hyphenation rules are not found for words beginning with a capital letter

[jschoen42]

when testing with german text, i noticed that despite 69.000 special rules for compound words,
many german words have the hyphen in the wrong place, although there are actually rules for them

e.g. all with wrong hyphens

• Fortschritt -> Forts-chritt
• Abendstern -> Abends-tern
• Morgenthau -> Mor-gent-hau
• Gastherme -> Gasther-me
• Nennwertherabsetzung -> Nenn-wer-ther-ab-set-zung

in PyHyphen there is a special handling of completely capitalized words (mode 2, 3), there is no handling for words where only the first letter is capitalized

my workaround for 'syllables' ('pairs' has the same problem)

hyphen = Hyphenator("de_DE", directory=DATA_DIR)

def syllables_patch( word ):
    mode = 0
    if word.istitle():
        word = word.lower()
        mode = 4

    result = hyphen.syllables( word )
    if len(result)>0 and mode == 4:
        result[0] = result[0].title()

    return result

with the patch

• Fortschritt -> Fort-schritt
• Abendstern -> Abend-stern
• Morgenthau -> Mor-gen-thau
• Gastherme -> Gas-ther-me
• Nennwertherabsetzung -> Nenn-wert-her-ab-set-zung

now all hyphens are correct

the problem affects all rules in all other languages, not just the german combound rules - but the error is clearly visible here

Jürgen

[jschoen42]

I have tested the patch with the german word lists in repo https://github.com/cpos/AlleDeutschenWoerter

result with the patch

• "Substantive": 14.609 complete, 1.130 with corrected/improved hyphens
• "Verben": 4.727 complete, 311 with corrected/improved hyphens, when the verb is at the beginning of a sentence
• "Adjektive": 6.837 complete, 477 with corrected/improved hyphens, when the adjective is at the beginning of a sentence

[benjamin-mueller]

I have a similar problem with "fill". It does not hypenate correctly.

h_de = Hyphenator('de_DE')
phrase = "bedrohliche Zwangslage eines Alpinisten"
print(fill(phrase, width=9, use_hyphenator=h_de).replace("\n","#))

gives me the result

bedrohli-#che Zwang#slage ei-#nes Alpi-#nisten

Unfortunatly, it does not matter if it is lower() or title()

bedrohli-#che zwang#slage ei-#nes alpi-#nisten

Thanks for the help

Benjamin