# For a comprehensive documentation of configuration parameters see the configuration manual
# ("Konfigurationshandbuch").

# debug mode: If set to true, token handling is disabled (i.e. no tokens are necessary).
debug = false

# Name of this mainzelliste distribution instance.
dist = Lokale Patientenliste der CCP-IT am Standort ML_SITE

# Database setup
db.driver = ML_DB_DRIVER
db.url = jdbc:ML_DB_TYPE://ML_DB_HOST:ML_DB_PORT/ML_DB_NAME
db.username = ML_DB_USER
db.password = ML_DB_PASS

# Logging
# By default, logs are written to stdout
gcp.audittrail = true

# Field definitions
field.Vorname.type = PlainTextField
field.Nachname.type = PlainTextField
field.Fruehere_Namen.type = PlainTextField
field.Geburtstag.type = IntegerField
field.Geburtsmonat.type = IntegerField
field.Geburtsjahr.type = IntegerField
field.Staatsangehoerigkeit.type = PlainTextField
field.Geschlecht.type = PlainTextField
field.locallyUniqueId.type = PlainTextField

# Server-side validation
validator.field.Vorname.required = true
validator.field.Nachname.required = true
validator.field.Geburtstag.required = true
validator.field.Geburtsmonat.required = true
validator.field.Geburtsjahr.required = true

validator.field.Vorname.format = [A-Za-zäÄöÖüÜß\\.\\-' ]*[A-Za-zäÄöÖüÜß]+[A-Za-zäÄöÖüÜß\\.\\-' ]*
validator.field.Nachname.format = [A-Za-zäÄöÖüÜß\\.\\-' ]*[A-Za-zäÄöÖüÜß]+[A-Za-zäÄöÖüÜß\\.\\-' ]*
validator.field.Fruehere_Namen.format = [A-Za-zäÄöÖüÜß\\.\\-' ]*[A-Za-zäÄöÖüÜß]+[A-Za-zäÄöÖüÜß\\.\\-' ]*

# Date validation:
validator.date.0.fields = Geburtstag, Geburtsmonat, Geburtsjahr
validator.date.0.format = ddMMyyyy

# Exchange groups
# An exchange group denotes a set of fields whose values are considered interchangeable.
# The matcher (currently only EpilinkMatcher) compares the fields in each exchange group to
# all of its permutations and uses the best matching combination for final decision.
exchangeGroup.0 = Vorname, Nachname, Fruehere_Namen

# Field transformations
#
# Transformations for a field {fieldname} are defined by
# field.{fieldname}.transformers = {transformerlist}, where
# transformerlist is a comma-separated list of names of subclasses of FieldTransformer.
field.Vorname.transformers = StringNormalizer, FirstNameDecomposer
field.Nachname.transformers = StringNormalizer, GermanLastNameDecomposer
field.Fruehere_Namen.transformers = StringNormalizer, GermanLastNameDecomposer

# field comparators
field.Vorname.comparator = NGramComparator
field.Nachname.comparator = NGramComparator
field.Fruehere_Namen.comparator = NGramComparator

field.Geburtstag.comparator = BinaryFieldComparator
field.Geburtsmonat.comparator = BinaryFieldComparator
field.Geburtsjahr.comparator = BinaryFieldComparator

field.Staatsangehoerigkeit.comparator = BinaryFieldComparator
field.Geschlecht.comparator = BinaryFieldComparator

# Matcher definition
# Property matcher defines the class name of the matcher to use, which must be a subclass
# of Matcher. Currently only EpilinkMatcher and NullMatcher are supported.
matcher = EpilinkMatcher

# Config for Epilink Matcher
#
# matcher.epilink.{fieldname}.frequency: Defines the assumed mean frequency of values
# for field {fieldname}. This corresponds to the u-probability in the Fellegi-Sunter-Model
# and can be estimated by the reciprocal of the number of distinct values (i.e. the
# frequency of "month" is 1 / 12).
matcher.epilink.Vorname.frequency = 0.000235
matcher.epilink.Nachname.frequency = 0.0000271
matcher.epilink.Fruehere_Namen.frequency = 0.0000271

# 1 / 30 (approximate average number of days in a month)
matcher.epilink.Geburtstag.frequency = 0.0333
# 1 / 12
matcher.epilink.Geburtsmonat.frequency = 0.0833
# example value: birth years for an age range of 35 years (40 - 75).
matcher.epilink.Geburtsjahr.frequency = 0.0286

