cmark_writer/writer/

cmark.rs

//! CommonMark writer implementation.
//!
//! This file contains the implementation of the CommonMarkWriter class, which serializes AST nodes to CommonMark-compliant text.

use crate::ast::{CustomNode, CustomNodeWriter, HeadingType, HtmlElement, ListItem, Node};
use crate::error::{WriteError, WriteResult};
use crate::options::WriterOptions;
use crate::CodeBlockType;
use std::fmt::{self};

use super::processors::{
    BlockNodeProcessor, CustomNodeProcessor, InlineNodeProcessor, NodeProcessor,
};

/// CommonMark writer
///
/// This struct is responsible for serializing AST nodes to CommonMark-compliant text.
#[derive(Debug)]
pub struct CommonMarkWriter {
    options: WriterOptions,
    buffer: String,
}

impl CommonMarkWriter {
    /// Create a new CommonMark writer with default options
    ///
    /// # Example
    ///
    /// ```
    /// use cmark_writer::writer::CommonMarkWriter;
    /// use cmark_writer::ast::Node;
    ///
    /// let mut writer = CommonMarkWriter::new();
    /// writer.write(&Node::Text("Hello".to_string())).unwrap();
    /// assert_eq!(writer.into_string(), "Hello");
    /// ```
    pub fn new() -> Self {
        Self::with_options(WriterOptions::default())
    }

    /// Create a new CommonMark writer with specified options
    ///
    /// # Parameters
    ///
    /// * `options` - Custom CommonMark formatting options
    ///
    /// # Example
    ///
    /// ```
    /// use cmark_writer::writer::CommonMarkWriter;
    /// use cmark_writer::options::WriterOptions;
    ///
    /// let options = WriterOptions {
    ///     strict: true,
    ///     hard_break_spaces: false,  // Use backslash for line breaks
    ///     indent_spaces: 2,          // Use 2 spaces for indentation
    /// };
    /// let writer = CommonMarkWriter::with_options(options);
    /// ```
    pub fn with_options(options: WriterOptions) -> Self {
        Self {
            options,
            buffer: String::new(),
        }
    }

    /// Whether the writer is in strict mode
    pub(crate) fn is_strict_mode(&self) -> bool {
        self.options.strict
    }

    /// Apply a specific prefix to multi-line text, used for handling container node indentation
    ///
    /// # Parameters
    ///
    /// * `content` - The multi-line content to process
    /// * `prefix` - The prefix to apply to each line
    /// * `first_line_prefix` - The prefix to apply to the first line (can be different from other lines)
    ///
    /// # Returns
    ///
    /// Returns a string with applied indentation
    fn apply_prefix(&self, content: &str, prefix: &str, first_line_prefix: Option<&str>) -> String {
        if content.is_empty() {
            return String::new();
        }

        let mut result = String::new();
        let lines: Vec<&str> = content.lines().collect();

        if !lines.is_empty() {
            let actual_prefix = first_line_prefix.unwrap_or(prefix);
            result.push_str(actual_prefix);
            result.push_str(lines[0]);
        }

        for line in &lines[1..] {
            result.push('\n');
            result.push_str(prefix);
            result.push_str(line);
        }

        result
    }

    /// Write an AST node as CommonMark format
    ///
    /// # Parameters
    ///
    /// * `node` - The AST node to write
    ///
    /// # Returns
    ///
    /// If writing succeeds, returns `Ok(())`, otherwise returns `Err(WriteError)`
    ///
    /// # Example
    ///
    /// ```
    /// use cmark_writer::writer::CommonMarkWriter;
    /// use cmark_writer::ast::Node;
    ///
    /// let mut writer = CommonMarkWriter::new();
    /// writer.write(&Node::Text("Hello".to_string())).unwrap();
    /// ```
    pub fn write(&mut self, node: &Node) -> WriteResult<()> {
        if let Node::Custom(_) = node {
            return CustomNodeProcessor.process(self, node);
        }

        if node.is_block() {
            BlockNodeProcessor.process(self, node)
        } else if node.is_inline() {
            InlineNodeProcessor.process(self, node)
        } else {
            // Keep this branch for future implementation needs
            Err(WriteError::UnsupportedNodeType)
        }
    }

