First Thoughts

Interview AI Assistant

2024-10-01T07:54:00.000-07:00

Interview Assistant helps you get
your dream job
using AI

Check the Best Real-time interview assistant on the web. Uses OpenAI Gpt-4-o model, does live transcription and can help you during your interviews.

Nice Ideas from Nice Discussions

2020-11-17T04:22:00.001-08:00

Neural Plasticity

Remarkable Points

✔ In order to change or update yourself:

First Act (Behavior First), then thoughts, feelings, and perceptions follow.

or as a moto "Mood follows Action"

✔ Use small rewards - punishments to move on

✔ In the phase of a physical threat, there are 3 options: freeze, retreat, or move forward. The moving forward response actually triggers activation of a connection in the brain to the dopamine circuitry of the brain and makes it more likely that you are going to be able to move forward in the future. Also important is that the highest level of agitation and stress help us to move forward. Stopping and relaxing is not helpful at all.

✔ Stress is designed to move us forward!!

source:

Regex to remove comments from SQL Stored Procedure

2018-10-18T02:09:00.003-07:00

SQL Remove comments

The following regex expression
/\/\*[\s\S]*?\*\/|([^:]|^)--.*$/gm
removes following comments:
1. Code block inside /* */
2. Rest of line --

In javascript you can simply do
return sql_string.replace(/\/\*[\s\S]*?\*\/|([^:]|^)--.*$/gm,'');

AWS S3 usefull utilities nodejs

2018-04-18T03:06:00.001-07:00

AWS S3 Helpful Utilities - Helper functions using node

I am using regularly the following snippets for downloading specific directory inside an S3 bucket.

// jshint esnext:true  
 // jshint node:true  
 'use strict';  
 const AWS = require('aws-sdk');  
 const path = require('path');  
 const rimraf = require('rimraf');  
 const fs = require('fs');  
 const EventEmitter = require('events').EventEmitter;  
 exports.createClient = function(options) {  
   return new Client(options);  
 };  
 function Client(options) {  
   options = options || {};  
   this.s3 = new AWS.S3(options.s3Options);  
   this.s3RetryCount = options.s3RetryCount || 3;  
   this.s3RetryDelay = options.s3RetryDelay || 1000;  
   console.log('Client Created');  
 }  
 Client.prototype.downloadDir = function(params) {  
   return downloadDir(this, params);  
 };  
 function downloadDir(self, params) {  
   //if folder exists delete it   
   var downloader = new EventEmitter();  
   var keys;  
   if (params.localDir === undefined || params.localDir === null) {  
     downloader.emit('error', 'localDir cannot be empty');  
     return downloader;  
   }  
   if (params.s3Params === undefined || params.s3Params === null) {  
     downloader.emit('error', 's3Params cannot be empty');  
     return downloader;  
   }  
   try {  
     initDirectorySync(params.localDir);  
   } catch (err) {  
     if (err) {  
       downloader.emit('error', err);  
       return downloader;  
     }  
   }  
   //lets get list items from S3   
   let items = [];  
   listKeys(self, params.s3Params, function(error, keys) {  
     console.log('Error = ' + JSON.stringify(error));  
     if (error) {  
       downloader.emit('error', error);  
       return downloader;  
     }  
     items = items.concat(keys);  
     items.forEach(function(item) {  
       var withoutPrefixKey = item.replace(params.s3Params.Prefix, '');  
       //Lets check each item - key  
       //if ends in '/' then we have a directory  
       //else it is a file and we should get and download  
       if (withoutPrefixKey) {  
         //console.log(withoutPrefixKey);  
         if (withoutPrefixKey.slice(-1) === '/') {  
           // console.log('Directory');  
           //lets create it   
           var fullpath = path.join(params.localDir, withoutPrefixKey);  
           fs.mkdirSync(fullpath);  
           console.log('Created Directory ' + fullpath);  
         } else {  
           downloadFile(self, params.s3Params.Bucket, item, path.join(params.localDir, withoutPrefixKey), function(err, result) {  
             if (err) {  
               console.log('Cannot create file ' + withoutPrefixKey);  
               console.log(err);  
             } else {  
               console.log('Created file ' + path.join(params.localDir, withoutPrefixKey));  
             }  
           });  
         }  
       }  
     });  
     downloader.emit('end');  
     return downloader;  
   });  
 }  
 /**  
  * List keys from the specified bucket.  
  *  
  * If providing a prefix, only keys matching the prefix will be returned.  
  *  
  * If providing a delimiter, then a set of distinct path segments will be  
  * returned from the keys to be listed. This is a way of listing "folders"  
  * present given the keys that are there.  
  *  
  * @param {Object} options  
  * @param {String} options.bucket - The bucket name.  
  * @param {String} [options.prefix] - If set only return keys beginning with  
  *  the prefix value.  
  * @param {String} [options.delimiter] - If set return a list of distinct  
  *  folders based on splitting keys by the delimiter.  
  * @param {Function} callback - Callback of the form function (error, string[]).  
  */  
 function listKeys(self, options, callback) {  
   var keys = [];  
   console.log('Options are ' + JSON.stringify(options));  
   /**  
    * Recursively list keys.  
    *  
    * @param {String|undefined} marker - A value provided by the S3 API  
    *  to enable paging of large lists of keys. The result set requested  
    *  starts from the marker. If not provided, then the list starts  
    *  from the first key.  
    */  
   function listKeysRecusively(marker) {  
     options.marker = marker;  
     listKeyPage(  
       self,  
       options,  
       function(error, nextMarker, keyset) {  
         if (error) {  
           return callback(error, keys);  
         }  
         keys = keys.concat(keyset);  
         if (nextMarker) {  
           listKeysRecusively(nextMarker);  
         } else {  
           callback(null, keys);  
         }  
       }  
     );  
   }  
   // Start the recursive listing at the beginning, with no marker.  
   listKeysRecusively();  
 }  
 /**  
    * List one page of a set of keys from the specified bucket.  
    *  
    * If providing a prefix, only keys matching the prefix will be returned.  
    *  
    * If providing a delimiter, then a set of distinct path segments will be  
    * returned from the keys to be listed. This is a way of listing "folders"  
    * present given the keys that are there.  
    *  
    * If providing a marker, list a page of keys starting from the marker  
    * position. Otherwise return the first page of keys.  
    *  
    * @param {Object} options  
    * @param {String} options.bucket - The bucket name.  
    * @param {String} [options.prefix] - If set only return keys beginning with  
    *  the prefix value.  
    * @param {String} [options.delimiter] - If set return a list of distinct  
    *  folders based on splitting keys by the delimiter.  
    * @param {String} [options.marker] - If set the list only a paged set of keys  
    *  starting from the marker.  
    * @param {Function} callback - Callback of the form  
     function (error, nextMarker, keys).  
    */  
 function listKeyPage(self, options, callback) { // jshint ignore:line  
   console.log('Options in listKeyPage are :' + JSON.stringify(options));  
   var params = {  
     Bucket: options.Bucket,  
     Delimiter: options.delimiter,  
     Marker: options.marker,  
     MaxKeys: 1000,  
     Prefix: options.Prefix  
   };  
   self.s3.listObjects(params, function(error, response) {  
     if (error) {  
       console.log(error);  
       return callback(error);  
     } else if (response.err) {  
       console.log(response.err);  
       return callback(new Error(response.err));  
     }  
     // Convert the results into an array of key strings, or  
     // common prefixes if we're using a delimiter.  
     var keys;  
     if (options.delimiter) {  
       // Note that if you set MaxKeys to 1 you can see some interesting  
       // behavior in which the first response has no response.CommonPrefix  
       // values, and so we have to skip over that and move on to the  
       // next page.  
       keys = response.CommonPrefixes.map(function(item) {  
         return item.Prefix;  
       });  
     } else {  
       keys = response.Contents.map(function(item) {  
         return item.Key;  
       });  
     }  
     // Check to see if there are yet more keys to be obtained, and if so  
     // return the marker for use in the next request.  
     var nextMarker;  
     if (response.IsTruncated) {  
       if (options.delimiter) {  
         // If specifying a delimiter, the response.NextMarker field exists.  
         nextMarker = response.NextMarker;  
       } else {  
         // For normal listing, there is no response.NextMarker  
         // and we must use the last key instead.  
         nextMarker = keys[keys.length - 1];  
       }  
     }  
     callback(null, nextMarker, keys);  
   });  
 }  
 // If directory exists delete it and create from scratch  
 // If directory doesnot exist, create it  
 function initDirectorySync(directory) {  
   try {  
     fs.statSync(directory);  
     //if directory exists, remove it   
     cleanDir(directory);  
     //create it clean  
     fs.mkdirSync(directory);  
   } catch (e) {  
     fs.mkdirSync(directory);  
   }  
 }  
 function cleanDir(dir) {  
   rimraf.sync(dir, {  
     maxBusyTries: 1000  
   });  
 }  
 function downloadFile(self, bucket, key, localPath, callback) {  
   var params = { Bucket: bucket, Key: key };  
   var file = fs.createWriteStream(localPath);  
   file.on('close', function() {  
     callback();  
   });  
   file.on('error', function(err) {  
     callback(err);  
   });  
   self.s3.getObject(params).createReadStream().pipe(file);  
 }

