1use crate::vocab::OpenMode;
9use std::cmp::Ordering;
10use std::collections::{BTreeMap, BTreeSet, HashMap};
11use std::fs::{File, OpenOptions};
12use std::io::{self, BufRead, BufReader, BufWriter, Read, Write};
13use std::ops::Bound;
14use std::path::PathBuf;
15use zarch::ebcdic::{self, CodePage};
16
17pub use crate::abend::FileStatus;
18
19#[derive(Clone, Copy, Debug, PartialEq, Eq)]
20pub enum Format {
21 Fixed,
22 Variable,
23 Text,
24}
25
26impl Format {
27 pub fn from_keyword(word: &str) -> Option<Format> {
28 match word.to_ascii_lowercase().as_str() {
29 "text" => Some(Format::Text),
30 "fixed" | "f" => Some(Format::Fixed),
31 "variable" | "v" => Some(Format::Variable),
32 _ => None,
33 }
34 }
35}
36
37#[derive(Clone, Debug, PartialEq, Eq)]
38pub struct Dd {
39 pub path: PathBuf,
40 pub format: Option<Format>,
41 pub append: bool,
43}
44
45#[derive(Clone, Debug, Default)]
47pub struct Dds {
48 given: HashMap<String, Dd>,
49 environment: bool,
50}
51
52fn dd_value(value: &str) -> Result<Dd, String> {
53 let (rest, append) = match value.rsplit_once(':') {
54 Some((rest, word)) if word.eq_ignore_ascii_case("mod") => (rest, true),
55 _ => (value, false),
56 };
57 let (path, format) = rest.rsplit_once(':').and_then(|(p, word)| Some((p, Some(Format::from_keyword(word)?)))).unwrap_or((rest, None));
58 if path.is_empty() {
59 return Err(format!("DD {value}: no path"));
60 }
61 Ok(Dd { path: PathBuf::from(path), format, append })
62}
63
64impl Dds {
65 pub fn new(specs: &[String], environment: bool) -> Result<Self, String> {
67 let mut given = HashMap::new();
68 for spec in specs {
69 let (name, value) = spec.split_once('=').ok_or_else(|| format!("--dd {spec}: expected NAME=path"))?;
70 given.insert(name.to_ascii_uppercase(), dd_value(value)?);
71 }
72 Ok(Self { given, environment })
73 }
74
75 pub fn get(&self, name: &str) -> Option<Dd> {
76 if let Some(dd) = self.given.get(name) {
77 return Some(dd.clone());
78 }
79 if !self.environment {
80 return None;
81 }
82 std::env::var(format!("DD_{name}")).ok().and_then(|v| dd_value(&v).ok())
83 }
84}
85
86enum Handle {
87 Reader(BufReader<File>),
88 Writer(BufWriter<File>),
89 Keyed(Box<Keyed>),
90 Empty,
91}
92
93#[derive(Clone, Copy, Debug, PartialEq, Eq)]
95pub struct KeySpan {
96 pub offset: usize,
97 pub len: usize,
98}
99
100impl KeySpan {
101 pub fn of(&self, record: &[u8]) -> Vec<u8> {
102 let mut key: Vec<u8> = record.iter().skip(self.offset).take(self.len).copied().collect();
103 key.resize(self.len, ebcdic::SPACE);
104 key
105 }
106}
107
108#[derive(Clone, Debug, PartialEq, Eq)]
111pub enum Keying {
112 Position,
114 Relative,
117 Indexed { prime: KeySpan, alternates: Vec<(KeySpan, bool)> },
119}
120
121pub const MAX_RELATIVE: u64 = 16_777_215;
125
126pub fn record_number(n: u64) -> Vec<u8> {
127 n.to_be_bytes().to_vec()
128}
129
130pub fn number_of(key: &[u8]) -> u64 {
131 u64::from_be_bytes(key.try_into().unwrap_or([0; 8]))
132}
133
134type Entry = (Vec<u8>, u64, Vec<u8>);
138
139struct Alternate {
140 span: KeySpan,
141 duplicates: bool,
142 index: BTreeSet<Entry>,
143 arrival: HashMap<Vec<u8>, u64>,
144}
145
146impl Alternate {
147 fn holders<'a>(&'a self, value: &'a [u8]) -> impl Iterator<Item = &'a Entry> {
148 self.index.range((value.to_vec(), 0, Vec::new())..).take_while(move |(v, ..)| v == value)
149 }
150}
151
152#[derive(Clone, Debug)]
154enum Cursor {
155 First,
