| 1 | use std::path::Path; |
| 2 | |
| 3 | use crate::{ |
| 4 | common::{CodepointIter, Codepoints, UcdFile, UcdFileByCodepoint}, |
| 5 | error::Error, |
| 6 | }; |
| 7 | |
| 8 | /// A single row in the `extracted/DerivedNumericValues.txt` file. |
| 9 | /// |
| 10 | /// This file gives the derived values of the Numeric_Value property. |
| 11 | #[derive (Clone, Debug, Default, Eq, PartialEq)] |
| 12 | pub struct DerivedNumericValues { |
| 13 | /// The codepoint or codepoint range for this entry. |
| 14 | pub codepoints: Codepoints, |
| 15 | /// The approximate Numeric_Value of the codepoints in this entry, |
| 16 | /// as a decimal. |
| 17 | pub numeric_value_decimal: String, |
| 18 | /// The exact Numeric_Value of the codepoints in this entry, as |
| 19 | /// a fraction. |
| 20 | pub numeric_value_fraction: String, |
| 21 | } |
| 22 | |
| 23 | impl UcdFile for DerivedNumericValues { |
| 24 | fn relative_file_path() -> &'static Path { |
| 25 | Path::new("extracted/DerivedNumericValues.txt" ) |
| 26 | } |
| 27 | } |
| 28 | |
| 29 | impl UcdFileByCodepoint for DerivedNumericValues { |
| 30 | fn codepoints(&self) -> CodepointIter { |
| 31 | self.codepoints.into_iter() |
| 32 | } |
| 33 | } |
| 34 | |
| 35 | impl std::str::FromStr for DerivedNumericValues { |
| 36 | type Err = Error; |
| 37 | |
| 38 | fn from_str(line: &str) -> Result<DerivedNumericValues, Error> { |
| 39 | let re_parts = regex!( |
| 40 | r"(?x) |
| 41 | ^ |
| 42 | \s*(?P<codepoints>[^\s;]+)\s*; |
| 43 | \s*(?P<numeric_value_decimal>[^\s;]+)\s*; |
| 44 | \s*; |
| 45 | \s*(?P<numeric_value_fraction>[^\s;]+)\s* |
| 46 | " , |
| 47 | ); |
| 48 | |
| 49 | let caps = match re_parts.captures(line.trim()) { |
| 50 | Some(caps) => caps, |
| 51 | None => return err!("invalid PropList line: ' {}'" , line), |
| 52 | }; |
| 53 | let codepoints = caps["codepoints" ].parse()?; |
| 54 | let numeric_value_decimal = caps["numeric_value_decimal" ].to_string(); |
| 55 | let numeric_value_fraction = |
| 56 | caps["numeric_value_fraction" ].to_string(); |
| 57 | |
| 58 | Ok(DerivedNumericValues { |
| 59 | codepoints, |
| 60 | numeric_value_decimal, |
| 61 | numeric_value_fraction, |
| 62 | }) |
| 63 | } |
| 64 | } |
| 65 | |
| 66 | #[cfg (test)] |
| 67 | mod tests { |
| 68 | use super::DerivedNumericValues; |
| 69 | |
| 70 | #[test ] |
| 71 | fn parse_single() { |
| 72 | let line = "0030 ; 0.0 ; ; 0 # Nd DIGIT ZERO \n" ; |
| 73 | let row: DerivedNumericValues = line.parse().unwrap(); |
| 74 | assert_eq!(row.codepoints, 0x0030); |
| 75 | assert_eq!(row.numeric_value_decimal, "0.0" ); |
| 76 | assert_eq!(row.numeric_value_fraction, "0" ); |
| 77 | } |
| 78 | |
| 79 | #[test ] |
| 80 | fn parse_range() { |
| 81 | let line = "11FC9..11FCA ; 0.0625 ; ; 1/16 # No [2] TAMIL FRACTION ONE SIXTEENTH-1..TAMIL FRACTION ONE SIXTEENTH-2 \n" ; |
| 82 | let row: DerivedNumericValues = line.parse().unwrap(); |
| 83 | assert_eq!(row.codepoints, (0x11FC9, 0x11FCA)); |
| 84 | assert_eq!(row.numeric_value_decimal, "0.0625" ); |
| 85 | assert_eq!(row.numeric_value_fraction, "1/16" ); |
| 86 | } |
| 87 | } |
| 88 | |