From b3a1dad64ae2f0d6310fffd6ff325c057d681510 Mon Sep 17 00:00:00 2001
From: Raphaela Heil <RaphaelaHeil@users.noreply.github.com>
Date: Mon, 25 Nov 2024 15:18:40 +0100
Subject: [PATCH] [fix] word-level search for line-level ALTO annotations

---
 src/builder/Digitized.ts |  8 +++++++-
 src/search/search.ts     | 15 +++++++++++++--
 2 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/src/builder/Digitized.ts b/src/builder/Digitized.ts
index 68f0fa8..bd3929f 100644
--- a/src/builder/Digitized.ts
+++ b/src/builder/Digitized.ts
@@ -115,7 +115,13 @@ export async function getAnnotationPage(item: RootItem, text: Text): Promise<Ann
                 const resource = Resource.createTextResource(word.content, text.language);
                 const annotation = new Annotation(annoUri(item.id, childItem.id, word.idx), resource, 'supplementing');
 
-                annotation.setTextGranularity('word');
+                if (word.content.split(/\s+/).filter(w => w.length > 0).length > 1){
+
+                    annotation.setTextGranularity('line');
+                }else{
+                    annotation.setTextGranularity('word');
+                }
+
                 annotation.setCanvas(canvas, {x: word.x, y: word.y, w: word.width, h: word.height});
 
                 annotations.push(annotation);
diff --git a/src/search/search.ts b/src/search/search.ts
index 846f7f0..ee00862 100644
--- a/src/search/search.ts
+++ b/src/search/search.ts
@@ -138,10 +138,21 @@ function mapMatches(text: Text, hl: string, matchExact: string | null): SearchRe
         .split(/[\t\r\n\s]+/)
         .filter(token => token.length > 0);
 
+    let id = 0;
+    const extractedWords: TextWord[] = [];
+    for (const word of words) {
+        const splitAttempt = word.content.split(/\s+/).filter(w => w.length > 0);
+        if (splitAttempt.length > 1) {
+            extractedWords.push(...splitAttempt.map(w => <TextWord>{ ...word, idx: id++, content: w, isHyphenated: false }));
+        } else {
+            extractedWords.push(<TextWord>{ ...word, idx: id++ });
+        }
+    }
+
     let tokenIdx = 0, wordIdx = 0, curMatch: SearchResultMatch | null = null;
     while (tokenIdx < tokens.length) {
         const token = tokens[tokenIdx];
-        const word = wordIdx < words.length ? words[wordIdx] : null;
+        const word = wordIdx < extractedWords.length ? extractedWords[wordIdx] : null;
 
         if (token.startsWith(PRE_TAG)) {
             if (curMatch == null)
@@ -159,7 +170,7 @@ function mapMatches(text: Text, hl: string, matchExact: string | null): SearchRe
 
         if (word?.isHyphenated) {
             wordIdx++;
-            curMatch?.words.push(words[wordIdx]);
+            curMatch?.words.push(extractedWords[wordIdx]);
         }
 
         if (curMatch && (!matchExact || matchExact === curMatch.match)) {