expry 0.4.0

#![allow(dead_code)]
#![cfg_attr(not(feature = "std"), no_std)]

use crate::termcolors::*;

use core::cmp::{min,max};
use core::fmt::Write;

#[cfg(not(feature = "std"))]
pub mod mystd {
    extern crate alloc;
    pub use alloc::format;
    pub use alloc::vec;
    pub use alloc::vec::Vec;
    pub use alloc::string::String;
    pub use alloc::boxed::Box;
    pub use alloc::string::ToString;
}

#[cfg(not(feature = "std"))]
use mystd::*;

pub struct ParserState<'b,T,E,Context=()>
where
T: PartialEq + core::fmt::Debug,
{
    token: Option<T>,
    token_count: usize, // token count
    token_pos: usize, // pos in input of current token
    #[allow(clippy::type_complexity)]
    tokenize: fn(reader: &mut &'b str, context: &'_ mut Context) -> Result<(T,usize), (E,usize,usize)>, // function returning a token and an offset of the token TO THE END (basically reader.len()), after whitespace processing, so better error messages can be generated). Same for errors
    pub reader: &'b str, // remaining string
    generate_backtrack_token_count: usize, // generate error if larger as this token count
    pub context: Context,
    depth_remaining: usize, // to avoid stack overflows
}
impl<'b,T,E,Context> core::fmt::Debug for ParserState<'b,T,E,Context>
where
T: core::fmt::Debug + PartialEq,
{
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        write!(f, "[{}={:?}@{:?} err_on={}]", self.token_count, self.token, self.token_pos, self.generate_backtrack_token_count)
    }
}

type TokenPos = (usize,usize); // from - to

#[derive(Debug)]
pub enum ParserStatus<E> {
    LookaheadBacktrack(),
    DepthLimitReached(),
    Error(E,TokenPos),
}

pub type ParserResult<T,E> = Result<T, ParserStatus<E>>;

#[derive(Copy,Clone)]
pub struct TokenInfo {
    token_count: usize,
    token_start_until_end_of_input: usize,
    token_end_until_end_of_input: usize,
}

impl TokenInfo {
    pub fn get_start(&self, length_of_whole_input: usize) -> usize {
        length_of_whole_input - self.token_start_until_end_of_input
    }
    pub fn get_end(&self, length_of_whole_input: usize) -> usize {
        length_of_whole_input - self.token_end_until_end_of_input
    }
    pub fn start_to_end_of_input(&self) -> usize {
        self.token_start_until_end_of_input
    }
    pub fn end_to_end_of_input(&self) -> usize {
        self.token_end_until_end_of_input
    }
    pub fn get_length(&self) -> usize {
        self.token_start_until_end_of_input - self.token_end_until_end_of_input
    }
    // useful if there is a more specific position inside the token to point an error to
    #[must_use]
    pub fn bound(&self, start: usize, end: usize) -> TokenInfo {
        TokenInfo {
            token_start_until_end_of_input: self.token_end_until_end_of_input + start,
            token_end_until_end_of_input: self.token_end_until_end_of_input + end,
            .. *self
        }
    }
}

type ParserPath<'b,T,E,S,Context> = fn (reader: &mut ParserState<'b,T,E,Context>) -> ParserResult<S,E>;
type ParserPathWithState<'b,T,E,S,Context,State> = fn (parser: &mut ParserState<'b,T,E,Context>, state: &mut State) -> ParserResult<S,E>;

impl<'b,T,E,Context> ParserState<'b,T,E,Context>
where
T: PartialEq + Copy + Clone + core::fmt::Debug,
E: Copy + Clone + core::fmt::Debug,
{
    pub fn repeat<P>(&mut self, mut path: P) -> ParserResult<(),E>
        where P: FnMut(&mut Self) -> ParserResult<bool,E>,
    {
        if self.depth_remaining == 0 {
            return Err(ParserStatus::DepthLimitReached());
        }
        self.depth_remaining -= 1;
        let getc = self.generate_backtrack_token_count;
        loop {
            self.generate_backtrack_token_count = self.token_count;
            match path(self) {
                Ok(false) => {
                    break;
                },
                Ok(_) => {
                    // if no progress can be made, stop
                    if self.token_count <= self.generate_backtrack_token_count {
                        break;
                    }
                },
                Err(ParserStatus::LookaheadBacktrack()) => {
                    break;
                }
                Err(e) => {
                    self.depth_remaining += 1;
                    return Err(e);
                }
            }
        }
        self.depth_remaining += 1;
        self.generate_backtrack_token_count = getc;
        Ok(())
    }

