html-to-markdown 3.4.0 → 3.6.0.pre.rc.24

data/vendor/html-to-markdown-rs/src/converter/form/mod.rs

@@ -1,82 +0,0 @@
-//! Form element handlers for HTML to Markdown conversion.
-//!
-//! This module provides specialized handlers for HTML form elements:
-//! - Container elements (form, fieldset, legend, label)
-//! - Input elements (input, textarea, select, option, optgroup, button)
-//! - Measurement elements (progress, meter, output, datalist)
-//!
-//! These handlers are designed to be extracted from the main `converter.rs`
-//! file and integrated through the dispatcher function.
-//!
-//! **Integration Pattern:**
-//! Each handler function takes the same signature:
-//! - `tag_name: &str` - The HTML tag being processed
-//! - `node_handle: &NodeHandle` - The DOM node handle
-//! - `parser: &Parser` - The HTML parser reference
-//! - `output: &mut String` - The output buffer to write to
-//! - `options: &ConversionOptions` - Conversion configuration
-//! - `ctx: &Context` - Processing context (state tracking)
-//! - `depth: usize` - Current DOM tree depth
-//! - `dom_ctx: &DomContext` - DOM context for tree relationships
-//!
-//! The main dispatcher function `dispatch_form_handler` routes tags to
-//! their appropriate handlers and returns a boolean indicating success.
-
-pub mod elements;
-
-// Re-export types from parent module for submodule access
-pub use super::walk_node;
-pub use super::{Context, DomContext};
-
-// Re-export handler function for direct use
-pub use elements::handle as handle_form_elements;
-
-// Re-exports are done via the dispatch function parameter types
-
-/// Dispatches form element handling to the appropriate handler.
-///
-/// This function routes form-related HTML elements to their specialized handlers
-/// based on tag name. It is designed to be called from the main `walk_node`
-/// function in `converter.rs`.
-///
-/// # Routing Table
-///
-/// The following tag routes are supported:
-/// - **Containers**: form, fieldset, legend, label
-/// - **Inputs**: input, textarea, select, option, optgroup, button
-/// - **Measurements**: progress, meter, output, datalist
-///
-/// # Returns
-///
-/// Returns `true` if the tag was successfully handled by a form handler,
-/// `false` if the tag is not a form element and requires other handling.
-///
-/// # Example
-///
-/// ```text
-/// if dispatch_form_handler(tag_name, &node_handle, &parser, output, options, ctx, depth, dom_ctx) {
-///     // Tag was handled
-/// } else {
-///     // Continue with other handlers
-/// }
-/// ```
-pub fn dispatch_form_handler(
-    tag_name: &str,
-    node_handle: &tl::NodeHandle,
-    parser: &tl::Parser,
-    output: &mut String,
-    options: &crate::options::ConversionOptions,
-    ctx: &super::Context,
-    depth: usize,
-    dom_ctx: &super::DomContext,
-) -> bool {
-    match tag_name {
-        // Form containers and metadata
-        "form" | "fieldset" | "legend" | "label" | "input" | "textarea" | "select" | "option" | "optgroup"
-        | "button" | "progress" | "meter" | "output" | "datalist" => {
-            handle_form_elements(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx);
-            true
-        }
-        _ => false,
-    }
-}

data/vendor/html-to-markdown-rs/src/converter/format/djot.rs

@@ -1,64 +0,0 @@
-//! Djot format renderer.
-
-use super::FormatRenderer;
-
-/// Renderer for Djot lightweight markup output.
-#[derive(Debug, Clone, Copy, Default)]
-pub struct DjotRenderer;
-
-impl FormatRenderer for DjotRenderer {
-    fn emphasis(&self, content: &str) -> String {
-        format!("_{content}_")
-    }
-
-    fn strong(&self, content: &str, _symbol: char) -> String {
-        // Djot always uses single asterisk for strong
-        format!("*{content}*")
-    }
-
-    fn strikethrough(&self, content: &str) -> String {
-        format!("{{-{content}-}}")
-    }
-
-    fn highlight(&self, content: &str) -> String {
-        format!("{{={content}=}}")
-    }
-
-    fn inserted(&self, content: &str) -> String {
-        format!("{{+{content}+}}")
-    }
-
-    fn subscript(&self, content: &str, _custom_symbol: &str) -> String {
-        // Djot has native subscript support
-        format!("~{content}~")
-    }
-
-    fn superscript(&self, content: &str, _custom_symbol: &str) -> String {
-        // Djot has native superscript support
-        format!("^{content}^")
-    }
-
-    fn span_with_attributes(&self, content: &str, classes: &[&str], id: Option<&str>) -> String {
-        if classes.is_empty() && id.is_none() {
-            return content.to_string();
-        }
-
-        let mut attrs = classes.iter().map(|c| format!(".{c}")).collect::<Vec<_>>();
-        if let Some(id_val) = id {
-            attrs.push(format!("#{id_val}"));
-        }
-        format!("[{content}]{{{}}}", attrs.join(" "))
-    }
-
-    fn div_with_attributes(&self, content: &str, classes: &[&str]) -> String {
-        if classes.is_empty() {
-            return content.to_string();
-        }
-        let class_str = classes.join(" ");
-        format!("::: {class_str}\n{content}\n:::")
-    }
-
-    fn is_djot(&self) -> bool {
-        true
-    }
-}