Airflow with Celery Executor (MacOs)

2018-03-13T03:53:00.001-07:00

Airflow - Celery Configuration

One of the best Apache projects (for me the best) is Airflow Workflow Platform. The ease of use, the extending documentation and the connectivity with so many existing tools, makes it one of the best tools for creating and monitoring tasks using DAGs.

On my first trials of the product I wanted to use Celery as worker machine. Airflow documentation describes the steps to do so, however I wanted to have more detailed instructions. In the following tutorial we are going to use as executor Celery. Celery will use RabbitMQ as Messaging Queue for the tasks.

(in brief instructions are located in https://airflow.apache.org/configuration.html?highlight=postgres#scaling-out-with-celery)

Install RabbitMQ
1. brew install rabbitmq
(If this fails to link installed file to /usr/local/sbin, the create dir and run brew link rabbitmq)
2. sudo nano ~/.bash_profile
3. Add to PATH :/usr/local/sbin
4. source ~/.bash_profile
5. sudo rabbitmq-server -detached

6. sudo rabbitmqctl add_user airflow airflow
7. Check status sudo rabbitmqctl status

Install Celery
1. pip install Celery==3.1.25 // version 4 seems to have issues
2. Check Celery installation: celery --version

Install Postgres

1. brew install postgresql
2. CREATE DATABASE airflow;

Create login user "airflow" with password "airflow"

4. pip install psycopg2

Run

5. airflow initdb

Config Airflow

1. nano ~/airflow/airflow.config
Change the following lines:
executor = CeleryExecutor
broker_url = amqp://airflow:airflow@localhost:5672/myvhost
celery_result_backend = db+postgres+psycopg2://airflow:airflow@127.0.0.1:5432/airflow
sql_alchemy_conn = postgres+psycopg2://airflow:airflow@127.0.0.1:5432/airflow

2. Run worker with
airflow worker

Tip.
Start scheduler and webserver
airflow webserver
airflow scheduler

References:

1. https://stlong0521.github.io/20161023%20-%20Airflow.html

2. Setup on Ubuntu machines: https://medium.com/a-r-g-o/installing-apache-airflow-on-ubuntu-aws-6ebac15db211

Deep Learning Essentials

2017-12-29T08:43:00.001-08:00

Deep Learning Essentials:

This page will become a reference page of most common used libraries when doing deep learning research in python.

Numpy

np.random
np.random.permutation
np.array([123.68, 116.779, 103.939], dtype=np.float32).reshape((3,1,1))

np.argmax(all_preds, axis=1)
np.array(preds)
np.newaxis
np.zeros

Scipy

from scipy import misc, ndimage

from scipy.ndimage.interpolation import zoom
from scipy.ndimage import imread

csv

with open('results.csv', 'wb') as csv_file:

writer = csv.writer(csv_file)

writer.writerow

for result in results:

for key, value in result.items():

writer.writerow([key, value])

Keras

from keras.preprocessing import image
image.load_img

Linear Model - Dense Layer (Fully Connexted: ax+b)

Deep Learning in Practice

2017-12-28T04:02:00.000-08:00

Lately, I am following and practicing (one or two hours per day) the amazing deep learning lessons from Jeremy Howard and Rachel Thomas. Their teaching methodology is rather "top-down" than "bottom up" so from the first hour you will be able to run successfully image classifications.

Jeremy has put a nice and helpful file with useful utility functions ('utils.py')

I will share here some useful scripts that maybe you will find useful.

