I'm graphic designer and programmer from Poland.

Homepage: https://jcubic.wordpress.com

Create javascript interface for all methods in cordova plugin

When you create a cordova plugin you need to create execute method and check for action parameter to do different task and in each task you need to create json object and call callbackContext. It would be nice if if you could just create a class with methods and each method is map to javascript method. Here is tutorial how to do this:

You need to create this ReflectService.java file:

package com.example.package;

import android.content.Context;

import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.io.StringWriter;
import java.io.PrintWriter;

import java.util.List;
import java.util.ArrayList;

import java.lang.reflect.Method;
import java.lang.reflect.Modifier;

import org.apache.cordova.CordovaPlugin;
import org.apache.cordova.CallbackContext;
import org.apache.cordova.CordovaInterface;
import org.apache.cordova.CordovaWebView;

import org.json.JSONArray;
import org.json.JSONException;
import org.json.JSONObject;

public class ReflectService extends CordovaPlugin {

    protected Method getMethod(String name) {
        Class aClass = this.getClass();
        Method[] methods = aClass.getMethods();
        Method method = null;
        for (int i=0; i<methods.length; ++i) {
            if (methods[i].getName().equals(name)) {
                method = methods[i];
        return method;

    public String[] getMethods() {
        Class aClass = this.getClass();
        Method[] methods = aClass.getDeclaredMethods();
        List<String> list = new ArrayList<String>();
        int len = methods.length;
        String[] result = new String[len];
        for (Method method : methods) {
            String name = method.getName();
            if (Modifier.isPublic(method.getModifiers()) &&
                !name.equals("initialize")) {
        return list.toArray(new String[list.size()]);

    protected String[] jsonArrayToString(JSONArray args) {
        String[] result = {};
        try {
            int len = args.length();
            result = new String[len];
            for (int i=0; i<len; ++i) {
                result[i] = args.getString(i);
        } catch (JSONException e) {}
        return result;

    protected String[] getStackTrace(Exception e) {
        StringWriter sw = new StringWriter();
        PrintWriter pw = new PrintWriter(sw);
        return sw.toString().split("\n");

    public boolean execute(String action,
                           JSONArray args,
                           final CallbackContext callbackContext) throws JSONException {
        if (!action.equals("execute")) {
            final Method method = this.getMethod(action);
            if (method == null) {
                return false;
            final Object[] arguments = this.jsonArrayToString(args);
            cordova.getThreadPool().execute(new Runnable() {
                public void run() {
                    JSONObject json;
                    Object result;
                    try {
                        result = method.invoke(ReflectService.this, arguments);
                        json = new JSONObject();
                        if (result instanceof Object[]) {
                            json.put("result", new JSONArray(result));
                        } else {
                            json.put("result", result);
                        json.put("error", null);
                    } catch(JSONException e) {
                    } catch(Exception e) {
                        try {
                            json = new JSONObject();
                            JSONObject error = new JSONObject();
                            error.put("error", "Exception");
                            error.put("code", 200);
                            error.put("message", e.getMessage());
                            String[] trace = ReflectService.this.getStackTrace(e);
                            error.put("trace", new JSONArray(trace));
                            json.put("error", error);
                            json.put("result", null);
                        } catch(JSONException ee) {
            return true;
        return false;

You probably don’t need to use threads but I’ve seen this in cordova-plugin-shell-exec.

and then you can write your Service like this:

package com.example.package;

import org.apache.cordova.CordovaInterface;
import org.apache.cordova.CordovaWebView;

public class Service extends ReflectService {

    public void initialize(CordovaInterface cordova, CordovaWebView webView) {
        super.initialize(cordova, webView);
        // your init code here
    public String echo(String input) {
        if (input.equals("ping")) {
            return "pong";
        } else {
            return null;

You will be able to execute each method you create from execute in ReflectService class.

Then you need to create js file to create function for each method in the class. We will use getMethods action (that will execute getMethods method) to get list of methods and create function for each name.

window.Service = function Service(callback) {
    function call(method, args, callback) {
        return cordova.exec(function(response) {
            callback(null, response);
        }, function(error) {
        }, "Service", method, args || []);
    var service = {};
    call("getMethods", [], function(err, response) {
        if (response.result instanceof Array) {
            response.result.forEach(function(method) {
                service[method] = function() {
                    var args = [].slice.call(arguments);
                    return function(callback) {
                        return call(method, args, function(err, response) {
                            err = err || response.error;
                            if (err) {
                            } else {
                                callback(null, response.result);

Next you need to create plugin.xml file that look like this:

<?xml version="1.0" encoding="UTF-8"?>
<plugin id="pl.jcubic.leash.service" version="1.0.0"
  <description>Apache Cordova Leash shell service plugin</description>
  <license>Apache 2.0</license>
  <js-module name="service" src="www/service.js">
    <clobbers target="service"/>
  <platform name="android">
    <config-file parent="/*" target="res/xml/config.xml">
      <feature name="Service">
        <param name="android-package" value="com.example.package.Service" />
        <param name="onload" value="true" />
    <source-file src="src/com/example/package/Service.java"
                 target-dir="src/com/example/package/Service" />
    <source-file src="src/com/example/package/ReflectService.java"
                 target-dir="src/com/example/package/Service" />

When you add this plugin you can call it in javascript like this:

window.Service(function(service) {
    service.echo('ping')(function(err, result) {
        var pre = document.createElement('pre');
        pre.innerHTML = JSON.stringify(result, null, 4);

NOTE: Instead of callbacks you can use promises.

