Support tokenizing arbitrary content.

Bug: 6110399
Change-Id: I37be63b68934fd451e6dffbf7d6079553619c0a3

2 files changed, 28 insertions, 5 deletions

diff --git a/include/utils/Tokenizer.h b/include/utils/Tokenizer.h
index c7db5fb..bb25f37 100644
--- a/include/utils/Tokenizer.h
+++ b/include/utils/Tokenizer.h
@@ -28,7 +28,8 @@ namespace android {
  * A simple tokenizer for loading and parsing ASCII text files line by line.
  */
 class Tokenizer {
-    Tokenizer(const String8& filename, FileMap* fileMap, char* buffer, size_t length);
+    Tokenizer(const String8& filename, FileMap* fileMap, char* buffer,
+            bool ownBuffer, size_t length);
 
 public:
     ~Tokenizer();
@@ -42,6 +43,15 @@ public:
     static status_t open(const String8& filename, Tokenizer** outTokenizer);
 
     /**
+     * Prepares to tokenize the contents of a string.
+     *
+     * Returns NO_ERROR and a tokenizer for the string, if successful.
+     * Otherwise returns an error and sets outTokenizer to NULL.
+     */
+    static status_t fromContents(const String8& filename,
+            const char* contents, Tokenizer** outTokenizer);
+
+    /**
      * Returns true if at the end of the file.
      */
     inline bool isEof() const { return mCurrent == getEnd(); }
@@ -111,6 +121,7 @@ private:
     String8 mFilename;
     FileMap* mFileMap;
     char* mBuffer;
+    bool mOwnBuffer;
     size_t mLength;
 
     const char* mCurrent;
diff --git a/libs/utils/Tokenizer.cpp b/libs/utils/Tokenizer.cpp
index efda2bf..7067533 100644
--- a/libs/utils/Tokenizer.cpp
+++ b/libs/utils/Tokenizer.cpp
@@ -35,15 +35,18 @@ static inline bool isDelimiter(char ch, const char* delimiters) {
     return strchr(delimiters, ch) != NULL;
 }
 
-Tokenizer::Tokenizer(const String8& filename, FileMap* fileMap, char* buffer, size_t length) :
+Tokenizer::Tokenizer(const String8& filename, FileMap* fileMap, char* buffer,
+        bool ownBuffer, size_t length) :
         mFilename(filename), mFileMap(fileMap),
-        mBuffer(buffer), mLength(length), mCurrent(buffer), mLineNumber(1) {
+        mBuffer(buffer), mOwnBuffer(ownBuffer), mLength(length),
+        mCurrent(buffer), mLineNumber(1) {
 }
 
 Tokenizer::~Tokenizer() {
     if (mFileMap) {
         mFileMap->release();
-    } else {
+    }
+    if (mOwnBuffer) {
         delete[] mBuffer;
     }
 }
@@ -65,6 +68,7 @@ status_t Tokenizer::open(const String8& filename, Tokenizer** outTokenizer) {
             size_t length = size_t(stat.st_size);
 
             FileMap* fileMap = new FileMap();
+            bool ownBuffer = false;
             char* buffer;
             if (fileMap->create(NULL, fd, 0, length, true)) {
                 fileMap->advise(FileMap::SEQUENTIAL);
@@ -77,6 +81,7 @@ status_t Tokenizer::open(const String8& filename, Tokenizer** outTokenizer) {
                 // The length we obtained from stat is wrong too (it will always be 4096)
                 // so we must trust that read will read the entire file.
                 buffer = new char[length];
+                ownBuffer = true;
                 ssize_t nrd = read(fd, buffer, length);
                 if (nrd < 0) {
                     result = -errno;
@@ -89,7 +94,7 @@ status_t Tokenizer::open(const String8& filename, Tokenizer** outTokenizer) {
             }
 
             if (!result) {
-                *outTokenizer = new Tokenizer(filename, fileMap, buffer, length);
+                *outTokenizer = new Tokenizer(filename, fileMap, buffer, ownBuffer, length);
             }
         }
         close(fd);
@@ -97,6 +102,13 @@ status_t Tokenizer::open(const String8& filename, Tokenizer** outTokenizer) {
     return result;
 }
 
+status_t Tokenizer::fromContents(const String8& filename,
+        const char* contents, Tokenizer** outTokenizer) {
+    *outTokenizer = new Tokenizer(filename, NULL,
+            const_cast<char*>(contents), false, strlen(contents));
+    return OK;
+}
+
 String8 Tokenizer::getLocation() const {
     String8 result;
     result.appendFormat("%s:%d", mFilename.string(), mLineNumber);