Fix bug in offsets-parser that led to a crash on stray ampersands in …

…the input
dbmdz · Jul 9, 2019 · daecb07 · daecb07
1 parent 16dfba2
commit daecb07
Showing 1 changed file with 4 additions and 1 deletion.
diff --git a/offsets-parser/src/main.rs b/offsets-parser/src/main.rs
@@ -77,7 +77,10 @@ fn main() -> Result<()> {
     };
     let word_re = get_regex(&ocr_text);
     for cap in word_re.captures_iter(&ocr_text) {
-        let text = decode_html_entities(&cap["text"]).unwrap();
+        let text = match decode_html_entities(&cap["text"]) {
+            Ok(t) => t,
+            Err(_) => (&cap["text"]).to_string()
+        };
         match cap.get(0) {
             Some(mat) => write!(out_writer, "{}{}{} ", text, delimiter, mat.start())?,
             None => ()