Changeset 2620

Timestamp:

Nov 7, 2012, 3:18:57 PM (6 years ago)

Author:

bhull

Message:

Changes to compilable version

File:

• proto/PDF/pdf.py (modified) (11 diffs)

proto/PDF/pdf.py

@@ -29 +29 @@
         Slash = 0
         Regular = 0
-        Delimiter = 0
         LAngle = 0
         RAngle = 0
         Hex = 0
         WS = 0
-        Digit= 0
+        Digit = 0
         Sign = 0
         Period = 0
@@ -40 +39 @@
         RBracket = 0
         LBracket = 0
+        Hash = 0
+        CR = 0
+        LF = 0
+        Special = 0
 
 
@@ -46 +49 @@
         UELParen = 0
         UERParen = 0
+        escapes = 0
 
 class Comments_Callouts() :
-        mask= 0
+        mask = 0
 
 class LiteralStrings_Callouts() :
         starts = 0
         ends = 0
+        escapes = 0
+        carriage_returns = 0
         mask = 0
         error = 0
 
 class Marker() :
-        mask=0
+        mask = 0
 
 class Names_Callouts() :
         starts = 0
         follows = 0
+        escapes = 0
 
 class HexStrings_Callouts() :
@@ -67 +74 @@
         ends = 0
         error = 0
+        hex = 0
 
 class Numeric_Callouts() :
@@ -75 +83 @@
 class Keywords_Callouts() :
         starts = 0
+        ends = 0
+        
+class SpecialChars_Callouts() :
+        starts = 0
         follows = 0
-        
-class Arrays_Callouts() :
-        starts = 0
-        ends = 0
 
