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

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

@@ -1,282 +0,0 @@
-//! Image element handler for HTML to Markdown conversion.
-//!
-//! Handles `<img>` elements including:
-//! - Basic image markdown output `![alt](src "title")`
-//! - Metadata collection for image extraction
-//! - Inline data URI image handling
-//! - Visitor callback integration
-
-use std::borrow::Cow;
-#[cfg(any(feature = "metadata", feature = "inline-images", feature = "visitor"))]
-use std::collections::BTreeMap;
-
-use crate::converter::Context;
-use crate::converter::dom_context::DomContext;
-#[cfg(feature = "visitor")]
-use crate::converter::utility::content::collect_tag_attributes;
-use crate::converter::utility::preprocessing::sanitize_markdown_url;
-use crate::options::ConversionOptions;
-
-#[cfg(feature = "inline-images")]
-use crate::converter::media::handle_inline_data_image;
-
-#[cfg(feature = "visitor")]
-use crate::converter::utility::serialization::serialize_node;
-
-#[cfg(feature = "metadata")]
-type ImageMetadataPayload = (BTreeMap<String, String>, Option<u32>, Option<u32>);
-
-/// Handle an `<img>` element and convert to Markdown.
-///
-/// This handler processes image elements including:
-/// - Extracting src, alt, and title attributes
-/// - Collecting metadata when the metadata feature is enabled
-/// - Handling inline data URIs when the inline-images feature is enabled
-/// - Invoking visitor callbacks when the visitor feature is enabled
-/// - Generating appropriate markdown output
-#[allow(clippy::too_many_arguments)]
-#[allow(clippy::too_many_lines)]
-#[cfg_attr(not(feature = "visitor"), allow(unused_variables))]
-pub fn handle_img(
-    node_handle: &tl::NodeHandle,
-    tag: &tl::HTMLTag,
-    parser: &tl::Parser,
-    output: &mut String,
-    options: &ConversionOptions,
-    ctx: &Context,
-    depth: usize,
-    dom_ctx: &DomContext,
-) {
-    let src = tag.attributes().get("src").flatten().map_or(Cow::Borrowed(""), |v| {
-        let s = v.as_utf8_str();
-        Cow::Owned(sanitize_markdown_url(&s).into_owned())
-    });
-
-    let alt = tag
-        .attributes()
-        .get("alt")
-        .flatten()
-        .map_or(Cow::Borrowed(""), |v| v.as_utf8_str());
-
-    let title = tag.attributes().get("title").flatten().map(|v| v.as_utf8_str());
-
-    // Collect metadata payload if metadata feature is enabled
-    #[cfg(feature = "metadata")]
-    #[allow(clippy::useless_let_if_seq)]
-    let mut metadata_payload: Option<ImageMetadataPayload> = None;
-    #[cfg(feature = "metadata")]
-    if ctx.metadata_wants_images {
-        let mut attributes_map = BTreeMap::new();
-        let mut width: Option<u32> = None;
-        let mut height: Option<u32> = None;
-        for (key, value_opt) in tag.attributes().iter() {
-            let key_str = key.to_string();
-            if key_str == "src" {
-                continue;
-            }
-            let value = value_opt.map(|v| v.to_string()).unwrap_or_default();
-            if key_str == "width" {
-                if let Ok(parsed) = value.parse::<u32>() {
-                    width = Some(parsed);
-                }
-            } else if key_str == "height" {
-                if let Ok(parsed) = value.parse::<u32>() {
-                    height = Some(parsed);
-                }
-            }
-            attributes_map.insert(key_str, value);
-        }
-        metadata_payload = Some((attributes_map, width, height));
-    }
-
-    // Handle inline data URI images
-    #[cfg(feature = "inline-images")]
-    if let Some(ref collector_ref) = ctx.inline_collector {
-        if src.trim_start().starts_with("data:") {
-            let mut attributes_map = BTreeMap::new();
-            for (key, value_opt) in tag.attributes().iter() {
-                let key_str = key.to_string();
-                let keep = key_str == "width"
-                    || key_str == "height"
-                    || key_str == "filename"
-                    || key_str == "aria-label"
