summaryrefslogtreecommitdiffstats
path: root/container-search/src/main/java/com/yahoo/search/yql
diff options
context:
space:
mode:
authorAlexey Chernyshev <aleksei@spotify.com>2022-04-04 16:23:07 +0200
committerAlexey Chernyshev <aleksei@spotify.com>2022-04-07 14:44:30 +0200
commit7e9b33401201db9a9e22971dd419247e268bbfaa (patch)
treef5032a82e9fa74247b2fdeb3dcde4dc6cf98ce89 /container-search/src/main/java/com/yahoo/search/yql
parentad7cc1d11f0c19baa2344a643377576c559555f7 (diff)
Propagating annotations for fuzzy query
Diffstat (limited to 'container-search/src/main/java/com/yahoo/search/yql')
-rw-r--r--container-search/src/main/java/com/yahoo/search/yql/VespaSerializer.java29
-rw-r--r--container-search/src/main/java/com/yahoo/search/yql/YqlParser.java21
2 files changed, 48 insertions, 2 deletions
diff --git a/container-search/src/main/java/com/yahoo/search/yql/VespaSerializer.java b/container-search/src/main/java/com/yahoo/search/yql/VespaSerializer.java
index e778798b0e5..4b511df5e5f 100644
--- a/container-search/src/main/java/com/yahoo/search/yql/VespaSerializer.java
+++ b/container-search/src/main/java/com/yahoo/search/yql/VespaSerializer.java
@@ -21,6 +21,7 @@ import static com.yahoo.search.yql.YqlParser.GEO_LOCATION;
import static com.yahoo.search.yql.YqlParser.HIT_LIMIT;
import static com.yahoo.search.yql.YqlParser.IMPLICIT_TRANSFORMS;
import static com.yahoo.search.yql.YqlParser.LABEL;
+import static com.yahoo.search.yql.YqlParser.MAX_EDIT_DISTANCE;
import static com.yahoo.search.yql.YqlParser.NEAR;
import static com.yahoo.search.yql.YqlParser.NEAREST_NEIGHBOR;
import static com.yahoo.search.yql.YqlParser.NORMALIZE_CASE;
@@ -31,6 +32,7 @@ import static com.yahoo.search.yql.YqlParser.ORIGIN_OFFSET;
import static com.yahoo.search.yql.YqlParser.ORIGIN_ORIGINAL;
import static com.yahoo.search.yql.YqlParser.PHRASE;
import static com.yahoo.search.yql.YqlParser.PREFIX;
+import static com.yahoo.search.yql.YqlParser.PREFIX_LENGTH;
import static com.yahoo.search.yql.YqlParser.RANGE;
import static com.yahoo.search.yql.YqlParser.RANK;
import static com.yahoo.search.yql.YqlParser.RANKED;
@@ -526,7 +528,8 @@ public class VespaSerializer {
@Override
boolean serialize(StringBuilder destination, FuzzyItem fuzzy) {
- String annotations = leafAnnotations(fuzzy);
+ String annotations = fuzzyAnnotations(fuzzy);
+
destination.append(normalizeIndexName(fuzzy.getIndexName())).append(" contains ");
if (annotations.length() > 0) {
@@ -543,6 +546,30 @@ public class VespaSerializer {
}
return false;
}
+
+ static String fuzzyAnnotations(FuzzyItem fuzzyItem) {
+ boolean isMaxEditDistanceSet = fuzzyItem.getMaxEditDistance() != FuzzyItem.DefaultMaxEditDistance;
+ boolean isPrefixLengthSet = fuzzyItem.getPrefixLength() != FuzzyItem.DefaultPrefixLength;
+ boolean anyAnnotationSet = isMaxEditDistanceSet || isPrefixLengthSet;
+
+ StringBuilder builder = new StringBuilder();
+ if (anyAnnotationSet) {
+ builder.append("[{");
+ }
+ if (isMaxEditDistanceSet) {
+ builder.append(MAX_EDIT_DISTANCE + ":").append(fuzzyItem.getMaxEditDistance());
+ }
+ if (isMaxEditDistanceSet && isPrefixLengthSet) {
+ builder.append(",");
+ }
+ if (isPrefixLengthSet) {
+ builder.append(PREFIX_LENGTH + ":").append(fuzzyItem.getPrefixLength());
+ }
+ if (anyAnnotationSet) {
+ builder.append("}]");
+ }
+ return builder.toString();
+ }
}
private static class ONearSerializer extends Serializer<ONearItem> {
diff --git a/container-search/src/main/java/com/yahoo/search/yql/YqlParser.java b/container-search/src/main/java/com/yahoo/search/yql/YqlParser.java
index 06ee0e706f3..fcb19dde10d 100644
--- a/container-search/src/main/java/com/yahoo/search/yql/YqlParser.java
+++ b/container-search/src/main/java/com/yahoo/search/yql/YqlParser.java
@@ -126,6 +126,8 @@ public class YqlParser implements Parser {
private static final String RANKED_DESCRIPTION = "setting for whether to use term for ranking";
private static final String STEM_DESCRIPTION = "setting for whether to use stem if field implies it";
private static final String USE_POSITION_DATA_DESCRIPTION = "setting for whether to use position data for ranking this item";
+ private static final String MAX_EDIT_DISTANCE_DESCRIPTION = "setting for an inclusive upper bound for a fuzzy edit-distance search";
+ private static final String PREFIX_LENGTH_DESCRIPTION = "setting for a prefix length that is considered frozen for a fuzzy search";
private static final String USER_INPUT_ALLOW_EMPTY = "allowEmpty";
private static final String USER_INPUT_DEFAULT_INDEX = "defaultIndex";
private static final String USER_INPUT_GRAMMAR = "grammar";
@@ -194,6 +196,9 @@ public class YqlParser implements Parser {
public static final String WEIGHT = "weight";
public static final String WEIGHTED_SET = "weightedSet";
public static final String FUZZY = "fuzzy";
+ public static final String MAX_EDIT_DISTANCE = "maxEditDistance";
+ public static final String PREFIX_LENGTH = "prefixLength";
+
private final IndexFacts indexFacts;
private final List<ConnectedItem> connectedItems = new ArrayList<>();
@@ -1313,7 +1318,21 @@ public class YqlParser implements Parser {
String wordData = getStringContents(args.get(0));
- FuzzyItem fuzzy = new FuzzyItem(field, true, wordData);
+ Integer maxEditDistance = getAnnotation(
+ ast,
+ MAX_EDIT_DISTANCE,
+ Integer.class,
+ FuzzyItem.DefaultMaxEditDistance,
+ MAX_EDIT_DISTANCE_DESCRIPTION);
+
+ Integer prefixLength = getAnnotation(
+ ast,
+ PREFIX_LENGTH,
+ Integer.class,
+ FuzzyItem.DefaultPrefixLength,
+ PREFIX_LENGTH_DESCRIPTION);
+
+ FuzzyItem fuzzy = new FuzzyItem(field, true, wordData, maxEditDistance, prefixLength);
return leafStyleSettings(ast, fuzzy);
}