如何使用String.contains在Java中最佳搜索字符串列表？

Question

Right now I'm working on an android application using voice recog. Basically, I'm wondering what the best search method is once I get String from voice recognition. I'm currently using a linear search on the list of packagename, using the following to get that list:

    pkgNames = new ArrayList<String>();
    pkgAppsList = (ArrayList<ApplicationInfo>) getPackageManager()
            .getInstalledApplications(PackageManager.GET_META_DATA);

    // List available packages on phone.
    for (ApplicationInfo appInfo : pkgAppsList) {

        if (!isSystemPackage(appInfo))
            pkgNames.add(appInfo.packageName);
    }

I've decide that it's probably better to use the ApplicationInfo list(pkgAppsList) and do aa search on that, but is there a faster way of searching that list than just a simple linear search and using the result to open the Application with an Intent. Right now, all I can think of doing is:

    for(ApplicationInfo ai: pkgAppsList){
        if((ai.name).contains(voice_recog_result))
           //open Launch Intent for ai.packageName
    }

Is there a way faster search method I can use with the contains method or a way I can do this without the contains method?

Answer 1

For small numbers of searched strings (application names), the simplest method (traversing a linked list) should be chosen.

Auto-growing hash tables are often used for similar tasks, but only if the set of searched strings is often large. The worst case performance of a hash table is the same as traversing a linked list (due to auto-growing, and due to the potential of mass hash value collisions), and it has non-negligible overhead. It is therefore not a great choice for "just in case" insurance against a longer set of strings.

The data structure that is theoretically fitted to this task is called trie . Tries are not part of JCL or Android libraries, but available implementations are available . The worst case performance of a trie is proportional to the length of the longest string, regardless of the number of strings. Tries however tend to take a lot of memory which makes them unsuitable for mobile environments.