Fixed bugs with Abbyy imports; added new test

scribeocr · Dec 19, 2024 · d4010d3 · d4010d3
1 parent fb26b14
commit d4010d3
Show file tree

Hide file tree

Showing 4 changed files with 384 additions and 3 deletions.
diff --git a/js/import/convertPageAbbyy.js b/js/import/convertPageAbbyy.js
@@ -337,7 +337,9 @@ export async function convertPageAbbyy({ ocrStr, n }) {
 
         lineObj.orientation = textOrientationFinal;
 
-        lineObj.raw = xmlText;
+        // This should be kept disabled as a rule unless debugging.
+        // Enabling has caused crashes with large documents.
+        if (debugMode) lineObj.raw = xmlText;
 
         let lettersKept = 0;
         for (let i = 0; i < text.length; i++) {

diff --git a/js/recognizeConvert.js b/js/recognizeConvert.js
@@ -322,8 +322,8 @@ export async function convertPageCallback({
   pageObj, dataTables, warn, langSet,
 }, n, mainData, engineName) {
   const fontPromiseArr = [];
-  if (langSet.has('chi_sim')) fontPromiseArr.push(loadChiSimFont());
-  if (langSet.has('rus') || langSet.has('ukr') || langSet.has('ell')) {
+  if (langSet && langSet.has('chi_sim')) fontPromiseArr.push(loadChiSimFont());
+  if (langSet && (langSet.has('rus') || langSet.has('ukr') || langSet.has('ell'))) {
     fontPromiseArr.push(loadBuiltInFontsRaw('all'));
   } else {
     fontPromiseArr.push(loadBuiltInFontsRaw());