#!/usr/bin/python
"""
FileStats.py
P. TenHoopen, 11/21/08

Read the statistics of files from a specified directory and subdirectories.
Display number of files, types and number of files, average size, smallest size, largest size
"""

# Import Modules
import os

# Global Variables
extensions = {}         # dictionary to hold extensions and counts
largestFileSz = 0       # size of largest file
largestFile = ""        # name of largest file
numFiles = 0            # number of files found
sumFileSizes = 0.0      # total size of files used to calc average file size

# Functions

def recordExtension(thisExtension):
    """This function keeps track of the extensions that are found."""

    # add to dictionary if new, or update count if existing
    if thisExtension not in extensions:
        extensions[thisExtension] = 1
        #print "New file type '%s' was added" % thisExtension  # debug
    else:
        extensions[thisExtension] += 1
        #print "Count of %s is %s" % (thisExtension, extensions[thisExtension] )  # debug

    return 1

def collectFileInfo(fileName):
    """This function examines file statistics to keep track of largest file and total file sizes."""
    
    # use global variables defined above
    global largestFileSz
    global largestFile
    global sumFileSizes

    # collect file statistics
    fileStats = os.stat(fileName)
    # get file size
    size = fileStats.st_size
    # keep size of all files
    sumFileSizes += size
    # is this the largest file found so far?
    if size > largestFileSz:
        largestFileSz = size
        largestFile = fileName

    return 1

def displayStats():
    """This function displays the collected statistics."""
    
    print "\n\nResults:"
    print "\nThere were %d files found." % numFiles
    print "\nCount \t File Ext"
    print "----- \t --------"
    for extension in extensions:
        print "%d \t %s" % (extensions[extension], extension)
    print "\n"
    print "Largest file:  %s at %d bytes" % (largestFile, largestFileSz)
    print "Total size of files is %d bytes" % (sumFileSizes)
    print "Average size of files is %.2f bytes" % (sumFileSizes / numFiles)
    print "\n"

    return 1

def main():
    """This function is called when the program starts.  It analyzes the files specified in the topDir variable."""

    # variables
    topDir = "."   # directory to search, default to current directory

    # use global variables defined above
    global numFiles
   
    # get top directory to analyze from user
    userDir = raw_input("Input the directory to analyze: ")
    if (len(userDir) > 0):  # weak error checking!
        topDir = userDir    # as long as the user enters something, use it otherwise use "."
 
    print "\nAnalyzing files in '%s'..." % topDir
    
    # walk through the files
    walklist = os.walk(topDir)

    # os.walk returns {dirpath, dirnames, filenames}

    for root, dirs, files in walklist:
        # examine each file found
        for eachFile in files:
            numFiles += 1
            # look for a extension delimiter (.) in the file name
            if "." in eachFile:
                # get the file extension using a string slice
                # extract all text after the period
                extension = eachFile[eachFile.index(".")+1:]
            else:
                # no period delimiter so no extension can be determined
                extension = "none"

            #print entry  # debug
            #print extension  # debug
            
            # keep track of extensions that are found
            recordExtension(extension)

            # get file name and path
            fileName = os.path.join(root, eachFile)e
            
            # examine files stats
            collectFileInfo(fileName)

    # display the results of the analysis
    displayStats()
    
    return 1

# This calls the 'main' function when this script is executed.
if __name__ == '__main__': main()