157 At { which: usize, at: Entry, inclusive: bool },
160 Undefined,
162}
163
164pub struct Found {
166 pub key: Vec<u8>,
167 pub record: Vec<u8>,
168 pub duplicate: bool,
170}
171
172pub struct Keyed {
176 pub keying: Keying,
177 records: BTreeMap<Vec<u8>, Vec<u8>>,
178 alternates: Vec<Alternate>,
179 arrivals: u64,
180 cursor: Cursor,
181 pub last_read: Option<Vec<u8>>,
184 dirty: bool,
185 path: Option<PathBuf>,
186 record_len: usize,
187 page: &'static CodePage,
188}
189
190impl Keyed {
191 fn new(keying: Keying, path: Option<PathBuf>, record_len: usize, page: &'static CodePage) -> Self {
192 let alternates = match &keying {
193 Keying::Indexed { alternates, .. } => {
194 alternates.iter().map(|&(span, duplicates)| Alternate { span, duplicates, index: BTreeSet::new(), arrival: HashMap::new() }).collect()
195 }
196 _ => Vec::new(),
197 };
198 Self { keying, records: BTreeMap::new(), alternates, arrivals: 0, cursor: Cursor::First, last_read: None, dirty: true, path, record_len, page }
199 }
200
201 pub fn prime_key(&self, record: &[u8]) -> Option<Vec<u8>> {
202 match &self.keying {
203 Keying::Indexed { prime, .. } => Some(prime.of(record)),
204 _ => None,
205 }
206 }
207
208 pub fn highest_key(&self) -> Option<&Vec<u8>> {
209 self.records.keys().next_back()
210 }
211
212 pub fn record(&self, key: &[u8]) -> Option<&Vec<u8>> {
213 self.records.get(key)
214 }
215
216 fn key_len(&self, which: usize) -> usize {
217 match (&self.keying, which) {
218 (Keying::Indexed { prime, .. }, 0) => prime.len,
219 (Keying::Indexed { .. }, n) => self.alternates[n - 1].span.len,
220 _ => 8,
221 }
222 }
223
224 pub fn get(&self, which: usize, value: &[u8]) -> Option<Found> {
226 if which == 0 {
227 return self.records.get(value).map(|r| Found { key: value.to_vec(), record: r.clone(), duplicate: false });
228 }
229 let mut hits = self.alternates[which - 1].holders(value);
230 let key = hits.next()?.2.clone();
231 Some(Found { record: self.records[&key].clone(), key, duplicate: hits.next().is_some() })
232 }
233
234 fn put(&mut self, key: Vec<u8>, record: Vec<u8>) -> Result<bool, FileStatus> {
238 let mut shared = false;
239 for alt in &self.alternates {
240 let value = alt.span.of(&record);
241 let others = alt.holders(&value).any(|(_, _, k)| *k != key);
242 if others && !alt.duplicates {
243 return Err(FileStatus::DuplicateKey);
244 }
245 shared |= others;
246 }
247 let old = self.records.get(&key).cloned();
248 let mut arrivals = self.arrivals;
249 for alt in &mut self.alternates {
250 let now = alt.span.of(&record);
251 if let Some(old) = &old {
252 let was = alt.span.of(old);
253 if was == now {
254 continue;
255 }
256 let seq = alt.arrival[&key];
257 alt.index.remove(&(was, seq, key.clone()));
258 }
259 alt.index.insert((now, arrivals, key.clone()));
260 alt.arrival.insert(key.clone(), arrivals);
261 arrivals += 1;
262 }
263 self.arrivals = arrivals;
264 self.records.insert(key, record);
265 self.dirty = true;
266 Ok(shared)
267 }
268
269 pub fn insert(&mut self, key: Vec<u8>, record: Vec<u8>) -> Result<bool, FileStatus> {
272 if self.records.contains_key(&key) {
273 return Err(FileStatus::DuplicateKey);
274 }
275 self.put(key, record)
276 }
277
278 pub fn replace(&mut self, key: Vec<u8>, record: Vec<u8>) -> Result<bool, FileStatus> {
280 if !self.records.contains_key(&key) {
281 return Err(FileStatus::NotFound);
282 }
283 self.put(key, record)