    pub fn opt<S, P>(&mut self, path: P) -> ParserResult<Option<S>,E>
        where P: FnOnce(&mut Self) -> ParserResult<S,E>,
    {
        let getc = self.generate_backtrack_token_count;
        self.generate_backtrack_token_count = self.token_count;
        match path(self) {
            Ok(v) => {
                Ok(Some(v))
            },
            Err(ParserStatus::LookaheadBacktrack()) => {
                self.generate_backtrack_token_count = getc;
                Ok(None)
            },
            Err(e) => {
                Err(e)
            },
        }
    }

    pub fn choose<S>(&mut self, paths: &[ParserPath<'b,T,E,S,Context>], err: fn () -> E) -> ParserResult<S,E> {
        debug_assert!(!paths.is_empty());
        if self.depth_remaining == 0 {
            return Err(ParserStatus::DepthLimitReached());
        }
        self.depth_remaining -= 1;
        let getc = self.generate_backtrack_token_count;
        self.generate_backtrack_token_count = self.token_count;
        for p in paths {
            match p(self) {
                Err(ParserStatus::LookaheadBacktrack()) => {
                },
                // check if no tokens are accepted, if so continue with next candidate
                Ok(_) if self.token_count <= self.generate_backtrack_token_count => {
                    // can happen when there is a rule that only has a repeat inside
                },
                v => {
                    self.depth_remaining += 1;
                    return v;
                },
            }
        }
        self.depth_remaining += 1;
        self.generate_backtrack_token_count = getc;
        if self.generate_backtrack_token_count == self.token_count {
            return Err(ParserStatus::LookaheadBacktrack());
        }
        Err(ParserStatus::Error(err(), (self.token_pos, self.reader.len())))
    }

    pub fn choose_with_state<S,State>(&mut self, paths: &[ParserPathWithState<'b,T,E,S,Context,State>], state: &mut State, err: fn () -> E) -> ParserResult<S,E> {
        debug_assert!(!paths.is_empty());
        if self.depth_remaining == 0 {
            return Err(ParserStatus::DepthLimitReached());
        }
        self.depth_remaining -= 1;
        let getc = self.generate_backtrack_token_count;
        self.generate_backtrack_token_count = self.token_count;
        for p in paths {
            match p(self, state) {
                Err(ParserStatus::LookaheadBacktrack()) => {
                },
                // check if no tokens are accepted, if so continue with next candidate
                Ok(_) if self.token_count <= self.generate_backtrack_token_count => {
                    // can happen when there is a rule that only has a repeat inside
                },
                v => {
                    self.depth_remaining += 1;
                    return v;
                },
            }
        }
        self.depth_remaining += 1;
        self.generate_backtrack_token_count = getc;
        if self.generate_backtrack_token_count == self.token_count {
            return Err(ParserStatus::LookaheadBacktrack());
        }
        Err(ParserStatus::Error(err(), (self.token_pos, self.reader.len())))
    }

    /// Useful for recursive functions (like for parsing negation and the unary minus).
    pub fn call<S>(&mut self, path: ParserPath<'b,T,E,S,Context>) -> ParserResult<S,E> {
        if self.depth_remaining == 0 {
            return Err(ParserStatus::DepthLimitReached());
        }
        self.depth_remaining -= 1;
        let result = path(self);
        self.depth_remaining += 1;
        result
    }

    fn _peek(&mut self) -> Result<T,ParserStatus<E>> {
        if let Some(x) = &self.token {
            return Ok(*x);
        }
        let (t,token_until_end) = (self.tokenize)(&mut self.reader, &mut self.context).map_err(|(x,error_start,error_end)| {
            self.token_pos = error_start;
            debug_assert!(self.token_pos >= self.reader.len());
            ParserStatus::Error(x, (error_start, error_end))
        })?;
        //eprintln!("token: {:?}", t);
        self.token = Some(t);
        self.token_pos = token_until_end;
        debug_assert!(self.token_pos >= self.reader.len());
        Ok(t)
    }

    fn _token_info(&self) -> TokenInfo {
        TokenInfo {
            token_count: self.token_count,
            token_start_until_end_of_input: self.token_pos,
            token_end_until_end_of_input: self.reader.len(),
        }
    }

    fn _next(&mut self) -> TokenInfo {
        let info = self._token_info();
        self.token = None;
        self.token_count += 1;
        info
    }