data/vendor/html-to-markdown-rs/src/converter/format/markdown.rs

@@ -1,59 +0,0 @@
-//! Markdown format renderer.
-
-use super::FormatRenderer;
-
-/// Renderer for standard Markdown output.
-#[derive(Debug, Clone, Copy, Default)]
-pub struct MarkdownRenderer;
-
-impl FormatRenderer for MarkdownRenderer {
-    fn emphasis(&self, content: &str) -> String {
-        format!("*{content}*")
-    }
-
-    fn strong(&self, content: &str, symbol: char) -> String {
-        format!("{symbol}{symbol}{content}{symbol}{symbol}")
-    }
-
-    fn strikethrough(&self, content: &str) -> String {
-        format!("~~{content}~~")
-    }
-
-    fn highlight(&self, content: &str) -> String {
-        format!("=={content}==")
-    }
-
-    fn inserted(&self, content: &str) -> String {
-        format!("++{content}++")
-    }
-
-    fn subscript(&self, content: &str, custom_symbol: &str) -> String {
-        if custom_symbol.is_empty() {
-            format!("~{content}~")
-        } else {
-            format!("{custom_symbol}{content}{custom_symbol}")
-        }
-    }
-
-    fn superscript(&self, content: &str, custom_symbol: &str) -> String {
-        if custom_symbol.is_empty() {
-            format!("^{content}^")
-        } else {
-            format!("{custom_symbol}{content}{custom_symbol}")
-        }
-    }
-
-    fn span_with_attributes(&self, content: &str, _classes: &[&str], _id: Option<&str>) -> String {
-        // Markdown doesn't support span attributes, just return content
-        content.to_string()
-    }
-
-    fn div_with_attributes(&self, content: &str, _classes: &[&str]) -> String {
-        // Markdown doesn't support div attributes, just return content
-        content.to_string()
-    }
-
-    fn is_djot(&self) -> bool {
-        false
-    }
-}

data/vendor/html-to-markdown-rs/src/converter/format/mod.rs

@@ -1,43 +0,0 @@
-//! Output format renderers for HTML to Markdown/Djot conversion.
-//!
-//! This module provides format-specific rendering through the `FormatRenderer` trait,
-//! allowing clean separation between Markdown and Djot output syntax.
-
-mod djot;
-mod markdown;
-
-/// Trait for format-specific rendering of inline elements.
-///
-/// Implementations provide the syntax for emphasis, strong, strikethrough, etc.
-/// in their respective output formats.
-pub trait FormatRenderer: Send + Sync {
-    /// Render emphasis (em, i elements)
-    fn emphasis(&self, content: &str) -> String;
-
-    /// Render strong emphasis (strong, b elements)
-    fn strong(&self, content: &str, symbol: char) -> String;
-
-    /// Render strikethrough (del, s elements)
-    fn strikethrough(&self, content: &str) -> String;
-
-    /// Render highlight (mark element)
-    fn highlight(&self, content: &str) -> String;
-
-    /// Render inserted text (ins element)
-    fn inserted(&self, content: &str) -> String;
-
-    /// Render subscript (sub element)
-    fn subscript(&self, content: &str, custom_symbol: &str) -> String;
-
-    /// Render superscript (sup element)
-    fn superscript(&self, content: &str, custom_symbol: &str) -> String;
-
-    /// Render span with attributes (for Djot: [text]{.class})
-    fn span_with_attributes(&self, content: &str, classes: &[&str], id: Option<&str>) -> String;
-
-    /// Render div with attributes (for Djot: ::: class)
-    fn div_with_attributes(&self, content: &str, classes: &[&str]) -> String;
-
-    /// Check if this is Djot format
-    fn is_djot(&self) -> bool;
-}