284 }
285
286 pub fn remove(&mut self, key: &[u8]) -> Option<Vec<u8>> {
287 let record = self.records.remove(key)?;
288 for alt in &mut self.alternates {
289 if let Some(seq) = alt.arrival.remove(key) {
290 alt.index.remove(&(alt.span.of(&record), seq, key.to_vec()));
291 }
292 }
293 self.dirty = true;
294 Some(record)
295 }
296
297 pub fn start(&mut self, which: usize, wanted: Ordering, or_equal: bool, value: &[u8]) -> bool {
301 let value = &value[..value.len().min(self.key_len(which))];
302 let fits = |k: &[u8]| {
303 let o = k[..value.len().min(k.len())].cmp(value);
304 o == wanted || (or_equal && o == Ordering::Equal)
305 };
306 let from = (value.to_vec(), 0, Vec::new());
307 let found = match which {
308 0 => self.records.range(from.0..).map(|(k, _)| k).find(|k| fits(k)).map(|k| (k.clone(), 0, Vec::new())),
309 n => self.alternates[n - 1].index.range(from..).find(|(k, ..)| fits(k)).cloned(),
310 };
311 self.last_read = None;
312 match found {
313 Some(at) => {
314 self.cursor = Cursor::At { which, at, inclusive: true };
315 true
316 }
317 None => {
318 self.lose_position();
319 false
320 }
321 }
322 }
323
324 pub fn lose_position(&mut self) {
326 self.cursor = Cursor::Undefined;
327 self.last_read = None;
328 }
329
330 pub fn read_at(&mut self, which: usize, key: &[u8]) {
332 let at = match which {
333 0 => (key.to_vec(), 0, Vec::new()),
334 n => {
335 let alt = &self.alternates[n - 1];
336 (alt.span.of(&self.records[key]), alt.arrival[key], key.to_vec())
337 }
338 };
339 self.cursor = Cursor::At { which, at, inclusive: false };
340 self.last_read = Some(key.to_vec());
341 }
342
343 pub fn seek(&self, from: &[u8], inclusive: bool, backward: bool) -> Option<(Vec<u8>, Vec<u8>)> {
346 let edge = if inclusive { Bound::Included(from.to_vec()) } else { Bound::Excluded(from.to_vec()) };
347 let mut hits = if backward { self.records.range((Bound::Unbounded, edge)) } else { self.records.range((edge, Bound::Unbounded)) };
348 let hit = if backward { hits.next_back() } else { hits.next() };
349 hit.map(|(k, r)| (k.clone(), r.clone()))
350 }
351
352 pub fn step(&mut self, backward: bool) -> Result<Option<Found>, &'static str> {
354 self.read_next(backward).map_err(FileStatus::as_str)
355 }
356
357 pub fn read_next(&mut self, backward: bool) -> Result<Option<Found>, FileStatus> {
359 let (which, bound) = match &self.cursor {
360 Cursor::Undefined => return Err(FileStatus::NoNextRecord),
361 Cursor::First => (0, Bound::Unbounded),
362 Cursor::At { which, at, inclusive: true } => (*which, Bound::Included(at.clone())),
363 Cursor::At { which, at, inclusive: false } => (*which, Bound::Excluded(at.clone())),
364 };
365 let range = if backward { (Bound::Unbounded, bound) } else { (bound, Bound::Unbounded) };
366 let hit = if which == 0 {
367 let mut keys = self.records.range((range.0.map(|(k, ..)| k), range.1.map(|(k, ..)| k))).map(|(k, _)| k);
368 let key = if backward { keys.next_back() } else { keys.next() };
369 key.map(|k| (k.clone(), false))
370 } else {
371 let mut entries = self.alternates[which - 1].index.range(range);
372 let entry = if backward { entries.next_back() } else { entries.next() };
373 entry.map(|(value, _, key)| {
374 let after = if backward { entries.next_back() } else { entries.next() };
375 (key.clone(), after.is_some_and(|(v, ..)| v == value))
376 })
377 };
378 let Some((key, duplicate)) = hit else {
