From 1f5b52dfd38a96e74628c91efaf3d6a25b02b050 Mon Sep 17 00:00:00 2001
From: 3y3 <3y3@ya.ru>
Date: Wed, 25 Sep 2024 17:32:45 +0300
Subject: [PATCH] fix: Improve phrase scoring

---
 src/worker/score.ts  | 61 ++++++++++++++++++++------------------------
 src/worker/search.ts | 22 ++++++++++++----
 2 files changed, 44 insertions(+), 39 deletions(-)

diff --git a/src/worker/score.ts b/src/worker/score.ts
index faf6435..e7ee50b 100644
--- a/src/worker/score.ts
+++ b/src/worker/score.ts
@@ -13,7 +13,7 @@ type ScoreState = {
     score: number;
     prev: ResultToken | null | undefined;
     curr: ResultToken;
-    phrase: string;
+    phrase: string[];
     position: Position;
 };
 
@@ -89,7 +89,7 @@ export function phrased(result: Index.Result, terms: string[]) {
             prev: null,
             curr: token,
             position: token.position.slice() as Position,
-            phrase: token.text,
+            phrase: [token.text],
         };
 
         return match;
@@ -101,7 +101,7 @@ export function phrased(result: Index.Result, terms: string[]) {
 
         state.score = 0;
         state.position = state.curr.position.slice() as Position;
-        state.phrase = state.curr.text;
+        state.phrase = [state.curr.text];
 
         if (!tokens.length) {
             return end;
@@ -117,20 +117,12 @@ export function phrased(result: Index.Result, terms: string[]) {
 
         state.prev = state.curr;
         state.curr = tokens.shift() as ResultToken;
-        state.phrase += ' ' + state.curr.text;
+        state.phrase.push(state.curr.text);
 
         return match;
     }
 
     function match() {
-        if (terms.includes(state.curr.text as string)) {
-            return scoreToken;
-        } else {
-            return scoreWildcard;
-        }
-    }
-
-    function scoreToken() {
         const {prev, curr} = state;
 
         state.score += 2;
@@ -139,12 +131,8 @@ export function phrased(result: Index.Result, terms: string[]) {
             return nextToken;
         }
 
-        // This is partially buggy, if phrase has more that one similar token
-        if (distance(prev.position, curr.position) <= MERGE_TOLERANCE) {
-            if (phrase.includes(state.phrase)) {
-                state.score += 10;
-            }
-
+        if (isPhrase(phrase, state.phrase, distance(prev.position, curr.position))) {
+            state.score += 10;
             state.position[1] = curr.position[1];
 
             return nextToken;
@@ -153,22 +141,6 @@ export function phrased(result: Index.Result, terms: string[]) {
         return nextScore;
     }
 
-    function scoreWildcard() {
-        const {prev, curr} = state;
-
-        state.score += 0.5;
-
-        if (prev && distance(prev.position, curr.position) <= MERGE_TOLERANCE) {
-            if (phrase.includes(state.phrase)) {
-                state.score += 0.5;
-            }
-
-            state.position[1] = state.curr.position[1];
-        }
-
-        return nextScore;
-    }
-
     function end() {
         results = dedupe(results);
         return null;
@@ -234,6 +206,27 @@ function dedupe(tokens: ScoreResult[]) {
     return result;
 }
 
+function isPhrase(phrase: string, tokens: string[], distance: number) {
+    if (distance > MERGE_TOLERANCE) {
+        return false;
+    }
+
+    tokens = tokens.slice();
+
+    let index = 0;
+    while (tokens.length && index > -1) {
+        const token = tokens.shift() as string;
+
+        index = phrase.indexOf(token, index);
+
+        if (index > -1) {
+            index += token.length;
+        }
+    }
+
+    return index > -1;
+}
+
 function isIntersection(a: Position, b: Position) {
     return (a[1] >= b[0] && a[1] <= b[1]) || (a[1] >= b[0] && a[1] <= b[1]);
 }
diff --git a/src/worker/search.ts b/src/worker/search.ts
index 05082f9..b4420c8 100644
--- a/src/worker/search.ts
+++ b/src/worker/search.ts
@@ -30,7 +30,7 @@ interface FixedClause extends Query.Clause {
 
 const makeStrategies = (tolerance: number, index: Index, clauses: FixedClause[], sealed: boolean) =>
     [
-        tolerance > -1 &&
+        tolerance >= 0 &&
             function precise(query: Query) {
                 query.clauses = clauses.slice();
 
@@ -42,7 +42,7 @@ const makeStrategies = (tolerance: number, index: Index, clauses: FixedClause[],
                     }
                 }
             },
-        tolerance > 0 &&
+        tolerance >= 1 &&
             function trailingWildcard(query: Query) {
                 query.clauses = clauses.map((clause) => {
                     if (clause.presence !== Query.presence.PROHIBITED) {
@@ -51,7 +51,7 @@ const makeStrategies = (tolerance: number, index: Index, clauses: FixedClause[],
                     return clause;
                 });
             },
-        tolerance > 1 &&
+        tolerance >= 2 &&
             function bothWildcard(query: Query) {
                 query.clauses = clauses.map((clause) => {
                     if (clause.presence !== Query.presence.PROHIBITED) {
@@ -102,14 +102,26 @@ export function search(
 }
 
 function wildcard(clause: FixedClause, mode: Query.wildcard) {
+    const requiredLength =
+        [
+            // eslint-disable-next-line no-bitwise
+            mode & Query.wildcard.TRAILING ? 2 : 0,
+            // eslint-disable-next-line no-bitwise
+            mode & Query.wildcard.LEADING ? 2 : 0,
+        ].reduce((a, b) => a + b, 0) + 1;
+
+    if (clause.term.length < requiredLength) {
+        return;
+    }
+
     // eslint-disable-next-line no-bitwise
     if (mode & Query.wildcard.TRAILING) {
-        clause.term = clause.term + '*';
+        clause.term = clause.term.slice(0, -1) + '*';
     }
 
     // eslint-disable-next-line no-bitwise
     if (mode & Query.wildcard.LEADING) {
-        clause.term = '*' + clause.term;
+        clause.term = '*' + clause.term.slice(1);
     }
 
     clause.wildcard = mode;