Changeset 108049 in vbox

Timestamp:

Feb 4, 2025 6:01:30 AM (2 weeks ago)

Author:

vboxsync

svn:sync-xref-src-repo-rev:

167328

Message:

Runtime/RTScriptLex*: Allow case insensitivity with converting all characters [a-z] to upper case [A-Z] internally, bugref:10733

Location:

trunk

Files:

• include/iprt/script.h (modified) (1 diff)
• src/VBox/Disassembler/testcase/tstDisasmArmv8-1.cpp (modified) (1 diff)
• src/VBox/Runtime/common/script/scriptlex.cpp (modified) (5 diffs)

trunk/include/iprt/script.h

@@ -309 +309 @@
 
 /** Default lexer config flags. */
-#define RTSCRIPT_LEX_CFG_F_DEFAULT          0
+#define RTSCRIPT_LEX_CFG_F_DEFAULT                0
 /** Case insensitive lexing, keywords and so on must be used lowercase to match
  * as the lexer will convert everything to lowercase internally. */
-#define RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE RT_BIT(0)
+#define RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_LOWER RT_BIT(0)
+/** Case insensitive lexing, keywords and so on must be used uppercase to match
+ * as the lexer will convert everything to uppercase internally. */
+#define RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_UPPER RT_BIT(1)
 
 

trunk/src/VBox/Disassembler/testcase/tstDisasmArmv8-1.cpp

@@ -127 +127 @@
     "ARMv8 disassembler lexer",
     /** fFlags */
-    RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE,
+    RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_LOWER,
     /** pszWhitespace */
     NULL,

trunk/src/VBox/Runtime/common/script/scriptlex.cpp

@@ -589 +589 @@
     AssertPtrReturn(pCfg, VERR_INVALID_POINTER);
 
+    /* Case insensitivity with internal lower or upper case conversion is mutually exclusive. */
+    AssertReturn(   (pCfg->fFlags & (RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_LOWER | RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_UPPER))
+                 != (RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_LOWER | RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_UPPER), VERR_INVALID_PARAMETER);
+
     if (!cchBuf)
         cchBuf = _16K;
@@ -853 +857 @@
     }
 
-    if (   (pThis->pCfg->fFlags & RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE)
-        && !(fFlags & RTSCRIPT_LEX_CONV_F_NOTHING))
-        ch = RT_C_TO_LOWER(ch);
+    if (!(fFlags & RTSCRIPT_LEX_CONV_F_NOTHING))
+    {
+        if (pThis->pCfg->fFlags & RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_LOWER)
+           ch = RT_C_TO_LOWER(ch);
+        else if (pThis->pCfg->fFlags & RTSCRIPT_LEX_CFG_F_CASE_INSENSITIVE_UPPER)
+           ch = RT_C_TO_UPPER(ch);
+    }
 
     return ch;
@@ -923 +931 @@
         /* Some hex prefix? */
         char chNext = RTScriptLexPeekCh(hScriptLex, 1);
-        if (chNext == 'x')
+        if (chNext == 'x' || chNext == 'X')
         {
             uBase = 16;
@@ -938 +946 @@
     {
         if (   (ch < '0' || ch > '9')
-            && (ch < 'a' || ch > 'f' || uBase == 10))
+            && (   (   !(ch >= 'a' && ch <= 'f') 
+                    && !(ch >= 'A' && ch <= 'F'))
+                || uBase == 10))
         {
             if (pTok->Type.Number.enmType == RTSCRIPTLEXTOKNUMTYPE_INTEGER)
@@ -955 +965 @@
             Assert(uBase == 16);
             u64 = (u64 << 4) + 10 + (ch - 'a');
+        }
+        else if (ch >= 'A' && ch <= 'F')
+        {
+            Assert(uBase == 16);
+            u64 = (u64 << 4) + 10 + (ch - 'A');
         }