pastebin/main.py

#!/usr/bin/env python3
# -- coding: utf-8 --

import web
import cgi
web.config.debug = True

import sys
import os
import time

import shutil
import hashlib
import magic
import mimetypes
from base64 import urlsafe_b64encode
from random import random
from io import StringIO

sys.path.append(os.path.dirname(__file__))
from config import *
import utils

render = web.template.render(CONF_TEMPLATE_DIR)
cgi.maxlen = CONF_MAX_UPLOAD_SIZE

urls = (
    '/', 'Greeter',
    '/p', 'Pastebin',
    '/info', 'FileInfo',
    '/cleanup', 'Cleanup',
    '/([a-zA-Z0-9_.-]*)', 'Pastebin',
    '/([a-zA-Z0-9_-]+)/(.*)', 'Pastebin'
    )

class Greeter:
  def GET(self):
    web.header("Content-Type", 'text/html')
    return render.index(web.ctx.host, CONF_MAX_UPLOAD_SIZE)

class Pastebin:
  def __init__(self):
    self.db = web.database(
        dbn  = CONF_SQL_DRIVER,
        db   = CONF_SQL_DATABASE,
        user = CONF_SQL_USER,
        pw   = CONF_SQL_PASSWORD)

    self.magic = magic.Magic(mime = True)

  # this function returns (mime type, file extension) for the file with the
  # given name or None, if the type is unknown
  def detect_file_type(self, filename):
    mime = self.magic.from_file(filename)

    # prevent weird extensions like ".ksh" for plain-text
    if mime == "text/plain":
      extension = ".txt"
    else:
      extension = mimetypes.guess_extension(mime)
    return mime, extension

  def POST(self, doc = None):
    web.header("Content-Type", 'text/plain')

    try:
      x = web.input(content = {})
    except ValueError:
      return "File too large!"

    if 'content' in x:
      # open a temporary file
      tempFileName = "/tmp/pastebin_" + web.ctx.ip + "_" + str(random())
      tempFile = open(tempFileName, "wb")

      # calculate md5 hash of the uploaded data while saving it to the
      # temporary file
      m = hashlib.md5()
      for line in x['content'].file:
        m.update(line)
        tempFile.write(line)

      tempFile.close()

      # generate the paste's id
      pasteid = urlsafe_b64encode(m.digest()[0:9]).decode('ascii')
      storeName = os.path.join(CONF_DATA_DIR, pasteid)

      basename = os.path.basename(x['content'].filename)
      pos = basename.find('.')
      if pos >= 0:
          basename = basename[:pos]

      # get the mime type and extension from the data
      mimeType, ext = self.detect_file_type(tempFileName)
      if not ext:
        ext = ""

      if not os.path.exists(storeName):
        # this file does not yet exist

        # create the file's database entry
        self.db.insert(CONF_SQL_FILESTABLE,
            hash = pasteid,
            detected_type = mimeType,
            create_time = time.time(),
            access_time = time.time())

        # store the document permanently
        shutil.move(tempFileName, storeName);
      else:
        # this file was uploaded before, so we just delete the temporary
        # file
        os.unlink(tempFileName);

      # return the generated URL
      return "http://" + web.ctx.host + "/" + pasteid + "/" + basename + ext + "\n"
    else:
      return "Nothing uploaded."

  def GET(self, paste_id = None, reqName=None):
    if paste_id:
      if reqName:
        fileName = reqName
      else:
        fileName = paste_id

      dotpos = paste_id.rfind('.')
      if dotpos >= 0:
        paste_id = paste_id[:dotpos]

      # set the content type from the user-specified file extension
      mimeType, encoding = mimetypes.guess_type(fileName);
      if mimeType:
        if mimeType[0:4] == "text":
          mimeType += "; charset=" + CONF_DEFAULT_CHARSET

        web.header("Content-Type", mimeType)
        if encoding:
          web.header("Content-Encoding", encoding)
      else:
        web.header("Content-Type", CONF_DEFAULT_TYPE)

      storeName = CONF_DATA_DIR + "/" + paste_id

      try:
        f = open(storeName, "rb")
        for line in f:
          yield line
        f.close()

