ironwork_compile/
picture.rs

1pub use rt::picture::Sym;
2use syntax::ast::{CurrencySign, Environment};
3
4#[derive(Clone, Copy, Debug, PartialEq, Eq)]
5pub enum Category {
6    Alphanumeric,
7    Numeric,
8    National,
9    /// G, with B: DBCS character positions, which USAGE DISPLAY-1 holds.
10    Dbcs,
11    NumericEdited,
12    AlphanumericEdited,
13}
14
15#[derive(Clone, Debug, PartialEq, Eq)]
16pub struct Picture {
17    pub category: Category,
18    /// Character positions: bytes for alphanumeric, digits for numeric, characters for national
19    /// and DBCS.
20    pub size: u32,
21    pub digits: u32,
22    pub scale: u32,
23    pub signed: bool,
24    pub edit: Option<Vec<Sym>>,
25    /// Scaling positions P to the right of the digits: the value is the digits times ten to this
26    /// power. Ps to the left of the digits raise `scale` past `digits` instead.
27    pub scaling: u32,
28    /// The currency sign value its currency symbol stands for.
29    pub currency: Option<String>,
30}
31
32/// What SPECIAL-NAMES changes in a PICTURE: whether the comma is the decimal point, and which
33/// characters are currency symbols, for which values.
34#[derive(Clone, Copy, Debug, Default)]
35pub struct Notation<'a> {
36    pub decimal_comma: bool,
37    pub currency: &'a [CurrencySign],
38}
39
40impl<'a> Notation<'a> {
41    pub fn of(environment: &'a Environment) -> Self {
42        Self { decimal_comma: environment.decimal_point_comma, currency: &environment.currency }
43    }
44
45    /// The value `symbol` stands for: with no CURRENCY SIGN clause or CURRENCY option, $ for $
46    /// (Language Reference SC27-8713-03, p. 212).
47    pub fn currency_value(&self, symbol: char) -> Option<&'a str> {
48        match self.currency {
49            [] => (symbol == '$').then_some("$"),
50            signs => signs.iter().find(|c| c.symbol == symbol).map(|c| c.value.as_str()),
51        }
52    }
53}
54
55/// An edited PICTURE is written out position by position, so its length is bounded tighter.
56const MAX_EDITED: u64 = 4096;
57
58/// Enterprise COBOL's limit on an elementary item's character positions.
59pub const MAX_POSITIONS: u64 = 134_217_727;
60
61pub fn analyse(text: &str) -> Result<Picture, String> {
62    analyse_with(text, Notation::default())
63}
64
65/// A PICTURE under `notation`. Under DECIMAL-POINT IS COMMA the comma is the decimal point and the
66/// period an insertion character (Language Reference SC27-8713-03, p. 208); a currency symbol is
67/// read as $, and the value it stands for kept.
68pub fn analyse_with(text: &str, notation: Notation) -> Result<Picture, String> {
69    let (runs, currency) = runs(text, notation)?;
70    let has = |symbol: char| runs.iter().any(|&(c, _)| c == symbol);
71    if has('G') || has('N') && has('B') {
72        return dbcs(text, &runs);
73    }
74    if runs.iter().any(|&(c, _)| matches!(c, 'Z' | '*' | '+' | '-' | '.' | ',' | 'B' | '0' | '/' | '$' | 'C' | 'R' | 'D')) {
75        return edited(text, &runs, notation.decimal_comma, currency);
76    }
77    let (mut digits, mut scale, mut signed, mut after_point) = (0u64, 0u64, false, false);
78    let (mut alnum, mut national) = (0u64, 0u64);
79    let (mut left, mut right) = (0u64, 0u64);
80    let misplaced = || Err(format!("PICTURE {text}: P must be one string of scaling positions at the left or right end of the digits"));
81    for (i, &(c, n)) in runs.iter().enumerate() {
82        match c {
83            '9' if right > 0 => return misplaced(),
