Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 7 additions & 3 deletions csv-core/src/reader.rs
Original file line number Diff line number Diff line change
Expand Up @@ -877,7 +877,7 @@ impl Reader {
if ends.is_empty() {
return (ReadRecordResult::OutputEndsFull, 0, 0, 0);
}
let (mut nin, mut nout, mut nend) = (0, self.output_pos, 0);
let (mut nin, mut nout, mut nend) = (0, 0, 0);
let mut state = self.nfa_state;
while nin < input.len() && nout < output.len() && nend < ends.len() {
let (s, io) = self.transition_nfa(state, input[nin]);
Expand All @@ -894,7 +894,7 @@ impl Reader {
}
state = s;
if state.is_field_final() {
ends[nend] = nout;
ends[nend] = self.output_pos + nout;
nend += 1;
if state != NfaState::EndFieldDelim {
break;
Expand All @@ -908,7 +908,11 @@ impl Reader {
nend >= ends.len(),
);
self.nfa_state = state;
self.output_pos = if res.is_record() { 0 } else { nout };
if res.is_record() {
self.output_pos = 0;
} else {
self.output_pos += nout;
}
(res, nin, nout, nend)
}

Expand Down
48 changes: 48 additions & 0 deletions csv-core/tests/repro_nfa_split.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,48 @@
use csv_core::{ReadRecordResult, ReaderBuilder};

fn read(nfa: bool, chunks: &[&[u8]]) -> Vec<String> {
let mut rdr = ReaderBuilder::new().nfa(nfa).build();
let mut out = [0u8; 64];
let mut ends = [0usize; 8];
let (mut outpos, mut endpos) = (0, 0);
let mut fields = vec![];
let mut chunks = chunks.iter().copied().chain(std::iter::once(&b""[..]));
let mut data = chunks.next().unwrap();
loop {
let (res, nin, nout, nend) =
rdr.read_record(data, &mut out[outpos..], &mut ends[endpos..]);
data = &data[nin..];
outpos += nout;
endpos += nend;
match res {
ReadRecordResult::InputEmpty => match chunks.next() {
Some(c) => data = c,
None => break,
},
ReadRecordResult::Record => {
let mut start = 0;
for &end in &ends[..endpos] {
fields.push(
String::from_utf8_lossy(&out[start..end]).into_owned(),
);
start = end;
}
outpos = 0;
endpos = 0;
}
ReadRecordResult::End => break,
r => panic!("{r:?}"),
}
}
fields
}

#[test]
fn repro_nfa_record_split_across_calls() {
let chunks: &[&[u8]] = &[b"ab,c", b"d,ef\n"];
let dfa = read(false, chunks);
let nfa = read(true, chunks);
println!("dfa={dfa:?}\nnfa={nfa:?}");
assert_eq!(dfa, ["ab", "cd", "ef"]);
assert_eq!(nfa, dfa);
}
Loading