task07_2_alternative

Second part of seventh task in EPAM BelSUT Training. Alternative version.

I learned recursion, recursive file renaming. In this task, I wrote a recursion with the transfer of work to new threads.

My idea was: do not wait for the end of sub-loops to make the program run faster (~2 seconds in this case, ha-ha).

---

↓ How it works ↓

---

The recursive loop is configured to define: a folder before it or a file. In any case, it creates a subcycle:

• for a folder: a new cycle will create a renamed folder (ADC_ + original name) in /out directory and then watch the contents of the folder.
• for a file: it will start changing the contents of the file and will save this file to /out directory under the edited name (with _ADC_ in name of file).

Eventually:

1: The first cycle starts looking at the main ADC folder. If the loop finds a folder, then it creates a new subcycle (2), passing the path of this subfolder to it and, without waiting for the subcycle to complete, looks at the rest of the files / folders. For each folder it finds, it creates a subcycle in a new thread (2) and looks further. When a file is found, it creates a thread for operations on the file (A) and immediately looks further. When the traversal of the ADC folder finishes, the main thread exits.

2: At the same time: each subcycle looks at its subfolder and creates a sub-subcycle (3) for folders or a new thread to work with each new file. It continues to run without waiting for the threads it has created to finish running.

3: Recursion.))

A: Each file is processed in an independent thread. Each cycle creates a destination folder, only then it finds the files and creates a thread for them for processing. Thus, work with the file begins after the presence of the directory of its destination.

---

In this task the main ADC folder contains many subfolders, they contain sub-subfolders... and they contain only one file each. Therefore, I think creating threads for subfolders is a good solution, but creating threads for each file is a dubious solution. With other initial data: if there are 100+ (a lot of) files in some folder, then the loop will need to sequentially process each file, which can be a long process. In such a case, splitting the file processing into separate threads can be a good solution.

P.S.: Such implicit recursion is more secure against stack overflow: the main loop does not know about the number of inner loops because it does not expect them = does not follow them.

P.P.S.: I understand that for small amounts of renamed data, the multithreaded implementation of the program is redundant.