Question

I would like app to add several (2k) words to UserDictionary.

I have experimented with ContentResolver().insert/bulk insert….

// array of words
    String[] words = getResources().getStringArray(R.array.word_array); 

    Long start,end = null;

    start = System.currentTimeMillis();


    int len = words.length;
    ContentValues[] mValueArray = new ContentValues[len];
    ContentValues mNewValues = new ContentValues();

    mNewValues.put(UserDictionary.Words.APP_ID, "com.my.example");
    mNewValues.put(UserDictionary.Words.LOCALE, "en");
    mNewValues.put(UserDictionary.Words.FREQUENCY, "100");

    for (int i = 0; i < len; ++i) {

        mNewValues.put(UserDictionary.Words.WORD, words[i]);

        mValueArray[i] = mNewValues;
    }

    getContentResolver().bulkInsert(
            UserDictionary.Words.CONTENT_URI,   // the user dictionary content URI
           mValueArray                       // the values to insert
        );
   end = System.currentTimeMillis();

   Toast toast = Toast.makeText(this, "Time for " + Integer.toString(len-1)+" words: " + Long.toString((end-start)) + "ms", 50);
   toast.show();

On my phone takes about 100ms per word if I do bulkinsert on batches of 100. 3+ minutes to insert 2k words.

Is anyone aware of a faster method to insert words or any optimization that can be done?

Side question: Is there an upper limit on UserDictionary size or does that depend on phone?

Any help appreciated

Michealster

Answer 1

Specific to your case:
Adding 2000 words in the dictionary at a stretch will take time and there is not much of an optimization which can be done. The time you see here are for the file operation and moving from Java->NDK->Native operation.

General:
Irrespective of the method used, one has to understand that these are nothing but file operations and is bound to take time. The APIs (including SQLite related) are just a wrapper to the intended file operations in the native code. Natively file write will always take a longer time then file read.