if vfst_word_class == "[Lp]":
entry = '[Lp]%s%s%s%s%s:%s%s EtuliitteenJatko_%s;' \
% (debug_info, rakenne, alkuWithTags, diacritics, infoFlags, alku, diacritics, get_prefix_jatko(word, altform))
else:
entry = '%s%s%s%s%s%s:%s%s %s%s_%s ;' \
% (vfst_word_class, debug_info, rakenne, infoFlags, injectBaseformToStructure(outputBaseform, alkuWithTags),
diacritics, alku, diacritics, vfst_class_prefix, jatko, vfst_vtype)
vocabularyFile.write(entry + "\n")
# Sanity check for alternative forms: if there are both multi part forms and single part forms
# then all multi part forms must end with a part contained in the single part set.
if singlePartForms:
for multiPartForm in multiPartForms:
lastPart = multiPartForm[max(multiPartForm.rfind(
"="), multiPartForm.rfind("|"), multiPartForm.rfind("-")) + 1:]
if lastPart not in singlePartForms:
sys.stderr.write(
"ERROR: suspicious alternative spelling: %s\n" %
multiPartForm)
sys.exit(1)
voikkoutils.process_wordlist(generate_lex_common.VOCABULARY_DATA + '/joukahainen.xml', \
handle_word, True)
for fileSuffix in vocabularyFileSuffixes:
vocabularyFiles[fileSuffix].write(
"\n\n"
) # Extra line feeds needed to avoid mixed lines in concatenated lexc file
vocabularyFiles[fileSuffix].close()
elif vtype == voikkoutils.VOWEL_BOTH: malaga_vtype = 'aä'
rakenne = generate_lex_common.get_structure(altform, malaga_word_class)
if baseform is None:
altBaseform = altform
else:
altBaseform = baseform
if malaga_word_class == "lyhenne":
perusmuotoEntry = ""
else:
perusmuotoEntry = 'perusmuoto: "%s", ' % altBaseform
entry = '[%salku: "%s", luokka: %s, jatko: %s, äs: %s%s%s%s];' \
% (perusmuotoEntry, alku, malaga_word_class, malaga_jatko, malaga_vtype, malaga_flags,
generate_lex_common.get_structure(altform, malaga_word_class),
additional_attributes)
generate_lex_common.write_entry(main_vocabulary, {}, word, entry)
# Sanity check for alternative forms: if there are both multi part forms and single part forms
# then all multi part forms must end with a part contained in the single part set.
if singlePartForms:
for multiPartForm in multiPartForms:
lastPart = multiPartForm[max(multiPartForm.rfind("="), multiPartForm.rfind("|"), multiPartForm.rfind("-")) + 1:]
if lastPart not in singlePartForms:
sys.stderr.write("ERROR: suspicious alternative spelling: %s\n" % multiPartForm)
sys.exit(1)
voikkoutils.process_wordlist(generate_lex_common.VOCABULARY_DATA + '/joukahainen.xml', \
handle_word, True)
main_vocabulary.close()