Lesson 1: fast-ai-utils.py

https://gist.githubusercontent.com/alexopoulos7/272d3e7d1f57e95d53d84abf45fcccbf/raw/a556dae82766bf3debb9109ed2c57be7863b7297/fast-ai-utils.py

 # fast.ai Lesson 1  
 def create_validation_set(from_path, to_path, percentage = 0.2):  
   count_files_from_path = len(os.listdir(from_path))  
   validation_count = percentage * count_files_from_path  
   print('Lets copy {}/{} files for validation'.format(validation_count, count_files_from_path))  
   for i in range(int(validation_count)):  
     random_path = random.choice(os.listdir(from_path))  
     os.rename(from_path+random_path, to_path+random_path)  
   
   print('Create {} Validation Set finished! '.format(percentage))    
   
 # create_validation_set('/home/ec2-user/courses/deeplearning1/nbs/data/train/cat/', '/home/ec2-user/courses/deeplearning1/nbs/data/validation/cat/')  
 # create_validation_set('/home/ec2-user/courses/deeplearning1/nbs/data/train/dog/', '/home/ec2-user/courses/deeplearning1/nbs/data/validation/dog/')  
   
   
 def move_files_to_categories(from_directory):  
   """  
   It will move labeled images to respective folders.  
   e.g. file cat_1.jpg will be moved to folder cat  
   """  
   for filename in os.listdir(from_directory):  
     parts = filename.split('.')  
     if len(parts) > 2 and parts[2] in ['jpg', 'jpeg', 'png']:  
       category = from_directory + parts[0]  
       if not os.path.exists(category):  
         os.makedirs(category)  
       print('Rename from {}'.format(from_directory+filename))  
       os.rename(from_directory+filename, os.path.join(category, filename))  
       print ('Moved file to {}'.format(os.path.join(category, filename)))  
       
 # move_files_to_categories('/home/ec2-user/courses/deeplearning1/nbs/data/train/')

import shutil
def create_sample_directory(from_directory, to_dir, sample_count = 1000):  
   """  
   It will create sample dir  
   """  
   for d in os.listdir(from_directory):  
        for i in range(sample_count):
            target_dir = to_dir+'validation/'+d
            if not os.path.exists(target_dir):  
                os.makedirs(target_dir)  
            random_path = random.choice(os.listdir(from_directory+d)) 
            
            #print ('Copy file from from {}'.format(os.path.join(from_directory+d, random_path)))  
            #print ('Copy file from to {}'.format(os.path.join(target_dir, random_path)))  
            shutil.copy2(os.path.join(from_directory+d, random_path), os.path.join(target_dir, random_path))
        
        for i in range(sample_count):
            target_dir = to_dir+'/train/'+d
            if not os.path.exists(target_dir):  
                os.makedirs(target_dir)  
            random_path = random.choice(os.listdir(from_directory+d)) 
            
            #print ('Copy file from from {}'.format(os.path.join(from_directory+d, random_path)))  
            #print ('Copy file from to {}'.format(os.path.join(target_dir, random_path)))  
            shutil.copy2(os.path.join(from_directory+d, random_path), os.path.join(target_dir, random_path))
       
create_sample_directory('/home/ec2-user/courses/deeplearning1/nbs/images/train/','/home/ec2-user/courses/deeplearning1/nbs/images/samples/')

  
 def show_image(image_path):  
   import matplotlib.image as mpimg    
   img=mpimg.imread(image_path)  
   imgplot = plt.imshow(img)

Matlab Matrices

2017-07-14T03:50:00.000-07:00

Lately, I have been taking the amazing and very famous Andrews NG Machine Learning Classes on Coursera platform!!

I am amazed on how everything is presented easily and simplified. The exercises and practical assignments come with a lot of code in tutorial code that we need to fill in the gaps.

My conclusion is that in order to master Matlab you have first to master Matlab matrices and vectors. If you want speed in calculation we have first to "matrixize" formulas.

Following are some of the best and most used tips on matrixes (from Quora question).
Always useful is Mathworks help article.

 x = zeros(n,m);     % initialize x to all zeros  
 x = ones(n,m) * 5;    % initialize all x to 5s  
 x(4,:) = 5;       % assign 5 to all elements of row #4  
 x(:,1) = x(:,1) + x(:,2);% add column #2 to column #1  
 y = x(2:5, 1:6);     % copy inner block rows 2-5 and columns 1-6 to y  
 z = x + y;        % add all elements of x and y, store in z  
 z = x * y;        % matrix multiply x and y, store in z  
 z = x .* y;       % individually multiply elements of x and y, store in z  
 z = x .^ 2;       % set each element of z to the same element of x squared  
 w = find(x == 5);    % search x for elements == 5, store integer indices in w  
 w = (x == 5);      % create binary matrix w that has a 1 anywhere x == 5  
 x(w) = 8;        % set value of all elements found above (either case) to 8  
 z = sum(x(:) < 1);     % count how many elements have value < 1  
 x(x < 0) = 0;      % set all negative elements of x to zero  
 y = mean(x, 1);     % set y(1,i) to the average of each column of x  
 y = mean(x(:));     % set y to the average of all elements of x  
 y = mean2(x);      % set y to the average of all elements of x (2D only)  
 x = [y, x];       % prepend columns of y to matrix x  
 x(:,end+1) = 5;     % add a column of 5s to the right side of x  
 z = sqrt(mean(x(:).^2)); % set z to the root-mean-square of all elements in x  
 if any(x(:) == 5)    % true if any element of x == 5  
 if all(x(:) == 0)    % true if all elements of x == 0  
 if any(all(x == 0, 2))  % true if any rows are all zeros  
 x = bsxfun(@times, x, 1./sum(x));  % normalize x so each column sums to 1  
 x = (x - min(x(:))) ./ (max(x(:)) - min(x(:))); % normalize x so min==0 and max==1

Search a directory tree for all files with same content

2017-03-15T07:39:00.001-07:00

 hashmap = {} # content signature -> list of matching filenames  
 for path, dirs, files in os.walk('.'): # walk current directory  
   for filename in files:  
     fullname = os.path.join(path, filename)  
     contents = open(fullname).read()  
     h = hashlib.md5(contents).hexdigest()  
     hashmap.setdefault(h,[]).append(fullname)  
 pprint.pprint(hashmap)

Parse Apache Logs

2017-03-15T07:02:00.000-07:00

Lets create a simple script to parse Apache Logs and print the 20 most common visited urls, I think this is a very useful script that can be easily ported in every case:

 counter = collections.Counter()  
 for filename in glob.glob('var/log/apache2/*.gz'):  
   for line in gzip.open(filename):  
     mo = re.search(r'GET (.*) HTTP/1', line)  
     if mo is not None:  
       url = mo.group(1)  
       counter[url] +=1  
 result = heapq.nlargest(20, counter.items(), key=lambda(url,cnt): cnt)  
 pprint result

Become awesome in Python!!

2017-03-14T06:20:00.000-07:00

Hello and welcome to my series of Python tips and tricks. I am going to list every single advice that I find interesting so that I can create a reference list.