You can find whole code in this gist.

, ,

Leave a comment

How to add vignette in GIMP

Besides wrting code I’m also amateur photographer (you can see my photos on flickr most of them released under creative commons attribution share alike license) and start adding Vignette in GIMP (the only option because I’m working on GNU/Linux).

To add vignette you can do this steps:

  1. Add ellipse selection
  2. Feather it by 1000 pixels (I have 24MP camera) for your case it may be larger or smaller
  3. Inverse the Mask
  4. Add new black layer
  5. Add layer mask from selection
  6. Make it 50% opacity
  7. remove the selection

And now you have vignette on the image. It was fine but I needed to repeat those steps on each image, and since GIMP have support for scripting using script-fu (variant of scheme), I thought that I add simple script for the steps, here is result:

(define (make-vignette size opacity img)
  (let* ((width (car (gimp-image-width img)))
         (height (car (gimp-image-height img)))
         (layer (car (gimp-layer-new img width height 0 "vignette" 100 DARKEN-ONLY-MODE))))
    (gimp-selection-clear img)
    (gimp-image-select-ellipse img 0 0 0 width height)
    (gimp-selection-feather img size)
    (gimp-selection-invert img)
    (gimp-layer-set-opacity layer opacity)
    (gimp-layer-add-mask layer (car (gimp-layer-create-mask layer ADD-SELECTION-MASK)))
    (gimp-selection-clear img)
    (gimp-image-insert-layer img layer 0 -1)))

  "Create Vignette for the image"
  "Jakub Jankiewicz"
  "Copyright (c) 2017 Jakub Jankiewicz <http://jcubic.pl/me>"
  "May 20, 2017"
  SF-VALUE "Size" "1000"
  SF-VALUE "Opacity" "50"
  SF-IMAGE "image" 0
(script-fu-menu-register "make-vignette" "<Image>/Filters/Light and Shadow")

I’ve save it in ~/gimp-2.8/scripts/vignette.scm and it added menu item Vignette to menu Light and Shadow in Filters menu

Vignette Menu

and after I’ve executed the script I’ve get the same vignette layer with mask like this:

Vignette Layer

, ,

Leave a comment

Execute callback in javascript when a div change visibility with html5

There is new html5 API that can be used to trigger a callback when element change visibility. That api is called IntersectionObserver it’s supported by Chrome, Edge and Opera, its similar to MutationObserver.

The code look like this:

var element = document.querySelector('selector');
if (window.IntersectionObserver) {
  var observer = new IntersectionObserver(function(entries) {
    if (entries[0].intersectionRatio) {
    } else {
  }, {
    root: document.body

NOTE: intersectionRatio will not work if the element have position: fixed so if you can use jQuery you can check $(element).is(':visible');.

If you don’t use jQuery you can take :visible code from jQuery which look like this:

elem.offsetWidth || elem.offsetHeight || elem.getClientRects().length;

To remove the observer you need to call:


The element will intersect with body only when it’s visible, and the observer is fired when intersection is changed, so it will fire when element will become visible or not visible. You can read about IntersectionObserver on MDN.

If you’re are using jQuery you can wrap this using special event:

if (window.IntersectionObserver) {
  $.event.special.visibility = {
    setup: function() {
      function event(visibility) {
        var e = $.Event("visibility");
        e.visible = visibility;
        return e;
      var element = this;
      var $element = $(this);
      var observer = new IntersectionObserver(function(entries) {
        var e = event($element.is(':visible'));
        ($.event.dispatch || $.event.handle).call(element, e);
      }, {
        root: document.body
      $.data(this, 'observer', observer);
    teardown: function() {
      var observer = $.data(this, 'observer');
      if (observer) {
        $.removeData(this, 'observer');

and call it like this:

$('div').on('visibility', function(e) {
  $('pre').html(e.visible ? 'visible' : 'hidden');

Here is Codepen DEMO.

The event will fire when intersection observer is added, if you don’t want that you may want to disable first check.

      var first_time = true;
      var observer = new IntersectionObserver(function(entries) {
        if (!first_time) {
          var e = event(!!entries[0].intersectionRatio);
          ($.event.dispatch || $.event.handle).call(element, e);
        first_time = false;
      }, {
        root: document.body


Leave a comment

How to detect if element is added or removed from DOM?