84            '9' => {
85                digits += n;
86                if after_point {
87                    scale += n;
88                }
89            }
90            'S' if i == 0 && n == 1 => signed = true,
91            'V' if left > 0 => return misplaced(),
92            'V' if !after_point && n == 1 => after_point = true,
93            'X' | 'A' => alnum += n,
94            'N' => national += n,
95            'P' if digits == 0 && runs.get(i + 1).is_some_and(|&(c, _)| matches!(c, 'P' | '9')) => left += n,
96            'P' if digits > 0 && left == 0 && !after_point => right += n,
97            'P' => return misplaced(),
98            _ => return Err(format!("PICTURE {text}: {c:?} is not a PICTURE symbol")),
99        }
100    }
101    if alnum + digits > MAX_POSITIONS || national > MAX_POSITIONS {
102        return Err(format!("PICTURE {text}: more than {MAX_POSITIONS} character positions"));
103    }
104    let positions = digits + left + right;
105    if left > 0 {
106        scale = left + digits;
107    }
108    let (digits, scale, alnum, national) = (digits as u32, scale as u32, alnum as u32, national as u32);
109    let scaled = left + right > 0;
110    match (digits > 0, alnum > 0, national > 0) {
111        (true, false, false) if positions <= 31 => Ok(Picture { category: Category::Numeric, size: digits, digits, scale, signed, edit: None, scaling: right as u32, currency: None }),
112        (true, false, false) => Err(format!("PICTURE {text}: more than 31 digits")),
113        (_, true, false) if !signed && !after_point && !scaled => Ok(Picture { category: Category::Alphanumeric, size: alnum + digits, digits: 0, scale: 0, signed, edit: None, scaling: 0, currency: None }),
114        (false, false, true) if !signed && !after_point && !scaled => Ok(Picture { category: Category::National, size: national, digits: 0, scale: 0, signed, edit: None, scaling: 0, currency: None }),
115        _ => Err(format!("PICTURE {text}: mixes symbols of different categories")),
116    }
117}
118
119/// A PICTURE of G, or of N with B: DBCS character positions, each B one that holds a DBCS space
120/// (Language Reference SC27-8713-03, pp. 214, 219). N with B is national-edited unless the item is
121/// DISPLAY-1, which the item's USAGE decides.
122fn dbcs(text: &str, runs: &[(char, u64)]) -> Result<Picture, String> {
123    let symbol = if runs.iter().any(|&(c, _)| c == 'G') { 'G' } else { 'N' };
124    if let Some(&(c, _)) = runs.iter().find(|&&(c, _)| c != symbol && c != 'B') {
125        return Err(format!("PICTURE {text}: {c:?} cannot be in a PICTURE of {symbol}, which takes {symbol} and B only"));
126    }
127    let size: u64 = runs.iter().map(|&(_, n)| n).sum();
128    let inserts = runs.iter().any(|&(c, _)| c == 'B');
129    if size > MAX_POSITIONS || inserts && size > MAX_EDITED {
130        return Err(format!("PICTURE {text}: more character positions than a DBCS item holds"));
131    }
132    let edit = inserts.then(|| runs.iter().flat_map(|&(c, n)| std::iter::repeat_n(if c == 'B' { Sym::Insert(' ') } else { Sym::Char }, n as usize)).collect());
133    let category = if symbol == 'G' { Category::Dbcs } else { Category::National };
134    Ok(Picture { category, size: size as u32, digits: 0, scale: 0, signed: false, edit, scaling: 0, currency: None })
135}
136
137/// A numeric PICTURE under BLANK WHEN ZERO, which makes the item numeric-edited (Language
138/// Reference SC27-8713-03, p. 195).