    /// Write a custom node using its implementation
    #[allow(clippy::borrowed_box)]
    pub(crate) fn write_custom_node(&mut self, node: &Box<dyn CustomNode>) -> WriteResult<()> {
        node.write(self)
    }

    /// Get context description for a node, used for error reporting
    pub(crate) fn get_context_for_node(&self, node: &Node) -> String {
        match node {
            Node::Text(_) => "Text".to_string(),
            Node::Emphasis(_) => "Emphasis".to_string(),
            Node::Strong(_) => "Strong".to_string(),
            Node::InlineCode(_) => "InlineCode".to_string(),
            Node::Link { .. } => "Link content".to_string(),
            Node::Image { .. } => "Image alt text".to_string(),
            Node::HtmlElement(_) => "HtmlElement content".to_string(),
            Node::Custom(_) => "Custom node".to_string(),
            _ => "Unknown inline element".to_string(),
        }
    }

    /// Check if the inline node contains a newline character and return an error if it does
    pub(crate) fn check_no_newline(&self, node: &Node, context: &str) -> WriteResult<()> {
        if Self::node_contains_newline(node) {
            return Err(WriteError::NewlineInInlineElement(context.to_string()));
        }
        Ok(())
    }

    /// Check if the inline node contains a newline character recursively
    fn node_contains_newline(node: &Node) -> bool {
        match node {
            Node::Text(s) | Node::InlineCode(s) => s.contains('\n'),
            Node::Emphasis(children) | Node::Strong(children) => {
                children.iter().any(Self::node_contains_newline)
            }
            Node::HtmlElement(element) => element.children.iter().any(Self::node_contains_newline),
            Node::Link { content, .. } => content.iter().any(Self::node_contains_newline),
            Node::Image { alt, .. } => alt.iter().any(Self::node_contains_newline),
            Node::SoftBreak | Node::HardBreak => true,
            // Custom nodes are handled separately
            Node::Custom(_) => false,
            _ => false,
        }
    }

    /// Writes text content with character escaping
    pub(crate) fn write_text_content(&mut self, content: &str) -> WriteResult<()> {
        let escaped = content
            .replace('\\', "\\\\")
            .replace('*', "\\*")
            .replace('_', "\\_")
            .replace('[', "\\[")
            .replace(']', "\\]")
            .replace('<', "\\<")
            .replace('>', "\\>")
            .replace('`', "\\`");

        self.write_str(&escaped)?;
        Ok(())
    }

    /// Writes inline code content
    pub(crate) fn write_code_content(&mut self, content: &str) -> WriteResult<()> {
        self.write_char('`')?;
        self.write_str(content)?;
        self.write_char('`')?;
        Ok(())
    }

    /// Helper function for writing content with delimiters
    pub(crate) fn write_delimited(&mut self, content: &[Node], delimiter: &str) -> WriteResult<()> {
        self.write_str(delimiter)?;

        for node in content {
            self.write(node)?;
        }

        self.write_str(delimiter)?;
        Ok(())
    }

    /// Write a document node
    pub(crate) fn write_document(&mut self, children: &[Node]) -> WriteResult<()> {
        for (i, child) in children.iter().enumerate() {
            if i > 0 {
                self.write_str("\n")?;
            }
            self.write(child)?;
        }
        Ok(())
    }

    /// Write a heading node
    pub(crate) fn write_heading(
        &mut self,
        level: u8,
        content: &[Node],
        heading_type: &HeadingType,
    ) -> WriteResult<()> {
        // 验证标题级别
        if level == 0 || level > 6 {
            return Err(WriteError::InvalidHeadingLevel(level));
        }

        match heading_type {
            // ATX heading, using # character
            HeadingType::Atx => {
                for _ in 0..level {
                    self.write_char('#')?;
                }
                self.write_char(' ')?;

                for node in content {
                    self.write(node)?;
                }

                self.write_char('\n')?;
            }

            HeadingType::Setext => {
                // First write the heading content
                for node in content {
                    self.write(node)?;
                }
                self.write_char('\n')?;

                // Add underline characters based on level
                // Setext only supports level 1 and 2 headings
                let underline_char = if level == 1 { '=' } else { '-' };

                // For good readability, we add underlines at least as long as the heading text
                // Calculate a reasonable underline length (at least 3 characters)
                let min_len = 3;

                // Write the underline characters
                for _ in 0..min_len {
                    self.write_char(underline_char)?;
                }

