Phpwcms Backend File Downloader

get-links.js

/**
 * Execute this on the phpwcms files page when all wanted folders are opened
 * (all visible download buttons will be added to the array)
 **/
var save_buttons = $('img[src="img/button/download_disc.gif"]');

function isFolder($tr) {
  return $tr.attr('bgcolor') == '#EBF2F4';
}

function getIndent($tr) {
  if(isFolder($tr)) {
    var spacer = $tr.find('td.msglist:eq(0) > img:eq(1)');
    var ind = spacer.attr('width');
  } else {
    var spacer = $tr.find('td.msglist');
    var ind = spacer.attr('width') - 15;
  }
  return ind ? ind : 999;
}

function getName($tr) {
  return $tr.find('strong').text();
}

function getLink($e) {
  return $e.closest('a').attr('href');
}

function getFileName($e) {
  return $e.closest('tr').find('td:eq(2) > a').text();
}

function getParentDir($e) {
  // get subtable
  /*while (!(closest = $e.closest('table')).is('#files')) {
    $e = closest;
  }*/
  if (!$e.is('tr')) {
   var $row = $e.closest('table').closest('tr');
  } else {
   var $row = $e;
  }
  var ind = getIndent($row);
  
  while($row.length && getIndent($row) >= ind) // skip boring tr's
    $row = $row.prev();
  return $row;
}

function getPath($e) {
  var path = "";
  while (($parent = getParentDir($e)).length) {
    path = getName($parent) + "/" + path;
    $e = $parent;
  }
  return path;
}



var data = {};
for(var save of saves) {
  var $save = $(save);
  data[getLink($save)] = [getFileName($save), getPath($save)];
}

// print result to console
console.log(JSON.stringify(data));

wcms-downloader.ipynb

wcms-downloader.py

# coding: utf-8

# This scripts downloads files from the phpwcms backend
# 
# ### Required data:

# In[19]:

# ENTER DATA HERE:
backend_links_file = 'wcms-files.json'  # contains download links+paths of wcms backend, generated by a JS script
target_dir = 'wcms-files/' # target directory for downloads

wcms_baseurl = 'http://www.your-site.org/'
cookies = {'PHPSESSID': 'XYZ'}  # for Phpwcms backend


# # Initialize

# In[5]:

import json
import codecs


# In[3]:

# small helper function to peek into variables
def glance(collection, n=3):
    print("%d entries" % len(collection))
    if type(collection) == dict:
        return list(collection.items())[:n]
    elif type(collection) == list:
        return collection[:n]


# # Load Data

# In[10]:

# generated by JS
with codecs.open(backend_links_file, encoding='utf-8') as f:
    wcms_links = json.load(f)
    
glance(wcms_links) # quick glance


# # Download

# In[15]:

import os
import requests

def dl(link, path, filename):
    r = requests.get(wcms_baseurl + link, cookies=cookies, stream=True)
    r.raise_for_status()
    
    os.makedirs(target_dir + path, exist_ok=True) # create directory if not existent
    with open(target_dir + path + filename, 'wb') as fd:
        for chunk in r.iter_content(1024):
            fd.write(chunk)


# In[18]:

done_count = 0
total_count = len(wcms_links)
for link, loc in wcms_links.items():
    print('Downloading [%d / %d]: %s -> %s' % (done_count+1, total_count, link, loc[1] + loc[0]))
    dl(link, loc[1], loc[0])
    done_count += 1