Riksdagsmonitor Intelligence Platform โ€” API Documentation - v1.0.30
    Preparing search index...
    SENTENCE_END_ABBREVIATIONS: readonly string[] = ...

    Common abbreviations that end with . followed by a space โ€” these must NOT be treated as sentence boundaries by truncateToSentenceBoundary, otherwise the description gets cut mid-sentence at e.g. โ€ฆ two propositions: the forestry deregulation (prop. (audit 2026-05-09 of news/2026-05-08-motions-en.html).

    Token comparison is case-insensitive โ€” abbreviations like prop. and Prop. are both treated as non-terminating. Matching is done by looking at the last whitespace-delimited word ending at the candidate sentence-end position; the word is lower-cased and looked up in SENTENCE_END_ABBREV_SET.

    Note: case-insensitive matching means Mr. will also match an (extremely unlikely) in-word Imr. ending, but the practical risk is negligible because the word match is anchored to a whitespace boundary on the left, not just any character.

    Keep the list short and high-signal โ€” false-negatives (we miss an abbreviation and cut early) only mean the description fragment is a bit shorter; false-positives (we treat a real sentence end as an abbreviation and overrun) would push the description past hardMax, which is unrecoverable downstream.