Upload naming convention

I'm creating a survey in which users can upload about 6 pictures. What would be the best way for me to store these in the file system? Users can do multiple surveys (thus tracking their improvement over time)...开发者_JAVA技巧so that means I can't just name them sequentially...

This is a CI/PHP app. :)

Or for knowledge's sake, how would you do it?

Wow, this is kinda open-ended... I think I'd probably structure a directory for each user, then name the images based upon the datetime they were uploaded. So it would look something like this:

/+
 |
 +--+ Kyle
    |
    +-- 20110113114903.png
    |
    +-- 20110113115122.png
    |
    +-- 20110114010304.png
 +--+ Kevin
    |
    +-- 20101114123456.png
    |
    +-- 20110102023648.png
    |
    +-- ...

However, this may not be as scalable as you like. You might want to make the directories deeper, so each person gets a directory per date, with images inside.

Mapping all this in a database to where you have them stored on the file system would be the best result in searchable surveys and users.

// Pseudo Database
// user table
user_id (pk)
user_name
user_filesystem_dir
...

// survey table
survey_id (pk)
survey_name
survey_filesystem_dir
...

// Image table
image_id (pk)
image_name
image_filesystem_dir
user_id_fk (this is the user_id from the user table)
survey_id_fk (this is the survey_id from the survey table)
...

// Find all the images for user id 123 and survey 456
SELECT * FROM image_table AS it, user_table AS ut, survey_table AS st
JOIN (it.user_id_fk = ut.user_id AND it.survey_id_fk = st.survey_id)
WHERE user_id_fk = 123
AND survey_id_fk = 456

// Pseudo Code
$user_id     = 123;
$survey_id   = 456;
$survey_name = 'Stack Survey #1';

$image_array = array(
    'image1.png',
    'image2.png',
    'image3.png',
    'image4.png',
    'image5.png',
    'image6.png'
);

// Folder Structure where User is most searched 
+-+123 <user_id>
  |
  +-467 <survey_id>
  | |
  | +-images
  |   |
  |   +-image1.jpg
  |   +-image2.jpg
  |   +-image3.jpg
  +-456 <survey_id>
    |
    +-images
      |
      +-image1.png
      +-image2.png
      +-image3.png
      +-image4.png
      +-image5.png
      +-image6.png

// Folder Structure where Survey is most searched 
+-+467 <survey_id>
| |
| +-123 <user_id>
|   |
|   +-images
|     |
|     +-image1.jpg
|     +-image2.jpg
|     +-image3.jpg
+-456 <survey_id>
    |
    +-123 <user_id>
      |
      +-images
        |
        +-image1.png
        +-image2.png
        +-image3.png
        +-image4.png
        +-image5.png
        +-image6.png

Consider sequential numbering:

user123_survey1_image6.jpg

The ID of the user being 123, and this being the user's sixth image, or even more easily:

user123_survey56_image899.jpg

where user 123's sixth image happens to have the ID 899.

If you track the individual users as is suggested by "thus tracking their improvement over time" I would start with that being the top level directory. However, even that can lead to a directory bloat as the number of users increases.

In one application I wrote where I had to create a file for every user in a system, I wrote a routine (was actually database side) that broke down their unique id into a structured set of folders.

For example: user 12345 upload #5 could be stored in UPLOAD_DIR/1/2/3/4/5/upload_5.png

If you wanted something a little less "guessable" you could still use something like the above but instead of just using their id, you could hash it but follow a similar pattern.

One alternative would be to name your files randomly with uniqid(). You would have to check to make sure that that uniqid does not already exist of course. Something like this.

do {
    $filename = uniqid().'.jpg';
} while (file_exists('images/'.$filename));