                // Add a newline to end the heading
                self.write_char('\n')?;
            }
        }

        Ok(())
    }

    /// Write a paragraph node
    pub(crate) fn write_paragraph(&mut self, content: &[Node]) -> WriteResult<()> {
        for node in content.iter() {
            self.write(node)?;
        }

        Ok(())
    }

    /// Write a blockquote node
    pub(crate) fn write_blockquote(&mut self, content: &[Node]) -> WriteResult<()> {
        // Create a temporary writer buffer to write all blockquote content
        let mut temp_writer = CommonMarkWriter::with_options(self.options.clone());

        // Write all content to temporary buffer
        for (i, node) in content.iter().enumerate() {
            if i > 0 {
                temp_writer.write_char('\n')?;
            }
            // Write all nodes uniformly
            temp_writer.write(node)?;
        }

        // Get all content
        let all_content = temp_writer.into_string();

        // Apply blockquote prefix "> " uniformly
        let prefix = "> ";
        let formatted_content = self.apply_prefix(&all_content, prefix, Some(prefix));

        // Write formatted content
        self.buffer.push_str(&formatted_content);
        Ok(())
    }

    /// Write a thematic break (horizontal rule)
    pub(crate) fn write_thematic_break(&mut self) -> WriteResult<()> {
        self.write_str("---")?;
        Ok(())
    }

    /// Write a code block node
    pub(crate) fn write_code_block(
        &mut self,
        language: &Option<String>,
        content: &str,
        block_type: &CodeBlockType,
    ) -> WriteResult<()> {
        match block_type {
            CodeBlockType::Indented => {
                let indent = "    ";
                let indented_content = self.apply_prefix(content, indent, Some(indent));
                self.buffer.push_str(&indented_content);
            }
            CodeBlockType::Fenced => {
                let max_backticks = content
                    .chars()
                    .fold((0, 0), |(max, current), c| {
                        if c == '`' {
                            (max.max(current + 1), current + 1)
                        } else {
                            (max, 0)
                        }
                    })
                    .0;

                let fence_len = std::cmp::max(max_backticks + 1, 3);
                let fence = "`".repeat(fence_len);

                self.write_str(&fence)?;
                if let Some(lang) = language {
                    self.write_str(lang)?;
                }
                self.write_char('\n')?;

                self.buffer.push_str(content);
                if !content.ends_with('\n') {
                    self.write_char('\n')?;
                }

                self.write_str(&fence)?;
            }
        }

        Ok(())
    }

    /// Write an unordered list node
    pub(crate) fn write_unordered_list(&mut self, items: &[ListItem]) -> WriteResult<()> {
        for (i, item) in items.iter().enumerate() {
            if i > 0 {
                self.write_char('\n')?;
            }
            self.write_list_item(item, "- ")?;
        }

        Ok(())
    }

    /// Write an ordered list node
    pub(crate) fn write_ordered_list(&mut self, start: u32, items: &[ListItem]) -> WriteResult<()> {
        // Track the current item number
        let mut current_number = start;

        for (i, item) in items.iter().enumerate() {
            if i > 0 {
                self.write_char('\n')?;
            }

            match item {
                // For ordered list items, check if there's a custom number
                ListItem::Ordered { number, content: _ } => {
                    if let Some(custom_num) = number {
                        // Use custom numbering
                        let prefix = format!("{}. ", custom_num);
                        self.write_list_item(item, &prefix)?;
                        // Next expected number
                        current_number = custom_num + 1;
                    } else {
                        // No custom number, use the current calculated number
                        let prefix = format!("{}. ", current_number);
                        self.write_list_item(item, &prefix)?;
                        current_number += 1;
                    }
                }
                // For other types of list items, still use the current number
                _ => {
                    let prefix = format!("{}. ", current_number);
                    self.write_list_item(item, &prefix)?;
                    current_number += 1;
                }
            }
        }

        Ok(())
    }

    /// Write a list item
    fn write_list_item(&mut self, item: &ListItem, prefix: &str) -> WriteResult<()> {
        match item {
            ListItem::Unordered { content } => {
                self.write_str(prefix)?;
                self.write_list_item_content(content, prefix.len())?;
            }
            ListItem::Ordered { number, content } => {
                if let Some(num) = number {
                    let custom_prefix = format!("{}. ", num);
                    self.write_str(&custom_prefix)?;
                    self.write_list_item_content(content, custom_prefix.len())?;
                } else {
                    self.write_str(prefix)?;
                    self.write_list_item_content(content, prefix.len())?;
                }
            }
        }