    // returns token pos in input, `info` is extra hint that is related
    pub fn accept<F>(&mut self, expected: T, info: Option<&TokenInfo>, err: F) -> ParserResult<TokenInfo,E>
        where F: Fn () -> E,
    {
        match self._peek()? {
            t if expected == t => {
                Ok(self._next())
            },
            _ if self.generate_backtrack_token_count == self.token_count => {
                Err(ParserStatus::LookaheadBacktrack())
            },
            _ => {
                //eprintln!("current token: {:?}", self.token);
                let (start, end) = if let Some(TokenInfo{token_count: _, token_start_until_end_of_input, token_end_until_end_of_input}) = info { (*token_start_until_end_of_input, *token_end_until_end_of_input) } else { (self.token_pos, self.reader.len()) };
                Err(ParserStatus::Error(err(), (start, end)))
            }
        }
    }
    
    pub fn get(&mut self) -> ParserResult<(T,TokenInfo),E> {
        let retval = self._peek()?;
        Ok((retval, self._next()))
    }

    pub fn undo_get(&mut self, token: T, info: TokenInfo) {
        debug_assert!(self.token.is_none());
        self.token = Some(token);
        self.token_count = info.token_count;
    }

    // generate an error directly related with the current token (possibly triggering a backtrace)
    pub fn error_token<F>(&mut self, token: T, info: TokenInfo, err: F) -> ParserStatus<E>
        where F: FnOnce(&mut Self) -> E,
    {
        if self.generate_backtrack_token_count == info.token_count {
            self.token = Some(token);
            self.token_count = info.token_count;
            ParserStatus::LookaheadBacktrack()
        } else {
            ParserStatus::Error(err(self), (info.token_start_until_end_of_input, info.token_end_until_end_of_input))
        }
    }

    // generate an error not related with the current token (so it will not backtrack)
    pub fn error_other(&mut self, info: &TokenInfo, err: E) -> ParserStatus<E> {
        ParserStatus::Error(err, (info.token_start_until_end_of_input, info.token_end_until_end_of_input))
    }

    pub fn token_info(&mut self) -> ParserResult<TokenInfo,E> {
        self._peek()?;
        Ok(self._token_info())
    }

    pub fn context(&mut self) -> &mut Context {
        &mut self.context
    }

    pub fn consume(self) -> Context {
        self.context
    }

    #[allow(clippy::type_complexity)]
    pub fn new_with(reader: &'b str, tokenize: fn(reader: &mut &'b str, context: &'_ mut Context) -> Result<(T,usize), (E,usize,usize)>, context: Context) -> Self {
        const MAX_DEPTH: usize = 128;
        Self {
            token: None,
            token_count: 0,
            token_pos: 0,
            reader,
            tokenize,
            generate_backtrack_token_count: usize::MAX,
            context,
            depth_remaining: MAX_DEPTH,
        }
    }

    #[allow(clippy::type_complexity)]
    pub fn parse<R>(&mut self, f: fn(&mut Self) -> ParserResult<R,E>, unexpected_token: E, depth_limit_reached: E) -> Result<R,(E,usize,usize,Option<(usize,usize)>)> {
        f(self).map_err(|x| {
            //eprintln!("error: {:?} on token {:?}", x, self);
            match x {
                ParserStatus::LookaheadBacktrack() => {
                    (unexpected_token, self.token_pos, self.reader.len(), None)
                },
                ParserStatus::Error(err, token_info) => {
                    let extra = if self.token_pos != token_info.0 {
                        Some((self.token_pos, self.reader.len()))
                    } else {
                        None
                    };
                    (err, token_info.0, token_info.1, extra)
                }
                ParserStatus::DepthLimitReached() => {
                    (depth_limit_reached, self.token_pos, self.reader.len(), None)
                },
            }
        })
    }
}