379 self.lose_position();
380 return Ok(None);
381 };
382 self.read_at(which, &key);
383 Ok(Some(Found { record: self.records[&key].clone(), key, duplicate }))
384 }
385
386 fn inbound(&self, raw: Vec<u8>, format: Format) -> Vec<u8> {
387 let mut record = match format {
388 Format::Text => self.page.encode_lossy(&String::from_utf8_lossy(&raw)),
389 _ => raw,
390 };
391 if format != Format::Variable {
392 record.resize(self.record_len.max(record.len()), ebcdic::SPACE);
393 }
394 record
395 }
396
397 fn outbound(&self, record: &[u8], format: Format) -> Vec<u8> {
398 match format {
399 Format::Text => format!("{}\n", self.page.decode(record).trim_end()).into_bytes(),
400 Format::Fixed => {
401 let mut r = record.to_vec();
402 r.resize(self.record_len, ebcdic::SPACE);
403 r
404 }
405 Format::Variable => record.to_vec(),
406 }
407 }
408
409 fn save(&self, format: Format) -> io::Result<()> {
410 let Some(path) = self.path.as_ref().filter(|_| self.dirty) else { return Ok(()) };
411 let mut out = Open { mode: OpenMode::Output, format, handle: Handle::Writer(BufWriter::new(File::create(path)?)), head: Head::Start, page: None };
412 let empty = match format {
413 Format::Fixed => vec![0; self.record_len],
414 Format::Variable => Vec::new(),
415 Format::Text => b"\n".to_vec(),
416 };
417 let mut next = 1;
418 for (key, record) in &self.records {
419 if self.keying == Keying::Relative {
420 for _ in next..number_of(key) {
421 out.write(&empty)?;
422 }
423 next = number_of(key) + 1;
424 }
425 out.write(&self.outbound(record, format))?;
426 }
427 out.close()
428 }
429}
430
431pub struct Open {
432 pub mode: OpenMode,
433 pub format: Format,
434 handle: Handle,
435 head: Head,
436 pub page: Option<crate::linage::Page>,
438}
439
440#[derive(Clone, Copy, Debug, PartialEq, Eq)]
442pub enum Move {
443 Lines(u64),
444 Page,
445}
446
447#[derive(Clone, Copy, Debug, PartialEq, Eq)]
450enum Head {
451 Start,
452 OnLine(bool),
453 Fresh,
454}
455
456pub enum Record {
458 Data(Vec<u8>),
459 WrongLength(Vec<u8>),
461 End,
462}
463
464pub fn open_keyed(dd: Option<&Dd>, mode: OpenMode, format: Format, keying: Keying, record_len: usize, page: &'static CodePage) -> io::Result<Open> {
467 let path = dd.filter(|_| mode != OpenMode::Input).map(|d| d.path.clone());
468 let mut keyed = Keyed::new(keying, path, record_len, page);
469 if let Some(dd) = dd.filter(|d| mode != OpenMode::Output && d.path.exists()) {
470 let mut reader = open(dd, OpenMode::Input, format)?;
471 let mut number = 0u64;
472 loop {
473 let raw = match reader.read(record_len)? {
474 Record::End => break,
475 Record::Data(r) | Record::WrongLength(r) => r,
476 };
477 number += 1;
478 if keyed.keying == Keying::Relative && raw.iter().all(|&b| b == 0) {
479 continue;
480 }
481 let record = keyed.inbound(raw, format);
482 let key = keyed.prime_key(&record).unwrap_or_else(|| record_number(number));
483 keyed.insert(key, record).map_err(|_| io::Error::new(io::ErrorKind::InvalidData, format!("{}: two records with one key", dd.path.display())))?;
484 }
485 keyed.dirty = false;
486 }
487 Ok(Open { mode, format, handle: Handle::Keyed(Box::new(keyed)), head: Head::Start, page: None })
488}
489
490pub fn open(dd: &Dd, mode: OpenMode, format: Format) -> io::Result<Open> {
491 let handle = match mode {
492 OpenMode::Input => Handle::Reader(BufReader::new(File::open(&dd.path)?)),