-                    || key_str.starts_with("data-");
-                if keep {
-                    let value = value_opt.map(|value| value.to_string()).unwrap_or_default();
-                    attributes_map.insert(key_str, value);
-                }
-            }
-            handle_inline_data_image(
-                collector_ref,
-                src.as_ref(),
-                alt.as_ref(),
-                title.as_deref(),
-                attributes_map,
-            );
-        }
-    }
-
-    let keep_as_markdown = ctx.in_heading && ctx.heading_allow_inline_images;
-
-    let should_use_alt_text =
-        !keep_as_markdown && (ctx.convert_as_inline || (ctx.in_heading && !ctx.heading_allow_inline_images));
-
-    // Generate image output with visitor integration
-    #[cfg(feature = "visitor")]
-    let image_output = if let Some(ref visitor_handle) = 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::Image,
-            tag_name: "img".to_string(),
-            attributes,
-            depth,
-            index_in_parent,
-            parent_tag,
-            is_inline: true,
-        };
-
-        let visit_result = {
-            let mut visitor = visitor_handle.borrow_mut();
-            visitor.visit_image(&node_ctx, &src, &alt, title.as_deref())
-        };
-        match visit_result {
-            VisitResult::Continue => Some(format_image_markdown(
-                &src,
-                &alt,
-                title.as_deref(),
-                should_use_alt_text,
-                options.link_style,
-                ctx.reference_collector.as_ref(),
-            )),
-            VisitResult::Custom(custom) => Some(custom),
-            VisitResult::Skip => None,
-            VisitResult::Error(err) => {
-                if ctx.visitor_error.borrow().is_none() {
-                    *ctx.visitor_error.borrow_mut() = Some(err);
-                }
-                None
-            }
-            VisitResult::PreserveHtml => Some(serialize_node(node_handle, parser)),
-        }
-    } else {
-        Some(format_image_markdown(
-            &src,
-            &alt,
-            title.as_deref(),
-            should_use_alt_text,
-            options.link_style,
-            ctx.reference_collector.as_ref(),
-        ))
-    };
-
-    #[cfg(not(feature = "visitor"))]
-    let image_output = Some(format_image_markdown(
-        &src,
-        &alt,
-        title.as_deref(),
-        should_use_alt_text,
-        options.link_style,
-        ctx.reference_collector.as_ref(),
-    ));
-
-    // Only output image if skip_images is not enabled
-    if !options.skip_images {
-        if let Some(img_text) = image_output {
-            output.push_str(&img_text);
-        }
-    }
-
-    // Add image to metadata collector
-    #[cfg(feature = "metadata")]
-    if ctx.metadata_wants_images {
-        if let Some(ref collector) = ctx.metadata_collector {
-            if let Some((attributes_map, width, height)) = metadata_payload {
-                if !src.is_empty() {
-                    let dimensions = match (width, height) {
-                        (Some(w), Some(h)) => Some((w, h)),
-                        _ => None,
-                    };
-                    collector.borrow_mut().add_image(
-                        src.to_string(),
-                        if alt.is_empty() { None } else { Some(alt.to_string()) },
-                        title.as_deref().map(std::string::ToString::to_string),
-                        dimensions,
-                        attributes_map,
-                    );
-                }
-            }
-        }
-    }
-
-    if let Some(ref sc) = ctx.structure_collector {
-        let src_opt = if src.is_empty() { None } else { Some(src.as_ref()) };
-        let alt_opt = if alt.is_empty() { None } else { Some(alt.as_ref()) };
-        sc.borrow_mut().push_image(src_opt, alt_opt);
-    }
-}
-
-/// Format an image as Markdown syntax.
-///
-/// If `use_alt_only` is true, returns just the alt text.
-/// Otherwise returns the full `![alt](src "title")` syntax.
