csv-core: fix comment handling

This commit fixes a bug in how comments were interpreted. In particular,
this causes the behavior to match the documentation, which is correct.
Comments should only be allowed at the beginning of a record, but the
parser permitted them to also appear at the beginning of a field. This
corrects the implementation by simply moving the transition from
StartField to StartRecord, and then adds a couple regression tests.

Fixes #137

csv-core/src/reader.rs

@@ -990,6 +990,8 @@ impl Reader {
  StartRecord => {
  if self.term.equals(c) {
  (StartRecord, NfaInputAction::Discard)
+ } else if self.comment == Some(c) {
+ (InComment, NfaInputAction::Discard)
  } else {
  (StartField, NfaInputAction::Epsilon)
  }
@@ -1004,8 +1006,6 @@ impl Reader {
  (EndFieldDelim, NfaInputAction::Discard)
  } else if self.term.equals(c) {
  (EndFieldTerm, NfaInputAction::Epsilon)
- } else if self.comment == Some(c) {
- (InComment, NfaInputAction::Discard)
  } else {
  (InField, NfaInputAction::CopyToOutput)
  }
@@ -1623,6 +1623,10 @@ mod tests {
  "foo\n# hi\nbar\n",
  csv![["foo"], ["# hi"], ["bar"]],
  |b: &mut ReaderBuilder| { b.comment(Some(b'\n')); });
+ parses_to!(comment_4, "foo,b#ar,baz", csv![["foo", "b#ar", "baz"]],
+ |b: &mut ReaderBuilder| { b.comment(Some(b'#')); });
+ parses_to!(comment_5, "foo,#bar,baz", csv![["foo", "#bar", "baz"]],
+ |b: &mut ReaderBuilder| { b.comment(Some(b'#')); });
 
  macro_rules! assert_read {
  (