Add rules for Swedish

src/rules/disallowed_words/sv.txt

@@ -0,0 +1,180 @@
+bl
+fr
+IUCN
+bg
+bh
+c:a
+cd
+d:r
+dr
+iaf
+iofs
+ISBN
+IVA
+jbo
+jb/o
+k:a
+kbfd
+KBM
+KFM
+mc
+mm
+mnkr
+Mkr
+msk
+mtp
+n/a
+N:o
+nr
+N:s
+obs
+omm
+oms
+pg
+pua
+RC
+R:dr
+Rdr
+R:gs
+RIP
+s:a
+SEK
+Sk:pund
+Sk
+sms
+sr
+ss
+ssk
+S:t
+S:ta
+tfn
+tgm
+tjf
+tjl
+tjm
+tkr
+t:r
+trpt
+trsp
+VAB
+vd
+vpl
+C/o
+OSA
+ZIP code
+Ftf
+AA
+AB
+AF
+ASEA
+ASEAN
+CSA
+CSA
+CSN
+DACO
+DAFA
+DCRI
+EG
+EMU
+FK
+FRA
+FN
+JK
+JO
+JämO
+KDU
+KF
+KO
+Komintern
+LO
+MUF
+NATO
+NF
+NBV
+NO
+OD
+OK
+PO
+RFoD
+RFSL
+RFSU
+ROKS
+SACO
+SAP
+SCB
+SEB
+SIFO
+SKF
+SKL
+SKR
+SKTF
+SR
+SSAB
+SSG
+SSR
+SSRS
+SSU
+TCO
+TRV
+Unesco
+Unicef
+WHO
+WWF
+sic
+AC
+AM
+ATB
+bps
+DC
+DIN
+DPF
+DSG
+DVD
+EAN
+EGR
+FM
+HF
+IF
+ISO
+KV
+LV
+LW
+MF
+MV
+MW
+MMS
+NOM
+PC
+PCM
+PWM
+RAM
+ROM
+rpm
+SQR
+SQRT
+SSB
+SW
+TDI
+UHF
+UKV
+VHF
+VLF
+DM
+DNF
+DNS
+efl
+FM
+GM
+GP
+GS
+GS
+IEM
+IVM
+JSM
+KO
+MMA
+MTB
+NM
+TKO
+USM
+VC