        Ok(())
    }

    /// Write list item content
    fn write_list_item_content(&mut self, content: &[Node], prefix_len: usize) -> WriteResult<()> {
        if content.is_empty() {
            return Ok(());
        }

        let mut temp_writer = CommonMarkWriter::with_options(self.options.clone());

        for (i, node) in content.iter().enumerate() {
            if i > 0 {
                temp_writer.write_char('\n')?;
            }

            temp_writer.write(node)?;
        }

        let all_content = temp_writer.into_string();

        let indent = " ".repeat(prefix_len);

        let formatted_content = self.apply_prefix(&all_content, &indent, Some(""));

        self.buffer.push_str(&formatted_content);

        Ok(())
    }

    /// Write a table
    pub(crate) fn write_table(&mut self, headers: &[Node], rows: &[Vec<Node>]) -> WriteResult<()> {
        // Write header
        self.write_char('|')?;
        for header in headers {
            self.check_no_newline(header, "Table Header")?;
            self.write_char(' ')?;
            self.write(header)?;
            self.write_str(" |")?;
        }
        self.write_char('\n')?;

        // Write alignment row
        self.write_char('|')?;
        for _ in 0..headers.len() {
            self.write_str(" --- |")?;
        }
        self.write_char('\n')?;

        // Write table content
        for row in rows {
            self.write_char('|')?;
            for cell in row {
                self.check_no_newline(cell, "Table Cell")?;
                self.write_char(' ')?;
                self.write(cell)?;
                self.write_str(" |")?;
            }
            self.write_char('\n')?;
        }

        Ok(())
    }

    /// Write a link
    pub(crate) fn write_link(
        &mut self,
        url: &str,
        title: &Option<String>,
        content: &[Node],
    ) -> WriteResult<()> {
        for node in content {
            self.check_no_newline(node, "Link Text")?;
        }
        self.write_char('[')?;

        for node in content {
            self.write(node)?;
        }

        self.write_str("](")?;
        self.write_str(url)?;

        if let Some(title_text) = title {
            self.write_str(" \"")?;
            self.write_str(title_text)?;
            self.write_char('"')?;
        }

        self.write_char(')')?;
        Ok(())
    }

    /// Write an image
    pub(crate) fn write_image(
        &mut self,
        url: &str,
        title: &Option<String>,
        alt: &[Node],
    ) -> WriteResult<()> {
        // Check for newlines in alt text content
        for node in alt {
            self.check_no_newline(node, "Image alt text")?;
        }

        self.write_str("![")?;

        // Write alt text content
        for node in alt {
            self.write(node)?;
        }

        self.write_str("](")?;
        self.write_str(url)?;

        if let Some(title_text) = title {
            self.write_str(" \"")?;
            self.write_str(title_text)?;
            self.write_char('"')?;
        }

        self.write_char(')')?;
        Ok(())
    }

    /// Write a soft line break
    pub(crate) fn write_soft_break(&mut self) -> WriteResult<()> {
        self.write_char('\n')?;
        Ok(())
    }

    /// Write a hard line break
    pub(crate) fn write_hard_break(&mut self) -> WriteResult<()> {
        if self.options.hard_break_spaces {
            self.write_str("  \n")?;
        } else {
            self.write_str("\\\n")?;
        }
        Ok(())
    }

    /// Write an HTML block
    pub(crate) fn write_html_block(&mut self, content: &str) -> WriteResult<()> {
        self.buffer.push_str(content);

        Ok(())
    }

    /// Write an autolink (URI or email address wrapped in < and >)
    pub(crate) fn write_autolink(&mut self, url: &str, is_email: bool) -> WriteResult<()> {
        // Autolinks shouldn't contain newlines
        if url.contains('\n') {
            return Err(WriteError::NewlineInInlineElement(
                "Autolink URL".to_string(),
            ));
        }

        // Write the autolink with < and > delimiters
        self.write_char('<')?;