 def Classify_bytes(basis_bits, lex):  
-        temp1 = (basis_bits.bit_0 | basis_bits.bit_1)
-        temp2 = (basis_bits.bit_2 &~ basis_bits.bit_3)
-        temp3 = (temp2 &~ temp1)
-        temp4 = (basis_bits.bit_4 &~ basis_bits.bit_5)
-        temp5 = (basis_bits.bit_6 | basis_bits.bit_7)
-        temp6 = (temp4 &~ temp5)
-        lex.LParen = (temp3 & temp6)
-        temp7 = (basis_bits.bit_7 &~ basis_bits.bit_6)
-        temp8 = (temp4 & temp7)
-        lex.RParen = (temp3 & temp8)
-        temp9 = (basis_bits.bit_1 &~ basis_bits.bit_0)
-        temp10 = (basis_bits.bit_3 &~ basis_bits.bit_2)
-        temp11 = (temp9 & temp10)
-        temp12 = (basis_bits.bit_4 & basis_bits.bit_5)
-        temp13 = (temp12 &~ temp5)
-        lex.Backslash = (temp11 & temp13)
-        temp14 = (temp5 &~ temp7)
-        temp15 = (temp4 &~ temp14)
-        temp16 = (temp3 & temp15)
-        temp17 = (basis_bits.bit_2 & basis_bits.bit_3)
-        temp18 = (temp17 &~ temp1)
-        temp19 = (temp18 & temp13)
-        temp20 = (temp16 | temp19)
-        temp21 = (basis_bits.bit_6 &~ basis_bits.bit_7)
-        temp22 = (temp12 & temp21)
-        temp23 = (temp18 & temp22)
-        temp24 = (temp20 | temp23)
-        temp25 = (basis_bits.bit_6 & basis_bits.bit_7)
-        temp26 = (temp4 & temp25)
-        temp27 = (temp11 & temp26)
-        temp28 = (temp24 | temp27)
-        temp29 = (temp12 & temp7)
-        temp30 = (temp11 & temp29)
-        temp31 = (temp28 | temp30)
-        temp32 = (temp9 & temp17)
-        temp33 = (temp32 & temp26)
-        temp34 = (temp31 | temp33)
-        temp35 = (temp32 & temp29)
-        temp36 = (temp34 | temp35)
-        temp37 = (basis_bits.bit_5 &~ basis_bits.bit_4)
-        temp38 = (temp37 & temp7)
-        temp39 = (temp3 & temp38)
-        temp40 = (temp36 | temp39)
-        temp41 = (temp12 & temp25)
-        temp42 = (temp3 & temp41)
-        lex.Delimiter = (temp40 | temp42)
-        lex.Percent = (temp3 & temp38)
-        lex.Period = (temp3 & temp22)
-        lex.LAngle = (temp18 & temp13)
-        lex.RAngle = (temp18 & temp22)
-        lex.LBracket = (temp11 & temp26)
-        lex.RBracket = (temp11 & temp29)
-        lex.Slash = (temp3 & temp41)
-        temp43 = (basis_bits.bit_2 | basis_bits.bit_3)
-        temp44 = (temp1 | temp43)
-        temp45 = (temp4 & temp21)
-        temp46 = (temp29 | temp45)
-        lex.EOL = (temp46 &~ temp44)
-        temp47 = (basis_bits.bit_4 | basis_bits.bit_5)
-        temp48 = (temp47 | temp5)
-        temp49 = (temp44 | temp48)
-        temp50 = (temp13 &~ temp44)
-        temp51 = (temp49 &~ temp50)
-        temp52 = (temp29 &~ temp44)
-        temp53 = (temp51 &~ temp52)
-        temp54 = (temp45 &~ temp44)
-        temp55 = (temp53 &~ temp54)
-        temp56 = (temp8 &~ temp44)
-        temp57 = (temp55 &~ temp56)
-        temp58 = (temp3 &~ temp48)
-        temp59 = (temp57 &~ temp58)
-        lex.WS = (~temp59)
-        temp60 = (basis_bits.bit_5 | basis_bits.bit_6)
-        temp61 = (basis_bits.bit_4 & temp60)
-        lex.Digit = (temp18 &~ temp61)
-        temp62 = (temp29 | temp26)
-        lex.Sign = (temp3 & temp62)
-        temp63 = (temp9 &~ temp43)
-        temp64 = (temp63 &~ basis_bits.bit_4)
-        temp65 = (~temp25)
-        temp66 = ((basis_bits.bit_5 & temp65)|(~(basis_bits.bit_5) & temp5))
-        temp67 = (temp64 & temp66)
-        temp68 = (lex.Digit | temp67)
-        temp69 = (temp9 & temp2)
-        temp70 = (temp69 &~ basis_bits.bit_4)
-        temp71 = (temp70 & temp66)
-        lex.Hex = (temp68 | temp71)
-        lex.Regular = ~lex.Delimiter&~lex.WS
+        temp1 = (basis_bits.bit_0 | basis_bits.bit_1)
+        temp2 = (basis_bits.bit_2 &~ basis_bits.bit_3)
+        temp3 = (temp2 &~ temp1)
+        temp4 = (basis_bits.bit_4 &~ basis_bits.bit_5)
+        temp5 = (basis_bits.bit_6 | basis_bits.bit_7)
+        temp6 = (temp4 &~ temp5)
+        lex.LParen = (temp3 & temp6)
+        temp7 = (basis_bits.bit_7 &~ basis_bits.bit_6)
+        temp8 = (temp4 & temp7)
+        lex.RParen = (temp3 & temp8)
+        temp9 = (basis_bits.bit_1 &~ basis_bits.bit_0)
+        temp10 = (basis_bits.bit_3 &~ basis_bits.bit_2)
+        temp11 = (temp9 & temp10)
+        temp12 = (basis_bits.bit_4 & basis_bits.bit_5)
+        temp13 = (temp12 &~ temp5)
+        lex.Backslash = (temp11 & temp13)
+        temp14 = (temp5 &~ temp7)
+        temp15 = (temp4 &~ temp14)
+        temp16 = (temp3 & temp15)
+        temp17 = (basis_bits.bit_2 & basis_bits.bit_3)
+        temp18 = (temp17 &~ temp1)
+        temp19 = (temp18 & temp13)
+        temp20 = (temp16 | temp19)
+        temp21 = (basis_bits.bit_6 &~ basis_bits.bit_7)
+        temp22 = (temp12 & temp21)
+        temp23 = (temp18 & temp22)
+        temp24 = (temp20 | temp23)
+        temp25 = (basis_bits.bit_6 & basis_bits.bit_7)
+        temp26 = (temp4 & temp25)
+        temp27 = (temp11 & temp26)
+        temp28 = (temp24 | temp27)
+        temp29 = (temp12 & temp7)
+        temp30 = (temp11 & temp29)
+        temp31 = (temp28 | temp30)
+        temp32 = (temp9 & temp17)
+        temp33 = (temp32 & temp26)
+        temp34 = (temp31 | temp33)
+        temp35 = (temp32 & temp29)
+        temp36 = (temp34 | temp35)
+        temp37 = (basis_bits.bit_5 &~ basis_bits.bit_4)
+        temp38 = (temp37 & temp7)
+        temp39 = (temp3 & temp38)
+        temp40 = (temp36 | temp39)
+        temp41 = (temp12 & temp25)
+        temp42 = (temp3 & temp41)
+        lex.Special = (temp40 | temp42)
+        lex.Percent = (temp3 & temp38)
+        lex.Period = (temp3 & temp22)
+        lex.LAngle = (temp18 & temp13)
+        lex.RAngle = (temp18 & temp22)
+        lex.LBracket = (temp11 & temp26)
+        lex.RBracket = (temp11 & temp29)
+        lex.Slash = (temp3 & temp41)
+        temp43 = (basis_bits.bit_2 | basis_bits.bit_3)
+        temp44 = (temp1 | temp43)
+        temp45 = (temp4 & temp21)
+        temp46 = (temp29 | temp45)
+        lex.EOL = (temp46 &~ temp44)
+        temp47 = (basis_bits.bit_4 | basis_bits.bit_5)
+        temp48 = (temp47 | temp5)
+        temp49 = (temp44 | temp48)
+        temp50 = (temp13 &~ temp44)
+        temp51 = (temp49 &~ temp50)
+        temp52 = (temp29 &~ temp44)
+        temp53 = (temp51 &~ temp52)
+        temp54 = (temp45 &~ temp44)
+        temp55 = (temp53 &~ temp54)
+        temp56 = (temp8 &~ temp44)
+        temp57 = (temp55 &~ temp56)
+        temp58 = (temp3 &~ temp48)
+        temp59 = (temp57 &~ temp58)
+        lex.WS = (~temp59)
+        temp60 = (basis_bits.bit_5 | basis_bits.bit_6)
+        temp61 = (basis_bits.bit_4 & temp60)
+        lex.Digit = (temp18 &~ temp61)
+        temp62 = (temp29 | temp26)
+        lex.Sign = (temp3 & temp62)
+        temp63 = (temp9 &~ temp43)
+        temp64 = (temp63 &~ basis_bits.bit_4)
+        temp65 = (~temp25)
+        temp66 = ((basis_bits.bit_5 & temp65)|(~(basis_bits.bit_5) & temp5))
+        temp67 = (temp64 & temp66)
+        temp68 = (lex.Digit | temp67)
+        temp69 = (temp9 & temp2)
+        temp70 = (temp69 &~ basis_bits.bit_4)
+        temp71 = (temp70 & temp66)
+        lex.Hex = (temp68 | temp71)
+        temp72 = (temp25 &~ temp47)
+        lex.Hash = (temp3 & temp72)
+        lex.CR = (temp29 &~ temp44)
+        lex.LF = (temp45 &~ temp44)
+        lex.Regular = ~lex.Special&~lex.WS
 
