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

ironwork_compile/
picture.rs