The advice will make our Python code

Beautiful (Beautiful is always good -- also it makes it easy maintainable)
Fast (Fast in processing, less in memory)
Pythonic (A lot of us come with different programming background, To master a language we have to go its way along)
Testable

1. Refactor (Extract) complex if clauses to variables to improve readibility.
e.g.

 if str.lower().endswith('r'):  
   print ('{0} ends with r'.format(str)

can become the more readable:

 str_ends_with_r = str.lower().endswith('r')  
 if str_ends_with_r:  
   print ('{0} ends with r'.format(str)

2. Refactor (Extract) variables into classes to improve testability.
e.g.

Lets create our class:

 class EndsWithR(object):  
   def __init__(self,str):  
     self.str = str  
     self.r = str.lower().endswith('r')  
     self.result = str.lower().endswith('r')  
   def __bool__(self): # __nonzero__   
     return self.result

Lets call our class

str_ends_with_r = EndsWitR(str)  
 if str_ends_with_r:   
      print ('{0} ends with r'.format(str)

So now it is very easy to test our code

 test = EndsWithR('some Example string that ends with r')  
 assert test  
 assert test.r

3. Refactor classes with a lot of properties by Splitting Classes.
Create a parent class and subclass.
Use @Property decorator

4. Exit Loops fast
e.g.

 def search_fast(haystack, needle):   
   for item in haystack:   
     if item == needle:   
       return True   
   return False


 def search_slow(haystack, needle):   
   return_value = False   
   for item in haystack:   
     if item == needle:   
       return_value = True   
   return return_value

5. Open and Write Files uniformly for Python 2 and Python 3

We add 'wb' type to indicate 'Binary mode', otherwise in Python 3 opening a file defaults to reading or writing UTF-8 strings.

 with open(‘/tmp/random.bin’, ‘wb’) as f:  
   f.write(os.urandom(10))

6. Use get method to access dictionary values
e.g. In this way we can get a default value in case dictionary key doesn't exist

dict = {'Name': 'Zabra', 'Age': 7}

print "Value : %s" %  dict.get('Age')
print "Value : %s" %  dict.get('Education', "Never")

Value : 7
Value : Never

7. Reverse a Sequence
You can use sequence slicing [start:end:stride] with stride -1
e.g.

 x = b’mongoose’  
 y = x[::-1]  
 print(y)

b’esoognom’

Be aware though that this will not work as expected for UTF-8 strings

8. Prefer Generators for Large Comprehensions
as they have minimum memory footprint
e.g.

 it = (len(x) for x in open(‘/tmp/my_file.txt’))  
 print(it)

 <generator object <genexpr> at 0x102b81480>

9. Use enumerate() if you need index in looping or begin counting from specific index:
e.g.

 for idx, flavor in enumerate(flavor_list, 2):  
     print(‘%d: %s’ % (idx, flavor))

10. Use zip() to process iterators in parallel.
e.g.

for name, count in zip(names, letters):  
    if count > max_letters:  
        longest_name = name  
        max_letters = count

Actually zip creates a tuple for the smallest list
If you need to iterate to the longest list then use: zip_longest
In Python 3 zip() is a generator function. In Python 2 it creates a new list of tuples. For python 2 we can use izip() to be more efficient.

11. Python can return Multiple Values
This gives extreme freedom and flexibility to return a Value and a Status of execution.
for example following code returns result of division as long the status of execution. The returned is a tuple:

def divide(a, b):  
    try:  
        return True, a / b  
    except ZeroDivisionError:  
        return False, None

# And we can use them as following

success, result = divide(x, y)
if not success:
    print(‘Invalid inputs’)

12. Return Generators instead of Lists
Using the keyword 'yield' we can return Generators instead of Lists so that we can maximize performance and memory consumption
The transformation of code from Lists to Generators is easy:
e.g.

def index_words(text):  
    result = []  
    if text:  
        result.append(0)  
    for index, letter in enumerate(text):  
        if letter == ‘ ‘:  
            result.append(index + 1)  
    return result

# The same code Can Become

 def index_words_iter(text):  
     if text:  
         yield 0  
      for index, letter in enumerate(text):  
          if letter == ‘ ‘:  
              yield index + 1

13. User repr for Classes for easier debugging

 def __repr__(self):  
   return '{0}'.format(self.__class__.__name__)

14. Use namedtuples
Instead of

 p = (170, 0.5, 0.6)

you can use namedtuples and make it much more readable

 from collections import namedtuple  
 Color = namedtuple('Color',['hue', 'saturation', 'luminosity'])  
 p = Color(170, 0.5, 0.6)  
 if p.saturation >0.5:  
   print ('That is bright')

References
1. https://www.youtube.com/watch?v=D_6ybDcU5gc
2. https://www.youtube.com/watch?v=wf-BqAjZb8M

Securing AWS Instances

2017-02-27T00:16:00.000-08:00

Lately I was trying to publish a Web application to Amazon EC2 Cloud. The application had the classic approach of DB - Server Architecture. To be more specific I thought of using 2 t2.medium EC2 instances, one for Web application and the other for the db. Those instances will be used for our staging environment.

As for the db we are using a PostgreSQL Server. We followed this approach and not use an Amazon RDS service as we wanted to host some other small applications in that specific machine instance. I would recommend however not to follow this approach in your cases.

The Application is served from Apache2 Server in port 80 that loads content from a django site. The Application instance is associated with an Elastic Ip, so that we give it to our client.

The most important part for the EC2 instances is the security and how we are going to be safe from the outer world. Luckily, security groups come very handy!

The Application server has opened the 80 port to our clients CIDR IP and our company's IPs. We have also opened SSH Port (22) to our company's IPs.

The difficult part was how to secure our db instance. We do not want any access from the outside, only from the Application Server. To do so at first we thought creating an Inbound Rule from Elastic IP to Port 5432. However, Amazon was not happy about this choice.

The reason as explained from Amazon Docs:

When you specify a security group as the source or destination for a rule, the rule affects all instances associated with the security group. For example, incoming traffic is allowed based on the private IP addresses of the instances that are associated with the source security group.

Solution:
We created a new Inbound Custom Rule using the security group name of the Application Server Instance. It worked like charm! Cheers :-)

Greedy Florist Solution in javascript

2016-12-10T04:29:00.002-08:00

Greedy Florist

Hackerrank solution in javascript

Problem

You and friends want to buy flowers. Each flower has some cost . The florist is greedy and wants to maximize his number of new customers, so he increases the sale price of flowers for repeat customers; more precisely, if a customer has already purchased flowers, price for is .

Find and print the minimum cost for your group to purchase flowers.

Note: You can purchase the flowers in any order.

Input Format

The first line contains two integers, (number of flowers to purchase) and (the size of your group of friends, including you).

