#!/usr/bin/env python3
#metadata_xml2po.py
#This shell has two work modes:
# - Reads a folder and its subfolders and generates a .po file with all the descriptions, summaries..etc.. founded on the metainfo folders
# - Updates the xml files of a folder (and its subfolders) with a given po

import os
import sys
import xml.etree.ElementTree as ET
import polib as PO

def _debug(msg):
	global dbg
	if dbg==1:
		print(str(msg))

def usage_help():
	print ("Usage:")
	print (sys.argv[0]+" --export|-e path_to_dirs" )
	print (sys.argv[0]+" --import|-i path_to_dirs path_to_po" )
	print("Options:")
	print("\t --debug|-d -> Enables debug mode")
	print("\t --xmlfix|-x -> When exporting fixes the original texts from xml files replacing \" with \' and \\n with spaces")
	print("\t --suffix|-s [suffix] -> Appends \"suffix\" to generated files")
	print("")
	exit(1)
#def usage_help

def parse_arguments():
	global mode
	global wrkDir
	global poFile
	global suffix
	global xmlFix
	sfx=0
	argList={"--export":"export","-e":"export","--import":"import","-i":"import","-d":"debug","--debug":"debug","--suffix":"suffix","-s":"suffix","-x":"xmlfix","--xmlfix":"correct"}
	args=sys.argv[1:]
	for arg in args:
		if sfx:
			suffix=arg
			sfx=0
		elif arg in argList.keys():
			if argList[arg]=='debug':
				global dbg
				dbg=1
				_debug("Debug enabled")
			elif argList[arg]=='suffix':
				sfx=1
			elif argList[arg]=='xmlfix':
				xmlFix=1
			else:
				mode=argList[arg]
		elif os.path.isdir(arg):
			wrkDir=arg
		elif os.path.isfile(arg):
			poFile=arg
	if ((not mode or not wrkDir) or (mode=="import" and not poFile)):
		usage_help()
#def parse_arguments

def get_package_dirs(path):
	candidatePaths=[]
	for basedir, dirs, files in os.walk(path):
		for subdir in dirs:
			if subdir.endswith("metainfo") and "fuentes/llx-resources" in basedir:
				_debug(("Adding %s to listdir")%(basedir+'/'+subdir))
				#Look for the xml
				candidatePaths.append(basedir+"/"+subdir)
	return candidatePaths
#def get_package_dirs

def generate_tmp_po(candidatePaths):
	global xmlFix
	for path in candidatePaths:
		dirContent = os.listdir(path)
		for xmlFile in dirContent:
			if not xmlFile.endswith(".xml"):
				continue
			try:
				xml = ET.parse(path+'/'+xmlFile)
				xmlRoot=xml.getroot()
				tagList=['name','summary','description']
				for tag in tagList:
					for element in xmlRoot.iter(tag):
						if tag=='description':
							for xmlDesc in element:
								if not (xmlDesc.attrib):
									msgid=str(xmlDesc.text).replace("\"","\'")
									msgid=str(msgid).replace("\n"," ")
									xmlDesc.text=msgid
									msgIdList.append(msgid)
								else:
									save_locale(xmlDesc.attrib['{http://www.w3.org/XML/1998/namespace}lang'],msgid,xmlDesc.text)
						else:
							if not (element.attrib):
								msgid=str(element.text).replace("\"","\'")
								msgid=str(msgid).replace("\n"," ")
								element.text=msgid
								msgIdList.append(msgid)
							else:
								save_locale(element.attrib['{http://www.w3.org/XML/1998/namespace}lang'],msgid,element.text)
				#Fixes the xml replacing " with ' and \n with spaces
				if xmlFix:
					_debug(("Attempting to fix the xmlFile %s")%(xmlFile))
					xml.write(path+'/'+xmlFile)
			except Exception as e:
				print(str(e))
				print (xmlFile)
#def generate_tmp_po

def save_locale(locale,msgid,msgstr=None):
	poString=str(msgstr).replace("\"","\'")
	poString=poString.replace("\n"," ")
	if "valencia" in locale or "qcv" in locale:
		locale="ca_ES.UTF-8@valencia"
	if locale in poDict.keys():
		poDict[locale].update({msgid:poString})
	else:
		poDict[locale]={msgid:poString}
	localeSet.append(locale)
#def save_locale