# Berechnungsgrundlage: 193 anerkannte + 13 umstrittene Staaten (Quelle: Wikipedia) -> ca. 1 / 200
matcher.epilink.Staatsangehoerigkeit.frequency = 0.005
matcher.epilink.Geschlecht.frequency = 0.5

# matcher.epilink.{fieldname}.error_rate defines the assumed error_rate for
# field {fieldname} (1 - m-probability)
#
# The supplied number stem from the evaluation of a German cancer registry
matcher.epilink.Vorname.errorRate = 0.01
matcher.epilink.Nachname.errorRate = 0.008
matcher.epilink.Fruehere_Namen.errorRate = 0.008
matcher.epilink.Geburtstag.errorRate = 0.005
matcher.epilink.Geburtsmonat.errorRate = 0.002
matcher.epilink.Geburtsjahr.errorRate = 0.004
matcher.epilink.Staatsangehoerigkeit.errorRate = 0.04
matcher.epilink.Geschlecht.errorRate = 0.04

# matcher.epilink.threshold.match: Defines the minimum weight (in the interval [0,1])
# for which a pair of records is considered a definite match.
#
# matcher.epilink.threshold_non_match: Defines the weight (in the interval [0,1])
# below which a pair of records is considered a definite non-match. Must be less than or equal
# to matcher.epilink.threshold_match.
#
# Record pairs with a weight w with threshold_non_match <= w < threshold_match are
# considered possible matches and lead to a tentative PID.
matcher.epilink.threshold_match = 0.95
matcher.epilink.threshold_non_match = 0.65

# Blocking
# The blocking strategy choice depends on the field types after the transformation.
# If compound fields are passed to the blocker, whether created by a field transformation or as input fields,
# the type of the subfields is the one that is relevant for the blocking.
# If exchange groups are used, only one of the fieldnames should be given here. The blocker is applied to
# the other fields as well.

# Use Soundex for PlainText fields
blocking.soundex.type = Soundex
blocking.soundex.fields = Vorname, Nachname, Fruehere_Namen

# Use FieldEquality e.g. for birthday based blocking
# This can be combined with Soundex blocking.
blocking.dob.type = FieldEquality
blocking.dob.fields = Geburtstag, Geburtsmonat, Geburtsjahr

# Servers (apiKey and permission definition)
servers.0.apiKey = ML_API_KEY
servers.0.permissions = createSession;showSessionIds;deleteAllPatients;createToken;tt_addPatient;tt_readPatients;tt_editPatient;readConfiguration
# TODO: Reduce to bridgeheads ip
servers.0.allowedRemoteAdresses = 0.0.0.0/0

# Allowed format of the callback
callback.allowedFormat = https://.*
callback.allowSelfsigned = false

# IdGenerators
idgenerators = BK_ML_SITE_L-ID, MDS_ML_SITE_L-ID, BK_ML_SITE_G-ID, locallyUniqueId, DKTK000001985_ML_SITE_L-ID, DKTK000001985_ML_SITE_G-ID, DKTK000001986_ML_SITE_L-ID, DKTK000001986_ML_SITE_G-ID, DKTK000001950_ML_SITE_L-ID, DKTK000001950_ML_SITE_G-ID, DKTK000001951_ML_SITE_L-ID, DKTK000001951_ML_SITE_G-ID, DKTK999999999_ML_SITE_L-ID, DKTK999999999_ML_SITE_G-ID, DKTK000002089_ML_SITE_L-ID, DKTK000002089_ML_SITE_G-ID

idgenerators.eagerGeneration = false

# Configuration of the local and global BK Pseudonymes
idgenerator.BK_ML_SITE_L-ID = PIDGenerator
idgenerator.BK_ML_SITE_L-ID.k1 = ML_BK_IDGENERATOR_RANDOM_1
idgenerator.BK_ML_SITE_L-ID.k2 = ML_BK_IDGENERATOR_RANDOM_2
idgenerator.BK_ML_SITE_L-ID.k3 = ML_BK_IDGENERATOR_RANDOM_3

idgenerator.BK_ML_SITE_G-ID = ExternalIDGenerator

# The unique id from the local datawarehouse
idgenerator.locallyUniqueId = ExternalIDGenerator

# Configuration for the local mds-id
idgenerator.MDS_ML_SITE_L-ID = PIDGenerator
idgenerator.MDS_ML_SITE_L-ID.k1 = ML_MDS_IDGENERATOR_RANDOM_1
idgenerator.MDS_ML_SITE_L-ID.k2 = ML_MDS_IDGENERATOR_RANDOM_2
idgenerator.MDS_ML_SITE_L-ID.k3 = ML_MDS_IDGENERATOR_RANDOM_3
idgenerator.MDS_ML_SITE_L-ID.eager = *
idgenerator.MDS_ML_SITE_L-ID.exportEncryption = mdsIdEncryption
# Encryption for export of mds
crypto.key.mdsIdPublicKey.type = RSA_PUBLIC
crypto.key.mdsIdPublicKey.uri = file:///run/secrets/centralSearchPublicKey
crypto.encryption.mdsIdEncryption.key = mdsIdPublicKey
crypto.encryption.mdsIdEncryption.type = RSA_ENCRYPT

