# squidge.py
# version 1.1.0
# dated 22 December 2000
# by Julius Welby
# for information see http://www.outwardlynormal.com/python/squidge.htm
# A Python module which supports two functions:
#
# A "squidge" takes a directory of HTML (or other) files and
# copies them into a single file, delimited by user defined tags
# (default is <SQUIDGEDFILE NAME="filename"> and </SQUIDGEDFILE>).
#
# An "unsquidge" takes the file created by a squidg and re-exports the
# sub-files into a new directory, adjacent to the original squidged file.
#
# Why? To allow a single operation (eg. a spellcheck)on the squidged
# file to be effectively sitewide once the site is unsquidged.
# This module has been written so that the squidged output is in XML,
# though by default the original file extension is preserved.
# 
# New in v1.1.0 - the files appear in the squidged file in name order.
# This is not important in terms of the squidge/unsquidge process,
# but may be useful if you need to display a squidged file in a browser.

import os, glob

path = ''

# option -you can name the root and first sub-elements whatever you want.
root_element = "SQUIDGEDFILES"
sub_element = "SQUIDGEDFILE"

# the following two declarations can be edited if necessary.
xml_declaration = '<?xml version="1.0" standalone = "yes"?>\n'
doctype_declaration = "<!DOCTYPE" + root_element + ">\n"

# create top and tail strings
xml_header = xml_declaration + doctype_declaration
xml_footer = "</" + root_element + ">"

# Functions to set-up the job

def welcome():
    # intro text
    print '\nWelcome to squidge.py'
    print '\nDo you wish to squidge a folder of files (option 1)'
    print 'or unsquidge a file to a folder (option 2)?'
    getoption()

def getoption():
    #squidge or unsquidge?
    option = raw_input('Please enter 1 or 2 --> ')
    if option == '1' or option == '2':
        getdir(option)
    else:
        print '\nYou must enter the digit 1 or the digit 2.'
        getoption()

def pathcheck(path, option):
    # is the working folder real?
    pathcheck = os.path.exists(path)
    if not pathcheck:
        print '\nThe folder path does not exist. Please re-enter.'
        getdir(option)
    if pathcheck:
        pass
        
def getparam(variable, prompt):
    # general text input function
    variable = raw_input(prompt)
    return variable

def getextn():
    # what type of files should we work with? (Only one type (per squidge) currently supported)
    print '\nPlease enter the relevant file extension. (e.g. htm)'
    ex = getparam('extn', 'File extension --> ')
    return ex

def getdir(option):
    # locate the relevant directory
    if option == '1':
        print '\nPlease enter the path of the folder containing the files.'
        path = getparam('path', 'Full path --> ')
        pathcheck(path,option)
        ex = getextn()
        squidge(path, ex)

    elif option == '2': 
        print '\nPlease enter the path of the folder containing the squidged file.'
        path = getparam('path', 'Full path --> ')
        pathcheck(path,option)
        ex = getextn()
        unsquidge(path, ex)

def squidge(path, ex):
    # squidge those files!
    store = ''
    # glob.glob1 looks into the folder (arg1) for matches to arg2, returns list of matches
    filenames = glob.glob1(path, '*'+ '.' + ex)  
    if 'squidged.' + ex in filenames:            # if there is already a squidged file, don't count it
        number = len(filenames) - 1
    else:
        number = len(filenames)                  # number of files to squidge
    if number == 0:
        print '\nThere are no .' + ex + ' files in the specified directory.'
        print 'Please re-enter.'
        getdir('1')
    elif number == 1:
        print '\nThere is only ' + `number` + ' files in the specified directory.'
        print 'squidging one file will do nothing. Lets just forget it, shall we?'
    else:
        print '\nThere are ' + `number` + ' squidgable files in this directory with the extension ' + ex + '.'
        print 'The squidged file will appear in the same folder as the files.'
        print
        store = makestore(filenames, ex, path)      # create the single string consisting of the wrapped individual files
        name = 'squidged.' + ex                     # name of squidged output file
        writefile(store, name, path)                # write the content into a new file adjacent to the originals
        print 'Squidge completed successfully.'
        # welcome()                                 # uncomment this line to restart on completion of squidge

def makestore(filenames, extn, path):
    # create the single string consisting of the wrapped individual files
    filedict = {}
    store = ''
    for file in filenames:
        if not file == ('squidged.' + extn):
            inputfile = os.path.join(path, file)
            infile = open(inputfile, 'r')
            content = infile.read()
            end = '\n</SQUIDGEDFILE>\n'
            filedict[file] = content
        else:
            pass
    thekeys = filedict.keys()
    thekeys.sort()                                   # sort keys to order the files in the squidged file as you would expect.
    for file in thekeys:
        start = '\n<SQUIDGEDFILE>\n<SQUIDGENAME name="' + file + '"/>\n'
        store = store + '\n' + start + filedict[file] + end
    store = xml_header + store + xml_footer # optional XML wrapping
    return store

def writefile(content, filename, folder):
    # output content to a file called filename in named folder
    output = open(os.path.join(folder, filename), 'w')
    output.write(content)
    output.close()

# the unsquidge function
def unsquidge(path, ex):
    name = 'squidged.' + ex
    sourcefile = open(os.path.join(path, name), 'r')
    sourcelist = sourcefile.readlines()
    i = 0
    startlist = []
    endlist = []
    filesdict = {}
    for item in sourcelist:
        if item == '<SQUIDGEDFILE>\012':
            startlist.append(i)
        elif item == '</SQUIDGEDFILE>\012':
            endlist.append(i)
        i = i + 1
    startlen = len(startlist)
    endlen = len(endlist)
    if not startlen == endlen:
        print 'There is a missing file start or end tag.'
        print 'File cannot be processed. Sorry.'
    else:
        for i in range(startlen):
            filestart = startlist[i] + 1
            fileend = endlist[i]
            filesdict[sourcelist[filestart]]=(sourcelist[filestart:fileend])
        output(path, filesdict)

def output(path, filesdict):
    try:
        outfolder = os.path.join(path, 'unsquidged')
        os.mkdir(outfolder)
        for nameline in filesdict.keys():
            filename = nameline[19:-4]
            content = listtostring(filesdict[nameline][1:])
            writefile(content, filename, outfolder)
        print '\nUnsquidge completed successfully.'
        print 'The unsquidged files will appear in a folder called "unsquidged" adjacent to the original files.'
    except OSError:
        print "It looks like there is already a folder of unsquidged files in that location. Please rename or delete it."
       
def listtostring(contentlist):
    string = ''
    length = len(contentlist)
    for i in range(length):
        string = string + contentlist[i]
    return string

welcome()