jogobot-red/red_fam.py

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

import hashlib
import re
from datetime import datetime

from mysql_red import MYSQL_RED_FAM

class RED_FAM:
	

	
	def __init__( self, fam_hash=None, articlesList=None, red_page_id=None, beginning=None, ending=None, status=0 ):
		"""
		Generates a new RED_FAM object
		
		@param		articlesList	list		List of articles of redundance family
		@param		beginning		datetime	Beginning date of redundance diskussion
		@param		ending			datetime	Ending date of redundance diskussion
		"""
		
		#if( beginning ):
		#	self.add_beginning( beginning )
		#	self._beginning = None
		
		#if( ending ):
		#	self.add_ending( ending )
		#else:
		#	self._ending = None
			
		#self._status = status # __TODO__ STATUS CODE
		
		#self._handle_db()



	def __repr__( self ):
		
		if( self._beginning ):
			beginning = ", beginning=" + repr( self._beginning )
		else:
			beginning = ""
			
		if( self._ending ):
			ending = ", ending=" + repr( self._ending )
		else:
			ending = ""
		
		__repr = "RED_FAM( " + repr( self._articlesList ) + beginning + ending + ", status=" + repr( self._status ) + " )"
		
		return __repr
		
class RED_FAM_PARSER( RED_FAM ):
	"""
	Provides an interface to RED_FAM for adding/updating redundance families while parsig redundance pages
	"""
	
	# Define the timestamp format
	__timestamp_format = "%H:%M,  %d. %b. %Y (%Z)"
	
	def __init__( self, red_fam_heading, red_page_id, red_page_archive, beginning, ending=None ):
		"""
		Creates a RED_FAM object based on data collected while parsing red_pages combined with possibly former known data from db
		
		@param		red_fam_heading		string		String with wikitext heading of redundance section
		@param		red_page_id			int			MediaWiki page_id of red_page containing red_fam
		@param		red_page_archive	bool		Is red_page an archive
		@param		beginning			datetime	Timestamp of beginning of redundance discussion
										string		Timestamp of beginning of redundance discussion as srftime parseable string
		@param		ending				datetime	Timestamp of ending of redundance discussion
										string		Timestamp of ending of redundance discussion as srftime parseable string
		"""
		## Set object attributes:
		self._red_page_id = red_page_id
		self._red_page_archive = red_page_archive
		
		# Method self.add_beginning sets self._beginning directly
		self.add_beginning( beginning )
		
		# Method self.add_ending sets self._ending directly
		if( ending ):
			self.add_ending( ending )
		else:
			#If no ending was provided set to None
			self._ending = None
		
		# Parse the provided heading of redundance section to set self._articlesList
		self.heading_parser( red_fam_heading )
		
		# Calculates the sha1 hash over self._articlesList to rediscover known redundance families
		self.fam_hash()
		
		# Open database connection, ask for data if existing, otherwise create entry
		self.__handle_db()
		
		# Check status changes
		self.status()
		
		# Since status change means something has changed, update database
		if( self._status != self.__mysql.data[ 'status' ] ):
			self.__mysql.update_fam( self._red_page_id, self._beginning, self._ending, self._status )
		
	def __handle_db( self ):
		"""
		Handles opening of db connection
		"""
		# We need a connection to our mysqldb
		self.__mysql = MYSQL_RED_FAM( self._fam_hash )
		
		if not self.__mysql.data:
			self.__mysql.add_fam( self._articlesList, self._red_page_id, self._beginning, self._ending )
	
	def heading_parser( self, red_fam_heading):
		"""
		Parses given red_fam_heading string and saves articles list
		"""
		
		# Predefine a pattern for wikilinks' destination
		wikilink_pat = re.compile( r"\[\[([^\[\]\|]*)(\]\]|\|)" )
	
		# We get the pages in first [0] element iterating over wikilink_pat.findall( line )
		self._articlesList = [ link[0] for link in wikilink_pat.findall( red_fam_heading ) ]

	def fam_hash( self ):
		"""
		Calculates the SHA-1 hash for the articlesList of redundance family.
		Since we don't need security SHA-1 is just fine.
		
		@returns	str					String with the hexadecimal hash digest
		"""
		
		h = hashlib.sha1()
		h.update( str( self._articlesList ).encode('utf-8') )
		
		self._fam_hash= h.hexdigest()

	def add_beginning( self, beginning ):
		"""
		Adds the beginning date of a redundance diskussion to the object and sets changed to True
		
		@param		datetime	datetime	Beginning date of redundance diskussion
		"""
		
		self._beginning = self.__datetime( beginning )
		
	def add_ending( self, ending ):
		"""
		Adds the ending date of a redundance diskussion to the object. Also sets the status to __TODO__ STATUS NUMBER and changed to True
		
		@param		datetime	datetime	Ending date of redundance diskussion
		"""
		
		self._ending = self.__datetime( ending )
		
		self._status = 1 #__TODO__ STATUS NUMBER
		self._changed = True
	
	def __datetime( self, timestamp ):
		"""
		Decides wether given timestamp is a parseable string or a datetime object and returns a datetime object in both cases
		
		@param		timestamp	datetime	Datetime object
								str			Parseable string with timestamp in format __timestamp_format
								
		@returns				datetime	Datetime object
		"""
		
		if( isinstance( timestamp, datetime ) ):
			return timestamp
		else:
			return datetime.strptime( timestamp, type( self ).__timestamp_format )
			
	def status( self ):
		"""
		Handles detection of correct status
		"""
		# Diskussion läuft --> Status ? --> 0			! _ending
		# Diskussion beendet --> Status 0 --> 1			 _ending
		# Diskussion archiviert --> Status 0/1 --> 2	???
		
		if not self._ending:
			self._status = 0
		else:
			if not self._red_page_archive:
				self._status = 1
			else:
				self._status = 2

class RED_FAM_WORKER( RED_FAM ):
	"""
	Handles working with redundance families stored in database where discussion is finished
	"""
	pass