139pub fn blank_when_zero(p: &Picture) -> Result<Picture, String> {
140    if p.signed {
141        return Err("BLANK WHEN ZERO cannot be given for a PICTURE with S".into());
142    }
143    let int = p.digits.saturating_sub(p.scale) as usize;
144    let mut syms = vec![Sym::Nine; int];
145    if p.scale > 0 {
146        syms.push(Sym::Implied);
147    }
148    syms.extend(std::iter::repeat_n(Sym::Nine, p.digits as usize - int));
149    Ok(Picture { category: Category::NumericEdited, edit: Some(syms), ..p.clone() })
150}
151
152fn edited(text: &str, runs: &[(char, u64)], decimal_comma: bool, currency: Option<&str>) -> Result<Picture, String> {
153    let total: u64 = runs.iter().map(|&(_, n)| n).sum();
154    if total > MAX_EDITED {
155        return Err(format!("PICTURE {text}: an edited PICTURE longer than {MAX_EDITED} positions"));
156    }
157    let chars: Vec<char> = runs.iter().flat_map(|&(c, n)| std::iter::repeat_n(c, n as usize)).collect();
158    let bad = |why: &str| Err(format!("PICTURE {text}: {why}"));
159    if chars.iter().any(|c| matches!(c, 'S' | 'N')) {
160        return bad("S and N are not allowed in an edited PICTURE");
161    }
162    if chars.iter().any(|c| matches!(c, 'X' | 'A')) {
163        let mut syms = Vec::new();
164        for &c in &chars {
165            syms.push(match c {
166                'X' | 'A' | '9' => Sym::Char,
167                'B' => Sym::Insert(' '),
168                '0' | '/' => Sym::Insert(c),
169                _ => return bad("an alphanumeric-edited PICTURE takes only X, A, 9, B, 0 and /"),
170            });
171        }
172        let size = syms.len() as u32;
173        return Ok(Picture { category: Category::AlphanumericEdited, size, digits: 0, scale: 0, signed: false, edit: Some(syms), scaling: 0, currency: None });
174    }
175    let (point, comma) = if decimal_comma { (',', '.') } else { ('.', ',') };
176    let floating: Vec<char> = ['+', '-', '$'].into_iter().filter(|f| chars.iter().filter(|c| *c == f).count() >= 2).collect();
177    if floating.len() > 1 {
178        return bad("two floating insertion strings");
179    }
180    let float = floating.first().copied();
181    let (mut syms, mut led, mut i) = (Vec::new(), false, 0);
182    let misplaced = "P must be one string of scaling positions at the left or right end of the digits";
183    let mut scaling: Option<(usize, u32, usize)> = None;
184    while i < chars.len() {
185        let c = chars[i];
186        let next = chars.get(i + 1).copied();
187        if c == 'P' {
188            match &mut scaling {
189                Some((at, n, _)) if *at + *n as usize == i => *n += 1,
190                Some(_) => return bad(misplaced),
191                None => scaling = Some((i, 1, syms.len())),
192            }
193            i += 1;
194            continue;
195        }
196        let sym = match c {
197            'C' if next == Some('R') => {
198                i += 1;
199                Sym::Cr
200            }
201            'D' if next == Some('B') => {
202                i += 1;
203                Sym::Db
204            }
205            c if Some(c) == float && !led => {
206                led = true;
207                Sym::FloatLead(c)
208            }
209            c if Some(c) == float => Sym::Float(c),
210            '+' | '-' => Sym::Sign(c),
211            '$' => Sym::Currency,
212            '9' => Sym::Nine,
213            'Z' => Sym::Z,
214            '*' => Sym::Star,
215            c if c == point => Sym::Point,
216            'V' => Sym::Implied,
217            'B' => Sym::Insert(' '),
218            '0' | '/' => Sym::Insert(c),
219            c if c == comma => Sym::Insert(c),
220            _ => return bad(&format!("{c:?} is not a numeric-edited symbol")),
221        };
222        syms.push(sym);
223        i += 1;
224    }
225    let points = syms.iter().filter(|s| matches!(s, Sym::Point | Sym::Implied)).count();
226    if points > 1 {
227        return bad("more than one decimal point");
228    }
229    let digits = syms.iter().filter(|s| s.is_digit()).count() as u32;
230    let mut scale = syms.iter().skip_while(|s| !matches!(s, Sym::Point | Sym::Implied)).filter(|s| s.is_digit()).count() as u32;
231    let mut right = 0;
232    if let Some((_, n, at)) = scaling {
233        let before = syms[..at].iter().filter(|s| s.is_digit()).count() as u32;
234        match (points, before) {
235            (0, 0) => scale = n + digits,
236            (0, b) if b == digits => right = n,
237            _ => return bad(misplaced),
238        }
239    }
240    if digits == 0 || digits + scaling.map_or(0, |(_, n, _)| n) > 31 {
241        return bad("a numeric-edited PICTURE needs 1 to 31 digit positions");
242    }
243    let widths: u32 = syms.iter().map(|s| s.width() as u32).sum();
244    let size = widths + currency.map_or(0, |v| v.chars().count() as u32 - 1);
245    Ok(Picture { category: Category::NumericEdited, size, digits, scale, signed: false, edit: Some(syms), scaling: right, currency: currency.map(str::to_owned) })
246}
247
248/// Runs of one PICTURE symbol and a count, and the currency sign value the PICTURE uses.
249type Runs<'a> = (Vec<(char, u64)>, Option<&'a str>);
250
251/// The PICTURE as runs of one symbol and a count, `9(4)` read as four nines without writing them
252/// out, its currency symbol as $, with the value that symbol stands for.
