bberry/src/core/parser.rs

use super::element::Element;

/// Parse a string input. Modifies the given `buffer` IN PLACE.
pub fn string_parser(buf: impl Iterator<Item = char>, buffer: &mut String) {
    let mut is_parsing_string: bool = false;
    let mut escape: bool = false;

    for char in buf {
        // string parser
        if is_parsing_string {
            if escape {
                buffer.push(char);
                escape = false;
                continue;
            }

            if char == '\\' {
                escape = true;
                continue;
            } else if char == '"' {
                break;
            }

            buffer.push(char);
        } else if char == ' ' {
            continue;
        }

        if char == '"' {
            // start string parser
            is_parsing_string = true;
        }
    }
}

/// Parse an attribute block.
///
/// # Returns
/// `(key, value)`
pub fn attr_parser(buf: &str) -> (String, String) {
    let mut key: String = String::new();
    let mut value: String = String::new();
    let mut buffer: String = String::new();
    let mut finished_key: bool = false;

    for _ in 0..2 {
        string_parser(
            if finished_key {
                // skip quote + key + quote chars to get to AFTER the first string
                buf.chars().skip(key.len() + 2)
            } else {
                // we haven't done the key yet, so let's just select from the start
                buf.chars().skip(0)
            },
            &mut buffer,
        );

        // at this point, `buffer` should contain a string since `string_parser` is blocking
        // until it fully parses a string
        if finished_key {
            value = buffer.clone();
        } else {
            key = buffer.clone();
            finished_key = true;
        }

        buffer.clear();
    }

    (key, value)
}

/// A parser for parsing an element expression. The element parser **only** parsers
/// element statements (stuff in parenthesis).
pub fn expr_parser(buf: &str) -> Element {
    let mut element: Element = Element::default();

    // state
    let mut skip_for: usize = 0;
    let mut finished_parsing_tag: bool = false;

    for (i, char) in buf.chars().enumerate() {
        if skip_for > 0 {
            // skip char
            skip_for -= 1;
            continue;
        }

        // parse tag first as it is the first thing
        if !finished_parsing_tag {
            if (char == ' ') | (char == '\n') | (char == '"') {
                finished_parsing_tag = true;
                continue;
            }

            if char != '(' {
                element.tag.push(char);
            }

            continue;
        }

        // special elements
        if (element.tag == "attr") | (element.tag == ":") | (element.tag.is_empty()) {
            let mut chars = (&buf[i..buf.len()]).to_string();

            if element.tag.is_empty() {
                chars.insert_str(0, "\"");
            }

            // parse
            let (k, v) = attr_parser(&chars);
            element.attrs.insert(k, v);
            element.tag = "attr".to_string();
            return element;
        } else if (element.tag == "text") | (element.tag.ends_with("'")) {
            let mut buffer: String = String::new();

            string_parser((&buf[i..buf.len()]).chars(), &mut buffer);
            if element.tag != "text" {
                // allows us to write `(h1' "Hello, world!")` instead of `(h1 [(text "Hello, world!")])`
                element.tag = element.tag.replace("'", "");

                let mut text_element = Element::default();
                text_element.tag = "text".to_string();
                text_element.attrs.insert("content".to_string(), buffer);

                element.children.push(text_element);
            } else {
                // just add attr
                element.attrs.insert("content".to_string(), buffer);
            }

            return element;
        }

        // everything else
        if char == '(' {
            // parse from here to the end (including the paren)
            let (element_, used) = element_parser(&buf[i..buf.len()]);
            skip_for = used;

            if element_.tag == "attr" {
                // element just holds attributes... merge with list
                for attr in element_.attrs {
                    element.attrs.insert(attr.0, attr.1);
                }
            } else {
                // element is a child
                element.children.push(element_);
            }
        }
    }

    // ...
    element
}

/// Parse a single element.
///
/// # Returns
/// `(element, number of characters consumed)`
pub fn element_parser(value: &str) -> (Element, usize) {
    let mut element: Element = Element::default();
    let mut consumed: usize = 0;

    // parser state
    let mut scope: i32 = 0; // the scope tells us when we can stop parsing junk...
    // basically, we stop parsing junk when the scope reaches the same scope we started
    // with when we began parsing junk
    let mut return_to_scope: i32 = 0;

    let mut buffer: String = String::new();
    let mut building_buffer_so_ignore_junk: bool = false;
    let mut comment: bool = false;

    for char in value.chars() {
        if comment {
            if char == '\n' {
                comment = false;
            }

            continue;
        }

        if building_buffer_so_ignore_junk {
            // everything here can be assumed to be junk to be parsed later
            if char == ')' {
                // if we're back at the scope we started when we first opened the block,
                // don't stop parsing as junk
                scope -= 1;
                if scope == return_to_scope {
                    // building_buffer_so_ignore_junk = false;

                    // parse actual stuff
                    consumed += buffer.len();
                    element = expr_parser(&buffer);

                    // reset the buffer and return first element IMMEDIATELY
                    // this allows expr_parser to continue parsing the next
                    // expression by skipping how many characters we parsed here
                    buffer.clear();
                    return (element, consumed);
                } else {
                    buffer.push(char);
                }
            } else {
                if char == '(' {
                    scope += 1;
                }

                buffer.push(char);
            }

            continue;
        }

        // everything here happens ONLY IF we aren't parsing any junk yet
        if char == ' ' {
            continue;
        }

        if char == '(' {
            // tell the parser that we're building a buffer to recursively parse,
            // so everything can currently be ignored
            scope += 1;

            if !building_buffer_so_ignore_junk {
                return_to_scope = scope - 1;
                building_buffer_so_ignore_junk = true;
            }
        }
    }

    // ...
    (element, consumed)
}

/// Parse a full document.
pub fn document(value: &str) -> Element {
    element_parser(&format!("(null? [{value}])")).0
}