#!/usr/bin/env python3
"""
=================================================================================
Simple tactical script to compare thumbspage results, sans date-generated lines.

Filters out two galleries' date-generated comments, and compares their folders.
Pass the pathnames of the two folders to compare as command-line arguments.
To compare folders, this uses the diffall program, if present; install it from
learning-python.com/mergeall.html, and set its install path in diffhome below.

This walks folder trees, because galleries can have nested gallery subfolders.
Spurious diffs in the host site include readme copies and ip-anon.py inserts.
An awk or sed script could probably do this too, but do we really want it to?
=================================================================================
"""

import os, sys, shutil
trace = print
diffall = True    # run diffall to compare results?
cleanup = True    # remove sans-dates temp folders?

# if diffall: number end lines to show, install path
diffshow = 50
diffhome = '/Users/me/MY-STUFF/Code/mergeall'    # edit me

if len(sys.argv) != 3:
    print('Usage: cmp-thumbspage-results.py gallerypath1 gallerypath2')
    sys.exit(1)

for i in (1, 2):
    if not os.path.isdir(sys.argv[i]):
        print('Gallery path %d is invalid:' % i, sys.argv[i])
        sys.exit(1)


# Replace date lines in HTML files

# clean temp dirs in '.'
temp1, temp2 = 'cmp-thumbspage-results-temp1', 'cmp-thumbspage-results-temp2'
for temp in (temp1, temp2):
    if os.path.exists(temp):
        shutil.rmtree(temp)

# copy thumbspage result folders
shutil.copytree(sys.argv[1], temp1)
shutil.copytree(sys.argv[2], temp2)

dateline = '<!-- Generated'
dropline = '<!-- date -->'

def changeone(path):
    """
    Strip date line in one file
    """
    lines = open(path, 'r', encoding='utf8').readlines()
    if any(line.startswith(dateline) for line in lines):
        lines = [dropline if line.startswith(dateline) else line for line in lines]
        open(path, 'w', encoding='utf8').writelines(lines)
        trace('...', os.sep.join(path.split(os.sep)[1:]))
        return 1
    return 0

def changeall(target):
    """
    Strip date lines in entire folder tree
    """
    changed = 0 
    for (adir, subs, files) in os.walk(target):
        for file in files:
            path = os.path.join(adir, file)
            if path.lower().endswith('.html'):
                changed += changeone(path)
    return changed

print('--Removing date lines--\n')

nchanged = changeall(temp1)
print('Files changed in path1 copy:', nchanged, end='\n\n')

nchanged = changeall(temp2)
print('Files changed in path2 copy:', nchanged, end='\n\n')


# Then diffall to compare?

if diffall:
    # Compare folder trees
    trace('--Comparing with diffall--\n')
    python = sys.executable or 'python3'
    if not os.path.exists(diffhome):
        print('diffall home does not exist: please correct or install')
    else:
        os.system('%s %s/diffall.py %s %s -skipcruft | tail -n %s' 
                         % (python, diffhome, temp1, temp2, diffshow))

if cleanup:
    shutil.rmtree(temp1)
    shutil.rmtree(temp2)



"""
================================================================================
OUTPUT:

--Removing date lines--

... index.html
... _thumbspage/2011-hartsfield-2.JPG.html
... _thumbspage/2010-losalamos-1.JPG.html
... _thumbspage/2004-colorado-2.JPG.html
....etc....
... _thumbspage/2003-sonyClie.jpg.html
... _thumbspage/2012-marysville.jpg.html
... _thumbspage/2017-pygadgets.png.html
... _thumbspage/2019-pyandroid-3.png.html
Files changed in path1 copy: 113

... index.html
... _thumbspage/2011-hartsfield-2.JPG.html
... _thumbspage/2010-losalamos-1.JPG.html
... _thumbspage/2004-colorado-2.JPG.html
....etc....
... _thumbspage/2003-sonyClie.jpg.html
... _thumbspage/2012-marysville.jpg.html
... _thumbspage/2017-pygadgets.png.html
... _thumbspage/2019-pyandroid-3.png.html
Files changed in path2 copy: 113

--Comparing with diffall--

2011-losalamos-1.JPG matches
2011-monterey.JPG matches
2006-chicago-2.JPG matches
2014-Frankenthon!.png matches
....etc....
2010-losalamos-2.JPG matches
2019-pyandroid-2.jpg matches
2018-mobile-site-2.png matches
2000-newmarket-3.jpg matches
================================================================================
Runtime hrs:mins:secs = 0:0:0.23
Dirs checked 8, Files checked: 467, Files skipped: 0
System metadata (cruft) files were skipped
Diffs found: 3
- files DIFFER at [cmp-thumbspage-results-temp1/xold-index.html] - [cmp-thumbspage-results-temp2/xold-index.html]
- files DIFFER at [cmp-thumbspage-results-temp1/_cut/Mark Lutz's Training Photos.htm] - [cmp-thumbspage-results-temp2/_cut/Mark Lutz's Training Photos.htm]
- items UNIQUE at [cmp-thumbspage-results-temp1/_cut/_RESTORED-mar1120-bad-shrink] - [cmp-thumbspage-results-temp2/_cut/_RESTORED-mar1120-bad-shrink]
End of report.
"""