yyyi_ru/src/mtgott/mod.rs

mod charclasses;

use serde_json;
use std::collections::HashMap;
use charclasses::*;

struct CallExpression {
    callee: Option<Box<Expression>>,
    arguments: Vec<Box<Expression>>,
}

enum Expression {
    Root(),
    Argument(u64),
    Get(Box<Expression>, Box<Expression>),
    Attribute(Box<Expression>, String),
    Call(Box<Expression>, Box<Expression>),
    Int(u64),
}

struct IfSubElement {
    branches: Vec<Element>,
    conditions: Vec<Expression>
}

struct ForSubElement {
    iterable: Expression,
    hold_key: bool,
    hold_value: bool,
    core: Element,
    /* Either "\n", " " or "" */
    join: String,
}

enum SubElement{
    Static(String),
    /*   ======== Other are dynamic ======== */
    If(IfSubElement),
    /* Both for {{}} and {[]} */
    InsertExpr(Expression),
    For(ForSubElement),
    Let(Expression, Element),
}

struct Element {
    argc: usize,
    sub_elements: Vec<SubElement>
}

enum Plemege {
    Element(Element),
    Package(Box<HashMap<String, Plemege>>),
}

pub enum FileParsingErrorKind {
    expected_pack_opening_or_element_opening_or_pack_ending,
    expected_pack_opening_or_element_opening_or_eof,
    unmatched_pack_ending_tag,
    expected_pack_name,
    illegal_pack_name,
    pack_member_name_already_occupied,
    expected_pack_opening_tag_end,
    expected_element_name,
    illegal_element_name,
    expected_argument_name_or_eldef_opening_tag_end,
    illegal_argument_name,
    repeated_argument_name,
    expected_command_name,
    incorrect_block_ending_tag_expected_normal,
    expected_write_tag_end_after_expression,
    expected_roughinsert_tag_end_after_expression,
    illegal_command_name,
    expected_cmd_tag_end,
}

use FileParsingErrorKind::*;

pub struct FileParsingError {
    kind: FileParsingErrorKind,
    p1: usize,
    p2: usize,
}

impl FileParsingError {
    fn new(kind: FileParsingErrorKind, p1: usize, p2: usize) -> Self {
        Self{kind, p1, p2}
    }
}

struct Parser<'a> {
    text: &'a str,
    p: usize
}

impl Parser {
    fn here(&self)->Option<char> {
        self.text[self.p..].chars().next()
    }

    fn is_ahead(&self, substr: &[u8])->bool {
        self.text[self.p..].starts_with(substr)
    }

    fn advance(&mut self) {
        self.p += self.text[self.p..].char_indices().next().unwrap().0;
    }

    fn skip_whitespace(&mut self) {
        loop {
            match self.here() {
                Some(ch ) => if !is_whitespace(ch) {
                    break
                } else { self.advance(); }
                None => break
            }
        }
    }

    fn skip_normal_word(&mut self){
        loop {
            match self.here() {
                Some(ch ) => if !is_normal_word_constituent(ch) {
                    break
                } else { self.advance(); }
                None => break
            }
        }
    }

    fn new_unexpected_char_error(&self, kind: FileParsingErrorKind) -> FileParsingError {
        match self.text[self.p..].char_indices().next() {
            Some((off, _)) => FileParsingError::new(kind, self.p, self.p + off),
            None => FileParsingError::new(kind, self.p, self.p),
        }
    }

