use super::*;
pub(in super::super) fn decode_graph(
graph: &Graph,
options: &LatticeOptions<'_>,
extra: Option<&Graph>,
) -> Vec<SentencePath> {
let n = graph.len();
let personal = active_personal(options);
let bigram_table = options.bigram.as_ref();
let bigram = options.bigram.as_deref();
let mut memo = options.bigram_memo.map(RefCell::borrow_mut);
let capacities = column_capacities(graph, extra, options.beam);
let mut columns: Vec<Vec<Hyp<'_>>> = capacities.into_iter().map(Vec::with_capacity).collect();
columns[0].push(Hyp {
score: 0.0,
prev: (0, 0),
edge: None,
typo_edges: 0,
});
let row_capacity = options.beam.max(options.nbest);
let mut bigram_rows: Vec<Vec<f32>> = Vec::new();
let mut bigram_row_indices: HashMap<&str, usize> = HashMap::new();
let mut outgoing: Vec<&Edge> = Vec::new();
for pos in 0..n {
keep_beam(&mut columns[pos], options.beam);
let (done, ahead) = columns.split_at_mut(pos + 1);
let current = &done[pos];
let mut used_bigram_rows = 0;
bigram_row_indices.clear();
outgoing.clear();
outgoing.extend(
graph[pos]
.iter()
.chain(extra.into_iter().flat_map(|extra| extra[pos].iter())),
);
for (index, hyp) in current.iter().enumerate() {
let previous = hyp.edge.map_or(SENTENCE_START, |edge| edge.word.as_str());
let bonuses = bigram_table.map(|table| {
let row = if let Some(&row) = bigram_row_indices.get(previous) {
row
} else {
let row = used_bigram_rows;
if row == bigram_rows.len() {
if bigram_rows.is_empty() {
bigram_rows.reserve_exact(row_capacity);
bigram_row_indices.reserve(row_capacity);
}
bigram_rows.push(Vec::new());
}
let scores = &mut bigram_rows[row];
scores.clear();
scores.reserve_exact(outgoing.len());
match memo.as_deref_mut() {
Some(memo) => scores.extend(
outgoing
.iter()
.map(|edge| score(memo, table, previous, &edge.word)),
),
None => scores.extend(
outgoing
.iter()
.map(|edge| table.bigram(previous, &edge.word)),
),
}
used_bigram_rows += 1;
bigram_row_indices.insert(previous, row);
row
};
&bigram_rows[row]
});
let context = personal.map(|personal| {
let earlier = match hyp.prev {
(column, at) if column > 0 => {
done[column][at].edge.map(|edge| edge.word.as_str())
}
_ => None,
};
personal.context(earlier, Some(previous))
});
for (at, &edge) in outgoing.iter().enumerate() {
let mut score = hyp.score + edge.log_prob;
let static_bonus = bonuses.map_or(0.0, |bonuses| f64::from(bonuses[at]));
if bigram.is_some() {
score += options.bigram_weight * static_bonus.max(0.0);
}
if let (Some(personal), Some(context)) = (personal, context.as_ref()) {
if context.confidence > 0.0 {
score += personal_transition(
personal,
context,
&edge.word,
static_bonus,
options,
);
}
}
ahead[edge.end - pos - 1].push(Hyp {
score,
prev: (pos, index),
edge: Some(edge),
typo_edges: hyp.typo_edges + usize::from(edge.typo),
});
}
}
}
let mut last = std::mem::take(&mut columns[n]);
if last.is_empty() {
return Vec::new();
}
keep_beam(&mut last, options.beam.max(options.nbest));
cxx_sort::sort(&mut last, &mut scores_higher);
let take = options.nbest.min(last.len());
let inline_unique = last.len() <= SMALL_SENTENCE_PATHS;
let mut paths: Vec<SentencePath> = Vec::with_capacity(last.len());
for hyp in &last {
if inline_unique && paths.len() == take {
break;
}
let mut words = Vec::with_capacity(n);
let mut keys = Vec::with_capacity(n);
let mut cursor = Some(hyp);
while let Some(step) = cursor {
let Some(edge) = step.edge else { break };
words.push(edge.word.clone());
keys.push(edge.key.as_str());
let (column, at) = step.prev;
cursor = Some(&columns[column][at]);
}
words.reverse();
keys.reverse();
let sentence = words.concat();
if sentence.is_empty()
|| (inline_unique && paths.iter().any(|path| path.sentence == sentence))
{
continue;
}
paths.push(SentencePath {
sentence,
key: keys.join("'"),
log_prob: hyp.score,
words,
typo_edges: hyp.typo_edges,
});
}
if !inline_unique {
retain_unique_sentences(&mut paths, take);
}
rescore_with_trigram(&mut paths, options);
paths
}
pub(super) fn score(
memo: &mut BigramMemo,
table: &Arc<NgramTable>,
previous: &str,
next: &str,
) -> f32 {
let Some(key) = table.bigram_key(previous, next) else {
return 0.0;
};
let address = Arc::as_ptr(table) as usize;
if memo.table != address {
memo.scores.clear();
memo.table = address;
}
if let Some(&score) = memo.scores.get(&key) {
return score;
}
if memo.scores.len() >= BIGRAM_MEMO_CAPACITY {
memo.scores.clear();
}
let score = table.score(key);
memo.scores.insert(key, score);
score
}