/usr/share/ucto/tokconfig-deu

version=0.2
[RULE-ORDER]
SUFFIX WORD-TOKEN ABBREVIATION-KNOWN
URL URL-WWW URL-DOMAIN E-MAIL
WORD-PARPREFIX-PARSUFFIX WORD-PARPREFIX WORD-PARSUFFIX WORD-COMPOUND
ABBREVIATION INITIAL SMILEY REVERSE-SMILEY PUNCTUATION-MULTI DATE-REVERSE DATE
NUMBER-YEAR TIME FRACNUMBER NUMBER CURRENCY WORD PUNCTUATION UNKNOWN

[META-RULES]
SPLITTER=%
#NUMBER-ORDINAL = \p{N}+-?(?: %ORDINALS% )(?:\Z|\P{Lu}|\P{Ll})$
ABBREVIATION-KNOWN = (?:\p{P}*)?(?:\A|[^\p{L}\.])((?:%ABBREVIATIONS%)\.)(?:\Z|\P{L})
WORD-TOKEN =(%TOKENS%)(?:\p{P}*)?$
#PREFIX = (?:\A|[^\p{Lu}\.]|[^\p{Ll}\.])(%PREFIXES% )(\p{L}+)
SUFFIX = ((?:\p{L})+)( %SUFFIXES% )(?:\Z|\P{L})

[RULES]
%include url
%include e-mail
%include smiley

#Ex (alt)-studente()
WORD-PARPREFIX-PARSUFFIX=(?:\p{Ps}\p{L}+[\p{Pc}\p{Pd}]?\p{Pe}[\p{Pc}\p{Pd}]?)\p{L}+(?:[\p{Pc}\p{Pd}]\p{L}+)*(?:[\p{Pc}\p{Pd}]?\p{Ps}[\p{Pc}\p{Pd}]?\p{L}+\p{Pe})

#Ex: (un)verstehbar,
WORD-PARPREFIX=(?:\p{Ps}\p{L}+[\p{Pc}\p{Pd}]?\p{Pe}[\p{Pc}\p{Pd}]?)\p{L}+(?:[\p{Pc}\p{Pd}]\p{L}+)*

#Ex: könig(in)
WORD-PARSUFFIX=\p{L}+(?:[\p{Pc}\p{Pd}]\p{L}+)*(?:[\p{Pc}\p{Pd}]?\p{Ps}[\p{Pc}\p{Pd}]?\p{L}+\p{Pe})

#Keep dash/underscore connected parts (even if they are in parenthesis)
WORD-COMPOUND=\p{L}+(?:[\p{Pc}\p{Pd}]\p{L}+)+

#Abbreviations with multiple periods
ABBREVIATION=^(\p{L}{1,3}(?:\.\p{L}{1,3})+\.?)(?:\Z|[,:;])

#retain initials
INITIAL=^(?:\p{Lt}|\p{Lu})\.$

#Homogeneous punctuation (ellipsis etc)
PUNCTUATION-MULTI=(?:\.|\-|[!\?]){2,}

#Date
DATE=\p{N}{1,2}[\.-]\p{Ps}?\p{N}{1,2}[\.-]\p{Ps}?\p{N}{2,4}
DATE-REVERSE=\p{N}{4}-\p{N}{1,2}[\.-]\p{N}{1,2}\.?

FRACNUMBER=\p{N}+(?:/\p{N}+)+

NUMBER-YEAR=(['`’‘´]\p{N}{2})(?:\P{N}|\z)

#Times
TIME=\p{N}{1,2}:\p{N}{1,2}(?::\p{N})?(?i:am|pm)?

#retain digits, including those starting with initial period (.22), and negative numbers
NUMBER=-?(?:[\.,]?\p{N}+)+

CURRENCY=\p{Sc}

WORD=[\p{L}\p{Mn}]+

PUNCTUATION=\p{P}

UNKNOWN=.

[PREFIXES]

[SUFFIXES]
\['`’‘´][sS]

[TOKENS]
\['`’‘´][sS]

[UNITS]
km
m
cm
mm
g
kg
C
l
s
sec
min
gb
mb
kb
St


[CURRENCY]
EUR
DM
USD

[ABBREVIATIONS]
Abb
Abf
Abk
Abs
Abt
abzgl
Adr
am
amtl
Anh
Anl
Anm
Aufl
b
Bd
beil
bes
Betr
Bez
Bhf
bzgl
bzw
ca
Chr
dgl
Dir
DM
Dr
dt
Dtzd
ehem
eigtl
eindschl
entspr
erb
erw
Erw
etc
ev
evtl
exkl
f
Fa
Fam
Ffm
Fr
Frl
frz
geb
Gebr
gedr
gegr
gek
Ges
gesch
gest
gez
ggf
ggfs
Hbf
Hg
hpts
Hptst
Hr
Hrn
Hrsg
Ing
Inh
inkl
jew
Jh
jhrl
Kap
kath
Kfm
kfm
kgl
Kl
l
led
Mio
möbl
Mr
Mrd
Msp
mtl
MwSt
MWSt
näml
Nr
o
Obb
od
österr
Pfd
Pl
Prof
r
Red
röm
röm\.-kath
S
s
Sa
schles
schwäb
schweiz
So
sog
Spvgg
St
Str
StR
südd
tägl
Tel
Ts
u
usw
Verf
verh
verw
vgl
vorm
z
zur
zus
Ztr
zzgl

[FILTER]
ﬂ fl
ﬀ ff
ﬃ ffi
ﬄ ffl
# also filter soft hyphen
\u00AD


[EOSMARKERS]
%include standard-eos

[QUOTES]
%include standard-quotes
%include exotic-quotes
uctodata 0.4-1 / usr / share / ucto / tokconfig-deu
