#!/usr/bin/python
# -*- coding: utf-8 -*-
import sys
import re
import codecs
import datetime
from xml.etree.ElementTree import ElementTree
from xml.etree.ElementTree import fromstring

SWEFN_URL = "swefn.xml"
SALDO_URL = "saldo.txt"

def read_csv_from_file(path, num_of_fields):
    with codecs.open(path, encoding='utf-8') as f:
        for line in f:
            e = [x for x in line[:-1].split('\t')]
            yield e[0:num_of_fields]

def escape(s):
    s = s.replace('&', '&')
    s = s.replace("'", ''')
    s = s.replace('<', '&lt;')
    s = s.replace('>', '&gt;')
    return s.replace('"', '&quot;')

if __name__ == '__main__':
    saldo_entries = set()
    for (saldoid, prim, sec, lem, gf, pos, paradigm) in read_csv_from_file(SALDO_URL, 7):
        saldo_entries.add(saldoid)
    
    tree = ElementTree()
    tree.parse(SWEFN_URL)

    reports = []

    entries = tree.find("Lexicon").findall("LexicalEntry")
    for le in entries:
        sense = le.find("Sense")
        new_saldo_lus = []
        missing_in_saldo = []
        if sense != None:
            frame_id = sense.get("id")
            createdBy = ""

            lus = []
            lu_suggestions = []
            empty_lu = False
            empty_lu_sugg = False

            feats = sense.findall("feat")
            for feat in feats:
                feat_att = feat.get("att")
                feat_val = feat.get("val")
                if feat_att == "semanticType":
                    pass#cemtypes.append(feat_val)
                elif feat_att == "domain":
                    pass#domains.append(feat_val)
                elif feat_att == "coreElement":
                    pass#c_elements.append(feat_val)
                elif feat_att == "inheritance":
                    pass#e.inheritance = feat_val
                elif feat_att == "peripheralElement":
                    pass#p_elements.append(feat_val)
                elif feat_att == "compound":
                    pass#comps.append(feat_val)
                elif feat_att == "compoundExample":
                    pass#comp_examples.append(feat_val)
                elif feat_att == "LU":
                    if feat_val.strip() != "":
                        lus.append(feat_val)
                    else:
                        empty_lu = True
                elif feat_att == "suggestionForLU":
                    if feat_val.strip() != "":
                        lu_suggestions.append(feat_val)
                    else:
                        empty_lu_sugg = True
                elif feat_att == "internal_comment": # New, but should maybe not be rendered out
                    pass
                elif feat_att == "comment": # Right now it's not a feat but I think it should be!
                    pass#e.comment = feat_val
                elif feat_att == "createdBy": # Right now it's not a feat but I think it should be!
                    createdBy = feat_val
                    pass#e.created_by = feat_val
                elif feat_att == "createdDate":
                    pass#e.createdDate = feat_val
                elif feat_att == "modifDate":
                    pass#e.modifDate = feat_val
                elif feat_att == "entry_status": # New field
                    pass#e.status = feat_val
                elif feat_att == "BFNID":
                    pass#e.bid = feat_val

            for lu in lus:
                if not lu in saldo_entries:
                    missing_in_saldo.append(lu)
            for sugg in lu_suggestions:
                if sugg in saldo_entries:
                    new_saldo_lus.append(u'<a href="http://spraakbanken.gu.se/karp/#search=sense%7C' + sugg + '">' + sugg + u'</a>');

            if new_saldo_lus or missing_in_saldo or empty_lu or empty_lu_sugg:
                report += '<h2>%s</h2><span style="font-size:small"> (%s) </span><ol>' % (frame_id.split("--")[1], createdBy)
                if new_saldo_lus:
                    report += u"<li>Ny identifierare finns nu i SALDO: " + ", ".join(new_saldo_lus) + "</li>"
                if missing_in_saldo:
                    report += u"<li>Lexem saknas: " + escape(", ".join(missing_in_saldo)) + "</li>"
                if empty_lu:
                    report += u"<li>Tom LU</li>"
                if empty_lu_sugg:
                    report += u"<li>Tomt LU-förslag</li>"
                report += '</ol>'
                reports.append(report)

        report = ''

    now = datetime.datetime.now()
    timestring = now.isoformat(' ').split(".")[0]
    print ('<html><body><h1>Error report for SweFN</h1>%s%s</body></html>' % (timestring, "<br/>".join(reports))).encode('utf-8')