vize_l1 0.653.0

mod compat;
mod literals;
mod regex;
mod script;

use alloc::borrow::Cow;
use memchr::memchr;
use vize_l0::FxHashMap;

pub(super) use super::{
    BlockAttrs, BlockEndSearch, BlockParseResult, advance_line, build_malformed_error,
    content_end_column, find_block_end, find_closing_tag_end, newlines,
};
use compat::is_void_block;
pub(super) use literals::{can_start_regex_literal, skip_script_string_literal};
pub(super) use regex::skip_regex_literal;
use script::find_script_block_end;

// Tag name bytes for fast comparison
pub(super) const TAG_TEMPLATE: &[u8] = b"template";
const TAG_SCRIPT: &[u8] = b"script";
const TAG_STYLE: &[u8] = b"style";

/// Fast tag name comparison using byte slices
#[inline(always)]
pub fn tag_name_eq(name: &[u8], expected: &[u8]) -> bool {
    name.len() == expected.len() && name.eq_ignore_ascii_case(expected)
}

/// Fast byte slice prefix check
#[inline(always)]
pub(super) fn starts_with_bytes(haystack: &[u8], needle: &[u8]) -> bool {
    haystack
        .get(..needle.len())
        .is_some_and(|prefix| prefix.eq_ignore_ascii_case(needle))
}

/// Fast tag name character check
#[inline(always)]
fn is_tag_name_char_fast(b: u8) -> bool {
    matches!(b, b'a'..=b'z' | b'A'..=b'Z' | b'0'..=b'9' | b'-' | b'_')
}

/// Fast whitespace check
#[inline(always)]
pub(super) fn is_whitespace_fast(b: u8) -> bool {
    matches!(b, b' ' | b'\t' | b'\n' | b'\r')
}

#[inline]
fn position_after(bytes: &[u8], line: usize, column: usize) -> (usize, usize) {
    let (count, last) = newlines(bytes);
    let end_column = last.map_or(column + bytes.len(), |offset| bytes.len() - offset);
    (line + count, end_column)
}

/// Parse a single block from the source using byte operations
/// Returns borrowed strings using Cow for zero-copy
///
/// - `Ok(Some(...))` — successfully parsed block.
/// - `Ok(None)` — no SFC block starts at this position.
/// - `Err(...)` — a block starts here but is incomplete or malformed.
pub fn parse_block_fast<'a>(
    bytes: &'a [u8],
    source: &'a str,
    start: usize,
    start_line: usize,
    start_column: usize,
) -> BlockParseResult<'a> {
    // This parser intentionally works on byte slices and returns borrowed `Cow`
    // values. SFC parsing sits on every compile/lint/check path, so avoiding
    // temporary strings for tag names, attrs, and block content has an outsized
    // effect on allocation profiles.
    let len = bytes.len();

    // Skip '<'
    let mut pos = start + 1;
    if pos >= len {
        return Ok(None);
    }

    // Parse tag name - find end of tag name
    let tag_start = pos;
    while bytes.get(pos).is_some_and(|&b| is_tag_name_char_fast(b)) {
        pos += 1;
    }

    if pos == tag_start {
        return Ok(None);
    }

    let tag_name = source.as_bytes().get(tag_start..pos).unwrap_or_default();

    // Parse attributes with zero-copy
    let mut attrs: BlockAttrs<'a> = FxHashMap::default();

    while bytes.get(pos).is_some_and(|&b| b != b'>') {
        // Skip whitespace
        while bytes.get(pos).is_some_and(|&b| is_whitespace_fast(b)) {
            pos += 1;
        }

        if bytes.get(pos).is_none_or(|&b| b == b'>' || b == b'/') {
            break;
        }

        // Parse attribute name
        let attr_start = pos;
        while let Some(&c) = bytes.get(pos) {
            if c == b'='
                || c == b' '
                || c == b'>'
                || c == b'/'
                || c == b'\t'
                || c == b'\n'
                || c == b'\r'
            {
                break;
            }
            pos += 1;
        }

        if pos == attr_start {
            pos += 1;
            continue;
        }

        // Zero-copy: borrow from source
        let attr_name: Cow<'a, str> =
            Cow::Borrowed(source.get(attr_start..pos).unwrap_or_default());