493 OpenMode::Output if !dd.append => Handle::Writer(BufWriter::new(File::create(&dd.path)?)),
494 OpenMode::Output | OpenMode::Extend => Handle::Writer(BufWriter::new(OpenOptions::new().append(true).create(true).open(&dd.path)?)),
495 OpenMode::InputOutput => return Err(io::Error::new(io::ErrorKind::Unsupported, "OPEN I-O of a line-sequential file")),
496 };
497 Ok(Open { mode, format, handle, head: Head::Start, page: None })
498}
499
500pub fn absent() -> Open {
502 Open { mode: OpenMode::Input, format: Format::Fixed, handle: Handle::Empty, head: Head::Start, page: None }
503}
504
505impl Open {
506 pub fn keyed(&mut self) -> Option<&mut Keyed> {
507 match &mut self.handle {
508 Handle::Keyed(k) => Some(k),
509 _ => None,
510 }
511 }
512
513 pub fn is_keyed(&self) -> bool {
514 matches!(self.handle, Handle::Keyed(_))
515 }
516
517 pub fn read(&mut self, fixed_len: usize) -> io::Result<Record> {
518 let reader = match &mut self.handle {
519 Handle::Reader(r) => r,
520 Handle::Empty => return Ok(Record::End),
521 Handle::Writer(_) | Handle::Keyed(_) => return Err(io::Error::other("a sequential READ of a file not open for it")),
522 };
523 match self.format {
524 Format::Fixed => {
525 let mut buf = Vec::with_capacity(fixed_len);
526 reader.by_ref().take(fixed_len as u64).read_to_end(&mut buf)?;
527 Ok(match buf.len() {
528 0 => Record::End,
529 n if n == fixed_len => Record::Data(buf),
530 _ => Record::WrongLength(buf),
531 })
532 }
533 Format::Variable => {
534 let mut rdw = [0u8; 4];
535 match reader.read_exact(&mut rdw) {
536 Err(e) if e.kind() == io::ErrorKind::UnexpectedEof => return Ok(Record::End),
537 other => other?,
538 }
539 let len = u16::from_be_bytes([rdw[0], rdw[1]]) as usize;
540 if len < 4 {
541 return Err(io::Error::new(io::ErrorKind::InvalidData, format!("an RDW of length {len}")));
542 }
543 let mut buf = vec![0u8; len - 4];
544 reader.read_exact(&mut buf)?;
545 Ok(Record::Data(buf))
546 }
547 Format::Text => {
548 let mut line = Vec::new();
549 if reader.read_until(b'\n', &mut line)? == 0 {
550 return Ok(Record::End);
551 }
552 while matches!(line.last(), Some(b'\n' | b'\r')) {
553 line.pop();
554 }
555 Ok(Record::Data(line))
556 }
557 }
558 }
559
560 pub fn write(&mut self, bytes: &[u8]) -> io::Result<()> {
561 let Handle::Writer(w) = &mut self.handle else {
562 return Err(io::Error::other("WRITE to a file not opened for output"));
563 };
564 match self.format {
565 Format::Fixed | Format::Text => w.write_all(bytes),
566 Format::Variable => {
567 let len = u16::try_from(bytes.len() + 4).map_err(|_| io::Error::new(io::ErrorKind::InvalidInput, "a record over 32,760 bytes"))?;
568 w.write_all(&len.to_be_bytes())?;
569 w.write_all(&[0, 0])?;
570 w.write_all(bytes)
571 }
572 }
573 }
574
575 pub fn print(&mut self, before: Option<Move>, line: &str, after: Option<Move>) -> io::Result<()> {
580 if let Some(m) = before {
581 self.feed(m)?;
582 }
583 let Handle::Writer(w) = &mut self.handle else {
584 return Err(io::Error::other("WRITE to a file not opened for output"));
585 };
586 let shown = matches!(self.head, Head::OnLine(true));
587 if shown && !line.is_empty() {
588 w.write_all(b"\r")?;
589 }
590 w.write_all(line.as_bytes())?;
591 self.head = Head::OnLine(shown || !line.is_empty());
592 match after {
593 Some(m) => self.feed(m),
594 None => Ok(()),
595 }
596 }
597
598 fn feed(&mut self, m: Move) -> io::Result<()> {
