161sh/CatgirlIntelligenceAgency
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Simplified query builders, preparation for a-tag inclusion.

Browse Source
This commit is contained in:
vlofgren 2022-08-01 20:29:15 +02:00
parent 9a4183a481
commit 3ccb1c6218
2 changed files with 35 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
marginalia_nu/src/main/java/nu/marginalia/wmsa/edge/index/reader/SearchIndexReader.java
View File

@ -57,18 +57,15 @@ public class SearchIndexReader implements AutoCloseable {
queryBuilders = new EnumMap<>(IndexBlock.class); queryBuilders = new EnumMap<>(IndexBlock.class);
underspecifiedQueryBuilders = new EnumMap<>(IndexBlock.class); underspecifiedQueryBuilders = new EnumMap<>(IndexBlock.class);
queryBuilders.put(IndexBlock.Words, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex, midIndex, lowIndex, namesIndex, wordsIndex), wordsIndex)); queryBuilders.put(IndexBlock.Words, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex, midIndex, lowIndex, linkIndex, namesIndex, wordsIndex), wordsIndex));
queryBuilders.put(IndexBlock.Low, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex, midIndex, lowIndex, namesIndex), wordsIndex)); queryBuilders.put(IndexBlock.Low, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex, midIndex, lowIndex, linkIndex), wordsIndex));
queryBuilders.put(IndexBlock.Middle, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex, midIndex), wordsIndex));
queryBuilders.put(IndexBlock.Top, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex), wordsIndex)); queryBuilders.put(IndexBlock.Top, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, topIndex), wordsIndex));
queryBuilders.put(IndexBlock.PositionWords, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, namesIndex, positionIndex), wordsIndex));
queryBuilders.put(IndexBlock.NamesWords, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, namesIndex), wordsIndex));
queryBuilders.put(IndexBlock.Link, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex, linkIndex), wordsIndex));
queryBuilders.put(IndexBlock.Title, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex), wordsIndex)); queryBuilders.put(IndexBlock.Title, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex, topicIndex, titleIndex), wordsIndex));
queryBuilders.put(IndexBlock.TitleKeywords, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex), wordsIndex)); queryBuilders.put(IndexBlock.TitleKeywords, new IndexQueryBuilder(listOfNonNulls(metaIndex, titleKeywordsIndex), wordsIndex));
underspecifiedQueryBuilders.put(IndexBlock.TitleKeywords, new IndexQueryBuilder(listOfNonNulls(titleKeywordsIndex, linkIndex, topicIndex, topIndex, midIndex, lowIndex, namesIndex, positionIndex, metaIndex), wordsIndex)); underspecifiedQueryBuilders.put(IndexBlock.TitleKeywords, new IndexQueryBuilder(listOfNonNulls(titleKeywordsIndex, linkIndex, topicIndex, topIndex, midIndex, lowIndex, namesIndex, positionIndex, metaIndex), wordsIndex));
underspecifiedQueryBuilders.put(IndexBlock.Link, new IndexQueryBuilder(listOfNonNulls(linkIndex, topicIndex, topIndex, midIndex, lowIndex, namesIndex, positionIndex, metaIndex), wordsIndex)); underspecifiedQueryBuilders.put(IndexBlock.Title, new IndexQueryBuilder(listOfNonNulls(titleIndex, topicIndex, linkIndex, topicIndex, topIndex, midIndex, lowIndex, namesIndex, positionIndex, metaIndex), wordsIndex));
underspecifiedQueryBuilders.put(IndexBlock.Top, new IndexQueryBuilder(listOfNonNulls(topIndex, linkIndex, midIndex, lowIndex, namesIndex, positionIndex, metaIndex), wordsIndex));
} }
@SafeVarargs @SafeVarargs
@ -121,7 +118,12 @@ public class SearchIndexReader implements AutoCloseable {
} }
public Query findWord(IndexBlock block, IndexSearchBudget budget, LongPredicate filter, int wordId) { public Query findWord(IndexBlock block, IndexSearchBudget budget, LongPredicate filter, int wordId) {
return queryBuilders.get(block).build(budget, filter, wordId); var builder = queryBuilders.get(block);
if (builder == null)
return Query.EMPTY;
return builder.build(budget, filter, wordId);
} }
@Override @Override
@ -135,13 +137,20 @@ public class SearchIndexReader implements AutoCloseable {
@SneakyThrows @SneakyThrows
public long numHits(IndexBlock block, int word) { public long numHits(IndexBlock block, int word) {
return numHitsCache.get(Pair.of(block, word), return numHitsCache.get(Pair.of(block, word), () -> numHitsForBlockWord(block, word));
() -> queryBuilders.get(block) }
private long numHitsForBlockWord(IndexBlock block, int word) {
IndexQueryBuilder builder = queryBuilders.get(block);
if (builder == null)
return 0L;
return builder
.getIndicies() .getIndicies()
.stream() .stream()
.mapToLong(idx -> idx.numUrls(word)) .mapToLong(idx -> idx.numUrls(word))
.sum() .sum();
);
} }
public IndexBlock getBlockForResult(int searchTerm, long urlId) { public IndexBlock getBlockForResult(int searchTerm, long urlId) {

11
marginalia_nu/src/main/java/nu/marginalia/wmsa/edge/index/reader/query/Query.java
View File

@ -3,6 +3,17 @@ package nu.marginalia.wmsa.edge.index.reader.query;
import java.util.stream.LongStream; import java.util.stream.LongStream;
public interface Query { public interface Query {
Query EMPTY = new Query() {
@Override
public Query also(int wordId) { return this; }
@Override
public Query not(int wordId) { return this; }
@Override
public LongStream stream() { return LongStream.empty(); }
};
Query also(int wordId); Query also(int wordId);
Query not(int wordId); Query not(int wordId);