-fn format_image_markdown(
-    src: &str,
-    alt: &str,
-    title: Option<&str>,
-    use_alt_only: bool,
-    link_style: crate::options::validation::LinkStyle,
-    reference_collector: Option<&crate::converter::reference_collector::ReferenceCollectorHandle>,
-) -> String {
-    if use_alt_only {
-        return alt.to_string();
-    }
-    if link_style == crate::options::validation::LinkStyle::Reference {
-        if let Some(collector) = reference_collector {
-            let ref_num = collector.borrow_mut().get_or_insert(src, title);
-            let mut buf = String::with_capacity(alt.len() + 10);
-            buf.push_str("![");
-            buf.push_str(alt);
-            buf.push_str("][");
-            buf.push_str(&ref_num.to_string());
-            buf.push(']');
-            return buf;
-        }
-    }
-    let mut buf = String::with_capacity(src.len() + alt.len() + 10);
-    buf.push_str("![");
-    buf.push_str(alt);
-    buf.push_str("](");
-
-    if src.is_empty() {
-        buf.push_str("<>");
-    } else if src.contains(' ') || src.contains('\n') {
-        buf.push('<');
-        buf.push_str(src);
-        buf.push('>');
-    } else {
-        let open_count = src.chars().filter(|&c| c == '(').count();
-        let close_count = src.chars().filter(|&c| c == ')').count();
-
-        if open_count == close_count {
-            buf.push_str(src);
-        } else {
-            let escaped_src = src.replace('(', "\\(").replace(')', "\\)");
-            buf.push_str(&escaped_src);
-        }
-    }
-
-    if let Some(title_text) = title {
-        buf.push_str(" \"");
-        buf.push_str(title_text);
-        buf.push('"');
-    }
-    buf.push(')');
-    buf
-}

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

@@ -1,316 +0,0 @@
-//! Link element handler for HTML to Markdown conversion.
-//!
-//! Handles `<a>` elements including:
-//! - Basic link markdown output `[text](href "title")`
-//! - Autolinks when text matches href
-//! - Links containing heading elements
-//! - Complex link content with mixed block/inline elements
-//! - Visitor callback integration
-//! - Link metadata collection
-
-#[cfg(any(feature = "metadata", feature = "visitor"))]
-use std::collections::BTreeMap;
-
-use crate::converter::Context;
-use crate::converter::block::heading::{find_single_heading_child, heading_allows_inline_images, push_heading};
-use crate::converter::dom_context::DomContext;
-use crate::converter::inline::link::append_markdown_link;
-use crate::converter::main::walk_node;
-#[cfg(feature = "visitor")]
-use crate::converter::utility::content::collect_tag_attributes;
-use crate::converter::utility::content::{
-    collect_link_label_text, escape_link_label, get_text_content, normalize_link_label, normalized_tag_name,
-};
-use crate::options::ConversionOptions;
-use crate::text;
-
-#[cfg(feature = "visitor")]
-use crate::converter::utility::serialization::serialize_node;
-
-/// Handle an `<a>` (link) element and convert to Markdown.
-///
-/// This handler processes link elements including:
-/// - Extracting href and title attributes
-/// - Detecting autolinks (where text equals href)
-/// - Handling links that contain heading elements
-/// - Processing complex link content (mixed block/inline)
-/// - Invoking visitor callbacks when the visitor feature is enabled
-/// - Collecting link metadata when the metadata feature is enabled
-/// - Generating appropriate markdown link output
-#[allow(clippy::too_many_arguments)]
-#[allow(clippy::too_many_lines)]
-#[cfg_attr(not(feature = "visitor"), allow(unused_variables))]
-pub fn handle_link(
-    node_handle: &tl::NodeHandle,
-    tag: &tl::HTMLTag,
-    parser: &tl::Parser,
-    output: &mut String,
-    options: &ConversionOptions,
-    ctx: &Context,
-    depth: usize,
-    dom_ctx: &DomContext,
-) {
-    let href_attr = tag
-        .attributes()
-        .get("href")
-        .flatten()
-        .map(|v| text::decode_html_entities(&v.as_utf8_str()));
-    let title = tag
-        .attributes()
-        .get("title")
-        .flatten()
-        .map(|v| v.as_utf8_str().to_string());
-
-    if let Some(href) = href_attr {
-        let raw_text = text::normalize_whitespace(&get_text_content(node_handle, parser, dom_ctx))
-            .trim()
-            .to_string();
-
-        let is_autolink = options.autolinks
-            && !options.default_title
-            && !href.is_empty()
-            && (raw_text == href || (href.starts_with("mailto:") && raw_text == href[7..]));
-
-        if is_autolink {
-            output.push('<');
-            if href.starts_with("mailto:") && raw_text == href[7..] {
-                output.push_str(&raw_text);
-            } else {
-                output.push_str(&href);
-            }
