File size distribution: Difference between revisions

Line 392:

=={{header|Haskell}}==

Uses a grouped frequency distribution. Program arguments are optional. Arguments include starting directory and initial frequency distribution group size. Distribution groups of 0 are removed. After the first frequency distribution is computed it further breaks it down for any group that exceeds 25% of the total file count, when possible.

<lang haskell>{-# LANGUAGE LambdaCase #-}

⚫

~~<lang haskell>~~import Control.Concurrent (forkIO, setNumCapabilities)

⚫

import Control.Concurrent (forkIO, setNumCapabilities)

import Control.Concurrent.Chan (Chan, newChan, readChan,

writeChan, writeList2Chan)

Line 424:

Line 426:

fileSizes :: [Item] -> [Integer]

fileSizes = foldr f []

fileSizes = foldr f [] where f (File _ n) acc = n:acc

⚫

f _ acc = acc

where

f (File _ n) acc = n:acc

⚫

f _ acc = acc

folders :: [Item] -> [FilePath]

folders = foldr f []

folders = foldr f [] where f (Folder p) acc = p:acc

⚫

f _ acc = acc

where

f (Folder p) acc = p:acc

⚫

f _ acc = acc

totalBytes :: [Item] -> Integer

Line 462:

Line 460:

placeGroups n fgMap = case findGroup n groupMinMax of

Just k -> Map.alter incrementCount k fgMap

Nothing -> ~~error~~ "Should never happen"

Nothing -> fgMap -- Should never happen.

expandGroups :: Int -- ^ Desired number of frequency groups.

-> [Integer] -- ^ List of collected file sizes~~. Must be sorted~~.

-> [Integer] -- ^ List of collected file sizes.

-> Integer -- ^ Computed frequency group limit.

-> FrequencyGroups -- ^ Expanded 'FrequencyGroups'

expandGroups gsize fileSizes groupThreshold

| groupThreshold > 0 = loop 15 $ frequencyGroups gsize ~~fileSizes~~

| groupThreshold > 0 = loop 15 $ frequencyGroups gsize sortedFileSizes

| otherwise = frequencyGroups gsize ~~fileSizes~~

| otherwise = frequencyGroups gsize sortedFileSizes

where

⚫

sortedFileSizes = sort fileSizes

loop 0 gs = gs -- break out in case we can't go below threshold

loop n gs

loop n gs | all (<= groupThreshold) $ Map.elems gs = gs

⚫

| otherwise = loop (pred n) (expand gs)

| all (<= groupThreshold) $ Map.elems gs = gs

⚫

| otherwise = loop (pred n) (expand gs)

expand :: FrequencyGroups -> FrequencyGroups

Line 491:

Line 489:

:: FGKey -- ^ Group Key

-> Integer -- ^ Count

-> Maybe (FGKey, FrequencyGroups) -- ^ ~~Tuple~~ ~~with key and~~ 'FrequencyGroups' to ~~replace~~ ~~the~~ ~~key~~

-> Maybe (FGKey, FrequencyGroups) -- ^ Returns expanded 'FrequencyGroups' with base key it replaces.

groupsFromGroup (min, max) count

| length range > 1 = Just ((min, max), frequencyGroups gsize range)

| otherwise = Nothing

where

range = filter (\n -> n >= min && n <= max) ~~fileSizes~~

range = filter (\n -> n >= min && n <= max) sortedFileSizes

displaySize :: Integer -> String

Line 533:

Line 531:

percentage :: Double

percentage = (realToFrac count / realToFrac filesCount) * 100

~~bars~~ = ~~replicate (~~round percentage~~) '█'~~

size = round percentage

bars | size == 0 = "▍"

| otherwise = replicate size '█'

parseArgs :: [String] -> Either String (FilePath, Int)

