Bug 1897647 - Generate URL_ analysis item for chrome:// and resource:// string literals in C++ code. r=asuth

Differential Revision: https://phabricator.services.mozilla.com/D210892

build/clang-plugin/mozsearch-plugin/MozsearchIndexer.cpp

@@ -122,6 +122,21 @@ static bool isValidIdentifier(std::string Input) {
   return true;
 }
 
+template <size_t N>
+static bool stringStartsWith(const std::string& Input,
+                             const char (&Prefix)[N]) {
+  return Input.length() > N - 1 && memcmp(Input.c_str(), Prefix, N - 1) == 0;
+}
+
+static bool isASCII(const std::string& Input) {
+  for (char C : Input) {
+    if (C & 0x80) {
+      return false;
+    }
+  }
+  return true;
+}
+
 struct RAIITracer {
   RAIITracer(const char *log) : mLog(log) {
     printf("<%s>\n", mLog);
@@ -486,6 +501,10 @@ private:
     return Filename;
   }
 
+  std::string mangleURL(std::string Url) {
+    return mangleFile(Url, FileType::Source);
+  }
+
   std::string mangleQualifiedName(std::string Name) {
     std::replace(Name.begin(), Name.end(), ' ', '_');
     return Name;
@@ -2244,6 +2263,35 @@ public:
     return true;
   }
 
+  bool VisitStringLiteral(StringLiteral *E) {
+    if (E->getCharByteWidth() != 1) {
+      return true;
+    }
+
+    StringRef sref = E->getString();
+    std::string s = sref.str();
+
+    if (!stringStartsWith(s, "chrome://") &&
+        !stringStartsWith(s, "resource://")) {
+      return true;
+    }
+
+    if (!isASCII(s)) {
+      return true;
+    }
+
+    SourceLocation Loc = E->getStrTokenLoc(0);
+    normalizeLocation(&Loc);
+
+    std::string symbol = std::string("URL_") + mangleURL(s);
+
+    visitIdentifier("use", "file", StringRef(s), Loc, symbol,
+                    QualType(), Context(),
+                    NotIdentifierToken | LocRangeEndValid);
+
+    return true;
+  }
+
   void enterSourceFile(SourceLocation Loc) {
     normalizeLocation(&Loc);
     FileInfo* newFile = getFileInfo(Loc);