599 let Handle::Writer(w) = &mut self.handle else {
600 return Err(io::Error::other("WRITE to a file not opened for output"));
601 };
602 match (m, self.head) {
603 (Move::Lines(0), _) => return Ok(()),
604 (Move::Lines(n), head) => {
605 for _ in 0..n - u64::from(head == Head::Start) {
606 w.write_all(b"\n")?;
607 }
608 }
609 (Move::Page, Head::OnLine(_)) => w.write_all(b"\n\x0c")?,
610 (Move::Page, _) => w.write_all(b"\x0c")?,
611 }
612 self.head = Head::Fresh;
613 Ok(())
614 }
615
616 pub fn close(self) -> io::Result<()> {
617 match self.handle {
618 Handle::Writer(mut w) => {
619 if matches!(self.head, Head::OnLine(_)) {
620 w.write_all(b"\n")?;
621 }
622 w.flush()
623 }
624 Handle::Keyed(k) => k.save(self.format),
625 _ => Ok(()),
626 }
627 }
628}
629
630#[cfg(test)]
631mod tests {
632 use super::*;
633
634 #[test]
635 fn dd_specs() {
636 let dds = Dds::new(&["IN=/tmp/a.txt:text".into(), "out=/tmp/b.dat".into()], false).unwrap();
637 assert_eq!(dds.get("IN"), Some(Dd { path: "/tmp/a.txt".into(), format: Some(Format::Text), append: false }));
638 assert_eq!(dds.get("OUT").unwrap().format, None);
639 assert!(dds.get("NONE").is_none());
640 assert!(Dds::new(&["NOEQUALS".into()], false).is_err());
641 }
642
643 #[test]
644 fn mod_suffix_follows_the_format_or_stands_alone() {
645 let dd = |value: &str| dd_value(value).unwrap();
646 assert_eq!(dd("a.dat:mod"), Dd { path: "a.dat".into(), format: None, append: true });
647 assert_eq!(dd("a.dat:text:mod"), Dd { path: "a.dat".into(), format: Some(Format::Text), append: true });
648 assert_eq!(dd("A.DAT:TEXT:MOD"), Dd { path: "A.DAT".into(), format: Some(Format::Text), append: true });
649 assert_eq!(dd("a.dat:text"), Dd { path: "a.dat".into(), format: Some(Format::Text), append: false });
650 assert_eq!(dd("c:\\a.dat"), Dd { path: "c:\\a.dat".into(), format: None, append: false });
651 assert!(dd_value(":mod").is_err());
652 }
653
654 #[test]
655 fn open_output_on_a_mod_dd_writes_after_the_records_there() {
656 let path = std::env::temp_dir().join(format!("ironwork-mod-{}", std::process::id()));
657 let write = |append: bool, records: &[&[u8]]| {
658 let mut out = open(&Dd { path: path.clone(), format: None, append }, OpenMode::Output, Format::Fixed).unwrap();
659 for record in records {
660 out.write(record).unwrap();
661 }
662 out.close().unwrap();
663 };
664 write(false, &[b"AA", b"BB"]);
665 write(true, &[b"CC"]);
666 assert_eq!(std::fs::read(&path).unwrap(), b"AABBCC");
667 write(false, &[b"DD"]);
668 assert_eq!(std::fs::read(&path).unwrap(), b"DD");
669 std::fs::remove_file(&path).unwrap();
670 }
671
672 #[test]
673 fn variable_records_round_trip_through_rdws() {
674 let path = std::env::temp_dir().join(format!("ironwork-vb-{}", std::process::id()));
675 let dd = Dd { path: path.clone(), format: None, append: false };
676 let mut out = open(&dd, OpenMode::Output, Format::Variable).unwrap();
677 out.write(b"AB").unwrap();
678 out.write(b"CDEF").unwrap();
679 out.close().unwrap();
680 assert_eq!(std::fs::read(&path).unwrap(), [0, 6, 0, 0, b'A', b'B', 0, 8, 0, 0, b'C', b'D', b'E', b'F']);
681 let mut input = open(&dd, OpenMode::Input, Format::Variable).unwrap();
682 assert!(matches!(input.read(0).unwrap(), Record::Data(d) if d == b"AB"));
683 assert!(matches!(input.read(0).unwrap(), Record::Data(d) if d == b"CDEF"));
684 assert!(matches!(input.read(0).unwrap(), Record::End));
685 }
686}