-  temp47 = (bit[2] | bit[3]);
-  temp48 = (temp1 | temp47);
-  control.CR = (temp20 &~ temp48);
-  temp49 = (temp9 & temp5);
-  control.LF = (temp49 &~ temp48);
-  temp50 = (temp9 & temp19);
-  control.HT = (temp50 &~ temp48);
-  control.SP = (temp3 &~ temp39);
-  temp51 = (temp20 | temp49);
-  temp52 = (temp51 | temp50);
-  temp53 = (temp52 &~ temp48);
-  lex.WS = (temp53 | control.SP);
-  temp54 = (bit[5] | bit[6]);
-  temp55 = (bit[4] & temp54);
-  lex.Digit = (temp8 &~ temp55);
-  temp56 = (temp16 &~ temp47);
-  temp57 = (temp56 &~ bit[4]);
-  temp58 = (~temp10);
-  temp59 = ((bit[5] & temp58)|(~(bit[5]) & temp13));
-  temp60 = (temp57 & temp59);
-  temp61 = (lex.Digit | temp60);
-  temp62 = (temp16 & temp2);
-  temp63 = (temp62 &~ bit[4]);
-  temp64 = (temp63 & temp59);
-  lex.Hex = (temp61 | temp64);
+  temp35 = (~temp34);
+  temp36 = (temp21 | temp13);
+  temp37 = ((bit[3] & temp35)|(~(bit[3]) & temp36));
+  temp38 = (temp32 & temp37);
+  temp39 = (temp31 | temp38);
+  temp40 = (temp16 & bit[2]);
+  temp41 = (temp40 & temp37);
+  lex.ASCII_name_start = (temp39 | temp41);
+  temp42 = (temp30 | lex.Hyphen);
+  temp43 = (temp3 & temp15);
+  temp44 = (temp42 | temp43);
+  temp45 = (temp8 &~ temp34);
+  temp46 = (temp44 | temp45);
+  temp47 = (temp46 | temp38);
+  lex.ASCII_name_char = (temp47 | temp41);
+  lex.NameScan = (lex.ASCII_name_char | bit[0]);
+  temp48 = (temp1 | bit[2]);
+  control.x00_x1F = (~temp48);
+  temp49 = (bit[2] | bit[3]);
+  temp50 = (temp1 | temp49);
+  control.CR = (temp20 &~ temp50);
+  control.LF = (temp29 &~ temp50);
+  temp51 = (temp9 & temp19);
+  control.HT = (temp51 &~ temp50);
+  control.SP = (temp3 &~ temp36);
+  temp52 = (temp20 | temp29);
+  temp53 = (temp52 | temp51);
+  temp54 = (temp53 &~ temp50);
+  lex.WS = (temp54 | control.SP);
+  temp55 = (bit[5] | bit[6]);
+  temp56 = (bit[4] & temp55);
+  lex.Digit = (temp8 &~ temp56);
+  temp57 = (temp16 &~ temp49);
+  temp58 = (temp57 &~ bit[4]);
+  temp59 = (~temp10);
+  temp60 = ((bit[5] & temp59)|(~(bit[5]) & temp13));
+  temp61 = (temp58 & temp60);
+  temp62 = (lex.Digit | temp61);
+  temp63 = (temp16 & temp2);
+  temp64 = (temp63 &~ bit[4]);
+  temp65 = (temp64 & temp60);
+  lex.Hex = (temp62 | temp65);
   u8.unibyte = (~bit[0]);
   u8.prefix = (bit[0] & bit[1]);
@@ -171 +179 @@
   u8.prefix4 = (u8.prefix & temp7);
   u8.suffix = (bit[0] &~ bit[1]);
-  temp65 = (u8.prefix &~ temp47);
-  temp66 = (temp21 | bit[6]);
-  temp67 = (temp65 &~ temp66);
-  temp68 = (bit[5] & temp13);
-  temp69 = (bit[4] | temp68);
-  temp70 = (u8.prefix4 & temp69);
-  u8.badprefix = (temp67 | temp70);
-  u8.xE0 = (u8.prefix3 &~ temp39);
+  temp66 = (u8.prefix &~ temp49);
+  temp67 = (temp21 | bit[6]);
+  temp68 = (temp66 &~ temp67);
+  temp69 = (bit[5] & temp13);
+  temp70 = (bit[4] | temp69);
+  temp71 = (u8.prefix4 & temp70);
+  u8.badprefix = (temp68 | temp71);
+  u8.xE0 = (u8.prefix3 &~ temp36);
   u8.xED = (u8.prefix3 & temp20);
-  u8.xF0 = (u8.prefix4 &~ temp39);
-  temp71 = (temp4 &~ temp13);
-  u8.xF4 = (u8.prefix4 & temp71);
+  u8.xF0 = (u8.prefix4 &~ temp36);
+  temp72 = (temp4 &~ temp13);
+  u8.xF4 = (u8.prefix4 & temp72);
   u8.xA0_xBF = (u8.suffix & bit[2]);
   u8.x80_x9F = (u8.suffix &~ bit[2]);
-  u8.x90_xBF = (u8.suffix & temp47);
-  u8.x80_x8F = (u8.suffix &~ temp47);
+  u8.x90_xBF = (u8.suffix & temp49);
+  u8.x80_x8F = (u8.suffix &~ temp49);
   u8.xEF = (u8.prefix3 & temp23);
-  temp72 = (u8.suffix & temp7);
-  u8.xBF = (temp72 & temp23);
-  u8.xBE = (temp72 & temp15);
+  temp73 = (u8.suffix & temp7);
+  u8.xBF = (temp73 & temp23);
+  u8.xBE = (temp73 & temp15);
+
   return (u8, control, lex)
 

proto/parabix2/parabix2.py

@@ -554 +554 @@
 
 
+#
+# Fully validate ASCII-based names and identify non-ASCII
+# positions within names.
+
+def prevalidate_names(lex, name_stream, nmtoken_stream):
+        """Fully validate ASCII-based names and identify non-ASCII positions within names.
+
+        >>> demo_prevalidate_names("<good -bad='hyphen'/><_OK/><:funny:butOK/><1problem a='b' d423='x'>")
+        input data: <good -bad='hyphen'/><_OK/><:funny:butOK/><1problem a='b' d423='x'>
+        names     : _1111_1111____________111___111111111111___11111111_1_____1111_____
+        name_check: ______1____________________________________1_______________________
+"""
+        name_check = name_stream &~ bitutil.Advance(name_stream) & ~lex.ASCII_name_start
+        name_check |= (name_stream | nmtoken_stream) & ~lex.ASCII_name_char
+        return name_check
+
+def demo_prevalidate_names(u8data):
+        lgth = len(u8data)
+        (bit, EOF_mask) = bitutil.transpose_streams(u8data)
+        (u8, control, lex) = byteclass.classify_bytes(bit)
+        lex = add_multiliterals(lex)
+        markup1 = parse_CtCDPI(lex, EOF_mask)
+        callouts = parse_tags(lex, markup1.CtCDPI_mask, EOF_mask)
+        name_stream = callouts.ElemNames | callouts.AttNames
+        name_check = prevalidate_names(lex, name_stream, 0)
+        bitutil.print_aligned_u8_byte_streams([('input data', u8data), 
+                              ('names', bitutil.bitstream2string(name_stream, lgth)),
+                              ('name_check', bitutil.bitstream2string(name_check, lgth))])
 
 def parabix_parse(u8data):