pub trait Spanner {
    fn next(&mut self, v: char) -> bool;
    fn valid(&mut self, _len: usize) -> bool {
        true
    }
    fn span<'b>(&mut self, reader: &mut &'b str) -> Option<&'b str> {
        let index = reader.char_indices().find(|(_pos, c)| !self.next(*c)).map(|(pos, _c)| pos).unwrap_or(reader.len());
        if !self.valid(index) {
            return None;
        }
        let value = &reader[0..index];
        *reader = &reader[index..];
        Some(value)
    }
}

pub struct StringLiteralSpanner {
    first: bool,
    end: bool,
    prev: char,
    pub unescape_needed: bool,
    delim: char,
}

impl Spanner for StringLiteralSpanner {
    fn next(&mut self, b: char) -> bool {
        if self.end {
            return false;
        }
        if self.first {
            self.first = false;
            return b == self.delim;
        }
        if b == self.delim && self.prev != '\\' {
            self.end = true;
            return true;
        }
        if self.prev == '\\' {
            self.prev = 0 as char;
            self.unescape_needed = true;
        } else {
            self.prev = b;
        }
        true
    }
    fn valid(&mut self, _len: usize) -> bool {
        self.end
    }
}
impl StringLiteralSpanner {
    pub fn new(delim: char) -> Self {
        Self {
            first: true,
            end: false,
            prev: 0 as char,
            unescape_needed: false,
            delim,
        }
    }
}

pub struct NumberSpanner {
    first: bool,
    prev: char,
    pub float: bool,
    has_digits: bool,
}

impl Spanner for NumberSpanner {
    fn next(&mut self, b: char) -> bool {
        if self.first && !(b.is_ascii_digit() || b == '-') {
            return false;
        }
        if !self.first && !(b.is_ascii_digit() || b == '.' || b == 'e' || b == 'E' || ((self.prev == 'e' || self.prev == 'E') && (b == '+' || b == '-'))) {
            return false;
        }
        self.first = false;
        self.float = self.float || b == '.' || b == 'e' || b == 'E';
        self.has_digits = self.has_digits || b.is_ascii_digit();
        self.prev = b;
        true
    }
    fn valid(&mut self, _len: usize) -> bool {
        self.has_digits
    }
}

impl NumberSpanner {
    pub fn new() -> Self {
        Self {
            first: true,
            prev: 0 as char,
            float: false,
            has_digits: false,
        }
    }
}

impl Default for NumberSpanner {
    fn default() -> Self {
        Self::new()
    }
}

// INCLUDE_OUTER_TAGS=false: match all contents of {{...}}, so matches up until a '}' is seen
// INCLUDE_OUTER+TAGS=true: match anything between '{' and '}' (supports nesting of '{'/'}' and
// strings)
pub struct HairyTemplateTagContentSpanner<const INCLUDE_OUTER_TAGS: bool> {
    prev_was_escape: bool,
    string: bool,
    nested: u32,
    len: usize,
}

impl<const INCLUDE_OUTER_TAGS: bool> Spanner for HairyTemplateTagContentSpanner<INCLUDE_OUTER_TAGS> {
    fn next(&mut self, b: char) -> bool {
        if INCLUDE_OUTER_TAGS && self.nested == 0 && !self.string && self.len > 0 {
            return false;
        }
        if !self.prev_was_escape && !self.string && b == '{' {
            self.nested += 1;
        }
        if !self.prev_was_escape && !self.string && b == '}' {
            if self.nested == 0 {
                return false;
            }
            self.nested -= 1;
        }
        if !self.prev_was_escape && b == '"' {
            self.string = !self.string;
        }
        self.prev_was_escape = !self.prev_was_escape && b == '\\';
        self.len += 1;
        true
    }
    fn valid(&mut self, _len: usize) -> bool {
        self.nested == 0 && !self.string && !self.prev_was_escape
    }
}
impl<const INCLUDE_OUTER_TAGS: bool> HairyTemplateTagContentSpanner<INCLUDE_OUTER_TAGS> {
    pub fn new() -> Self {
        Self {
            prev_was_escape: false,
            string: false,
            nested: 0,
            len: 0,
        }
    }
}