253fn runs<'a>(text: &str, notation: Notation<'a>) -> Result<Runs<'a>, String> {
254    let mut out: Vec<(char, u64)> = Vec::new();
255    let (mut chars, mut currency) = (text.chars(), None);
256    while let Some(c) = chars.next() {
257        if c == '(' {
258            let count: String = chars.by_ref().take_while(|&d| d != ')').collect();
259            let n: u64 = count.parse().ok().filter(|&n| (1..=MAX_POSITIONS).contains(&n)).ok_or_else(|| format!("PICTURE {text}: bad repetition ({count})"))?;
260            let last = out.last_mut().ok_or_else(|| format!("PICTURE {text}: a repetition with nothing to repeat"))?;
261            last.1 += n - 1;
262        } else if let Some(value) = notation.currency_value(c) {
263            match currency {
264                Some((symbol, _)) if symbol != c => return Err(format!("PICTURE {text}: two different currency symbols")),
265                _ => currency = Some((c, value)),
266            }
267            out.push(('$', 1));
268        } else if c == '$' {
269            return Err(format!("PICTURE {text}: '$' is not a currency symbol of this program, whose CURRENCY SIGN clauses or CURRENCY option name others"));
270        } else {
271            out.push((c.to_ascii_uppercase(), 1));
272        }
273    }
274    Ok((out, currency.map(|(_, value)| value)))
275}
276
277#[cfg(test)]
278mod tests {
279    use super::*;
280
281    const COMMA: Notation = Notation { decimal_comma: true, currency: &[] };
282
283    #[test]
284    fn numeric_pictures() {
285        assert_eq!(analyse("S9(3)V99").unwrap(), Picture { category: Category::Numeric, size: 5, digits: 5, scale: 2, signed: true, edit: None, scaling: 0, currency: None });
286        assert_eq!(analyse("9(18)").unwrap().digits, 18);
287        assert_eq!(analyse("SV9").unwrap().scale, 1);
288    }
289
290    #[test]
291    fn edited_pictures() {
292        let p = analyse("$$,$$9.99CR").unwrap();
293        assert_eq!((p.category, p.size, p.digits, p.scale), (Category::NumericEdited, 11, 6, 2));
294        assert_eq!(p.edit.as_ref().unwrap()[0], Sym::FloatLead('$'));
295        let z = analyse("-ZZ,ZZ9").unwrap();
296        assert_eq!(z.edit.unwrap()[0], Sym::Sign('-'));
297        assert_eq!(analyse("XXBXX/99").unwrap().category, Category::AlphanumericEdited);
298    }
299
300    #[test]
301    fn alphanumeric_and_national() {
302        assert_eq!(analyse("X(512)").unwrap().size, 512);
303        assert_eq!(analyse("N(256)").unwrap().category, Category::National);
304    }
305
306    #[test]
307    fn unsupported_and_invalid_pictures_say_why() {
308        assert!(analyse("ZZ9.99.9").unwrap_err().contains("decimal point"));
309        assert!(analyse("9(32)").unwrap_err().contains("31 digits"));
310        assert!(analyse("XN").unwrap_err().contains("categories"));
311        assert!(analyse("X(999999999)").unwrap_err().contains("repetition"));
312        assert!(analyse("X(134217727)X").unwrap_err().contains("character positions"));
313        assert_eq!(analyse("XX99").unwrap(), Picture { category: Category::Alphanumeric, size: 4, digits: 0, scale: 0, signed: false, edit: None, scaling: 0, currency: None });
314    }
315
316    #[test]
317    fn scaling_positions_left_of_the_digits_raise_the_scale_and_right_of_them_scale_the_value() {
318        let shape = |pic: &str| analyse(pic).map(|p| (p.category, p.size, p.digits, p.scale, p.scaling));
319        assert_eq!(shape("SP(8)9"), Ok((Category::Numeric, 1, 1, 9, 0)));
320        assert_eq!(shape("VPP99"), Ok((Category::Numeric, 2, 2, 4, 0)));
321        assert_eq!(shape("S999PP"), Ok((Category::Numeric, 3, 3, 0, 2)));
322        assert_eq!(shape("99P(6)V"), Ok((Category::Numeric, 2, 2, 0, 6)));
323        assert_eq!(shape("ZZZPP"), Ok((Category::NumericEdited, 3, 3, 0, 2)));