def print_locale(outputFile):
	global suffix
	for msgId in msgIdList:
		#Open locale po and append the info
		for localization in localeSet:
			f=open(outputFile+"_"+localization+".po"+suffix,'a')
			if f.tell()==0:
				f.write("msgid \"\"\nmsgstr \"\"\n\"Content-Type: text/plain; charset=utf-8\"\n")
			f.write("msgid \""+	msgId+"\"\n")
			if msgId in poDict[localization]:
				f.write("msgstr \""+poDict[localization][msgId]+"\"\n")
			else:
				f.write("msgstr \"\"\n")
			f.close
#def print_locale

def get_locale_from_po(poFile):
	locale=os.path.basename(poFile)
	locale=locale.replace('catalogue_','')
	locale=locale.replace('.po','')
	if "valencia" in locale or "qcv" in locale:
		locale="ca_ES.UTF-8@valencia"
	_debug("Locale set to "+locale)
	return(locale)
#def get_locale_from_po

def load_po(poFile,locale):
	global poDict
	global fileEncoding
	fileEncoding=PO.detect_encoding(poFile)
	poDict={}
	_debug(("Loading po %s with %s encoding")%(poFile,fileEncoding))
	po = PO.pofile(poFile,encoding=fileEncoding)
	for entry in po:
		if entry.msgid in poDict.keys():
			poDict[entry.msgid].update({locale:entry.msgstr})
		else:
			poDict[entry.msgid]={locale:entry.msgstr}
#def load_po

def get_translations(msgid):
	translationDict={}
	_debug("Searching translations for " + msgid)
	if msgid in poDict.keys():
		translationDict=poDict[msgid].copy()
	_debug("Get "+str(translationDict))
	return translationDict
#def get_translations

def merge_translations(candidatePaths,locale):
	global suffix
	for path in candidatePaths:
		dirContent = os.listdir(path)
		for xmlFile in dirContent:
			if not xmlFile.endswith(".xml"):
				continue
			xml = ET.parse(path+'/'+xmlFile)
			xmlRoot=xml.getroot()
			tagList=['name','summary','description']
			for tag in tagList:
				for element in xmlRoot.iter(tag):
					if tag=='description':
						#1st loop, get the msgid
						for xmlDesc in element:
							if not (xmlDesc.attrib):
								translationDict=get_translated_msgid(xmlDesc)
								break
						#2dn loop, modify translations for descriptions
						for xmlDesc in element:
							if not (xmlDesc.attrib):
								continue
							modify_translation(xmlDesc,translationDict)
						#Generate new translations
						for locale,translation in translationDict.items():
							add_new_translation(element,'p',translation)
					else:
						if not (element.attrib):
							translationDict=get_translated_msgid(element)
						else:
							modify_translation(element,translationDict)
				#Generate new translations
				for locale,translation in translationDict.items():
					if tag=='description':
						continue
					add_new_translation(xmlRoot,tag,translation)
			xml.write(path+'/'+xmlFile+suffix,encoding=fileEncoding,xml_declaration=True)

def get_translated_msgid(element):
	msgid=str(element.text).replace("\"","\'")
	msgid=str(msgid).replace("\n"," ")
	translationDict=get_translations(msgid)
	return translationDict
#def get_translated_msgid

def modify_translation(element,translationDict):
	_debug(("Modifying %s translation")%(locale))
	translation=element.text
	localeTrans=element.attrib['{http://www.w3.org/XML/1998/namespace}lang']
	if localeTrans in translationDict.keys():
		translation=translationDict[localeTrans]
		if translation:
			element.text=translation
		del translationDict[localeTrans]
	return (translation)
#def modify_translation

def add_new_translation(element,tag,text):
	if text!='':
		newDesc=ET.SubElement(element, tag)
		newDesc.tail="\n"
		newDesc.set('xml:lang',locale)
		newDesc.text=text
		_debug(("Added new tag %s with %s translation")%(tag,locale))
#def add_new_translation

#### MAIN PROGRAM ####
dbg=0
xmlFix=0
suffix=''
fileEncoding=''
mode=''
wrkDir=''
poFile=''
parse_arguments()
print(("Analyzing %s for action %s") % (wrkDir,mode))
poDict={}
pathList=get_package_dirs(wrkDir)
if mode=='export':
	localeSet=[]
	msgIdList=[]
	outputFile="catalogue"
	generate_tmp_po(pathList)
	localeSet=set(localeSet)
	print_locale(outputFile)
	print("Po files generated for "+str(localeSet))
else:
	locale=get_locale_from_po(poFile)
	load_po(poFile,locale)
	merge_translations(pathList,locale)
	print("Updated metadata xml in "+str(pathList))