    fn parse_pack_plus_ending(&mut self, top: bool) -> Result<Plemege::Package, FileParsingError> {
        let mut res: HashMap<String, Plemege> = HashMap::new();
        loop {
            self.skip_whitespace();
            if self.p == self.text.len() {
                return if top {
                    Ok(Plemege::Package(Box::new(res)))
                } else {
                    Err(self.new_unexpected_char_error(expected_pack_opening_or_element_opening_or_pack_ending))
                }
            }
            if self.is_ahead(&[b'{', b'$', b'}']) {
                if top {
                    return Err(FileParsingError::new(unmatched_pack_ending_tag, self.p, self.p + 3))
                } else {
                    self.p += 3;
                    return Ok(res);
                }
            } else if self.is_ahead(&[b'{', b'$']) {
                self.p += 2;
                self.skip_whitespace();
                let p1 = self.p;
                self.skip_normal_word();
                if self.p == p1 {
                    return Err(self.new_error(expected_pack_name))
                }
                let child_name: &str = &self.text[p1..self.p];
                if !is_illegal_name(child_name) {
                    return Err(FileParsingError::new(illegal_pack_name, p1, self.p))
                }
                if let Some(_) = res.get(child_name) {
                    return Err(FileParsingError::new(pack_member_name_already_occupied, p1, self.p))
                }
                self.skip_normal_word();
                if !self.is_ahead(&[b'$', b'}']) {
                    return Err(self.new_unexpected_char_error(expected_pack_opening_tag_end))
                }
                self.p += 2;
                res.insert(String::from(child_name), self.parse_pack_plus_ending(false));
            } else if self.is_ahead(&[b'{', b'@']) {
                self.p += 2;
                self.skip_whitespace();
                let p1 = self.p;
                self.skip_normal_word();
                if p1 == self.p {
                    return Err(FileParsingError::new(expected_element_name, p1, self.p))
                }
                let child_name = &self.text[p1..self.p];
                if is_illegal_name(child_name) {
                    return Err(FileParsingError::new(illegal_element_name, p1, self.p))
                }
                if let Some(_) = res.get(child_name) {
                    return Err(FileParsingError::new(pack_member_name_already_occupied, p1, self.p))
                }
                let mut arg_names: Vec<&str> = Vec::new();
                loop {
                    self.skip_whitespace();
                    if self.is_ahead(&[b'@', b'}']) {
                        self.p += 2;
                        break
                    }
                    let p1 = self.p;
                    self.skip_normal_word();
                    if p1 == self.p {
                        return Err(FileParsingError::new(expected_argument_name_or_eldef_opening_tag_end, p1, self.p))
                    }
                    let arg_name: &str = &self.text[p1..self.p];
                    if is_illegal_name(arg_name) {
                        return Err(FileParsingError::new(illegal_argument_name, p1, self.p))
                    }
                    if arg_names.iter().any(|b: &str| b == arg_name) {
                        return Err(FileParsingError::new(repeated_argument_name, p1, self.p))
                    }
                    arg_names.push(arg_name);
                }
                let (child_el, end_cmd): (Element, ReasonOfElementEnd) = self.parse_element_plus_ending(arg_names)?;
                if end_cmd.cmd != BlockEndingCmdTag::NORMAL {
                     return Err(FileParsingError::new(incorrect_block_ending_tag_expected_normal, end_cmd.p1, self.p))
                }
                res.insert(child_name, child_el);
            } else {
                self.new_unexpected_char_error(if top {
                    expected_pack_opening_or_element_opening_or_eof
                } else {
                    expected_pack_opening_or_element_opening_or_pack_ending
                })
            }
        }
    }
}

enum BlockEndingCmdTag {
    NORMAL,
    LF,
    GAP,
    NOGAP,
    ENDLOOP,
    ELSE_IF,
    ELSE,
    ENDIF
}

struct ReasonOfElementEnd {
    p1: usize,
    cmd: BlockEndingCmdTag,
}

impl Parser {
    /* If BlockEndingCmdTag::ELSE_IF is returned, the ending tag won't be read completely,
     * But in other case it would be read to the end */
    fn parse_element_plus_ending_tag(&mut self, arg_names: Vec<&str>) -> Result<(Element, ReasonOfElementEnd), FileParsingError> {
        let mut res: Vec<SubElement> = Vec::new();
        let mut tp1 = self.p;

        let fin_static = || {
            if tp1 < self.p {
                res.push(SubElement::Static(String::from(&self.text[tp1..self.p])))
            }
        };

        /* Fixes whitespaces in  */
        let finishing_touches = |ree: ReasonOfElementEnd| -> Result<(Element, ReasonOfElementEnd), FileParsingError> {
            Ok((Element{ argc: arg_names.count(), sub_elements: res }, ree))
        };