Line 563:

main :: IO ()

main = parseArgs <$> getArgs >>= \case

main = do

args <- getArgs

case parseArgs args of

Left errorMessage -> hPutStrLn stderr errorMessage

Right (path, groupSize) -> do

items <- parallelItemCollector path

let (fileCount, folderCount) = counts items

printf "Total files: %d\n" fileCount

printf "Total files: %d\nTotal folders: %d\n" fileCount folderCount

printf "Total folders: %d\n" folderCount

printf "Total size: %s\n" $ displaySize $ totalBytes items

~~putStrLn~~ "\nDistribution:\n"

printf "\nDistribution:\n\n%9s <-> %9s %7s\n" "From" "To" "Count"

printf "%9s <-> %9s %7s\n" "From" "To" "Count"

putStrLn $ replicate 46 '-'

let results = expandGroups groupSize (~~sizes~~ items) (groupThreshold fileCount)

let results = expandGroups groupSize (fileSizes items) (groupThreshold fileCount)

mapM_ (displayFrequency fileCount) $ Map.assocs results

where

⚫

~~sizes~~ = sort . fileSizes

groupThreshold = round . (*0.25) . realToFrac</lang>

<pre style="height: 50rem;">$ filedist ~/Music ~~1 ↵~~

<pre style="height: 50rem;">$ filedist ~/Music

Using 4 worker threads

Total files: 688

Line 597:

Line 592:

243B <-> 323B = 99 14.390%: ██████████████

323B <-> 645B = 23 3.343%: ███

646B <-> 968B = 2 0.291%:

646B <-> 968B = 2 0.291%: ▍

969B <-> 1.26KB = 1 0.145%:

969B <-> 1.26KB = 1 0.145%: ▍

3.19KB <-> 6.38KB = 12 1.744%: ██

6.38KB <-> 9.58KB = 22 3.198%: ███

Line 608:

Line 603:

108.41KB <-> 162.61KB = 23 3.343%: ███

162.61KB <-> 216.81KB = 8 1.163%: █

236.46KB <-> 472.93KB = 3 0.436%:

236.46KB <-> 472.93KB = 3 0.436%: ▍

709.39KB <-> 945.85KB = 44 6.395%: ██████

3.30MB <-> 4.96MB = 4 0.581%: █

Line 614:

Line 609:

6.67MB <-> 13.33MB = 72 10.465%: ██████████

13.33MB <-> 20.00MB = 6 0.872%: █

20.00MB <-> 26.66MB = 1 0.145%:

20.00MB <-> 26.66MB = 1 0.145%: ▍

$ filedist ~/Music 10

Line 631:

Line 626:

267B <-> 355B = 57 8.285%: ████████

356B <-> 444B = 20 2.907%: ███

801B <-> 889B = 2 0.291%:

801B <-> 889B = 2 0.291%: ▍

959B <-> 1.87KB = 1 0.145%:

959B <-> 1.87KB = 1 0.145%: ▍

3.75KB <-> 4.68KB = 1 0.145%:

3.75KB <-> 4.68KB = 1 0.145%: ▍

4.68KB <-> 5.62KB = 1 0.145%:

4.68KB <-> 5.62KB = 1 0.145%: ▍

5.62KB <-> 6.55KB = 11 1.599%: ██

6.56KB <-> 7.49KB = 10 1.453%: █

Line 650:

Line 645:

94.59KB <-> 189.17KB = 42 6.105%: ██████

189.17KB <-> 283.76KB = 4 0.581%: █

283.76KB <-> 378.35KB = 2 0.291%:

283.76KB <-> 378.35KB = 2 0.291%: ▍

851.28KB <-> 945.87KB = 44 6.395%: ██████

2.67MB <-> 5.33MB = 5 0.727%: █

Line 656:

Line 651:

8.00MB <-> 10.67MB = 35 5.087%: █████

10.67MB <-> 13.33MB = 16 2.326%: ██

13.33MB <-> 16.00MB = 3 0.436%:

13.33MB <-> 16.00MB = 3 0.436%: ▍

16.00MB <-> 18.67MB = 3 0.436%:

16.00MB <-> 18.67MB = 3 0.436%: ▍

24.00MB <-> 26.66MB = 1 0.145%:

24.00MB <-> 26.66MB = 1 0.145%: ▍

</pre>