The second line contains space-separated positive integers describing the cost () for each flower .

 function processData(input) {  
   input = input.split('\n');  
   let n = parseInt(input[0].split(' ')[0]);  
   let k = parseInt(input[0].split(' ')[1]);  
   let flowers = input[1].split(' ').map(Number).sort((a,b)=>{return b-a});  
   let price = 0;  
   let x =0;  
   for (let i=0; i<n;i++){  
     if(i%k===0 && i!==0){  
       x++;  
     }  
     price +=((x+1)*flowers[i]);  
   }  
   //console.log('Final Price '+price);  
     
   console.log(price);  
     
 }   
   
 process.stdin.resume();  
 process.stdin.setEncoding("ascii");  
 _input = "";  
 process.stdin.on("data", function (input) {  
   _input += input;  
 });  
   
 process.stdin.on("end", function () {  
   processData(_input);  
 });

Mark and Toys Solution in Javascript

2016-12-10T03:05:00.002-08:00

Mark and Toys

Solution in Javascript

problem from Hackerrank.com

Problem

Mark and Jane are very happy after having their first kid. Their son is very fond of toys, so Mark wants to buy some. There are different toys lying in front of him, tagged with their prices, but he has only . He wants to maximize the number of toys he buys with this money.

Now, you are Mark's best friend and have to help him buy as many toys as possible.

Input Format

The first line contains two integers, and , followed by a line containing space separated integers indicating the products' prices

The key thing to the solution is just to sort toys.

 function processData(input) {  
   input = input.split('\n');  
   n = parseInt(input[0].split(' ')[0]);  
   money = parseInt(input[0].split(' ')[1]);  
   toys = input[1].split(' ').map(Number).sort((a,b)=>{return a-b});  
   //console.log('Toys sorted are '+toys);  
   let countToys = 0;  
   let moneyLeft = money;  
   let i = 0;  
   while(n-- && moneyLeft>0){  
     moneyLeft-=toys[i];  
     if(moneyLeft>=0){  
       countToys++;  
       //console.log('Buy item '+i +' with price '+toys[i]+' and moneyLeft = '+moneyLeft);  
       i++;  
     }  
     else{  
       break;  
     }  
   }  
   console.log(countToys);  
 }   
   
 process.stdin.resume();  
 process.stdin.setEncoding("ascii");  
 _input = "";  
 process.stdin.on("data", function (input) {  
   _input += input;  
 });  
   
 process.stdin.on("end", function () {  
   processData(_input);  
 });

Largest Permutation Javascript Solution

2016-12-10T02:43:00.001-08:00

Largest Permutation Solution

from hackerrank.com

Problem:

You are given an array of integers which is a permutation of the first natural numbers. You can swap any two elements of the array. You can make at most swaps. What is the largest permutation, in numerical order, you can make?

Input Format

The first line of the input contains two integers, and , the size of the input array and the maximum swaps you can make, respectively. The second line of the input contains a permutation of the first natural numbers.

Output Format

Print the lexicographically largest permutation you can make with at most swaps.

Solution in Javascript

 function swap(list, index1,index2){  
   //console.log('SWAP '+index1 +' with index '+index2);  
   var b = list[index2];  
   list[index2] = list[index1];  
   list[index1] = b;  
 }  
   
 function processData(input) {  
   input = input.split('\n');  
   let list = input[1].split(' ').map(Number);  
   let n = parseInt(input[0].split(' ')[0]);  
   let k = parseInt(input[0].split(' ')[1]);  
   let result_list = [];  
   let j = 0;  
   let list_index = [];  
   //console.log('List is '+list);  
   for(let i=0; i<n; i++){  
     //console.log(i, list[i]);  
     list_index[list[i]] = i;  
   }  
   let exitMainLoop = false;  
   // console.log('List Index is '+list_index);  
   while (k--) {  
     //console.log('K is '+k);  
     for(let i=j;i<n;i++){  
       //console.log('i = '+i+' j= '+j);  
       //console.log('List[i] = '+list[i]);  
       //console.log('n-i ='+(n-i));  
      if(list[i]!==n-i){  
        let temp = list_index[n-i];  
        //console.log('Temp is '+temp);  
        let t = list_index[list[i]];  
        //console.log('list_index before swapping is '+list_index);  
        swap(list_index,list[i],list[temp]);  
        //console.log('list_index after swapping is '+list_index);  
        //console.log('list before swapping is '+list);  
        swap(list,i,temp);  
        //console.log('list after swapping is '+list);  
          
        break;  
      }  
       // console.log('In order to avoid timeouts we should exit main While loop as soon we have finished swapping all elements');  
       if(i>n-2){  
          exitMainLoop = true;  
        }  
         
     }  
     if(exitMainLoop){  
       break;  
     }  
     j++;  
   }  
     
   console.log(list.join(' '));  
     
 }   
   
 process.stdin.resume();  
 process.stdin.setEncoding("ascii");  
 _input = "";  
 process.stdin.on("data", function (input) {  
   _input += input;  
 });  
   
 process.stdin.on("end", function () {  
   processData(_input);  
 });

Spidering the Web

2016-12-09T04:54:00.002-08:00

Spidering the Web

Spidering (or better crawling) the Web is such a common task for crawlers that a lot of ready to use good quality tools are already available in the web. For example check this page of Top 50 crawlers.

If you are interested on what these programs do in their core as me, you can find a simple spider in my git hub account.

As you can see in code with Python and only in a few lines you can have an honest spider that saves a list of urls links in a text file using all cores of your pc. Parsing a page is a kind of cpu intensive process so spawing this process to all cores we can have production performance.

As a reference check code and comments below:

 from multiprocessing import Pool  
 import bs4 as bs  
 import requests  
   
   
 urls = ['http://martinfowler.com', 'https://www.tutorialspoint.com']  
   
   
 def handle_local_links(url, link):  
   print('Url:{}, Link {}'.format(url,link))  
   if link.startswith('//'):  
     return url  
   elif link.startswith('/'):  
     return ''.join([url, link])  
   elif link.startswith('http'):  
     return link  
   elif link.startswith('#'):  
     return ''  
   else:  
     return ''.join([url, link])  
   
   
 def get_links(url):  
   try:  
     print('Start parsing URL: {}'.format(url))  
     response = requests.get(url)  
     soup = bs.BeautifulSoup(response.text, 'lxml')  
     body = soup.body  
     links = [link.get('href') for link in body.find_all('a')]  
     links = [handle_local_links(url, link) for link in links]  
     # Encode  
     links = [str(link.encode('ascii')) for link in links]  
     return links  
   except TypeError as type_error:  
     # Iterate over None  
     print (type_error)  
     print ('Got a Type Error, probably a None that we tried to iterate')  
     return []  
   except IndexError as index_error:  
     print (index_error)  
     print ('We probably did not find any useful links, returning empty list')  
     return []  
   except AttributeError as attribute_error:  
     print (attribute_error)  
     print ('Probably we got None for links, so we return empty list')  
     return []  
   except Exception as exception:  
     print (str(exception))  
     return []  
   
   
 def main():  
   how_many = 100  
   p = Pool(processes=how_many)  
   data = p.map(get_links, [link for link in urls])  
   print ('Data after multiprocessing {}'.format(data))  
   # data is now a list of lists  
   # lets flatten it to a list  
   data = [url for url_list in data for url in url_list]  
   print ('Data is {}'.format(data))  
   sorted_list = sorted(data)  
   print ('Sorted List is {}'.format(sorted_list))  
   p.close()  
   with open('urls.txt', 'w') as f:  
     f.write('\n'.join(sorted_list))  
   
   
 if __name__ == '__main__':  
   main()