        // Skip whitespace
        while matches!(bytes.get(pos), Some(b' ' | b'\t')) {
            pos += 1;
        }

        let attr_value: Cow<'a, str> = if bytes.get(pos) == Some(&b'=') {
            pos += 1;

            // Skip whitespace
            while matches!(bytes.get(pos), Some(b' ' | b'\t')) {
                pos += 1;
            }

            if let Some(&quote_char) = bytes.get(pos).filter(|&&b| b == b'"' || b == b'\'') {
                pos += 1;
                let value_start = pos;

                // Use memchr for fast quote finding
                if let Some(quote_pos) = memchr(quote_char, bytes.get(pos..).unwrap_or_default()) {
                    pos += quote_pos;
                    let value = Cow::Borrowed(source.get(value_start..pos).unwrap_or_default());
                    pos += 1; // Skip closing quote
                    value
                } else {
                    // No closing quote found
                    while bytes.get(pos).is_some_and(|&b| b != quote_char) {
                        pos += 1;
                    }
                    let value = Cow::Borrowed(source.get(value_start..pos).unwrap_or_default());
                    if pos < len {
                        pos += 1;
                    }
                    value
                }
            } else {
                // Unquoted value
                let value_start = pos;
                while let Some(&c) = bytes.get(pos) {
                    if c == b' ' || c == b'>' || c == b'/' || c == b'\t' || c == b'\n' {
                        break;
                    }
                    pos += 1;
                }
                Cow::Borrowed(source.get(value_start..pos).unwrap_or_default())
            }
        } else {
            // Boolean attribute
            Cow::Borrowed("")
        };

        if !attr_name.is_empty() {
            attrs.insert(attr_name, attr_value);
        }
    }

    // Handle self-closing tag.
    let is_self_closing = bytes.get(pos) == Some(&b'/');

    if is_self_closing {
        pos += 1;
        while bytes.get(pos).is_some_and(|&b| is_whitespace_fast(b)) {
            pos += 1;
        }
        if bytes.get(pos) != Some(&b'>') {
            return Err(build_malformed_error(
                tag_name,
                "the self-closing tag is incomplete",
            ));
        }
        pos += 1;
        let (content_line, content_column) = position_after(
            bytes.get(start..pos).unwrap_or_default(),
            start_line,
            start_column,
        );
        return Ok(Some((
            tag_name,
            attrs,
            Cow::Borrowed(""),
            pos,
            pos,
            pos,
            content_line,
            content_column,
        )));
    }

    // Skip '>'
    if bytes.get(pos) == Some(&b'>') {
        pos += 1;
    } else {
        return Err(build_malformed_error(
            tag_name,
            "the opening tag is incomplete",
        ));
    }

    let content_start = pos;
    let (content_start_line, content_start_column) = position_after(
        bytes.get(start..content_start).unwrap_or_default(),
        start_line,
        start_column,
    );

    if is_void_block(tag_name) {
        return Ok(Some((
            tag_name,
            attrs,
            Cow::Borrowed(""),
            content_start,
            content_start,
            pos,
            content_start_line,
            content_start_column,
        )));
    }

    // Find closing tag based on tag type

    // Handle known tags with static closing tags
    if tag_name.eq_ignore_ascii_case(TAG_TEMPLATE) {
        return super::template_boundary::find_template_block_end(BlockEndSearch {
            bytes,
            source,
            tag_name,
            pos,
            content_start,
            start_line: content_start_line,
            start_column: content_start_column,
            initial_last_newline: content_start,
            attrs,
        });
    }

    if tag_name.eq_ignore_ascii_case(TAG_STYLE) {
        return find_block_end(BlockEndSearch {
            bytes,
            source,
            tag_name,
            pos,
            content_start,
            start_line: content_start_line,
            start_column: content_start_column,
            initial_last_newline: content_start,
            attrs,
        });
    }

    // Custom block: need to find closing tag dynamically
    if !tag_name.eq_ignore_ascii_case(TAG_SCRIPT) {
        return find_block_end(BlockEndSearch {
            bytes,
            source,
            tag_name,
            pos,
            content_start,
            start_line: content_start_line,
            start_column: content_start_column,
            initial_last_newline: content_start,
            attrs,
        });
    }

    find_script_block_end(BlockEndSearch {
        bytes,
        source,
        tag_name,
        pos,
        content_start,
        start_line: content_start_line,
        start_column: content_start_column,
        initial_last_newline: content_start,
        attrs,
    })
}