 
@@ -185 +197 @@
         even_final = pablo.ScanThru(even_start, lex.Backslash) 
         escape = even_final & odd
-   
+        escape_char = even & pablo.SpanUpTo(even_start, even_final)
+        
         odd_start = start & odd
         odd_final = pablo.ScanThru(odd_start, lex.Backslash)
         escape = escape | (odd_final & even)
-                   
+        escape_char |= odd & pablo.SpanUpTo(odd_start, odd_final)
+        
+        escaped_Callouts.escapes = escape_char
         escaped_Callouts.UELParen = lex.LParen &~ escape
         escaped_Callouts.UERParen = lex.RParen &~ escape
@@ -239 +254 @@
         comment_starts=0
         comment_ends=0
-        scan_characters = escaped_Callouts.UELParen |escaped_Callouts.UERParen | lex.Percent
-        literalStrings_Callouts.starts = 0
-        literalStrings_Callouts.ends = 0
-        literalStrings_Callouts.error = 0
+        scan_characters = escaped_Callouts.UELParen | escaped_Callouts.UERParen | lex.Percent
         cursor = pablo.ScanToFirst(scan_characters)
         while pablo.inFile(cursor) :
@@ -252 +264 @@
                         comment_ends |=comment_cursor
                         cursor = pablo.ScanTo(pablo.Advance(comment_cursor), scan_characters)
-   
-
-                
+                
                 if string_cursor :
                         if (string_cursor & escaped_Callouts.UELParen) :
@@ -270 +280 @@
                                         literalStrings_Callouts.ends|=string_cursor
                                         cursor = pablo.ScanTo(pablo.Advance(string_cursor), scan_characters)
-                                if lt_zero():
+                                elif lt_zero():
                                         literalStrings_Callouts.error|=string_cursor
                                         cursor = pablo.ScanTo(pablo.Advance(string_cursor), scan_characters)
-                                if ~(is_zero() | lt_zero()):
+                                else:
                                         cursor = pablo.ScanTo(pablo.Advance(string_cursor), escaped_Callouts.UELParen | escaped_Callouts.UERParen)
-                                #literalStrings_Callouts.ends |= string_cursor & is_zero()      
-                                #literalStrings_Callouts.error |= string_cursor & lt_zero()
-                                #cursor = pablo.ScanTo(pablo.Advance(string_cursor), scan_characters ^ (lex.Percent & (is_zero() | lt_zero())))
-                                                                        
-                
-                
-        #if (depth>0):
-                #error
+                                        
+
         comments_Callouts.mask = pablo.SpanUpTo(comment_starts, comment_ends)
-        literalStrings_Callouts.mask = pablo.SpanUpTo(literalStrings_Callouts.starts,literalStrings_Callouts.ends)
+        literalStrings_Callouts.mask = pablo.InclusiveSpan(literalStrings_Callouts.starts,literalStrings_Callouts.ends)
+        literalStrings_Callouts.escapes = escaped_Callouts.escapes & literalStrings_Callouts.mask
+        literalStrings_Callouts.carriage_returns = lex.CR & literalStrings_Callouts.mask &~ literalStrings_Callouts.escapes
         marker.mask = comments_Callouts.mask | literalStrings_Callouts.mask
 