Priyanka and Toys Solution in JS

2016-12-08T14:33:00.004-08:00

Priyanka and Toys Solution

Hackerrank.com

solution in Javascript

Hierarchical Clustering Algorithms Work!!

2016-12-02T03:58:00.000-08:00

Today I was playing with Hierarchical Clustering Algorithms. Hierarchical Clustering Algorithms belongs to Unsupervised Learning algorithms. It creates clusters of similar objects.

So I tried to create clusters of the following blog sites:

http://feeds.feedburner.com/37signals/beMH
http://feeds.feedburner.com/blogspot/bRuz
http://battellemedia.com/index.xml
http://blog.guykawasaki.com/index.rdf
http://blog.outer-court.com/rss.xml
http://feeds.searchenginewatch.com/sewblog
http://blog.topix.net/index.rdf
http://blogs.abcnews.com/theblotter/index.rdf
http://feeds.feedburner.com/ConsumingExperienceFull
http://flagrantdisregard.com/index.php/feed/
http://featured.gigaom.com/feed/
http://gizmodo.com/index.xml
http://gofugyourself.typepad.com/go_fug_yourself/index.rdf
http://googleblog.blogspot.com/rss.xml
http://feeds.feedburner.com/GoogleOperatingSystem
http://headrush.typepad.com/creating_passionate_users/index.rdf
http://feeds.feedburner.com/hotair/main
http://feeds.feedburner.com/instapundit/main
http://jeremy.zawodny.com/blog/rss2.xml
http://joi.ito.com/index.rdf
http://journals.aol.com/thecoolerblog/AOLNewsCooler/rss.xml
http://feeds.feedburner.com/Mashable
http://michellemalkin.com/index.rdf
http://moblogsmoproblems.blogspot.com/rss.xml
http://newsbusters.org/node/feed
http://beta.blogger.com/feeds/27154654/posts/full?alt=rss
http://feeds.feedburner.com/paulstamatiou
http://powerlineblog.com/index.rdf
http://feeds.feedburner.com/Publishing20
http://radar.oreilly.com/index.rdf
http://scienceblogs.com/pharyngula/index.xml
http://scobleizer.wordpress.com/feed/
http://sethgodin.typepad.com/seths_blog/index.rdf
http://rss.slashdot.org/Slashdot/slashdot
http://thinkprogress.org/feed/
http://feeds.feedburner.com/andrewsullivan/rApM
http://wilwheaton.typepad.com/wwdnbackup/index.rdf
http://www.43folders.com/feed/
http://www.456bereastreet.com/feed.xml
http://www.autoblog.com/rss.xml
http://www.bloggersblog.com/rss.xml
http://www.bloglines.com/rss/about/news
http://www.blogmaverick.com/rss.xml
http://www.boingboing.net/index.rdf
http://www.buzzmachine.com/index.xml
http://www.captainsquartersblog.com/mt/index.rdf
http://www.coolhunting.com/index.rdf
http://feeds.copyblogger.com/Copyblogger
http://feeds.feedburner.com/crooksandliars/YaCP
http://feeds.dailykos.com/dailykos/index.xml
http://www.deadspin.com/index.xml
http://www.downloadsquad.com/rss.xml
http://www.engadget.com/rss.xml
http://www.gapingvoid.com/index.rdf
http://www.gawker.com/index.xml
http://www.gothamist.com/index.rdf
http://www.huffingtonpost.com/raw_feed_index.rdf
http://www.hyperorg.com/blogger/index.rdf
http://www.joelonsoftware.com/rss.xml
http://www.joystiq.com/rss.xml
http://www.kotaku.com/index.xml
http://feeds.kottke.org/main
http://www.lifehack.org/feed/
http://www.lifehacker.com/index.xml
http://littlegreenfootballs.com/weblog/lgf-rss.php
http://www.makezine.com/blog/index.xml
http://www.mattcutts.com/blog/feed/
http://xml.metafilter.com/rss.xml
http://www.mezzoblue.com/rss/index.xml
http://www.micropersuasion.com/index.rdf
http://www.neilgaiman.com/journal/feed/rss.xml
http://www.oilman.ca/feed/
http://www.perezhilton.com/index.xml
http://www.plasticbag.org/index.rdf
http://www.powazek.com/rss.xml
http://www.problogger.net/feed/
http://feeds.feedburner.com/QuickOnlineTips
http://www.readwriteweb.com/rss.xml
http://www.schneier.com/blog/index.rdf
http://scienceblogs.com/sample/combined.xml
http://www.seroundtable.com/index.rdf
http://www.shoemoney.com/feed/
http://www.sifry.com/alerts/index.rdf
http://www.simplebits.com/xml/rss.xml
http://feeds.feedburner.com/Spikedhumor
http://www.stevepavlina.com/blog/feed
http://www.talkingpointsmemo.com/index.xml
http://www.tbray.org/ongoing/ongoing.rss
http://feeds.feedburner.com/TechCrunch
http://www.techdirt.com/techdirt_rss.xml
http://www.techeblog.com/index.php/feed/
http://www.thesuperficial.com/index.xml
http://www.tmz.com/rss.xml
http://www.treehugger.com/index.rdf
http://www.tuaw.com/rss.xml
http://www.valleywag.com/index.xml
http://www.we-make-money-not-art.com/index.rdf
http://www.wired.com/rss/index.xml
http://www.wonkette.com/index.xml
http://feeds.feedburner.com/blogspot/eXYjt

Using Pythons PIL Library I have created the dendogram for the clusters created:

Another interesting thing is the Word Clustering Dendogram:

Lets zoom in a small sub cluster to get a better idea:

Trump and Elections

Now if you want to see for yourself or play with the code, please visit my git hub account here: https://github.com/alexopoulos7/collective-intelligence/tree/master/CollectiveIntelligence/Unsupervised%20Learning/Word%20Vectors