324        for bad in ["P9P", "9P9", "PPV9", "9V9P", "99VPP", "Z.ZPP", "P(31)9"] {
325            assert!(analyse(bad).is_err(), "{bad}");
326        }
327        assert!(analyse("P(30)9").is_ok());
328    }
329
330    #[test]
331    fn under_decimal_point_is_comma_the_comma_is_the_point_and_the_period_an_insertion() {
332        let p = analyse_with("Z.ZZ9,99", COMMA).unwrap();
333        assert_eq!((p.category, p.size, p.digits, p.scale), (Category::NumericEdited, 8, 6, 2));
334        assert_eq!(p.edit.as_ref().unwrap()[1], Sym::Insert('.'));
335        assert_eq!(p.edit.as_ref().unwrap()[5], Sym::Point);
336        assert_eq!(analyse_with("ZZ9.99", Notation::default()).unwrap().scale, 2);
337        assert_eq!(analyse_with("ZZ9.99", COMMA).unwrap().scale, 0);
338        assert!(analyse_with("9,99,9", COMMA).unwrap_err().contains("decimal point"));
339    }
340
341    #[test]
342    fn a_picture_can_end_in_an_insertion_comma_or_the_decimal_point() {
343        let edit = |pic: &str, notation: Notation, value: u128| {
344            let p = analyse_with(pic, notation).unwrap();
345            let point = if notation.decimal_comma { ',' } else { '.' };
346            ((p.category, p.size, p.digits, p.scale), rt::edit::numeric(p.edit.as_ref().unwrap(), p.digits, false, value, false, point, ""))
347        };
348        assert_eq!(edit("9,9,9,", Notation::default(), 123), ((Category::NumericEdited, 6, 3, 0), "1,2,3,".into()));
349        assert_eq!(edit("999.", Notation::default(), 7), ((Category::NumericEdited, 4, 3, 0), "007.".into()));
350        assert_eq!(edit("9.9.9,", COMMA, 123), ((Category::NumericEdited, 6, 3, 0), "1.2.3,".into()));
351        assert_eq!(edit("999.", COMMA, 7), ((Category::NumericEdited, 4, 3, 0), "007.".into()));
352        assert_eq!(analyse_with("999.", COMMA).unwrap().edit.unwrap()[3], Sym::Insert('.'));
353    }
354
355    #[test]
356    fn a_currency_sign_clause_names_the_symbol_and_the_value_it_stands_for() {
357        let signs = [CurrencySign { value: "W".into(), symbol: 'W', hex: None }, CurrencySign { value: "EUR ".into(), symbol: 'e', hex: None }];
358        let notation = Notation { decimal_comma: true, currency: &signs };
359        let w = analyse_with("W9.999,99", notation).unwrap();
360        assert_eq!((w.category, w.size, w.digits, w.scale, w.currency.as_deref()), (Category::NumericEdited, 9, 6, 2, Some("W")));
361        assert_eq!(w.edit.as_ref().unwrap()[0], Sym::Currency);
362        let floating = analyse_with("eeee9,99", notation).unwrap();
363        assert_eq!((floating.size, floating.digits, floating.currency.as_deref()), (11, 6, Some("EUR ")));
364        assert_eq!(floating.edit.as_ref().unwrap()[0], Sym::FloatLead('$'));
365        assert!(analyse_with("$$9", notation).unwrap_err().contains("'$'"));
366        assert!(analyse_with("We9", notation).unwrap_err().contains("two different currency symbols"));
367        assert!(analyse_with("E9", notation).unwrap_err().contains("not a PICTURE symbol"));
368        assert_eq!(analyse("$9").unwrap().currency.as_deref(), Some("$"));
369    }
370
371    #[test]
372    fn blank_when_zero_makes_a_numeric_picture_numeric_edited() {
373        let p = blank_when_zero(&analyse("99V9").unwrap()).unwrap();
374        assert_eq!((p.category, p.size, p.digits, p.scale), (Category::NumericEdited, 3, 3, 1));
375        assert_eq!(p.edit.unwrap(), [Sym::Nine, Sym::Nine, Sym::Implied, Sym::Nine]);
376        assert_eq!(blank_when_zero(&analyse("VPP99").unwrap()).unwrap().edit.unwrap(), [Sym::Implied, Sym::Nine, Sym::Nine]);
377        assert!(blank_when_zero(&analyse("S99").unwrap()).unwrap_err().contains("with S"));
378    }
379}
ironwork_compile/picture.rs

ironwork_compile/
picture.rs
