use crate::components::Component;
use crate::context::ParsingContext;
use crate::dictionaries::nl::{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 PATTERN: LazyLock<Regex> = LazyLock::new(|| {
Regex::new(
r"(?ix)
(?:
(?P<weekday>zondag|zon|zo|maandag|maan|ma|dinsdag|dins|di|woensdag|woens|wo|donderdag|donder|do|vrijdag|vrij|vr|zaterdag|zater|za)
(?:\s*,?\s*)?
)?
(?P<day>\d{1,2})(?:e|ste|de)?\s*
(?:
(?:(?:tot|\-|–)\s*(?P<end_day>\d{1,2})(?:e|ste|de)?\s*)?
)?
(?P<month>januari|jan\.?|februari|feb\.?|maart|mrt\.?|april|apr\.?|mei|juni|jun\.?|juli|jul\.?|augustus|aug\.?|september|sep\.?|sept\.?|oktober|okt\.?|november|nov\.?|december|dec\.?)
(?:
(?:\s*[\-/,]?\s*)?
(?P<year>\d{1,4}(?!:))?
(?:\s*(?P<era>
(?:voor|v\.?)\s*(?:Christus|Chr\.?)
|(?:na|n\.?)\s*(?:Christus|Chr\.?)
))?
)?
(?=\W|$)
"
).unwrap()
});
static ABBREV_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
Regex::new(
r"(?ix)(?P<weekday>zondag|zon|zo|maandag|maan|ma|dinsdag|dins|di|woensdag|woens|wo|donderdag|donder|do|vrijdag|vrij|vr|zaterdag|zater|za)\.?\s+(?P<day>\d{1,2})(?:e|ste|de)?\s+(?P<month>januari|jan\.?|februari|feb\.?|maart|mrt\.?|april|apr\.?|mei|juni|jun\.?|juli|jul\.?|augustus|aug\.?|september|sep\.?|sept\.?|oktober|okt\.?|november|nov\.?|december|dec\.?)(?:\s*(?P<year>\d{2,4}))?(?=\W|$)"
).unwrap()
});
pub struct NLMonthNameParser;
impl NLMonthNameParser {
pub fn new() -> Self {
Self
}
fn parse_year_with_era(year_str: Option<&str>, era_str: Option<&str>) -> Option<i32> {
let year_text = year_str?;
let mut year: i32 = year_text.parse().ok()?;
if year < 100 && era_str.is_none() {
year = if year > 50 { 1900 + year } else { 2000 + year };
}
if let Some(era) = era_str {
let era_lower = era.to_lowercase();
if era_lower.contains("voor") || era_lower.starts_with('v') {
return Some(-year);
}
}
Some(year)
}
}
impl Default for NLMonthNameParser {
fn default() -> Self {
Self::new()
}
}
impl Parser for NLMonthNameParser {
fn name(&self) -> &'static str {
"NLMonthNameParser"
}
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 ABBREV_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 weekday_str = captures.name("weekday").map(|m| m.as_str().to_lowercase());
let day_str = captures.name("day").map(|m| m.as_str()).unwrap_or("1");
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 month_clean = month_str.trim_end_matches('.');
let Some(month) = get_month(month_clean) else {
start = match_end;
continue;
};
let day: i32 = day_str.parse().unwrap_or(1);
if !(1..=31).contains(&day) {
start = match_end;
continue;
}
let mut components = context.create_components();
if let Some(y) = year_str {
let year = Self::parse_year_with_era(Some(y), None).unwrap_or(ref_date.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 Some(weekday) = get_weekday(wd_str)
{
components.assign(Component::Weekday, weekday as i32);
}
if !components.is_valid_date() {
start = match_end;
continue;
}
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 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 weekday_str = captures.name("weekday").map(|m| m.as_str().to_lowercase());
let day_str = captures.name("day").map(|m| m.as_str()).unwrap_or("1");
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 era_str = captures.name("era").map(|m| m.as_str());
let end_day_str = captures.name("end_day").map(|m| m.as_str());
let month_clean = month_str.trim_end_matches('.');
let Some(month) = get_month(month_clean) else {
start = match_end;
continue;
};
let day: i32 = day_str.parse().unwrap_or(1);
if !(1..=31).contains(&day) {
start = match_end;
continue;
}
let mut components = context.create_components();
if year_str.is_some() || era_str.is_some() {
let year = Self::parse_year_with_era(year_str, era_str).unwrap_or(ref_date.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 Some(weekday) = get_weekday(wd_str)
{
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() || era_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)
}
}