Similarity Functions Benchmark in Python

2016-11-28T23:55:00.004-08:00

Lets create a benchmark between 4 of the most common similarity formulas used in recommendations

Euclidean Distance
Pearson Coefficients
Tanimoto
Cosine

The data we are going to use is this simple dictionary of user ratings of movies (sample data is taken from "Programming Collective Intelligence" book by Toby Segaran)

 # Load a dictionary of movie critics and their ratings of a small set of movies  
 critics = {  
   'Lisa Rose': {  
     'Lady in the Water': 2.5,  
     'Snakes on a Plane': 3.5,  
     'Just My Luck': 3.0,  
     'Superman Returns': 3.5,  
     'You, Me and Dupree': 2.5,  
     'The Night Listener': 3.0  
   },  
   'Gene Seymour': {  
     'Snakes on a Plane': 3.5,  
     'Lady in the Water': 3.0,  
     'Just My Luck': 1.5,  
     'Superman Returns': 5.0,  
     'You, Me and Dupree': 3.5,  
     'The Night Listener': 3.0  
   },  
   'Michael Philips': {  
     'Snakes on a Plane': 3.0,  
     'Lady in the Water': 2.5,  
     'Superman Returns': 3.5,  
     'The Night Listener': 4.5  
   },  
   'Claudia Puig': {  
     'Snakes on a Plane': 3.5,  
     'Just My Luck': 3.0,  
     'Superman Returns': 4.0,  
     'You, Me and Dupree': 2.5,  
     'The Night Listener': 4.5  
   },  
   'Mick LaSalle': {  
     'Snakes on a Plane': 4.0,  
     'Lady in the Water': 3.0,  
     'Just My Luck': 1.5,  
     'Superman Returns': 3.0,  
     'You, Me and Dupree': 2.0,  
     'The Night Listener': 3.0  
   },  
   'Jack Matthews': {  
     'Snakes on a Plane': 4.0,  
     'Lady in the Water': 3.0,  
     'Superman Returns': 5.0,  
     'You, Me and Dupree': 3.5,  
     'The Night Listener': 3.0  
   },  
   'Toby': {  
     'Snakes on a Plane': 4.5,  
     'Superman Returns': 4.0,  
     'You, Me and Dupree': 1.0  
   }  
 }

OUTPUT BENCHMARK

Lisa Rose:	[2.5, 3.5, 3.5, 3.0]
Michael Philips:	[2.5, 3.0, 3.5, 4.5]

Euclidean Distance Score between Lisa Rose and Michael Philips is 0.387425886723
Pearson Correlation Coefficient Score between Lisa Rose and Michael Philips is 0.254823595719
Tanimoto Correlation Coefficient Score between Lisa Rose and Michael Philips is 0.666666666667
Cosine Similarity Score between Lisa Rose and Michael Philips is 0.975

Programming Collective Intelligence

2016-11-28T07:33:00.001-08:00

Lately I have been reading the amazing book "Programming Collective Intelligence" book by Toby Segaran

Its a fascinating book if you are trying to understand Machine Learning Algorithms using Python from the scratch.

I have been playing with the code so I am publishing to my git account code examples that I have used from the book.

For the moment I have posted code in one big recommendations.py file algorithms for
user-based and item-based filtering
using

Euclidean Distance
Pearson Approach
Tanimoto
Cosine Similarity

You can simple download or clone repo. Afterwards just run python recommendations.py
Please enjoy

iphone PhotoStream

2016-11-23T08:34:00.002-08:00

https://flic.kr/ps/2X5PvQ

Django Forms Choice Tutorial Sample Code

2016-11-23T04:27:00.006-08:00

Lately I have been playing with Django!! Its a really amazing Python Web Framework that can help anybody with small experience in web frameworks and python language to create amazing web sites.

One of the biggest advantages of using Django is the really detailed documentation. Please check yourself at https://docs.djangoproject.com/en/1.10/.

In this tutorial I am going to show a quick example of creating a simple form with 3 comboboxes that will take some data from 3 tables and save the choices to a 4th table. The sample output will be as following:

Before we start:

For the above scenario we are going to have 3 simple models - tables: Category, Service, Criteria and a more complex one which will store 3 foreign keys for those 3 tables.

Implementation:

1. models.py

The trick here is to have __unicode__ function defined. This will help our comboboxes to populate a unicode string representation of our model.

The __str__ function is used by our Django Admin site

I usually add created_at and updated_at fields for every model I create. I have found it usefull in many cases.

from django.db import models


class Service(models.Model):
    """    Service Model.     """    name = models.TextField(max_length=200, help_text='Service Name')
    short_code = models.TextField(max_length=15, help_text='Service Short Code') 
    created_at = models.DateTimeField(auto_now_add=True, editable=False)
    updated_at = models.DateTimeField(auto_now=True, editable=False)

    def __str__(self):
        return self.name + ' code: ' + self.short_code  

    def __unicode__(self):
        return self.name + ' code: ' + self.short_code  


class Category(models.Model):
    """    # Template Messages Categories (e.g. Help, Bulk, etc    """    name = models.TextField(max_length=200, help_text='Template Message Category. e.g. HELP, STOP, etc')
    description = models.TextField(max_length=1000, help_text='Template Message Category Description')
    created_at = models.DateTimeField(auto_now_add=True, editable=False)
    updated_at = models.DateTimeField(auto_now=True, editable=False)

    def __str__(self):
        return self.name

    def __unicode__(self):
        return self.name


class SubscriptionCriteria(models.Model):
    """    Criteria for Selecting Users to bulk    """    name = models.CharField(max_length=50, help_text='Name your Subscription Criteria')
    description = models.TextField(max_length=1000, help_text='Describe your Subscription Criteria')
    created_at = models.DateTimeField(auto_now_add=True, editable=False)
    updated_at = models.DateTimeField(auto_now=True, editable=False)

    @property    def __str__(self):
        return self.name

    def __unicode__(self):
        return self.name
    
    
class BulkWorker(models.Model):
    """    A Bulk Worker saves the progress of a worker .....     """    task_id = models.TextField(primary_key=True, max_length=150, help_text='Celery task Id')
    launched_at = models.DateTimeField(auto_now_add=True, editable=False)
    charged_at = models.DateTimeField(editable=True, default=None)
    state = models.TextField(max_length=1000)  # TO ADD Exceptions    percent = models.IntegerField(default=0)
    service = models.ForeignKey(Service)
    message_category = models.ForeignKey(Category, default=1)
    criteria = models.ForeignKey(SubscriptionCriteria, default=1)
    created_at = models.DateTimeField(auto_now_add=True, editable=False)
    updated_at = models.DateTimeField(auto_now=True, editable=False)

    def __str__(self):
        return "Worker started at " + str(self.launched_at) + " for service " + str(self.service)

2. views.py

Now that we have our models we can create the views.py.

