use crate::components::Component;
use crate::context::ParsingContext;
use crate::dictionaries::uk::{get_month, get_weekday};
use crate::error::Result;
use crate::parsers::Parser;
use crate::results::ParsedResult;
use chrono::Datelike;
use fancy_regex::Regex;
use std::sync::LazyLock;
static DOT_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
Regex::new(r"(?<![0-9])(?P<day>\d{1,2})\.(?P<month>\d{1,2})\.(?P<year>\d{4}|\d{2})(?![0-9])")
.unwrap()
});
static MONTH_NAME_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
Regex::new(
r"(?ix)
(?:
(?P<weekday>понеділок|вівторок|середа|середу|четвер|п'ятниця|п'ятницю|пятниця|субота|суботу|неділя|неділю|пн|вт|ср|чт|пт|сб|нд)(?:\.|,)?\s*
(?:,\s*)?
)?
(?P<day>\d{1,2})
(?:
\s*(?:-|–|до)\s*
(?P<end_day>\d{1,2})
)?
\s+
(?P<month>січня?|лютого?|лют\.?|березня?|бер\.?|квітня?|квіт\.?|травня?|трав\.?|червня?|черв\.?|липня?|лип\.?|серпня?|серп\.?|вересня?|вер\.?|жовтня?|жовт\.?|листопада?|лист\.?|грудня?|груд\.?)
(?:
\s+
(?P<year>\d{4}|\d{2})
(?:\s*р\.?)?
)?
(?![а-яА-ЯіїєґІЇЄҐ])"
).unwrap()
});
pub struct UKMonthNameParser;
impl UKMonthNameParser {
pub fn new() -> Self {
Self
}
}
impl Default for UKMonthNameParser {
fn default() -> Self {
Self::new()
}
}
impl Parser for UKMonthNameParser {
fn name(&self) -> &'static str {
"UKMonthNameParser"
}
fn should_apply(&self, _context: &ParsingContext) -> bool {
true
}
fn parse(&self, context: &ParsingContext) -> Result<Vec<ParsedResult>> {
let mut results = Vec::new();
let ref_date = context.reference.instant;
let mut start = 0;
while start < context.text.len() {
let search_text = &context.text[start..];
let captures = match DOT_PATTERN.captures(search_text) {
Ok(Some(caps)) => caps,
Ok(None) => break,
Err(_) => break,
};
let full_match = match captures.get(0) {
Some(m) => m,
None => break,
};
let match_start = start + full_match.start();
let match_end = start + full_match.end();
let day: i32 = captures
.name("day")
.and_then(|m| m.as_str().parse().ok())
.unwrap_or(0);
let month: i32 = captures
.name("month")
.and_then(|m| m.as_str().parse().ok())
.unwrap_or(0);
let year_str = captures.name("year").map(|m| m.as_str());
if !(1..=31).contains(&day) || !(1..=12).contains(&month) {
start = match_end;
continue;
}
let mut components = context.create_components();
if let Some(y) = year_str {
let mut year: i32 = y.parse().unwrap_or(ref_date.year());
if year < 100 {
year = if year > 50 { 1900 + year } else { 2000 + year };
}
components.assign(Component::Year, year);
} else {
components.imply(Component::Year, ref_date.year());
}
components.assign(Component::Month, month);
components.assign(Component::Day, day);
if components.is_valid_date() {
results.push(context.create_result(match_start, match_end, components, None));
}
start = match_end;
}
start = 0;
while start < context.text.len() {
let search_text = &context.text[start..];
let captures = match MONTH_NAME_PATTERN.captures(search_text) {
Ok(Some(caps)) => caps,
Ok(None) => break,
Err(_) => break,
};
let full_match = match captures.get(0) {
Some(m) => m,
None => break,
};
let match_start = start + full_match.start();
let match_end = start + full_match.end();
let overlaps = results.iter().any(|r| {
(match_start >= r.index && match_start < r.index + r.text.len())
|| (r.index >= match_start && r.index < match_end)
});
if overlaps {
start = match_end;
continue;
}
let weekday_str = captures.name("weekday").map(|m| m.as_str().to_lowercase());
let day: i32 = captures
.name("day")
.and_then(|m| m.as_str().parse().ok())
.unwrap_or(0);
let month_str = captures
.name("month")
.map(|m| m.as_str().to_lowercase())
.unwrap_or_default();
let year_str = captures.name("year").map(|m| m.as_str());
let end_day_str = captures.name("end_day").map(|m| m.as_str());
let clean_month = month_str.trim_end_matches('.');
let month = get_month(clean_month).unwrap_or(0);
if month == 0 || !(1..=31).contains(&day) {
start = match_end;
continue;
}
let mut components = context.create_components();
if let Some(y) = year_str {
let mut year: i32 = y.parse().unwrap_or(ref_date.year());
if year < 100 {
year = if year > 50 { 1900 + year } else { 2000 + year };
}
components.assign(Component::Year, year);
} else {
components.imply(Component::Year, ref_date.year());
}
components.assign(Component::Month, month as i32);
components.assign(Component::Day, day);
if let Some(ref wd_str) = weekday_str {
let clean_wd = wd_str.trim_end_matches('.').trim_end_matches(',');
if let Some(weekday) = get_weekday(clean_wd) {
components.assign(Component::Weekday, weekday as i32);
}
}
if !components.is_valid_date() {
start = match_end;
continue;
}
let end_components = if let Some(end_day_text) = end_day_str {
let end_day: i32 = end_day_text.parse().unwrap_or(0);
if end_day > 0 && end_day <= 31 {
let mut end_comp = context.create_components();
if let Some(start_year) = components.get(Component::Year) {
if year_str.is_some() {
end_comp.assign(Component::Year, start_year);
} else {
end_comp.imply(Component::Year, start_year);
}
}
end_comp.assign(Component::Month, month as i32);
end_comp.assign(Component::Day, end_day);
if end_comp.is_valid_date() {
Some(end_comp)
} else {
None
}
} else {
None
}
} else {
None
};
results.push(context.create_result(match_start, match_end, components, end_components));
start = match_end;
}
Ok(results)
}
}