-
 def Parse_Names(lex, marker, names_Callouts) :
-        names_Callouts.starts = pablo.Advance(lex.Slash&~marker.mask)
-        names_Callouts.follows = pablo.ScanThru(names_Callouts.starts, lex.Regular)
-        marker.mask |=names_Callouts.starts
+        names_Callouts.starts =lex.Slash&~marker.mask
+        firstCharacter = pablo.Advance(names_Callouts.starts)
+        names_Callouts.follows = pablo.ScanThru(firstCharacter, lex.Regular)
+        names_Callouts.escapes = (pablo.Advance(pablo.Advance(lex.Hash))&pablo.Advance(lex.Hex)&lex.Hex)&pablo.SpanUpTo(names_Callouts.starts, names_Callouts.follows)
+        marker.mask |=names_Callouts.starts|(firstCharacter&~names_Callouts.follows)
 
 def Parse_HexStrings(lex, marker, hexStrings_Callouts) :
-        #Haven't confirmed that this is the correct behaviour with bad input
         hexStrings_Callouts.starts = lex.LAngle &~marker.mask
         hexStrings_Callouts.ends = pablo.ScanThru(pablo.Advance(hexStrings_Callouts.starts),lex.Hex|lex.WS)
         hexStrings_Callouts.error = hexStrings_Callouts.ends &~ lex.RAngle
-        marker.mask |= pablo.SpanUpTo(hexStrings_Callouts.starts,hexStrings_Callouts.ends)
+        mask = pablo.InclusiveSpan(hexStrings_Callouts.starts,hexStrings_Callouts.ends)
+        hexStrings_Callouts.hex = lex.Hex & mask
+        marker.mask |= mask
 
 def Parse_Numeric(lex, marker, numeric_Callouts) :
@@ -306 +314 @@
         marker.mask |= numeric_Callouts.starts
 
-        # aftersign = pablo.Advance(numeric_Callouts.starts & lex.Sign) | (numeric_Callouts.starts &~ lex.Sign)
-        # temp = pablo.ScanThru(aftersign, lex.Digit)
-        # afterperiod = pablo.Advance(temp & lex.Period) | (temp &~ lex.Period)
-        # numeric_Callouts.follows = pablo.ScanThru(afterperiod, lex.Digit)
-        # numeric_Callouts.error = numeric_Callouts.follows &~ (lex.WS | lex.Delimiter)
-        # marker.mask |= numeric_Callouts.starts
-
 def Parse_Keywords(lex, marker, keywords_Callouts) :
         keywords_Callouts.starts = (lex.Regular &~ pablo.Advance(lex.Regular)) &~ marker.mask
         keywords_Callouts.follows = pablo.ScanThru(keywords_Callouts.starts, lex.Regular)
         
-def Parse_Arrays(lex, marker, arrays_Callouts) :
-        arrays_Callouts.starts =  lex.LBracket &~ marker.mask
-        arrays_Callouts.ends = lex.RBracket &~ marker.mask
+def Parse_SpecialChars(lex, marker, specialChars_Callouts) :
+        specialChars_Callouts.starts =  (lex.Special) &~ marker.mask
+        specialChars_Callouts.follows = pablo.Advance(specialChars_Callouts.starts)
         
 def Main(basis_bits, lex, parity, escaped_Callouts):
@@ -330 +331 @@
                 Parse_Numeric(lex, marker, numeric_Callouts)
                 Parse_Keywords(lex, marker, keywords_Callouts)
-                Parse_Arrays(lex, marker, arrays_Callouts)
-        
+                Parse_SpecialChars(lex, marker, specialChars_Callouts)
+