Trim input, update TODO

The inputs are now trimmed before conversion.

The TODO items are updated for possible correct formats.

TODO.md

@@ -9,3 +9,16 @@
 - [X] Ordinals (e.g. `twentieth`, `first`, `two millionth`)
   - [ ] Implement other `fractionType`s, implement `lazy` fractions, e.g. `0.05` => `zero point zero five`,
     will implement `ratio` (`zero and five over one hundred`) and `part` (`zero and five hundredths`)
+- [ ] Special names? (e.g. `googol` for `1e+100`)
+
+Some inputs to consider:
+
+1.  `1/2` converts to `one half`. When only `lazy` is enabled, it converts to `one over two`.
+2.  `5/10` or `0.5` or `.5` (on en-US or en-PH) converts to `five tenths`. When `simplify` is `true`, it converts to
+    `one half`. When only `lazy` is enabled, it converts to `five over ten`.
+3.  `1e-2` converts to `one hundredth`. It follows by `one thousandth`, `one millionth`, `one billionth`, `one trillionth` etc.
+4.  `2/1e+2` converts to `two hundredths`. `2/2e+2` converts to `two two-hundredths`.
+5.  Should we enable `1e+(1e+100)` inputs? "googolplex"
+    1.  Should we enable `1e+1/10`?
+    2.  Should we enable `1e+1/1e+2`?
+    3.  Should we enable `1e+(1e+1)/10`?

packages/core/src/utils/numeric.ts

@@ -20,7 +20,7 @@ export const normalizeNumeric = (x: Numeric): string => {
       return x.toString(10);
     case 'string':
       // TODO assume not all strings follow a correct format
-      return x;
+      return x.trim();
       // return new BigNumber(x).toString(10)
     case 'object':
       return x.toString(10);