impl<const INCLUDE_OUTER_TAGS: bool> Default for HairyTemplateTagContentSpanner<INCLUDE_OUTER_TAGS> {
    fn default() -> Self {
        Self::new()
    }
}
pub trait Acceptor<'a> {
    fn accept(self, expected: &str) -> bool;
    fn span_fn<M>(self, matcher: &mut M) -> Option<&'a str>
    where
        M: FnMut(char) -> bool;
    fn span<P: Spanner>(self, spanner: &mut P) -> Option<&'a str>;
}
impl<'a> Acceptor<'a> for &mut &'a str {
    fn accept(self, expected: &str) -> bool {
        if let Some(remaining) = self.strip_prefix(expected) {
            *self = remaining;
            true
        } else {
            false
        }
    }

    // missing from the standard api
    fn span_fn<M>(self, matcher: &mut M) -> Option<&'a str>
    where
        M: FnMut(char) -> bool,
    {
        let index = self.char_indices().find(|(_pos, c)| !matcher(*c)).map(|(pos, _c)| pos).unwrap_or(self.len());
        if index == 0 {
            return None;
        }
        let value = &self[0..index];
        *self = &self[index..];
        Some(value)
    }
    fn span<P: Spanner>(self, spanner: &mut P) -> Option<&'a str> {
        spanner.span(self)
    }
}

#[derive(Debug, Clone)]
pub struct LineContext {
    offsets: Vec<u32>,
}

impl LineContext {
    pub fn empty() -> Self {
        Self { offsets: Vec::new(), }
    }
    pub fn new(reader: &str) -> Self {
        let mut offsets : Vec<u32> = Vec::new();
        offsets.push(0);
        for (i,c) in reader.char_indices() {
            if c == '\n' {
                offsets.push(i as u32 + 1);
            }
        }
        offsets.push(reader.len() as u32 + 1);
        Self {
            offsets
        }
    }
    pub fn format_error_context_short(&self, reader: &str, start: usize, end: usize) -> Result<String,core::fmt::Error> {
        // convert from distance to end to position from start
        let (start, end) = (reader.len() - start, reader.len() - end);
        let (_, _, line_start, line_end) = self.position_to_line_info(start as u32);
        let line = &reader[line_start as usize..line_end as usize];
        let (start_in_line, end_in_line) = (start - line_start as usize, max(start+1, min(end, line_end as usize)) - line_start as usize);

        const UNDERLINE : &str = "↑"; //▔↑⬆️

        let mut retval = String::new();

        let mut highlight = false;
        for (i,c) in line.char_indices() {
            if i >= end_in_line {
                if highlight {
                    write!(retval, "")?;
                    highlight = false;
                }
            } else if i >= start_in_line && !highlight {
                write!(retval, "")?;
                highlight = true;
            }
            write!(retval, "{}", c)?;
        }
        writeln!(retval)?;

        let mut pos = 0;
        for (i,c) in line.char_indices() {
            pos = i+1;
            if i >= end_in_line {
                break;
            } else if i >= start_in_line {
                write!(retval, "{}", UNDERLINE)?;
            } else if c == '\t' {
                write!(retval, "\t")?;
            } else {
                write!(retval, " ")?;
            }
        }
        for _ in pos..end_in_line {
            write!(retval, "{}", UNDERLINE)?;
        }
        Ok(retval)
    }
    pub fn format_error_context_html(&self, reader: &str, start: usize, end: usize, extra_line_no: u32) -> Result<(u32, String),core::fmt::Error> {
        // convert from distance to end to position from start
        let (start, end) = (reader.len() - start, reader.len() - end);
        let (line_no, _, line_start, line_end) = self.position_to_line_info(start as u32);
        let line = &reader[line_start as usize..line_end as usize];
        let (start_in_line, end_in_line) = (start - line_start as usize, max(start+1, min(end, line_end as usize)) - line_start as usize);

        let line_no_text = format!("{}", line_no+extra_line_no);
        let mut retval = String::new();

