## macros to be used in RegExps rules
<Macros>
ALPHA      [^\]<>[(\.,";:?!¿¡«»'`)^@~|}{_/\\+=&$#*+%\s\d\-]     
ALPHANUM   [^\]<>[(\.,";:?!¿¡«»'`)^@~|}{_/\\+=&$#*+%\s\-]
NOALPHANUM [\]<>[(\.,";:?!¿¡«»'`)^@~|}{_/\\+=&$#*+%\s\-]
PARTNUM    [^\]<>[(";:?!¿¡«»'`)^@~|}{\s]
OTHERS     [\]<>[(\.,";:?!¿¡«»'`)^@~|}{_/\\+=&$#*+%\-]
</Macros>

## Tokenization rules. They are applied in the order of definition.
## The first matching the *beginning* of the line is applied
## and a token built. The process is repeated until the line 
## has been completely processed.
##    -The first field in the rule is the rule name. If it starts
##     with a "*", the RegExp will only produce a token if the 
##     match is found in abbreviation list <Abbreviations> below.
##    -The second field in the rule is the substring to form the token/s with
##     It may be 0 (the match of the whole expression) or any number
##     from 1 to the number of substrings (up to 9). A token will be
##     created for each substring from 1 to the specified value.
##    -The third field is the regexp to match against the line
## 
<RegExps>
INDEX_SEQUENCE  0  (\.{4,}|-{2,}|\*{2,}|_{2,}|/{2,})
INITIALS1       1  ([A-Z](\.[A-Z])+)(\.\.\.)
INITIALS2 	0  ([A-Z]\.)+
TIMES           0  (([01]?[0-9]|2[0-4]):[0-5][0-9])
NAMES_CODES	0  ({PARTNUM}*[0-9]{PARTNUM}*{ALPHANUM})
THREE_DOTS 	0  (\.\.\.)
QUOTES	        0  (``|<<|>>|'')
MAILS 	        0  {ALPHANUM}+([\._]{ALPHANUM}+)*@{ALPHANUM}+([\._]{ALPHANUM}+)*
URLS 	        0  ((mailto:|(news|http|https|ftp|ftps)://)\S+|^(www(\.\S+)+))
KEEP_COMPOUNDS  0  {ALPHA}+(['_\-]{ALPHA}+)+
*ABREVIATIONS1  0  (({ALPHA}+\.)+)(?!\.\.)
*ABREVIATIONS2  0  ({ALPHA}+\.)(?!\.\.)
WORD            0  {ALPHANUM}+
OTHERS_C        0  {OTHERS}
</RegExps>

## Abbreviations.  The dot is not tokenized separately
## in the cases listed below.
<Abbreviations>
a.c.
aa.rr.
abrev.
adj.
adm.
admón.
afma.
afmas.
afmo.
afmos.
ag.
am.
ap.
apdo.
art.
arts.
arz.
arzbpo.
assn.
atte.
av.
avda.
bros.
bv.
cap.
caps.
cg.
cgo.
cia.
cit.
cl.
cm.
co.
col.
corp.
cos.
cta.
cte.
ctra.
cts.
cía.
d.c.
dcha.
dept.
depto.
dg.
dl.
dm.
doc.
docs.
dpt.
dpto.
dr.
dra.
dras.
dres.
dto.
dupdo.
ed.
ee.uu.
ej.
emma.
emmas.
emmo.
emmos.
entlo.
entpo.
esp.
etc.
ex.
excm.
excma.
excmas.
excmo.
excmos.
fasc.
fdo.
fig.
figs.
fol.
fra.
gb.
gral.
ha.
hnos.
hros.
hz.
ib.
ibid.
ibíd.
id.
ilm.
ilma.
ilmas.
ilmo.
ilmos.
iltre.
inc.
intr.
izq.
izqda.
izqdo.
jr.
kc.
kcal.
kg.
khz.
kl.
km.
kw.
lda.
ldo.
lib.
lim.
loc.
ltd.
ltda.
lám.
ma.
mg.
mhz.
min.
mons.
mr.
mrs.
ms.
mss.
mtro.
máx.
mín.
ntra.
ntro.
núm.
ob.
obpo.
op.
pd.
ph.
pje.
pl.
plc.
pm.
pp.
ppal.
pral.
prof.
prov.
pról.
ps.
pta.
ptas.
pte.
pts.
pza.
pág.
págs.
párr.
rda.
rdo.
ref.
reg.
rel.
rev.
revda.
revdo.
rma.
rmo.
rte.
s.
sdad.
sec.
secret.
seg.
sg.
sig.
smo.
sr.
sra.
sras.
sres.
srs.
srta.
ss.mm.
sta.
sto.
sust.
tech.
tel.
telf.
teléf.
ten.
tfono.
tlf.
tít.
ud.
uds.
vda.
vdo.
vid.
vol.
vols.
vra.
vro.
vta.
íd.
ít.
mm.
mms.
ms.
pulg.
yda.
mi.
Ha.
ac.
ml.
dl.
hl.
ac-pie.
oz.
qt.
gal.
pk.
bu.
cr.
crt.
tz.
pt.
mpa.
pa.
psi.
lb.
mmhg.
cmhg.
mhg.
mol.
mg.
gr.
grs.
kg.
kgs.
mgr.
oz.
lb.
ton.
tm.
milgal.
lt.
lps.
gps.
gpm.
gph.
gpd.
mgd.
gal.
gpcd.
mph.
lbf.
yb.
zb.
eb.
pb.
tb.
gb.
mb.
kb.
wb.
cd.
rad.
sr.
hz.
lm.
lx.
nq.
gy.
sv.
</Abbreviations>