        loop {
            if self.is_ahead(&[b'{', b'{']) {
                fin_static();
                self.p += 2;
                let (expr, tt) = self.parse_expression()?;
                if tt != ExpressionEndingTagEnd::Write {
                    return Err(FileParsingError::new(expected_write_tag_end_after_expression, self.p - 2, self.p))
                }
                res.push(SubElement::InsertExpr(
                    Expression::Call(
                        Box::new(Expression::Attribute(
                            Box::new(Expression::Root()), "sanitize"
                        )),
                        expr)
                ));
                tp1 = self.p;
            } else if self.is_ahead(&[b'{', b'[']) {
                fin_static();
                self.p += 2;
                let (expr, tt) = self.parse_expression()?;
                if tt != ExpressionEndingTagEnd::RoughInsert {
                    return Err(FileParsingError::new(expected_roughinsert_tag_end_after_expression, self.p - 2, self.p))
                }
                res.push(SubElement::InsertExpr(expr));
            } else if self.is_ahead(&[b'{', b'%', b'}']) {
                fin_static();
                self.p += 3;
                return finishing_touches(ReasonOfElementEnd{p1: self.p - 3, cmd: BlockEndingCmdTag::NORMAL});
            } else if self.is_ahead(&[b'{', b'%']) {
                fin_static();
                /* Might be needed if this is the ENDING cmd tag */
                let p1 = self.p;
                self.p += 2;
                self.skip_whitespace();
                let pb = self.p;
                self.skip_normal_word();
                if pb == self.p {
                    return Err(self.new_unexpected_char_error(expected_command_name))
                }
                let cmd = &self.text[pb..self.p];

                /* Read space + expect %} and do finishing_touches */
                let just_one_thing = |cmd: BlockEndingCmdTag| -> Result<(Element, ReasonOfElementEnd), FileParsingError> {
                    self.skip_whitespace();
                    if !self.is_ahead(&[b'%', b'}']) {
                        return self.new_unexpected_char_error(expected_cmd_tag_end);
                    }
                    self.p += 2;
                    finishing_touches(ReasonOfElementEnd{p1, cmd})
                };

                match cmd {
                    "lf" => return just_one_thing(BlockEndingCmdTag::LF),
                    "gap" => return just_one_thing(BlockEndingCmdTag::GAP),
                    "nogap" => return just_one_thing(BlockEndingCmdTag::NOGAP),
                    "else" => {
                        self.skip_whitespace();
                        let ps = self.p;
                        self.skip_normal_word();
                        if ps == self.p {
                            return just_one_thing(BlockEndingCmdTag::ELSE)
                        } else if self.text[ps..self.p] != "if" {
                            return Err(FileParsingError::new(illegal_command_name, pb, self.p))
                        }
                        return finishing_touches(ReasonOfElementEnd{p1, cmd: BlockEndingCmdTag::ELSE_IF})
                    }
                    "endif" => return just_one_thing(BlockEndingCmdTag::ENDIF),
                    "endloop" => return just_one_thing(BlockEndingCmdTag::ENDLOOP),
                    "for" => res.push(self.parse_let(&arg_names)?),
                    "if" => res.push(self.parse_if(&arg_names)?),
                    "let" => res.push(self.parse_let(&arg_names)?),
                    _ => return Err(FileParsingError::new(illegal_command_name, pb, self.p)),
                }
            } else {
                self.advance();
            }
        }
    }

    /* It turned out to be so complex I put it in a separate function.
     * It parses expr %} block {% else if expr %} block {% else %} block {%}   */
    fn parse_if(&mut self, arg_names: &Vec<&str>) -> Result<SubElement::If, FileParsingError> {
        // todo
    }

    fn parse_let(&mut self, arg_names: &Vec<&str>) -> Result<SubElement::Let, FileParsingError> {
        self.skip_whitespace();
        let p1 = self.p;
        self.skip_normal_word();
        if p1 == self.p {

        }
    }

    fn parse_for(&mut self, arg_names: &Vec<&str>) -> Result<SubElement::For, FileParsingError> {
        // todo
    }
}

enum ExpressionEndingTagEnd {
    Write, RoughInsert, Cmd,
}

impl Parser {
    fn parse_expression_plus_tag_end(&mut self) -> Result<(Expression, ExpressionEndingTagEnd), FileParsingError> {
        self.skip_whitespace();
        return Err(self.new_unexpected_char_error(expected_pack_name)) // todo
        // todo
    }
}

fn parse_one_file(text: &str) -> Result<Plemege::Package, FileParsingError> {
    let mut parser: Parser = Parser{text, p: 0};
    parser.parse_pack_plus_ending(true)
}

#[cfg(test)]
mod tests{
    use super::*;

    #[test]
    fn t1 () {

    }
}