[Mangler] Add "$" (dollar sign) to the list of legal encoding characters.

Author: nadavrot

lib/ABI/CBCTables.h

@@ -4,12 +4,12 @@
 #include "llvm/ADT/APInt.h"
 using APInt = llvm::APInt;
 // This file is autogenerated. Do not modify this file.
-// Processing text files: /Users/nadav/swift_devel/repo/mangling/inputs/SZCompressed.sz
+// Processing text files: CBC_Compressed.txt
 namespace Huffman {
 // The charset that the fragment indices can use:
-unsigned CharsetLength = 63;
-unsigned LongestEncodingLength = 10;
-const char *Charset = "0123456789_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
+unsigned CharsetLength = 64;
+unsigned LongestEncodingLength = 11;
+const char *Charset = "0123456789_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ$";
 char variable_decode(APInt &num) {
  uint64_t tailbits = *num.getRawData();
 if ((tailbits & 1) == 0) {
@@ -481,23 +481,31 @@ if ((tailbits & 1) == 1) {
       if ((tailbits & 1) == 0) {
        tailbits/=2;
        if ((tailbits & 1) == 0) {
+        tailbits/=2;
+        num = num.lshr(8);
+        return 'j';
+       }
+       if ((tailbits & 1) == 1) {
         tailbits/=2;
         if ((tailbits & 1) == 0) {
          tailbits/=2;
-         num = num.lshr(9);
-         return 'Q';
+         if ((tailbits & 1) == 0) {
+          tailbits/=2;
+          num = num.lshr(10);
+          return '$';
+         }
+         if ((tailbits & 1) == 1) {
+          tailbits/=2;
+          num = num.lshr(10);
+          return 'Q';
+         }
         }
         if ((tailbits & 1) == 1) {
          tailbits/=2;
          num = num.lshr(9);
          return 'H';
         }
        }
-       if ((tailbits & 1) == 1) {
-        tailbits/=2;
-        num = num.lshr(8);
-        return 'j';
-       }
       }
       if ((tailbits & 1) == 1) {
        tailbits/=2;
@@ -572,9 +580,10 @@ if (ch == 'g') {/*0101111*/ num = num.shl(1); num = num.shl(1); num = ++num; num
 if (ch == '7') {/*1101111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
 if (ch == 'C') {/*0011111*/ num = num.shl(1); num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
 if (ch == 'z') {/*1011111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
-if (ch == 'Q') {/*000111111*/ num = num.shl(1); num = num.shl(1); num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
-if (ch == 'H') {/*100111111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
-if (ch == 'j') {/*10111111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
+if (ch == 'j') {/*00111111*/ num = num.shl(1); num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
+if (ch == '$') {/*0010111111*/ num = num.shl(1); num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
+if (ch == 'Q') {/*1010111111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
+if (ch == 'H') {/*110111111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
 if (ch == '0') {/*1111111*/ num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; num = num.shl(1); num = ++num; return;}
 assert(false);
 }

lib/ABI/HuffTables.h

@@ -25,6 +25,7 @@ const char* TestValues[] = {"AA", "r", "J", "Swift","A", "ArrayStringPrintable",
   "__TMPVs15ContiguousArray",
   "__TIF14StdlibUnittest13checkSequenceu0_Rxs14CollectionType_s12SequenceTypeWx9Generator7Element_zW_9GeneratorS3__rFTxq",
   "__TTSg5VSS13CharacterViewS_s14CollectionTypes_GVs17IndexingGeneratorS__GS1_S__s13GeneratorTypes_VS_5IndexS3_s16Forwar",
+  "__TMANameWith$dollar$inIt",
   ""};
 
 // Test that the code book compression round trips.

unittests/Basic/CompressionTests.cpp

@@ -49,7 +49,7 @@ def addLine(line):
 escape_char1 = 'J'
 
 # notice that Y and J are missing because they are escape chars:
-charset = r"0123456789_abcdefghijklmnopqrstuvwxyzABCDEFGHIKLMNOPQRSTUVWXZ"
+charset = r"0123456789_abcdefghijklmnopqrstuvwxyzABCDEFGHIKLMNOPQRSTUVWXZ$"
 encoders = [c for c in charset] # alphabet without the escape chars.
 enc_len = len(encoders)
 

utils/name-compression/CBCGen.py

@@ -86,7 +86,7 @@ def generate_encoder(self, stack):
     return sb
 
 # Only accept these characters into the tree.
-charset = r"0123456789_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
+charset = r"0123456789_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ$"
 charser_length = str(len(charset))
 
 # Convert the characters and frequencies to a list of trees