mirror of
https://github.com/bspeice/dtparse
synced 2024-11-14 09:58:09 -05:00
Add ParserInfo struct
This commit is contained in:
parent
58e3b05b45
commit
5beff61366
146
src/lib.rs
146
src/lib.rs
@ -1,5 +1,8 @@
|
|||||||
extern crate chrono;
|
extern crate chrono;
|
||||||
|
|
||||||
|
use chrono::Datelike;
|
||||||
|
use chrono::Local;
|
||||||
|
use std::collections::HashMap;
|
||||||
use std::vec::Vec;
|
use std::vec::Vec;
|
||||||
|
|
||||||
#[cfg(test)]
|
#[cfg(test)]
|
||||||
@ -216,3 +219,146 @@ pub fn tokenize(parse_string: &str) -> Vec<Token> {
|
|||||||
let tokenizer = Tokenizer::new(parse_string.to_owned());
|
let tokenizer = Tokenizer::new(parse_string.to_owned());
|
||||||
tokenizer.collect()
|
tokenizer.collect()
|
||||||
}
|
}
|
||||||
|
|
||||||
|
fn parse_info(vec: Vec<Vec<&str>>) -> HashMap<String, usize> {
|
||||||
|
let mut m = HashMap::new();
|
||||||
|
|
||||||
|
if vec.len() == 1 {
|
||||||
|
for (i, val) in vec.get(0).unwrap().into_iter().enumerate() {
|
||||||
|
m.insert(val.to_lowercase(), i);
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
for (i, val_vec) in vec.into_iter().enumerate() {
|
||||||
|
for val in val_vec.into_iter() {
|
||||||
|
m.insert(val.to_lowercase(), i);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
m
|
||||||
|
}
|
||||||
|
|
||||||
|
struct ParserInfo {
|
||||||
|
jump: HashMap<String, usize>,
|
||||||
|
weekday: HashMap<String, usize>,
|
||||||
|
months: HashMap<String, usize>,
|
||||||
|
hms: HashMap<String, usize>,
|
||||||
|
ampm: HashMap<String, usize>,
|
||||||
|
utczone: HashMap<String, usize>,
|
||||||
|
pertain: HashMap<String, usize>,
|
||||||
|
tzoffset: HashMap<String, usize>,
|
||||||
|
dayfirst: bool,
|
||||||
|
yearfirst: bool,
|
||||||
|
year: u32,
|
||||||
|
century: u32,
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
impl Default for ParserInfo {
|
||||||
|
fn default() -> Self {
|
||||||
|
let year = Local::now().year();
|
||||||
|
let century = year / 100 * 100;
|
||||||
|
|
||||||
|
ParserInfo {
|
||||||
|
jump: parse_info(vec![vec![
|
||||||
|
" ", ".", ",", ";", "-", "/", "'",
|
||||||
|
"at", "on", "and", "ad", "m", "t", "of",
|
||||||
|
"st", "nd", "rd", "th"
|
||||||
|
]]),
|
||||||
|
weekday: parse_info(vec![
|
||||||
|
vec!["Mon", "Monday"],
|
||||||
|
vec!["Tue", "Tues", "Tuesday"],
|
||||||
|
vec!["Wed", "Wednesday"],
|
||||||
|
vec!["Thu", "Thurs", "Thursday"],
|
||||||
|
vec!["Fri", "Friday"],
|
||||||
|
vec!["Sat", "Saturday"],
|
||||||
|
vec!["Sun", "Sunday"],
|
||||||
|
]),
|
||||||
|
months: parse_info(vec![
|
||||||
|
vec!["Jan", "January"],
|
||||||
|
vec!["Feb", "February"],
|
||||||
|
vec!["Mar", "March"],
|
||||||
|
vec!["Apr", "April"],
|
||||||
|
vec!["May"],
|
||||||
|
vec!["Jun", "June"],
|
||||||
|
vec!["Jul", "July"],
|
||||||
|
vec!["Aug", "August"],
|
||||||
|
vec!["Sep", "Sept", "September"],
|
||||||
|
vec!["Oct", "October"],
|
||||||
|
vec!["Nov", "November"],
|
||||||
|
vec!["Dec", "December"],
|
||||||
|
]),
|
||||||
|
hms: parse_info(vec![
|
||||||
|
vec!["h", "hour", "hours"],
|
||||||
|
vec!["m", "minute", "minutes"],
|
||||||
|
vec!["s", "second", "seconds"],
|
||||||
|
]),
|
||||||
|
ampm: parse_info(vec![
|
||||||
|
vec!["am", "a"],
|
||||||
|
vec!["pm", "p"],
|
||||||
|
]),
|
||||||
|
utczone: parse_info(vec![vec![
|
||||||
|
"UTC", "GMT", "Z"
|
||||||
|
]]),
|
||||||
|
pertain: parse_info(vec![vec!["of"]]),
|
||||||
|
tzoffset: parse_info(vec![vec![]]),
|
||||||
|
dayfirst: false,
|
||||||
|
yearfirst: false,
|
||||||
|
year: year as u32,
|
||||||
|
century: century as u32,
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl ParserInfo {
|
||||||
|
fn get_jump(&self, name: &str) -> bool {
|
||||||
|
self.jump.contains_key(&name.to_lowercase())
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_weekday(&self, name: &str) -> Option<usize> {
|
||||||
|
self.weekday.get(&name.to_lowercase()).cloned() // TODO: Why do I have to clone a primitive?
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_month(&self, name: &str) -> Option<usize> {
|
||||||
|
self.months.get(&name.to_lowercase()).map(|u| u + 1)
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_hms(&self, name: &str) -> Option<usize> {
|
||||||
|
self.hms.get(&name.to_lowercase()).cloned()
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_ampm(&self, name: &str) -> Option<usize> {
|
||||||
|
self.ampm.get(&name.to_lowercase()).cloned()
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_pertain(&self, name: &str) -> bool {
|
||||||
|
self.pertain.contains_key(&name.to_lowercase())
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_utczone(&self, name: &str) -> bool {
|
||||||
|
self.utczone.contains_key(&name.to_lowercase())
|
||||||
|
}
|
||||||
|
|
||||||
|
fn get_tzoffset(&self, name: &str) -> Option<usize> {
|
||||||
|
if self.utczone.contains_key(&name.to_lowercase()) {
|
||||||
|
Some(0)
|
||||||
|
} else {
|
||||||
|
self.tzoffset.get(&name.to_lowercase()).cloned()
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
fn convertyear(&self, year: u32, century_specified: bool) -> u32 {
|
||||||
|
let mut year = year;
|
||||||
|
|
||||||
|
if year < 100 && !century_specified {
|
||||||
|
year += self.century;
|
||||||
|
if year >= self.year + 50 {
|
||||||
|
year -= 100;
|
||||||
|
} else if year < self.year - 50 {
|
||||||
|
year += 100
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
year
|
||||||
|
}
|
||||||
|
}
|
Loading…
Reference in New Issue
Block a user