fix(regex): Add custom regex-like parsing for version extraction

Signed-off-by: royalpinto007 <[email protected]>

fix(regex): Add custom regex-like parsing

Signed-off-by: royalpinto007 <[email protected]>

compiler/src/dmd/sarif.d

@@ -154,6 +154,95 @@ string formatErrorMessage(const(char)* format, va_list ap) nothrow {
     return buffer[0 .. buffer.length].dup;
 }
 
+/**
+Extracts a version number from a string using custom parsing rules.
+
+Params:
+  toolVersion = The version string to extract from.
+
+Returns:
+  A cleaned version string, or an empty string if no valid version is found.
+*/
+string extractVersionWithCustomRegex(string toolVersion) nothrow {
+    size_t index = 0;
+    char[] cleanedVersion;
+
+    // Match optional "v" prefix (equivalent to `v?`)
+    if (index < toolVersion.length && toolVersion[index] == 'v') {
+        index++; // Move past 'v' if present
+    }
+
+    // Match one or more digits (equivalent to `\d+`)
+    bool matchedDigits = matchDigits(index, toolVersion, cleanedVersion);
+    if (!matchedDigits) {
+        return ""; // Return empty if no digits found
+    }
+
+    // Match zero or more dot-separated digits (equivalent to `(\.\d+)*`)
+    matchDotSeparatedDigits(index, toolVersion, cleanedVersion);
+
+    return cast(string) cleanedVersion;
+}
+
+/**
+Matches a sequence of digits in the input string.
+
+Params:
+  index = The current position in the input string (updated by reference).
+  input = The input string to match against.
+  buffer = The buffer to store matched digits.
+
+Returns:
+  `true` if digits were matched; otherwise `false`.
+*/
+bool matchDigits(ref size_t index, string input, ref char[] buffer) nothrow {
+    bool hasMatched = false;
+    while (index < input.length && isDigit(input[index])) {
+        buffer ~= input[index];
+        index++;
+        hasMatched = true;
+    }
+    return hasMatched;
+}
+
+/**
+Matches a dot followed by one or more digits in the input string.
+
+Params:
+  index = The current position in the input string (updated by reference).
+  input = The input string to match against.
+  buffer = The buffer to store matched characters.
+*/
+void matchDotSeparatedDigits(ref size_t index, string input, ref char[] buffer) nothrow {
+    while (index < input.length && input[index] == '.') {
+        size_t tempIndex = index + 1;
+        char[] tempBuffer;
+
+        // Check if there's a sequence of digits after the dot
+        if (matchDigits(tempIndex, input, tempBuffer)) {
+            buffer ~= '.';
+            buffer ~= tempBuffer;
+            index = tempIndex;
+        } else {
+            break; // No digits after dot, stop matching
+        }
+    }
+}
+
+/**
+Checks if a character is a digit.
+
+Params:
+  c = The character to check.
+
+Returns:
+  `true` if the character is a digit; otherwise `false`.
+*/
+bool isDigit(char c) nothrow
+{
+    return c >= '0' && c <= '9';
+}
+
 /**
 Generates a SARIF (Static Analysis Results Interchange Format) report and prints it to `stdout`.
 
@@ -165,7 +254,7 @@ Params:
   format = A format string for constructing the error message.
   ap = A variable argument list used with the format string.
   kind = The kind of error (error, warning, deprecation, note, or message).
-  executionSuccessful = `true` if execution succeeded; includes an empty `results` array in the SARIF report.
+  executionSuccessful = `true` for an empty `results` array; `false` for detailed errors.
 
 Throws:
   This function is marked as `nothrow` and does not throw exceptions.
@@ -179,19 +268,9 @@ void generateSarifReport(const ref SourceLoc loc, const(char)* format, va_list a
     OutBuffer ob;
     ob.doindent = true;
 
-    // Extract and clean the version string
+    // Use extractVersionWithCustomRegex to get the cleaned version string
     string toolVersion = global.versionString();
-    // Remove 'v' prefix if it exists
-    if (toolVersion.length > 0 && toolVersion[0] == 'v') {
-        toolVersion = toolVersion[1 .. $];
-    }
-    // Find the first non-numeric character after the version number
-    size_t length = toolVersion.length;
-    const(char)* nonNumeric = strchr(toolVersion.ptr, '-');
-    if (nonNumeric) {
-        length = cast(size_t)(nonNumeric - toolVersion.ptr);
-    }
-    string cleanedVersion = toolVersion[0 .. length];
+    string cleanedVersion = extractVersionWithCustomRegex(toolVersion);
 
     // Build SARIF report
     ob.level = 0;