        # update the corresponding access time and count
        items = self.db.select(CONF_SQL_FILESTABLE,
            where = "hash=" + web.db.sqlquote(paste_id),
            what = "access_count")

        if items:
          oldcount = items[0]['access_count']

          self.db.update(CONF_SQL_FILESTABLE,
              where = "hash=" + web.db.sqlquote(paste_id),
              access_time = time.time(),
              access_count = oldcount + 1)
        else:
          # The count could not be read.
          # Probably there is no entry for this file -> create one
          self.db.insert(CONF_SQL_FILESTABLE,
              hash = paste_id,
              detected_type = mimeType,
              create_time = os.path.getmtime(storeName),
              access_time = time.time(),
              access_count = 1)
      except IOError:
        yield "Cannot read \"" + paste_id + "\"."

    else:
      raise web.seeother("/")

class FileInfo:
  def __init__(self):
    self.db = web.database(
        dbn  = CONF_SQL_DRIVER,
        db   = CONF_SQL_DATABASE,
        user = CONF_SQL_USER,
        pw   = CONF_SQL_PASSWORD)

  def GET(self, doc = None):
    web.header("Content-Type", 'text/plain')

    # generate a list of all files
    items = self.db.select(CONF_SQL_FILESTABLE,
        order = "access_time DESC")

    if not items:
      return "There are no files in the database."
    else:
      # table header
      outbuf = StringIO()
      outbuf.write("%-12s | %-20s | %-8s | %-25s | %-16s | %-10s |\n" % (
        "Hash", "Detected Type", "Size", "Created", "Last Access", "# Accessed"))
      outbuf.write("-"*12 + "-+-" + "-"*20 + "-+-" + "-"*8 + "-+-" + "-"*25
          + "-+-" + "-"*16 + "-+-" + "-"*10 + "-|\n")

      now = time.time()

      for i in items:
        # determine file size (and existance)
        storeName = CONF_DATA_DIR + "/" + i["hash"]
        try:
          fileSize = os.path.getsize(storeName)
          sizeStr = utils.readable_size(fileSize)
        except OSError:
          sizeStr = "N/A"

        lastacc_delay = (now - i['access_time'])
        lastacc_days = lastacc_delay / 86400
        lastacc_hours = (lastacc_delay / 3600) % 24
        lastacc_minutes = (lastacc_delay / 60) % 60

        outbuf.write("%12s | %-20s | %8s | %25s | %16s | %10u |\n" % (
          i["hash"],
          i["detected_type"],
          sizeStr,
          time.strftime(CONF_TIME_FORMAT, time.localtime(i['create_time'])),
          "%ud, %2u:%02u ago" % (lastacc_days, lastacc_hours, lastacc_minutes),
          i['access_count']))
      return outbuf.getvalue()

class Cleanup:
  def __init__(self):
    self.db = web.database(
        dbn  = CONF_SQL_DRIVER,
        db   = CONF_SQL_DATABASE,
        user = CONF_SQL_USER,
        pw   = CONF_SQL_PASSWORD)

  def GET(self):
    web.header("Content-Type", 'text/html')

    getdata = web.input(doit = False)

    if not getdata.doit:
      return render.cleanup(stage = 1, result = None)
    else:
      # generate a list of all files
      items = self.db.select(CONF_SQL_FILESTABLE,
          order = "access_time DESC")

      if not items:
        return render.cleanup(stage = 2, result = "There are no files in the database.")
      else:
        deleted_entries = 0
        for i in items:
          # determine file existance
          storeName = CONF_DATA_DIR + "/" + i["hash"]
          if not os.path.exists(storeName):
            # this entry has no corresponding file -> delete it
            deleted_entries += self.db.delete(
                CONF_SQL_FILESTABLE,
                where = "hash=" + web.db.sqlquote(i["hash"]))

        return render.cleanup(
            stage = 2,
            result = "Cleanup successful, " + str(deleted_entries) + " entries deleted.")


app = web.application(urls, globals())

if __name__ == "__main__":
  app.run()

application = app.wsgifunc()