In home view we just load an empty BulkForm. BulkForm is created below in forms.py

We also add bulk Task to save form data

# coding=utf-8# Create your views here.from __future__ import print_function

from django.contrib.auth.decorators import user_passes_test
from django.shortcuts import render_to_response
from django.template import RequestContext
from django.views.decorators.csrf import csrf_exempt

from .forms import BulkForm


@user_passes_test(lambda u: u.is_superuser)
@csrf_exemptdef home(request):
    """    Main View Page for Bulk Tasks    :param request:    :return:    """    form = BulkForm()
    return render_to_response('bulk.html', {'form': form}, context_instance=RequestContext(request))


def add_bulk_task(request):
    """    Add in db a bulk task    Send to start task    :param request: POST object should be a Request object    :return: HttpResponse    """    context = RequestContext(request)

    if request.method == 'POST':
        form = BulkForm(request.POST)

        if form.is_valid():
            form.save() # commit = True by default            json_data = 'Bulk Task saved fine'        else:
            json_data = 'Form is not valid'    else:
        json_data = 'Only POST requests are allowed'    return HttpResponse(json_data, content_type='application/json')

3. forms.py

In our forms.py we use ModelChoiseField to load the foreign keys. As widget we use forms.Select as we only want to select one value. We query and get all objects populated for all referenced tables in queryset

from django import forms
from .models import BulkWorker, Service, SubscriptionCriteria, Category


class BulkForm(forms.ModelForm):
    """    Create Form for Sending Bulk Requests    """    service = forms.ModelChoiceField(widget=forms.Select, queryset=Service.objects.all())
    message_category = forms.ModelChoiceField(widget = forms.Select, queryset=Category.objects.all())
    criteria = forms.ModelChoiceField(widget = forms.Select, queryset=SubscriptionCriteria.objects.all())

    class Meta:
        model = BulkWorker

4. templates/bulk.html

Lastly we should add the template code so that django will replace data from view.

With Django its the easiest thing to do!

We just need 5 lines of code.

Note: csrf_token is used to prevent cross site forgery attacks

so the line that does all the work is {{form.as_p}}

we cound have used of course the simple {{form}} with the same results, I just wanted to put the form in a paragraph (<p> element)

<div id="bulk-form">    <form id="BulkForm" action="save_bulk" method="post">        {% csrf_token %}
        {{ form.as_p }}
     <input type="submit" name="submit" value="Create BULK Task" />    </form></div>

5. urs.py

Lastly we should add our page in urls.py as simple as following

url(r'^bulk$', 'amusely.views.home', name='bulk'),

url(r'^save_bulk$', 'amusely.views.add_bulk_task', name='save_task'),

As always comments are welcomed. I would be glad to see what you think or help you. CU next time

Internet in Greece

2016-05-20T09:35:00.001-07:00

Stay connected during your vacations in Greece with portable wifi hotspot.

Connect up to 10 devices without extreme fees!!!

Visit

http://www.hotspotonthego.com

Welcome and Enjoy..

Homework Feedback | Algorithms: Design and Analysis, Part I

2013-01-31T11:34:00.001-08:00

Homework Feedback | Algorithms: Design and Analysis, Part I:

Feedback — Programming Question-1

The answer of the 1st programming question is ... 2407905288

Web Site Performance Test

2012-02-13T09:59:00.000-08:00

At last! You have finished that project that made you loose your sleep, friends, or I don't know what else... A really "boring" and very time consuming step is ... yes you know... Testing!

There are a lot of things that you want to test. I will complete the list through time... But now I would like to notice 2 very important tests:

1. Put your enemies to test your site. Make them find whatever bug, fault anything that they don't like. Make sure you Thank them then.

2. MAKE a performance test: If you
want to start with the big picture—how well the server copes with
demand—software like ApacheBench and Siege
will run benchmarks on your Web
server, reporting on how many requests can be handled per second, which is
the standard measuring tool for a site’s performance. Once you start checking your site’s performance, you will fi nd that big, systemwide changes you make
will have the greatest impact. These include:
■ Changing the server hardware: increasing memory, installing faster hard
drives, and using faster processors
■ Changing the demands on the server: disabling unnecessary features,
putting fewer users or sites on a single server, and balancing loads across
multiple servers
■ Caching the PHP output
■ Caching the PHP execution
■ Caching the database results

If you think about the process involved for handling the request of a PHPMySQL
based page, you’ll see three areas where caching can be applied
(Figure 1.4). First, if the database or PHP is caching the results of a database
query, then that query will not need to be executed with each request.
Second, by default, each request of a PHP script requires that the PHP code
be executed as if it had never been run before. By applying an opcode cache
such as the Alternative PHP Cache (APC), the PHP code
itself is cached by the system, making that execution faster. Finally, the end
result is that HTML is sent to the Web browser. If you can cache the dynamically
generated HTML, then no PHP code will be executed at all, no database
queries are required, and the request itself becomes as fast as a request for
a static HTML page.

First Thoughts

Interview AI Assistant

Interview Assistant helps you getyour dream jobusing AI

Nice Ideas from Nice Discussions

Neural Plasticity

Regex to remove comments from SQL Stored Procedure

SQL Remove comments

AWS S3 usefull utilities nodejs

AWS S3 Helpful Utilities - Helper functions using node

Airflow with Celery Executor (MacOs)

Deep Learning Essentials

Deep Learning Essentials:

Deep Learning in Practice

Matlab Matrices

Search a directory tree for all files with same content

Parse Apache Logs

Become awesome in Python!!

Securing AWS Instances

Greedy Florist Solution in javascript

Mark and Toys Solution in Javascript

Mark and Toys

Solution in Javascript

Problem

Largest Permutation Javascript Solution

p.p1 {margin: 0.0px 0.0px 0.0px 0.0px; font: 36.0px 'Avenir Next'; color: #39424e; -webkit-text-stroke: #39424e} span.s1 {font-kerning: none} Largest Permutation Solution

from hackerrank.com

Problem:

Solution in Javascript

Spidering the Web

Spidering the Web

Priyanka and Toys Solution in JS

Hierarchical Clustering Algorithms Work!!

Similarity Functions Benchmark in Python

Programming Collective Intelligence

iphone PhotoStream

Django Forms Choice Tutorial Sample Code

Internet in Greece

Homework Feedback | Algorithms: Design and Analysis, Part I

Feedback — Programming Question-1

Web Site Performance Test

Interview Assistant helps you get
your dream job
using AI

Largest Permutation Solution