There is new HTML5 API called MutationObserver that allow to add events when DOM is changing including a case when element is added or removed from DOM. Mutation observer support is pretty good.

Here is example to detect if element is added/removed from DOM with little help from jQuery

var in_dom = !!element.closest('body').length;
var observer = new MutationObserver(function(mutations) {
    if (element.closest('body').length) {
        if (!in_dom) {
            console.log("element inserted");
        in_dom = true;
    } else if (in_dom) {
        in_dom = false;
        console.log("element removed");

observer.observe(document.body, {childList: true});

to stop listening to changes you can run:


, ,

1 Comment

How to refresh the coverage badge in github README after coverage change

In my project jQuery Terminal I commited a change that make coverage was changed from 79% to 80% (the project is using jasmine, travis and coveralls service) but my badge in README still was showing 79%, even after removing cache. I’ve fixed the issue by adding md5 hash (I know it have been broken, but we don’t use anything secure here) of the spec file after url of my badge, so each time I add something to spec file I’ve get different url, so github wont cache the file.

I’m using make as my build script and I have two README files. Source with .in extension and result with .md extension so I’ve put another variable (I’m already using BRANCH and VERSION in README):


and in my Makefile I’ve added:

SPEC_CHECKSUM=`md5sum spec/terminalSpec.js | cut -d' ' -f 1`

    $(SED) -e "s/{{VER}}/$(VERSION)/g" -e "s/{{BRANCH}}/$(BRANCH)/g" -e "s/{{CHECKSUM}}/$(SPEC_CHECKSUM)/" < README.in > README.md

This is simplification, actually I have more code, you can see in my Makefile

if you have more then one spec file you can use this to generate checksum

SPEC_CHECKSUM=`cat spec/*.js | md5sum | cut -d' ' -f 1`

This will also work if you have different language and different build system.


Leave a comment

Dropdown menu in R Shiny

Shiny framework use bootstrap 3 that support drop down menus, all you have to do to have drop down menu in Shiny is to prepare html and the bootstrap will handle style and logic for opening and closing the menu. Here is the handy function that will create drop down menu for you:

dropdownMenu <- function(label=NULL, icon=NULL, menu=NULL) {
  ul <- lapply(names(menu), function(id) {
    if (is.character(menu[[id]])) {
      tags$li(actionLink(id, menu[[id]]))
    } else {
      args <- menu[[id]]
      args$inputId <- id
      tags$li(do.call(actionLink, args))
  ul$class <- "dropdown-menu"
    class = "dropdown",
      class = "btn btn-default dropdown-toggle",
      type = "button",
      `data-toggle` = "dropdown",
      `if`(!is.null(icon), icon, tags$span(class="caret"))
    do.call(tags$ul, ul)

The function will create drop down menu with caret like on the demo on getbootstrap.com or if you provide icon you will get only that icon the label is optional so you can have dropdown menu with just an icon. It use actionLink to create links so you don't need to to create custom input widget

You can use this function like this, to create hamburger menu:

  icon = icon("bars"),
  menu = list(edit = "edit item", rename = list(label = "address", icon = icon("id-card"))

and it will create two inputs input$edit and input$rename so you can add observeEvent to listen to on click.

, ,

Leave a comment

Execute javascript code when R shiny output have been rendered

If you have reactive observer like this:

output$resultsTable <- renderTable({

and you want to modify the table from javascript you can’t put shinyjs::runjs after table is rendered because the table data need to be the last expression of the reactive renderTable observer. The solution is to use new feature of JavaScript, implemented in browsers, which is mutation observer. The browser support is very good, you can check it on caniuse. The solution that will work, look like this (using jQuery plugin):

$.fn.onRender = function(callback, options) {
  if (arguments[0] == 'cancel') {
    return this.each(function() {
      var self = $(this);
      var render = self.data('render');
      if (render) {
  } else {
    var settings = $.extend({
      oneTime: true,
      observer: {childList: true, subtree: true}
    }, options);
    return this.each(function() {
      var node = $(this);
      var render = node.data('render');
      if (!render) {
        render = {
          callbacks: $.Callbacks(),
          observer: new MutationObserver(function(mutations) {
            if (settings.oneTime) {
        render.observer.observe(node[0], settings.observer);
        node.data('render', render);

You can use it like this:

function changeTable(selector) {
   $(selector).onRender(function() {
      $(this).find('thead th').css('color', 'red');

if you want to execute code once and use it outside of shiny observer you can use option oneTime: false like this:

function updateTableWhenRender(selector) {
   $(selector).onRender(function() {
      $(this).find('thead th').css('color', 'red');
   }, {oneTime: false});

then the function can be put outside of observer like this:

output$resultsTable <- renderTable({

To cancel the mutation observer, that’s create with the plugin, you can execute this line of code:


If you have multiple renders and have onRender on each render call you should call cancel before you call onRender again.

, ,

Leave a comment

%d bloggers like this: