1use std::cmp::Ordering;
8use std::collections::{BTreeMap, BTreeSet, HashMap};
9use std::fs::{File, OpenOptions};
10use std::io::{self, BufRead, BufReader, BufWriter, Read, Write};
11use std::ops::Bound;
12use std::path::PathBuf;
13use syntax::ast::OpenMode;
14use zarch::ebcdic::{self, CodePage};
15
16#[derive(Clone, Copy, Debug, PartialEq, Eq)]
17pub enum Format {
18 Fixed,
19 Variable,
20 Text,
21}
22
23#[derive(Clone, Debug, PartialEq, Eq)]
24pub struct Dd {
25 pub path: PathBuf,
26 pub format: Option<Format>,
27}
28
29#[derive(Clone, Debug, Default)]
31pub struct Dds {
32 given: HashMap<String, Dd>,
33 environment: bool,
34}
35
36fn dd_value(value: &str) -> Result<Dd, String> {
37 let (path, format) = match value.rsplit_once(':') {
38 Some((p, "text")) => (p, Some(Format::Text)),
39 Some((p, "fixed" | "f" | "F")) => (p, Some(Format::Fixed)),
40 Some((p, "variable" | "v" | "V")) => (p, Some(Format::Variable)),
41 _ => (value, None),
42 };
43 if path.is_empty() {
44 return Err(format!("DD {value}: no path"));
45 }
46 Ok(Dd { path: PathBuf::from(path), format })
47}
48
49impl Dds {
50 pub fn new(specs: &[String], environment: bool) -> Result<Self, String> {
52 let mut given = HashMap::new();
53 for spec in specs {
54 let (name, value) = spec.split_once('=').ok_or_else(|| format!("--dd {spec}: expected NAME=path"))?;
55 given.insert(name.to_ascii_uppercase(), dd_value(value)?);
56 }
57 Ok(Self { given, environment })
58 }
59
60 pub fn get(&self, name: &str) -> Option<Dd> {
61 if let Some(dd) = self.given.get(name) {
62 return Some(dd.clone());
63 }
64 if !self.environment {
65 return None;
66 }
67 std::env::var(format!("DD_{name}")).ok().and_then(|v| dd_value(&v).ok())
68 }
69}
70
71enum Handle {
72 Reader(BufReader<File>),
73 Writer(BufWriter<File>),
74 Keyed(Box<Keyed>),
75 Empty,
76}
77
78#[derive(Clone, Copy, Debug, PartialEq, Eq)]
80pub struct KeySpan {
81 pub offset: usize,
82 pub len: usize,
83}
84
85impl KeySpan {
86 pub fn of(&self, record: &[u8]) -> Vec<u8> {
87 let mut key: Vec<u8> = record.iter().skip(self.offset).take(self.len).copied().collect();
88 key.resize(self.len, ebcdic::SPACE);
89 key
90 }
91}
92
93#[derive(Clone, Debug, PartialEq, Eq)]
96pub enum Keying {
97 Position,
99 Relative,
102 Indexed { prime: KeySpan, alternates: Vec<(KeySpan, bool)> },
104}
105
106pub const MAX_RELATIVE: u64 = 16_777_215;
110
111pub fn record_number(n: u64) -> Vec<u8> {
112 n.to_be_bytes().to_vec()
113}
114
115pub fn number_of(key: &[u8]) -> u64 {
116 u64::from_be_bytes(key.try_into().unwrap_or([0; 8]))
117}
118
119type Entry = (Vec<u8>, u64, Vec<u8>);
123
124struct Alternate {
125 span: KeySpan,
126 duplicates: bool,
127 index: BTreeSet<Entry>,
128 arrival: HashMap<Vec<u8>, u64>,
129}
130
131impl Alternate {
132 fn holders<'a>(&'a self, value: &'a [u8]) -> impl Iterator<Item = &'a Entry> {
133 self.index.range((value.to_vec(), 0, Vec::new())..).take_while(move |(v, ..)| v == value)
134 }
135}
136
137#[derive(Clone, Debug)]
139enum Cursor {
140 First,
142 At { which: usize, at: Entry, inclusive: bool },
145 Undefined,
147}
148
149pub struct Found {
151 pub key: Vec<u8>,
152 pub record: Vec<u8>,
153 pub duplicate: bool,
155}
156
157pub struct Keyed {
161 pub keying: Keying,
162 records: BTreeMap<Vec<u8>, Vec<u8>>,
163 alternates: Vec<Alternate>,
164 arrivals: u64,
165 cursor: Cursor,
166 pub last_read: Option<Vec<u8>>,
169 dirty: bool,
170 path: Option<PathBuf>,
171 record_len: usize,
172 page: &'static CodePage,
173}
174
175impl Keyed {
176 fn new(keying: Keying, path: Option<PathBuf>, record_len: usize, page: &'static CodePage) -> Self {
177 let alternates = match &keying {
178 Keying::Indexed { alternates, .. } => {
179 alternates.iter().map(|&(span, duplicates)| Alternate { span, duplicates, index: BTreeSet::new(), arrival: HashMap::new() }).collect()
180 }
181 _ => Vec::new(),
182 };
183 Self { keying, records: BTreeMap::new(), alternates, arrivals: 0, cursor: Cursor::First, last_read: None, dirty: true, path, record_len, page }
184 }
185
186 pub fn prime_key(&self, record: &[u8]) -> Option<Vec<u8>> {
187 match &self.keying {
188 Keying::Indexed { prime, .. } => Some(prime.of(record)),
189 _ => None,
190 }
191 }
192
193 pub fn highest_key(&self) -> Option<&Vec<u8>> {
194 self.records.keys().next_back()
195 }
196
197 pub fn record(&self, key: &[u8]) -> Option<&Vec<u8>> {
198 self.records.get(key)
199 }
200
201 fn key_len(&self, which: usize) -> usize {
202 match (&self.keying, which) {
203 (Keying::Indexed { prime, .. }, 0) => prime.len,
204 (Keying::Indexed { .. }, n) => self.alternates[n - 1].span.len,
205 _ => 8,
206 }
207 }
208
209 pub fn get(&self, which: usize, value: &[u8]) -> Option<Found> {
211 if which == 0 {
212 return self.records.get(value).map(|r| Found { key: value.to_vec(), record: r.clone(), duplicate: false });
213 }
214 let mut hits = self.alternates[which - 1].holders(value);
215 let key = hits.next()?.2.clone();
216 Some(Found { record: self.records[&key].clone(), key, duplicate: hits.next().is_some() })
217 }
218
219 fn put(&mut self, key: Vec<u8>, record: Vec<u8>) -> Result<bool, &'static str> {
223 let mut shared = false;
224 for alt in &self.alternates {
225 let value = alt.span.of(&record);
226 let others = alt.holders(&value).any(|(_, _, k)| *k != key);
227 if others && !alt.duplicates {
228 return Err("22");
229 }
230 shared |= others;
231 }
232 let old = self.records.get(&key).cloned();
233 let mut arrivals = self.arrivals;
234 for alt in &mut self.alternates {
235 let now = alt.span.of(&record);
236 if let Some(old) = &old {
237 let was = alt.span.of(old);
238 if was == now {
239 continue;
240 }
241 let seq = alt.arrival[&key];
242 alt.index.remove(&(was, seq, key.clone()));
243 }
244 alt.index.insert((now, arrivals, key.clone()));
245 alt.arrival.insert(key.clone(), arrivals);
246 arrivals += 1;
247 }
248 self.arrivals = arrivals;
249 self.records.insert(key, record);
250 self.dirty = true;
251 Ok(shared)
252 }
253
254 pub fn insert(&mut self, key: Vec<u8>, record: Vec<u8>) -> Result<bool, &'static str> {
257 if self.records.contains_key(&key) {
258 return Err("22");
259 }
260 self.put(key, record)
261 }
262
263 pub fn replace(&mut self, key: Vec<u8>, record: Vec<u8>) -> Result<bool, &'static str> {
265 if !self.records.contains_key(&key) {
266 return Err("23");
267 }
268 self.put(key, record)
269 }
270
271 pub fn remove(&mut self, key: &[u8]) -> Option<Vec<u8>> {
272 let record = self.records.remove(key)?;
273 for alt in &mut self.alternates {
274 if let Some(seq) = alt.arrival.remove(key) {
275 alt.index.remove(&(alt.span.of(&record), seq, key.to_vec()));
276 }
277 }
278 self.dirty = true;
279 Some(record)
280 }
281
282 pub fn start(&mut self, which: usize, wanted: Ordering, or_equal: bool, value: &[u8]) -> bool {
286 let value = &value[..value.len().min(self.key_len(which))];
287 let fits = |k: &[u8]| {
288 let o = k[..value.len().min(k.len())].cmp(value);
289 o == wanted || (or_equal && o == Ordering::Equal)
290 };
291 let from = (value.to_vec(), 0, Vec::new());
292 let found = match which {
293 0 => self.records.range(from.0..).map(|(k, _)| k).find(|k| fits(k)).map(|k| (k.clone(), 0, Vec::new())),
294 n => self.alternates[n - 1].index.range(from..).find(|(k, ..)| fits(k)).cloned(),
295 };
296 self.last_read = None;
297 match found {
298 Some(at) => {
299 self.cursor = Cursor::At { which, at, inclusive: true };
300 true
301 }
302 None => {
303 self.lose_position();
304 false
305 }
306 }
307 }
308
309 pub fn lose_position(&mut self) {
311 self.cursor = Cursor::Undefined;
312 self.last_read = None;
313 }
314
315 pub fn read_at(&mut self, which: usize, key: &[u8]) {
317 let at = match which {
318 0 => (key.to_vec(), 0, Vec::new()),
319 n => {
320 let alt = &self.alternates[n - 1];
321 (alt.span.of(&self.records[key]), alt.arrival[key], key.to_vec())
322 }
323 };
324 self.cursor = Cursor::At { which, at, inclusive: false };
325 self.last_read = Some(key.to_vec());
326 }
327
328 pub fn seek(&self, from: &[u8], inclusive: bool, backward: bool) -> Option<(Vec<u8>, Vec<u8>)> {
331 let edge = if inclusive { Bound::Included(from.to_vec()) } else { Bound::Excluded(from.to_vec()) };
332 let mut hits = if backward { self.records.range((Bound::Unbounded, edge)) } else { self.records.range((edge, Bound::Unbounded)) };
333 let hit = if backward { hits.next_back() } else { hits.next() };
334 hit.map(|(k, r)| (k.clone(), r.clone()))
335 }
336
337 pub fn step(&mut self, backward: bool) -> Result<Option<Found>, &'static str> {
339 let (which, bound) = match &self.cursor {
340 Cursor::Undefined => return Err("46"),
341 Cursor::First => (0, Bound::Unbounded),
342 Cursor::At { which, at, inclusive: true } => (*which, Bound::Included(at.clone())),
343 Cursor::At { which, at, inclusive: false } => (*which, Bound::Excluded(at.clone())),
344 };
345 let range = if backward { (Bound::Unbounded, bound) } else { (bound, Bound::Unbounded) };
346 let hit = if which == 0 {
347 let mut keys = self.records.range((range.0.map(|(k, ..)| k), range.1.map(|(k, ..)| k))).map(|(k, _)| k);
348 let key = if backward { keys.next_back() } else { keys.next() };
349 key.map(|k| (k.clone(), false))
350 } else {
351 let mut entries = self.alternates[which - 1].index.range(range);
352 let entry = if backward { entries.next_back() } else { entries.next() };
353 entry.map(|(value, _, key)| {
354 let after = if backward { entries.next_back() } else { entries.next() };
355 (key.clone(), after.is_some_and(|(v, ..)| v == value))
356 })
357 };
358 let Some((key, duplicate)) = hit else {
359 self.lose_position();
360 return Ok(None);
361 };
362 self.read_at(which, &key);
363 Ok(Some(Found { record: self.records[&key].clone(), key, duplicate }))
364 }
365
366 fn inbound(&self, raw: Vec<u8>, format: Format) -> Vec<u8> {
367 let mut record = match format {
368 Format::Text => {
369 let unknown = self.page.encode_char('?').unwrap_or(0x6F);
370 String::from_utf8_lossy(&raw).chars().map(|c| self.page.encode_char(c).unwrap_or(unknown)).collect()
371 }
372 _ => raw,
373 };
374 if format != Format::Variable {
375 record.resize(self.record_len.max(record.len()), ebcdic::SPACE);
376 }
377 record
378 }
379
380 fn outbound(&self, record: &[u8], format: Format) -> Vec<u8> {
381 match format {
382 Format::Text => format!("{}\n", self.page.decode(record).trim_end()).into_bytes(),
383 Format::Fixed => {
384 let mut r = record.to_vec();
385 r.resize(self.record_len, ebcdic::SPACE);
386 r
387 }
388 Format::Variable => record.to_vec(),
389 }
390 }
391
392 fn save(&self, format: Format) -> io::Result<()> {
393 let Some(path) = self.path.as_ref().filter(|_| self.dirty) else { return Ok(()) };
394 let mut out = Open { mode: OpenMode::Output, format, handle: Handle::Writer(BufWriter::new(File::create(path)?)) };
395 let empty = match format {
396 Format::Fixed => vec![0; self.record_len],
397 Format::Variable => Vec::new(),
398 Format::Text => b"\n".to_vec(),
399 };
400 let mut next = 1;
401 for (key, record) in &self.records {
402 if self.keying == Keying::Relative {
403 for _ in next..number_of(key) {
404 out.write(&empty)?;
405 }
406 next = number_of(key) + 1;
407 }
408 out.write(&self.outbound(record, format))?;
409 }
410 out.close()
411 }
412}
413
414pub struct Open {
415 pub mode: OpenMode,
416 pub format: Format,
417 handle: Handle,
418}
419
420pub enum Record {
422 Data(Vec<u8>),
423 WrongLength(Vec<u8>),
425 End,
426}
427
428pub fn open_keyed(dd: Option<&Dd>, mode: OpenMode, format: Format, keying: Keying, record_len: usize, page: &'static CodePage) -> io::Result<Open> {
431 let path = dd.filter(|_| mode != OpenMode::Input).map(|d| d.path.clone());
432 let mut keyed = Keyed::new(keying, path, record_len, page);
433 if let Some(dd) = dd.filter(|d| mode != OpenMode::Output && d.path.exists()) {
434 let mut reader = open(dd, OpenMode::Input, format)?;
435 let mut number = 0u64;
436 loop {
437 let raw = match reader.read(record_len)? {
438 Record::End => break,
439 Record::Data(r) | Record::WrongLength(r) => r,
440 };
441 number += 1;
442 if keyed.keying == Keying::Relative && raw.iter().all(|&b| b == 0) {
443 continue;
444 }
445 let record = keyed.inbound(raw, format);
446 let key = keyed.prime_key(&record).unwrap_or_else(|| record_number(number));
447 keyed.insert(key, record).map_err(|_| io::Error::new(io::ErrorKind::InvalidData, format!("{}: two records with one key", dd.path.display())))?;
448 }
449 keyed.dirty = false;
450 }
451 Ok(Open { mode, format, handle: Handle::Keyed(Box::new(keyed)) })
452}
453
454pub fn open(dd: &Dd, mode: OpenMode, format: Format) -> io::Result<Open> {
455 let handle = match mode {
456 OpenMode::Input => Handle::Reader(BufReader::new(File::open(&dd.path)?)),
457 OpenMode::Output => Handle::Writer(BufWriter::new(File::create(&dd.path)?)),
458 OpenMode::Extend => Handle::Writer(BufWriter::new(OpenOptions::new().append(true).create(true).open(&dd.path)?)),
459 OpenMode::InputOutput => return Err(io::Error::new(io::ErrorKind::Unsupported, "OPEN I-O of a line-sequential file")),
460 };
461 Ok(Open { mode, format, handle })
462}
463
464pub fn absent() -> Open {
466 Open { mode: OpenMode::Input, format: Format::Fixed, handle: Handle::Empty }
467}
468
469impl Open {
470 pub fn keyed(&mut self) -> Option<&mut Keyed> {
471 match &mut self.handle {
472 Handle::Keyed(k) => Some(k),
473 _ => None,
474 }
475 }
476
477 pub fn is_keyed(&self) -> bool {
478 matches!(self.handle, Handle::Keyed(_))
479 }
480
481 pub fn read(&mut self, fixed_len: usize) -> io::Result<Record> {
482 let reader = match &mut self.handle {
483 Handle::Reader(r) => r,
484 Handle::Empty => return Ok(Record::End),
485 Handle::Writer(_) | Handle::Keyed(_) => return Err(io::Error::other("a sequential READ of a file not open for it")),
486 };
487 match self.format {
488 Format::Fixed => {
489 let mut buf = Vec::with_capacity(fixed_len);
490 reader.by_ref().take(fixed_len as u64).read_to_end(&mut buf)?;
491 Ok(match buf.len() {
492 0 => Record::End,
493 n if n == fixed_len => Record::Data(buf),
494 _ => Record::WrongLength(buf),
495 })
496 }
497 Format::Variable => {
498 let mut rdw = [0u8; 4];
499 match reader.read_exact(&mut rdw) {
500 Err(e) if e.kind() == io::ErrorKind::UnexpectedEof => return Ok(Record::End),
501 other => other?,
502 }
503 let len = u16::from_be_bytes([rdw[0], rdw[1]]) as usize;
504 if len < 4 {
505 return Err(io::Error::new(io::ErrorKind::InvalidData, format!("an RDW of length {len}")));
506 }
507 let mut buf = vec![0u8; len - 4];
508 reader.read_exact(&mut buf)?;
509 Ok(Record::Data(buf))
510 }
511 Format::Text => {
512 let mut line = Vec::new();
513 if reader.read_until(b'\n', &mut line)? == 0 {
514 return Ok(Record::End);
515 }
516 while matches!(line.last(), Some(b'\n' | b'\r')) {
517 line.pop();
518 }
519 Ok(Record::Data(line))
520 }
521 }
522 }
523
524 pub fn write(&mut self, bytes: &[u8]) -> io::Result<()> {
525 let Handle::Writer(w) = &mut self.handle else {
526 return Err(io::Error::other("WRITE to a file not opened for output"));
527 };
528 match self.format {
529 Format::Fixed | Format::Text => w.write_all(bytes),
530 Format::Variable => {
531 let len = u16::try_from(bytes.len() + 4).map_err(|_| io::Error::new(io::ErrorKind::InvalidInput, "a record over 32,760 bytes"))?;
532 w.write_all(&len.to_be_bytes())?;
533 w.write_all(&[0, 0])?;
534 w.write_all(bytes)
535 }
536 }
537 }
538
539 pub fn close(self) -> io::Result<()> {
540 match self.handle {
541 Handle::Writer(mut w) => w.flush(),
542 Handle::Keyed(k) => k.save(self.format),
543 _ => Ok(()),
544 }
545 }
546}
547
548#[cfg(test)]
549mod tests {
550 use super::*;
551
552 #[test]
553 fn dd_specs() {
554 let dds = Dds::new(&["IN=/tmp/a.txt:text".into(), "out=/tmp/b.dat".into()], false).unwrap();
555 assert_eq!(dds.get("IN"), Some(Dd { path: "/tmp/a.txt".into(), format: Some(Format::Text) }));
556 assert_eq!(dds.get("OUT").unwrap().format, None);
557 assert!(dds.get("NONE").is_none());
558 assert!(Dds::new(&["NOEQUALS".into()], false).is_err());
559 }
560
561 #[test]
562 fn variable_records_round_trip_through_rdws() {
563 let path = std::env::temp_dir().join(format!("ironwork-vb-{}", std::process::id()));
564 let dd = Dd { path: path.clone(), format: None };
565 let mut out = open(&dd, OpenMode::Output, Format::Variable).unwrap();
566 out.write(b"AB").unwrap();
567 out.write(b"CDEF").unwrap();
568 out.close().unwrap();
569 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']);
570 let mut input = open(&dd, OpenMode::Input, Format::Variable).unwrap();
571 assert!(matches!(input.read(0).unwrap(), Record::Data(d) if d == b"AB"));
572 assert!(matches!(input.read(0).unwrap(), Record::Data(d) if d == b"CDEF"));
573 assert!(matches!(input.read(0).unwrap(), Record::End));
574 }
575}