Create file system file from file stored in Microsoft SQL database - php

Question

The title sounds confusing, but I don't know how to describe it better.
I've been given the task to do a database migration from program A to program B.
Program A uses a MSSQL database and stores all its files in the database.
Program B handles files with stored in the "normal" way, meaning in the file system.
Now I have to extract, transform and download the database stored files to file system files with PHP, but I was unsuccessful in my attempts to convert them.
For testing purposes, I created a simple .txt file with the content Test document for migration and program A stores it like this in the database:
0x5465737420646F63756D656E7420666F72206D6967726174696F6E'
What format is that, and how do I convert it to a normal document.txt file ?
Edit
Lot of thanks to #PanagiotisKanavos. This now does work with a stream:
$query = "select top(1) DESCRIPTION, FILETYPE, DOCUMENT from dbo.Documents;";
$stmt = sqlsrv_query($this->sqlsrv_conn, $query);
if (sqlsrv_fetch($stmt)) {
$document = sqlsrv_get_field($stmt, 2, SQLSRV_PHPTYPE_STREAM(SQLSRV_ENC_BINARY));
$fileName = sqlsrv_get_field($stmt, 0, SQLSRV_PHPTYPE_STRING(SQLSRV_ENC_CHAR));
$ext = sqlsrv_get_field($stmt, 1, SQLSRV_PHPTYPE_STRING(SQLSRV_ENC_CHAR));
file_put_contents(
$fileName . '.' . $ext,
stream_get_contents($document),
);
}
Now what is the most efficient way to do this with ALL the files? Do I have to execute a query for each and every row?
With PDO I could use $stmt->fetchAll(FETCH_ASSOC) which gave me a nice array of assoc arrays with data inside.
sqlsrv has a similar function sqlsrv_fetch_array which is explained php.net and [docs.microsoft] with the following example:
while( $row = sqlsrv_fetch_array( $stmt, SQLSRV_FETCH_ASSOC) ) {
echo $row['LastName'].", ".$row['FirstName']."<br />";
}
But as much as I could search, I couldn't find a way to loop over the result set without fetching it and then fetching each row individually with the stream and string type mixed. sqlsrv_fetch_array accepts only SQLSRV_FETCH_NUMERIC, SQLSRV_FETCH_ASSOC and SQLSRV_FETCH_BOTH and then the result set is already fetched, and I can't use sqlsrv_get_field to set the type of each field.
I can't be the first person ever that needs something like this, but I'm unable to find anything about it. Probably I'm searching wrongly, or I misunderstood a concept.

Answer 1

I found the solution!
I tested multiple online tool to decode strings and find out what the encoding was.
Even tried hex2bin() but all the tools told me that it wasn't a valid hex string.
Until I stumbled upon this godsent tool which translated the invalid hex to a ? but translated the rest of it which resulted in:
?Test document for migration
From then on it was easy to deduct that 0x was the troublemaker. After removing it, the conversion works like a charm and I could "convert" even complexer files like .doc. Here is the code:
file_put_contents(
// 'DESCRIPTION' is the file name
'files/' . $dbDocument['DESCRIPTION'] .
// 'FILETYPE' is the extension
'.' . mb_strtolower($dbDocument['FILETYPE']),
// 'DOCUMENT' is the document content hex encoded with prepended '0x'
hex2bin(str_replace('0x', '', $dbDocument['DOCUMENT']))
);

Answer 2

This isn't a format, it's how some database administration tools, eg SSMS, display binary data. The data is already binary and doesn't need to be converted.
Reading Large Objects (LOBs) as if they were numbers or strings is really slow though, caching the entire document in both server and client memory, even though the object won't be reused, and doesn't even need to be held in memory. After all, a BLOB in SQL Server can be 2GB and more. That's why almost all databases and data access libraries allow handling LOBs as file streams.
Microsoft's PHP doc examples show how to read LOBs as file streams both for PDO and SQLSRV.
Copying from the example, this parameterized query will search for a user's picture:
/* Get the product picture for a given product ID. */
$tsql = "SELECT LargePhoto
FROM Production.ProductPhoto AS p
JOIN Production.ProductProductPhoto AS q
ON p.ProductPhotoID = q.ProductPhotoID
WHERE ProductID = ?";
$params = array(&$_REQUEST['productId']);
/* Execute the query. */
$stmt = sqlsrv_query($conn, $tsql, $params);
Instead of reading the entire picture as a single value though, it's loaded as a file stream :
$getAsType = SQLSRV_PHPTYPE_STREAM(SQLSRV_ENC_BINARY);
if ( sqlsrv_fetch( $stmt ) )
{
$image = sqlsrv_get_field( $stmt, 0, $getAsType);
fpassthru($image);
}
else
{
echo "Error in retrieving data.</br>";
die(print_r( sqlsrv_errors(), true));
}
$getAsType = SQLSRV_PHPTYPE_STREAM(SQLSRV_ENC_BINARY); specifies that the data will be retrieved as a stream.
$image = sqlsrv_get_field( $stmt, 0, $getAsType); retrieves the 1st field using the specified type, in this case a stream. This doesn't load the actual contents.
fpassthru copies the stream contents directly to the output. The picture may be 2GB but it will never be held in the web server's memory.