data/vendor/html-to-markdown-rs/src/converter/handlers/blockquote.rs

@@ -1,173 +0,0 @@
-//! Blockquote element handler for HTML to Markdown conversion.
-//!
-//! Handles `<blockquote>` elements including:
-//! - Basic blockquote markdown output with `> ` prefix
-//! - Nested blockquotes
-//! - Citation URLs via `cite` attribute
-//! - Visitor callback integration
-
-use crate::converter::Context;
-use crate::converter::dom_context::DomContext;
-use crate::converter::main::walk_node;
-use crate::options::ConversionOptions;
-
-#[cfg(feature = "visitor")]
-#[cfg(feature = "visitor")]
-use crate::converter::utility::content::collect_tag_attributes;
-#[cfg(feature = "visitor")]
-use std::collections::BTreeMap;
-
-#[cfg(feature = "visitor")]
-use crate::converter::utility::serialization::serialize_node_to_html;
-
-/// Handle a `<blockquote>` element and convert to Markdown.
-///
-/// This handler processes blockquote elements including:
-/// - Converting inline blockquotes by processing children as inline
-/// - Handling nested blockquotes via blockquote_depth tracking
-/// - Processing citation URLs from cite attribute
-/// - Invoking visitor callbacks when the visitor feature is enabled
-/// - Adding proper spacing and blockquote prefix formatting
-#[allow(clippy::too_many_arguments)]
-#[allow(clippy::too_many_lines)]
-#[cfg_attr(not(feature = "visitor"), allow(unused_variables))]
-pub fn handle_blockquote(
-    node_handle: &tl::NodeHandle,
-    tag: &tl::HTMLTag,
-    parser: &tl::Parser,
-    output: &mut String,
-    options: &ConversionOptions,
-    ctx: &Context,
-    depth: usize,
-    dom_ctx: &DomContext,
-) {
-    // If in inline conversion mode, just process children inline
-    if ctx.convert_as_inline {
-        let children = tag.children();
-        {
-            for child_handle in children.top().iter() {
-                walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
-            }
-        }
-        return;
-    }
-
-    // Extract cite attribute if present
-    let cite = tag
-        .attributes()
-        .get("cite")
-        .flatten()
-        .map(|v| v.as_utf8_str().to_string());
-
-    // Create context for nested blockquote processing
-    let blockquote_ctx = Context {
-        blockquote_depth: ctx.blockquote_depth + 1,
-        ..ctx.clone()
-    };
-
-    // Process blockquote content
-    let mut content = String::with_capacity(256);
-    let children = tag.children();
-    {
-        for child_handle in children.top().iter() {
-            walk_node(
-                child_handle,
-                parser,
-                &mut content,
-                options,
-                &blockquote_ctx,
-                depth + 1,
-                dom_ctx,
-            );
-        }
-    }
-
-    let trimmed_content = content.trim();
-
-    // Handle visitor integration
-    #[cfg(feature = "visitor")]
-    if let Some(ref visitor) = ctx.visitor {
-        use crate::visitor::{NodeContext, NodeType, VisitResult};
-
-        let attributes: BTreeMap<String, String> = collect_tag_attributes(tag);
-
-        let node_id = node_handle.get_inner();
-        let parent_tag = dom_ctx.parent_tag_name(node_id, parser);
-        let index_in_parent = dom_ctx.get_sibling_index(node_id).unwrap_or(0);
-
-        let node_ctx = NodeContext {
-            node_type: NodeType::Blockquote,
-            tag_name: "blockquote".to_string(),
-            attributes,
-            depth,
-            index_in_parent,
-            parent_tag,
-            is_inline: false,
-        };
-
-        let mut visitor_ref = visitor.borrow_mut();
-        match visitor_ref.visit_blockquote(&node_ctx, trimmed_content, ctx.blockquote_depth) {
-            VisitResult::Continue => {}
-            VisitResult::Custom(custom) => {
-                output.push_str(&custom);
-                return;
-            }
-            VisitResult::Skip => return,
-            VisitResult::PreserveHtml => {
-                let mut html_output = String::new();
-                serialize_node_to_html(node_handle, parser, &mut html_output);
-                output.push_str(&html_output);
-                return;
-            }
-            VisitResult::Error(err) => {
-                if ctx.visitor_error.borrow().is_none() {
-                    *ctx.visitor_error.borrow_mut() = Some(err);
-                }
-                return;
-            }
-        }
-    }
-
-    // Output blockquote if content is not empty
-    if !trimmed_content.is_empty() {
-        // Add proper spacing based on blockquote depth
-        if ctx.blockquote_depth > 0 {
-            output.push_str("\n\n\n");
-        } else if !output.is_empty() {
-            if output.ends_with("\n\n") {
-                // Paragraph already added \n\n; blockquote needs just \n
-                output.truncate(output.len() - 1);
-            } else if !output.ends_with('\n') {
-                output.push_str("\n\n");
-            } else if !output.ends_with("\n\n") {
-                output.push('\n');
-            }
-        }
-
-        // Add blockquote prefix to each line
-        let prefix = "> ";
-
-        for line in trimmed_content.lines() {
-            output.push_str(prefix);
-            output.push_str(line.trim());
-            output.push('\n');
-        }
-
-        // Add citation if present
-        if let Some(url) = cite {
-            output.push('\n');
-            output.push_str("— <");
-            output.push_str(&url);
-            output.push_str(">\n\n");
-        }
-
-        // Add trailing newlines only when appropriate for proper spacing
-        // (matching paragraph conditional logic for CommonMark compliance)
-        if !ctx.convert_as_inline && !ctx.in_table_cell && !ctx.in_list_item {
-            while output.ends_with('\n') {
-                output.truncate(output.len() - 1);
-            }
-            output.push_str("\n\n");
-        }
-    }
-}