getResource method about unifiedarchive HOT 9 CLOSED

I've added getFileResource() in 0.2 branch that will be released soon. Now it supports native streams only for zip/rar and one-file archives, all other types simulate streaming.

from unifiedarchive.

There is no need to open access to underlying resources in API, so you can extract all files and then open them as you want.

from unifiedarchive.

why do extra work to extract all files when just one is needed ?

from unifiedarchive.

Because splitting text files into array of lines is just one case, and create another method in API to access file stream is overhead for api. I'll think about this, but for now this opportunity is not necessary.
If server configuration allow you read all file in memory, then why use streams? Just for fgets()?
If file is too big, just unpack it on storage and work like with normal file.

from unifiedarchive.

I was just suggesting this method to avoid overhead. What you propose does this:

1. read archive from disk
2. write unpacked archive to RAM (this is part of unpacking)
3. write file to disk
4. open file pointer to file (in case of using streams)
5. write split lines to RAM (php strings)
6. remove file from disk

If you can use streams then you can go directly from RAM to RAM (step 2 to step 5) and you can skip step 3 and step 6. This should be a lot faster (also depending on what kind of disk you use).

The getResource method won't be called in other use cases, so the overhead you mention is only when parsing the PHP file to generate opcodes. If you have opcache on then this overhead is gone. You still need a few bytes in RAM to register a new method (to have the opcodes loaded).

from unifiedarchive.

If other people agree that this useful method, I will add it.

from unifiedarchive.

So, anybody else thinks it's a good addition?

from unifiedarchive.

Please don't implement it just for me, it's not worth to do it for 1 person.

from unifiedarchive.

Ok, may be in future I will rethink this, but now it's closed.

from unifiedarchive.