        // For email autolinks, we don't need to add any prefix
        // For URI autolinks, ensure it has a scheme
        if !is_email && !url.contains(':') {
            // Default to https if no scheme is provided
            self.write_str("https://")?;
        }

        self.write_str(url)?;
        self.write_char('>')?;

        Ok(())
    }

    /// Write a link reference definition
    pub(crate) fn write_link_reference_definition(
        &mut self,
        label: &str,
        destination: &str,
        title: &Option<String>,
    ) -> WriteResult<()> {
        // Format: [label]: destination "optional title"
        self.write_char('[')?;
        self.write_str(label)?;
        self.write_str("]: ")?;
        self.write_str(destination)?;

        if let Some(title_text) = title {
            self.write_str(" \"")?;
            self.write_str(title_text)?;
            self.write_char('"')?;
        }

        Ok(())
    }

    /// Write a reference link
    pub(crate) fn write_reference_link(
        &mut self,
        label: &str,
        content: &[Node],
    ) -> WriteResult<()> {
        // Check for newlines in content
        for node in content {
            self.check_no_newline(node, "Reference Link Text")?;
        }

        // If content is empty or exactly matches the label (as plain text),
        // this is a shortcut reference link: [label]
        if content.is_empty() {
            self.write_char('[')?;
            self.write_str(label)?;
            self.write_char(']')?;
            return Ok(());
        }

        // Check if content is exactly the same as the label (to use shortcut syntax)
        let is_shortcut =
            content.len() == 1 && matches!(&content[0], Node::Text(text) if text == label);

        if is_shortcut {
            // Use shortcut reference link syntax: [label]
            self.write_char('[')?;
            self.write_str(label)?;
            self.write_char(']')?;
        } else {
            // Use full reference link syntax: [content][label]
            self.write_char('[')?;

            for node in content {
                self.write(node)?;
            }

            self.write_str("][")?;
            self.write_str(label)?;
            self.write_char(']')?;
        }

        Ok(())
    }

    /// Write an HTML element
    pub(crate) fn write_html_element(&mut self, element: &HtmlElement) -> WriteResult<()> {
        self.write_char('<')?;
        self.write_str(&element.tag)?;

        for attr in &element.attributes {
            self.write_char(' ')?;
            self.write_str(&attr.name)?;
            self.write_str("=\"")?;
            self.write_str(&attr.value)?; // Assume attributes are pre-escaped if needed
            self.write_char('"')?;
        }

        if element.self_closing {
            self.write_str(" />")?;
            return Ok(());
        }

        self.write_char('>')?;

        for child in &element.children {
            // HTML element content can contain newlines, so no strict check here
            self.write(child)?;
        }

        self.write_str("</")?;
        self.write_str(&element.tag)?;
        self.write_char('>')?;
        Ok(())
    }

    /// Get the generated CommonMark format text
    ///
    /// Consumes the writer and returns the generated string
    ///
    /// # Example
    ///
    /// ```
    /// use cmark_writer::writer::CommonMarkWriter;
    /// use cmark_writer::ast::Node;
    ///
    /// let mut writer = CommonMarkWriter::new();
    /// writer.write(&Node::Text("Hello".to_string())).unwrap();
    /// let result = writer.into_string();
    /// assert_eq!(result, "Hello");
    /// ```
    pub fn into_string(self) -> String {
        self.buffer
    }
    /// Ensure content ends with a newline (for consistent handling at the end of block nodes)
    ///
    /// Adds a newline character if the content doesn't already end with one; does nothing if it already ends with a newline
    pub(crate) fn ensure_trailing_newline(&mut self) -> WriteResult<()> {
        if !self.buffer.ends_with('\n') {
            self.write_char('\n')?;
        }
        Ok(())
    }
}

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

// Implement Display trait for Node structure
impl fmt::Display for Node {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let mut writer = CommonMarkWriter::new();
        match writer.write(self) {
            Ok(_) => write!(f, "{}", writer.into_string()),
            Err(e) => write!(f, "Error writing Node: {}", e),
        }
    }
}

// Implement CustomNodeWriter trait for CommonMarkWriter
impl CustomNodeWriter for CommonMarkWriter {
    fn write_str(&mut self, s: &str) -> fmt::Result {
        self.buffer.push_str(s);
        Ok(())
    }

    fn write_char(&mut self, c: char) -> fmt::Result {
        self.buffer.push(c);
        Ok(())
    }
}