# Configuration of the study specific id-generators

# Configuration for project DKTK000001985
idgenerator.DKTK000001985_ML_SITE_L-ID = PIDGenerator
idgenerator.DKTK000001985_ML_SITE_L-ID.k1 = ML_DKTK000001985_IDGENERATOR_RANDOM_1
idgenerator.DKTK000001985_ML_SITE_L-ID.k2 = ML_DKTK000001985_IDGENERATOR_RANDOM_2
idgenerator.DKTK000001985_ML_SITE_L-ID.k3 = ML_DKTK000001985_IDGENERATOR_RANDOM_3

idgenerator.DKTK000001985_ML_SITE_G-ID = ExternalIDGenerator

# Configuration for project DKTK000001986
idgenerator.DKTK000001986_ML_SITE_L-ID = PIDGenerator
idgenerator.DKTK000001986_ML_SITE_L-ID.k1 = ML_DKTK000001986_IDGENERATOR_RANDOM_1
idgenerator.DKTK000001986_ML_SITE_L-ID.k2 = ML_DKTK000001986_IDGENERATOR_RANDOM_2
idgenerator.DKTK000001986_ML_SITE_L-ID.k3 = ML_DKTK000001986_IDGENERATOR_RANDOM_3

idgenerator.DKTK000001986_ML_SITE_G-ID = ExternalIDGenerator

# Configuration for project DKTK000001950
idgenerator.DKTK000001950_ML_SITE_L-ID = PIDGenerator
idgenerator.DKTK000001950_ML_SITE_L-ID.k1 = ML_DKTK000001950_IDGENERATOR_RANDOM_1
idgenerator.DKTK000001950_ML_SITE_L-ID.k2 = ML_DKTK000001950_IDGENERATOR_RANDOM_2
idgenerator.DKTK000001950_ML_SITE_L-ID.k3 = ML_DKTK000001950_IDGENERATOR_RANDOM_3

idgenerator.DKTK000001950_ML_SITE_G-ID = ExternalIDGenerator

# Configuration for project DKTK000001951
idgenerator.DKTK000001951_ML_SITE_L-ID = PIDGenerator
idgenerator.DKTK000001951_ML_SITE_L-ID.k1 = ML_DKTK000001951_IDGENERATOR_RANDOM_1
idgenerator.DKTK000001951_ML_SITE_L-ID.k2 = ML_DKTK000001951_IDGENERATOR_RANDOM_2
idgenerator.DKTK000001951_ML_SITE_L-ID.k3 = ML_DKTK000001951_IDGENERATOR_RANDOM_3

idgenerator.DKTK000001951_ML_SITE_G-ID = ExternalIDGenerator

# Configuration for project DKTK999999999
idgenerator.DKTK999999999_ML_SITE_L-ID = PIDGenerator
idgenerator.DKTK999999999_ML_SITE_L-ID.k1 = ML_DKTK999999999_IDGENERATOR_RANDOM_1
idgenerator.DKTK999999999_ML_SITE_L-ID.k2 = ML_DKTK999999999_IDGENERATOR_RANDOM_2
idgenerator.DKTK999999999_ML_SITE_L-ID.k3 = ML_DKTK999999999_IDGENERATOR_RANDOM_3

idgenerator.DKTK999999999_ML_SITE_G-ID = ExternalIDGenerator

# Configuration for project DKTK000002089
idgenerator.DKTK000002089_ML_SITE_L-ID = PIDGenerator
idgenerator.DKTK000002089_ML_SITE_L-ID.k1 = ML_DKTK000002089_IDGENERATOR_RANDOM_1
idgenerator.DKTK000002089_ML_SITE_L-ID.k2 = ML_DKTK000002089_IDGENERATOR_RANDOM_2
idgenerator.DKTK000002089_ML_SITE_L-ID.k3 = ML_DKTK000002089_IDGENERATOR_RANDOM_3

idgenerator.DKTK000002089_ML_SITE_G-ID = ExternalIDGenerator