        write!(retval, "<code><span class=\"lineno\">{}</span>", line_no_text)?;
        let mut highlight = false;
        for (i,c) in line.char_indices() {
            if i >= end_in_line {
                if highlight {
                    write!(retval, "</span>")?;
                    highlight = false;
                }
            } else if i >= start_in_line && !highlight {
                write!(retval, "<span class=\"highlight\">")?;
                highlight = true;
            }
            write!(retval, "{}", c)?;
        }
        if highlight {
            write!(retval, "</span>")?;
        }
        write!(retval, "</code>")?;
        Ok((line_no+extra_line_no, retval))
    }
    pub fn format_error_context_console(&self, reader: &str, start: usize, end: usize, extra_line_no: u32) -> Result<(u32, String,String),core::fmt::Error> {
        // convert from distance to end to position from start
        let (start, end) = (reader.len() - start, reader.len() - end);
        let (line_no, _, line_start, line_end) = self.position_to_line_info(start as u32);
        let line = &reader[line_start as usize..line_end as usize];
        let (start_in_line, end_in_line) = (start - line_start as usize, max(start+1, min(end, line_end as usize)) - line_start as usize);

        const DELIM : &str = " │ "; // │ | "; //⸾⌇
        const DELIM_ALT : &str = " ┿ "; // ┥ ┤│ | "; //⸾⌇
        const DELIM_ALT2 : &str = " ├ "; // │ | "; //⸾⌇
        const UNDERLINE : &str = "▔"; //↑

        let line_no_text = format!("{:>3}", line_no+extra_line_no);
        let prefix = format!("{}{TERM_BRIGHT_BLACK}{}{TERM_RESET}", [' '].iter().cycle().take(line_no_text.len()).collect::<String>(), DELIM);
        let mut retval = String::new();

        write!(retval, "{TERM_BRIGHT_BLACK}{}{}{TERM_RESET}", line_no_text, DELIM_ALT)?;
        let mut highlight = false;
        for (i,c) in line.char_indices() {
            if i >= end_in_line {
                if highlight {
                    write!(retval, "{TERM_RESET}{TERM_DIM_DEFAULT}")?;
                    highlight = false;
                }
            } else if i >= start_in_line && !highlight {
                write!(retval, "{TERM_BRIGHT_YELLOW}")?;
                highlight = true;
            }
            write!(retval, "{}", c)?;
        }
        writeln!(retval, "{TERM_RESET}")?;

        write!(retval, "{}{TERM_BRIGHT_RED}", prefix)?;
        let mut pos = 0;
        for (i,c) in line.char_indices() {
            pos = i+1;
            if i >= end_in_line {
                break;
            } else if i >= start_in_line {
                write!(retval, "{}", UNDERLINE)?;
            } else if c == '\t' {
                write!(retval, "\t")?;
            } else {
                write!(retval, " ")?;
            }
        }
        for _ in pos..end_in_line {
            write!(retval, "{}", UNDERLINE)?;
        }
        writeln!(retval, "{TERM_RESET}")?;
        Ok((line_no+extra_line_no, prefix, retval))
    }
    pub fn remaining_to_line_info(&self, remaining: u32) -> (u32, u32, u32, u32) {
        if let Some(last) = self.offsets.last() {
            self.position_to_line_info(last - 1 - remaining)
        } else {
            (0, 0, 0, 0)
        }
    }
    // returns (line_no, column_no, offset_start_of_line, offset_end_of_line)
    pub fn position_to_line_info(&self, pos: u32) -> (u32, u32, u32, u32) {
        let i = self.offsets.partition_point(|x| *x <= pos);
        if i > 0 && i < self.offsets.len() {
            let start_of_this_line = self.offsets[i-1];
            let start_of_next_line = self.offsets[i];
            return (i as u32, pos - start_of_this_line, start_of_this_line, start_of_next_line-1) // +1 because line numbers start at 1, and +1 in begin position to skip enter
        }
        (0, 0, 0, 0) // should not occur because line_no_offsets always should have a last extra entry
    }

}


#[cfg(test)]
mod tests {
    use crate::*;
    #[test]
    fn number_spanner() {
        let mut spanner = NumberSpanner::new();
        let mut reader = "3.14";
        assert!(spanner.span(&mut reader).is_some());
        assert!(spanner.float);
    }
}