-            output.push('>');
-            return;
-        }
-
-        if let Some((heading_level, heading_handle)) = find_single_heading_child(*node_handle, parser) {
-            if let Some(heading_node) = heading_handle.get(parser) {
-                if let tl::Node::Tag(heading_tag) = heading_node {
-                    let heading_name = normalized_tag_name(heading_tag.name().as_utf8_str()).into_owned();
-                    let mut heading_text = String::new();
-                    let heading_ctx = Context {
-                        in_heading: true,
-                        convert_as_inline: true,
-                        heading_allow_inline_images: heading_allows_inline_images(
-                            &heading_name,
-                            &ctx.keep_inline_images_in,
-                        ),
-                        ..ctx.clone()
-                    };
-                    walk_node(
-                        &heading_handle,
-                        parser,
-                        &mut heading_text,
-                        options,
-                        &heading_ctx,
-                        depth + 1,
-                        dom_ctx,
-                    );
-                    let trimmed_heading = heading_text.trim();
-                    if !trimmed_heading.is_empty() {
-                        let escaped_label = escape_link_label(trimmed_heading);
-                        let mut link_buffer = String::new();
-                        append_markdown_link(
-                            &mut link_buffer,
-                            &escaped_label,
-                            href.as_str(),
-                            title.as_deref(),
-                            raw_text.as_str(),
-                            options,
-                            ctx.reference_collector.as_ref(),
-                        );
-                        push_heading(output, ctx, options, heading_level, link_buffer.as_str());
-                        return;
-                    }
-                }
-            }
-        }
-
-        let children: Vec<_> = tag.children().top().iter().copied().collect();
-        let (inline_label, _block_nodes, saw_block) = collect_link_label_text(&children, parser, dom_ctx);
-        let mut label = if saw_block {
-            let mut content = String::new();
-            let link_ctx = Context {
-                inline_depth: ctx.inline_depth + 1,
-                convert_as_inline: true,
-                ..ctx.clone()
-            };
-            for child_handle in &children {
-                let mut child_buf = String::new();
-                walk_node(
-                    child_handle,
-                    parser,
-                    &mut child_buf,
-                    options,
-                    &link_ctx,
-                    depth + 1,
-                    dom_ctx,
-                );
-                if !child_buf.trim().is_empty()
-                    && !content.is_empty()
-                    && !content.chars().last().is_none_or(char::is_whitespace)
-                    && !child_buf.chars().next().is_none_or(char::is_whitespace)
-                {
-                    content.push(' ');
-                }
-                content.push_str(&child_buf);
-            }
-            if content.trim().is_empty() {
-                normalize_link_label(&inline_label)
-            } else {
-                normalize_link_label(&content)
-            }
-        } else {
-            let mut content = String::new();
-            let link_ctx = Context {
-                inline_depth: ctx.inline_depth + 1,
-                ..ctx.clone()
-            };
-            for child_handle in &children {
-                walk_node(
-                    child_handle,
-                    parser,
-                    &mut content,
-                    options,
-                    &link_ctx,
-                    depth + 1,
-                    dom_ctx,
-                );
-            }
-            normalize_link_label(&content)
-        };
-
-        if label.is_empty() && saw_block {
-            let fallback = text::normalize_whitespace(&get_text_content(node_handle, parser, dom_ctx));
-            label = normalize_link_label(&fallback);
-        }
-
-        if label.is_empty() && !raw_text.is_empty() {
-            label = normalize_link_label(&raw_text);
-        }
-
-        if label.is_empty() && !href.is_empty() && !children.is_empty() {
-            label = href.clone();
-        }
-
-        // Normalize Wikipedia-style back-reference links: <a href="#cite_ref-N">^</a>
-        // These produce `[^](#cite_ref-N)` which is confusing (looks like a footnote).
-        // Convert to `[↑](#cite_ref-N)` to avoid ambiguity with markdown footnote syntax.
