sync error-handling split logic with syntax regex and RBQL

mechatroner · Sep 8, 2018 · a294a47 · a294a47
1 parent 240f115
commit a294a47
Showing 1 changed file with 33 additions and 37 deletions.
diff --git a/rainbow_utils.js b/rainbow_utils.js
@@ -1,48 +1,44 @@
+function extract_next_field(src, dlm, preserve_quotes, cidx, result) {
+    var warning = false;
+    if (src.charAt(cidx) === '"') {
+        var uidx = src.indexOf('"', cidx + 1);
+        while (uidx != -1 && uidx + 1 < src.length && src.charAt(uidx + 1) == '"') {
+            uidx = src.indexOf('"', uidx + 2);
+        }
+        if (uidx != -1 && (uidx + 1 == src.length || src.charAt(uidx + 1) == dlm)) {
+            if (preserve_quotes) {
+                result.push(src.substring(cidx, uidx + 1));
+            } else {
+                result.push(src.substring(cidx + 1, uidx).replace(/""/g, '"'));
+            }
+            return [uidx + 2, false];
+        }
+        warning = true;
+    }
+    var uidx = src.indexOf(dlm, cidx);
+    if (uidx == -1)
+        uidx = src.length;
+    var field = src.substring(cidx, uidx);
+    warning = warning || field.indexOf('"') != -1;
+    result.push(field);
+    return [uidx + 1, warning];
+}
+
+
 function split_quoted_str(src, dlm, preserve_quotes=false) {
-    if (src.indexOf('"') == -1)
+    if (src.indexOf('"') == -1) // Optimization for most common case
         return [src.split(dlm), false];
     var result = [];
     var cidx = 0;
+    var warning = false;
     while (cidx < src.length) {
-        if (src.charAt(cidx) === '"') {
-            var uidx = cidx + 1;
-            while (true) {
-                uidx = src.indexOf('"', uidx);
-                if (uidx == -1) {
-                    result.push(src.substring(cidx));
-                    return [result, true];
-                } else if (uidx + 1 == src.length || src.charAt(uidx + 1) == dlm) {
-                    if (preserve_quotes) {
-                        result.push(src.substring(cidx, uidx + 1));
-                    } else {
-                        result.push(src.substring(cidx + 1, uidx).replace(/""/g, '"'));
-                    }
-                    cidx = uidx + 2;
-                    break;
-                } else if (src.charAt(uidx + 1) == '"') {
-                    uidx += 2; 
-                    continue;
-                } else {
-                    result.push(src.substring(cidx));
-                    return [result, true];
-                }
-            }
-        } else {
-            var uidx = src.indexOf(dlm, cidx);
-            if (uidx == -1)
-                uidx = src.length;
-            var field = src.substring(cidx, uidx);
-            if (field.indexOf('"') != -1) {
-                result.push(src.substring(cidx));
-                return [result, true];
-            }
-            result.push(field);
-            cidx = uidx + 1;
-        }
+        var extraction_report = extract_next_field(src, dlm, preserve_quotes, cidx, result);
+        cidx = extraction_report[0];
+        warning = warning || extraction_report[1];
     }
     if (src.charAt(src.length - 1) == dlm)
         result.push('');
-    return [result, false];
+    return [result, warning];
 }