src/rules/sv.toml

@@ -0,0 +1,43 @@
+min_trimmed_length = 3
+min_word_count = 2
+max_word_count = 14
+min_characters = 2
+may_end_with_colon = false
+quote_start_with_letter = true
+needs_punctuation_end = true
+needs_letter_start = true
+needs_uppercase_start = true
+# This should cover most common Swedish words
+allowed_symbols_regex = "[a-zåäöA-ZÅÄÖé,.?!: ]"
+broken_whitespace = ["  ", " ,", " .", " ?", " !", " ;"]
+
+# No roman numerals (common in texts about Swedish kings etc., but hard to read out)
+# The :s is for genitive constructions (Karl XII:s)
+other_patterns = ["[MDCLXVI]+(:s)?"]
+
+# We don’t allow quotes or parenthesis symbols. No other use for matching
+# (Swedish quotes are actually opening "”" and closing "”", so impossible to
+# match anyway)
+# matching_symbols = [ ]
+
+#expand some commmon abbreviations that will probably be reasonable and unambigous to speak out
+replacements = [
+  ["bl.a.", "bland annat"],
+  ["ca.", "cirka"],
+  ["d.v.s.", "det vill säga"],
+  ["fr.o.m.", "från och med"],
+  ["i.o.m.", "i och med"],
+  ["m.m.", "med mera"],
+  ["m.fl.", "med flera"],
+  ["o.d.", "och dylikt"],
+  ["o.dyl.", "och dylikt"],
+  ["o.s.v.", "och så vidare"],
+  ["p.g.a.", "på grund av"],
+  ["resp.", "respektive"],
+  ["t.ex.", "till exempel"],
+  ["t.o.m.", "till och med"],
+]
+
+# Filter out lots of other abbreviations
+# Taken from list: https://sv.wikipedia.org/wiki/Lista_%C3%B6ver_f%C3%B6rkortningar
+abbreviation_patterns = [ "art\\.|bl\\.a\\.|B\\.V\\.|civ\\.ek\\.|civ\\.ing\\.|doc\\.|d\\.v\\.s\\.|d\\.y\\.|d\\.ä\\.|ekon\\.|farm\\.|f\\.d\\.|fig\\.|fil\\.|fr\\.o\\.m\\.|Ibid\\.|ib\\.|i\\.o\\.m\\.|i\\.s\\.f\\.|lb\\.|lic\\.|lisp\\.|mag\\.|med\\.|m\\.h\\.p\\.p\\.|min\\.|m\\.m\\.|m\\.m\\.d\\.|mom\\.|m\\.v\\.h\\.|möjl\\.|m ö\\.h\\.|n\\.b\\.|näml\\.|nästk\\.|obs\\.|o\\.d\\.|odont\\.|o\\.dyl\\.|o\\.k\\.s\\.|omkr\\.|o\\.m\\.s\\.|op\\.|ordf\\.|org\\.nr|o\\.s\\.v\\.|pers\\.|p\\.g\\.a\\.|pol\\.|prel\\.|prof\\.|prov\\.|rc\\.|ref\\.|resp\\.|R\\.I\\.P\\.|rst\\.|s\\.a\\.s\\.|sek\\.|sekr\\.|s\\.g\\.s\\.|sid\\.|sign\\.|sistl\\.|s\\.k\\.|sk\\.|skepp\\.|skålp\\.|s\\.m\\.|s\\.m\\.s\\.|sp\\.|spec\\.|s\\.st\\.|st\\.|St\\.|stud\\.|särsk\\.|tab\\.|tekn\\.|tel\\.|temp\\.|teol\\.|t\\.ex\\.|tf\\.|t\\.h\\.|tim\\.|t\\.o\\.m\\.|trol\\.|t\\.v\\.|ung\\.|u\\.a\\.|u\\.f\\.a\\.|u\\.p\\.a\\.|urspr\\.|usk\\.|utg\\.|å\\.k\\.|äv\\.|ö\\.a\\.|övers\\.anm\\.|ö\\.h\\.|ö\\.h\\.t\\.|ök\\.|övers\\.|att\\.|Avs\\.|b\\.v\\.|D\\.S\\.|n\\.b\\.|o\\.s\\.a\\.|P\\.P\\.S\\.|P\\.S\\.|tr\\.|ö\\.g\\.|A\\.D\\.|e\\.Kr\\.|e\\.v\\.t\\.|g\\.s\\.|mån\\.|s\\.å\\.|civ\\.ek\\.|civ\\.ing\\.|ekon\\.dr|ekon\\.mag\\.|ekon\\.kand\\.|fil\\.dr|fil\\.lic\\.|fil\\.kand|fil\\.mag|jur\\.kand\\.|jur\\.utr\\.kand\\.|jur\\.lic\\.|jur\\.dr|med\\.dr|med\\.lic\\.|med\\.kand\\.|odont\\.kand\\.|odont\\.lic\\.|odont\\.dr|pol\\.kand\\.|pol\\.mag\\.|pol\\.dr|tekn\\.dr|tekn\\.lic\\.|teol\\.kand\\.|teol\\.lic\\.|teol\\.dr|a\\.a\\.|e\\.g\\.|e\\.o\\.|ibid\\.|id\\.|i\\.e\\.|L\\.s\\.|l\\.s\\.|m\\.p\\.|N\\.N\\.|op\\.cit\\.|alban\\.|arab\\.|aram\\.|armen\\.|assyr\\.|avest\\.|babyl\\.|bret\\.|bulg\\.|dan\\.|egypt\\.|eng\\.|est\\.|fa\\.|fd\\.|fenic\\.|fe\\.|feng\\.|ffrans\\.|ffris\\.|fgutn\\.|fht\\.|fi\\.|fin\\.|find\\.|flfran\\.|fnor\\.|fpers\\.|fpreuss\\.|fr\\.|frans\\.|frank\\.|fris\\.|fsax\\.|fslav\\.|fsv\\.|fvn\\.|fär\\.|gael\\.|gall\\.|georg\\.|got\\.|gr\\.|grek\\.|hebr\\.|hett\\.|hind\\.|hindost\\.|holl\\.|ieur\\.|ir\\.|irl\\.|isl\\.|it\\.|ital\\.|jap\\.|kelt\\.|kines\\.|korn\\.|kymr\\.|kyrkslav\\.|lat\\.|lap\\.|lapp\\.|lett\\.|lit\\.|lt\\.|mag\\.|meng\\.|mfr\\.|mholl\\.|mht\\.|mlat\\.|mlt\\.|mnl\\.|mnt\\.|moes\\.|nfris\\.|nht\\.|nl\\.|no\\.|nor\\.|ns\\.|nt\\.|nyfris\\.|nygr\\.|nyheb\\.|nyisl\\.|nysv\\.|osset\\.|osk\\.|per\\.|pers\\.|pol\\.|port\\.|prt\\.|pt\\.|provenç\\.|prov\\.|rundan\\.|runsv\\.|ry\\.|ryss\\.|sam\\.|sard\\.|skr\\.|slav\\.|semit\\.|sengr\\.|senlat\\.|shet\\.|slovak\\.|sloven\\.|slov\\.|sorb\\.|sp\\.|span\\.|sv\\.|syr\\.|tam\\.|tjeck\\.|tochar\\.|ty\\.|ung\\.|urgerm\\.|urn\\.|wal\\.|vall\\.|wogul\\.|zig\\.|æthiop\\.|disk\\.|s\\.d\\.|w\\.o\\." ]