-        if label == "^" && href.starts_with('#') {
-            label = "↑".to_string();
-        }
-
-        let escaped_label = escape_link_label(&label);
-
-        #[cfg(feature = "visitor")]
-        let link_output = if let Some(ref visitor_handle) = 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::Link,
-                tag_name: "a".to_string(),
-                attributes,
-                depth,
-                index_in_parent,
-                parent_tag,
-                is_inline: true,
-            };
-
-            let visit_result = {
-                let mut visitor = visitor_handle.borrow_mut();
-                visitor.visit_link(&node_ctx, &href, &label, title.as_deref())
-            };
-            match visit_result {
-                VisitResult::Continue => {
-                    let mut buf = String::new();
-                    append_markdown_link(
-                        &mut buf,
-                        &escaped_label,
-                        href.as_str(),
-                        title.as_deref(),
-                        label.as_str(),
-                        options,
-                        ctx.reference_collector.as_ref(),
-                    );
-                    Some(buf)
-                }
-                VisitResult::Custom(custom) => Some(custom),
-                VisitResult::Skip => None,
-                VisitResult::Error(err) => {
-                    if ctx.visitor_error.borrow().is_none() {
-                        *ctx.visitor_error.borrow_mut() = Some(err);
-                    }
-                    None
-                }
-                VisitResult::PreserveHtml => Some(serialize_node(node_handle, parser)),
-            }
-        } else {
-            let mut buf = String::new();
-            append_markdown_link(
-                &mut buf,
-                &escaped_label,
-                href.as_str(),
-                title.as_deref(),
-                label.as_str(),
-                options,
-                ctx.reference_collector.as_ref(),
-            );
-            Some(buf)
-        };
-
-        #[cfg(not(feature = "visitor"))]
-        let link_output = {
-            let mut buf = String::new();
-            append_markdown_link(
-                &mut buf,
-                &escaped_label,
-                href.as_str(),
-                title.as_deref(),
-                label.as_str(),
-                options,
-                ctx.reference_collector.as_ref(),
-            );
-            Some(buf)
-        };
-
-        if let Some(link_text) = link_output {
-            output.push_str(&link_text);
-        }
-
-        #[cfg(feature = "metadata")]
-        if ctx.metadata_wants_links {
-            if let Some(ref collector) = ctx.metadata_collector {
-                let rel_attr = tag
-                    .attributes()
-                    .get("rel")
-                    .flatten()
-                    .map(|v| v.as_utf8_str().to_string());
-                let mut attributes_map = BTreeMap::new();
-                for (key, value_opt) in tag.attributes().iter() {
-                    let key_str = key.to_string();
-                    if key_str == "href" {
-                        continue;
-                    }
-
-                    let value = value_opt.map(|v| v.to_string()).unwrap_or_default();
-                    attributes_map.insert(key_str, value);
-                }
-                collector
-                    .borrow_mut()
-                    .add_link(href.clone(), label, title.clone(), rel_attr, attributes_map);
-            }
-        }
-    } else {
-        let children = tag.children();
-        {
-            for child_handle in children.top().iter() {
-                walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
-            }
-        }
-    }
-}

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

@@ -1,26 +0,0 @@
-//! Element handlers extracted from the main conversion pipeline.
-//!
-//! This module contains handler functions for specific HTML elements,
-//! allowing the main walk_node function to delegate to specialized handlers.
-//!
-//! Each handler takes the standard set of parameters:
-//! - `node_handle`: Reference to the DOM node
-//! - `tag`: The HTML tag being processed
-//! - `parser`: The DOM parser
-//! - `output`: The output string buffer
-//! - `options`: Conversion options
-//! - `ctx`: Conversion context
-//! - `depth`: Current tree depth
-//! - `dom_ctx`: DOM context cache
-
-pub mod blockquote;
-pub mod code_block;
-pub mod graphic;
-pub mod image;
-pub mod link;
-
-pub use blockquote::handle_blockquote;
-pub use code_block::{handle_code, handle_pre};
-pub use graphic::handle_graphic;
-pub use image::handle_img;
-pub use link::handle_link;