Feature/attr tracking (#78)

* fix metadata tracking bug
* attribute source tracking
* full source attribute tracking; bugfix for tracking infostring source info

Author: cscheid

crates/quarto-markdown-pandoc/src/filters.rs

@@ -291,10 +291,21 @@ impl_inline_filterable_terminal!(
     Math,
     RawInline,
     Shortcode,
-    NoteReference,
-    Attr
+    NoteReference
 );
 
+// Attr is special because it has two fields (Attr, AttrSourceInfo)
+// We need a custom impl that preserves attr_source
+// However, filters don't actually work on Attr values directly,
+// so this is just a placeholder that should never be called
+impl InlineFilterableStructure for (pandoc::Attr, crate::pandoc::attr::AttrSourceInfo) {
+    fn filter_structure(self, _: &mut Filter) -> Inline {
+        // Note: This should not be called in practice because Attr inlines
+        // are stripped during postprocessing before filters run
+        Inline::Attr(self.0, self.1)
+    }
+}
+
 macro_rules! impl_inline_filterable_simple {
     ($($variant:ident),*) => {
         $(
@@ -350,6 +361,7 @@ impl InlineFilterableStructure for pandoc::Cite {
                     mode: cit.mode,
                     note_num: cit.note_num,
                     hash: cit.hash,
+                    id_source: cit.id_source,
                 })
                 .collect(),
             content: topdown_traverse_inlines(self.content, filter),
@@ -641,8 +653,22 @@ pub fn topdown_traverse_inline(inline: Inline, filter: &mut Filter) -> Inlines {
         Inline::NoteReference(note_ref) => {
             handle_inline_filter!(NoteReference, note_ref, note_reference, filter)
         }
-        Inline::Attr(attr) => {
-            handle_inline_filter!(Attr, attr, attr, filter)
+        Inline::Attr(attr, attr_source) => {
+            // Special handling for Attr since it has two fields and filters don't actually work on Attr tuples
+            // Attr inlines should be stripped during postprocessing before filters run
+            // So this branch should rarely be hit
+            if let Some(f) = &mut filter.inline {
+                let inline = Inline::Attr(attr, attr_source);
+                match f(inline.clone()) {
+                    FilterReturn::Unchanged(_) => vec![inline],
+                    FilterReturn::FilterResult(result, _should_recurse) => result,
+                }
+            } else {
+                vec![traverse_inline_structure(
+                    Inline::Attr(attr, attr_source),
+                    filter,
+                )]
+            }
         }
         Inline::Insert(ins) => {
             handle_inline_filter!(Insert, ins, insert, filter)
@@ -827,6 +853,7 @@ fn traverse_inline_nonterminal(inline: Inline, filter: &mut Filter) -> Inline {
                     mode: cit.mode,
                     note_num: cit.note_num,
                     hash: cit.hash,
+                    id_source: cit.id_source,
                 })
                 .collect(),
             content: topdown_traverse_inlines(c.content, filter),
@@ -837,12 +864,16 @@ fn traverse_inline_nonterminal(inline: Inline, filter: &mut Filter) -> Inline {
             target: l.target,
             content: topdown_traverse_inlines(l.content, filter),
             source_info: l.source_info,
+            attr_source: l.attr_source,
+            target_source: l.target_source,
         }),
         Inline::Image(i) => Inline::Image(crate::pandoc::Image {
             attr: i.attr,
             target: i.target,
             content: topdown_traverse_inlines(i.content, filter),
             source_info: i.source_info,
+            attr_source: i.attr_source,
+            target_source: i.target_source,
         }),
         Inline::Note(note) => Inline::Note(crate::pandoc::Note {
             content: topdown_traverse_blocks(note.content, filter),
@@ -852,6 +883,7 @@ fn traverse_inline_nonterminal(inline: Inline, filter: &mut Filter) -> Inline {
             attr: span.attr,
             content: topdown_traverse_inlines(span.content, filter),
             source_info: span.source_info,
+            attr_source: span.attr_source,
         }),
         _ => panic!("Unsupported inline type: {:?}", inline),
     }
@@ -870,7 +902,7 @@ pub fn traverse_inline_structure(inline: Inline, filter: &mut Filter) -> Inline
         // extensions
         Inline::Shortcode(_) => inline,
         Inline::NoteReference(_) => inline,
-        Inline::Attr(_) => inline,
+        Inline::Attr(_, _) => inline,
         _ => traverse_inline_nonterminal(inline, filter),
     }
 }
@@ -893,6 +925,7 @@ fn traverse_caption(
         long: caption
             .long
             .map(|long| topdown_traverse_blocks(long, filter)),
+        source_info: caption.source_info,
     }
 }
 

crates/quarto-markdown-pandoc/src/pandoc/attr.rs

@@ -3,6 +3,8 @@
  * Copyright (c) 2025 Posit, PBC
  */
 
+use quarto_source_map::SourceInfo;
+use serde::{Deserialize, Serialize};
 use std::collections::HashMap;
 
 pub fn empty_attr() -> Attr {
@@ -14,3 +16,50 @@ pub type Attr = (String, Vec<String>, HashMap<String, String>);
 pub fn is_empty_attr(attr: &Attr) -> bool {
     attr.0.is_empty() && attr.1.is_empty() && attr.2.is_empty()
 }
+
+/// Source location information for Attr components.
+///
+/// Attr is a tuple: (id: String, classes: Vec<String>, attributes: HashMap<String, String>)
+/// This struct tracks source locations for each component:
+/// - id: Source location of the id string (None if id is empty "")
+/// - classes: Source locations for each class string
+/// - attributes: Source locations for each key-value pair (both key and value)
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct AttrSourceInfo {
+    pub id: Option<SourceInfo>,
+    pub classes: Vec<Option<SourceInfo>>,
+    pub attributes: Vec<(Option<SourceInfo>, Option<SourceInfo>)>,
+}
+
+impl AttrSourceInfo {
+    /// Creates an empty AttrSourceInfo with no source tracking.
+    pub fn empty() -> Self {
+        AttrSourceInfo {
+            id: None,
+            classes: Vec::new(),
+            attributes: Vec::new(),
+        }
+    }
+}
+
+/// Source location information for Target components.
+///
+/// Target is a tuple: (url: String, title: String)
+/// This struct tracks source locations for each component:
+/// - url: Source location of the URL string (None if url is empty "")
+/// - title: Source location of the title string (None if title is empty "")
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct TargetSourceInfo {
+    pub url: Option<SourceInfo>,
+    pub title: Option<SourceInfo>,
+}
+
+impl TargetSourceInfo {
+    /// Creates an empty TargetSourceInfo with no source tracking.
+    pub fn empty() -> Self {
+        TargetSourceInfo {
+            url: None,
+            title: None,
+        }
+    }
+}

crates/quarto-markdown-pandoc/src/pandoc/block.rs

@@ -4,7 +4,7 @@
  */
 
 use crate::pandoc::MetaValueWithSourceInfo;
-use crate::pandoc::attr::Attr;
+use crate::pandoc::attr::{Attr, AttrSourceInfo};
 use crate::pandoc::caption::Caption;
 use crate::pandoc::inline::Inlines;
 use crate::pandoc::list::ListAttributes;
@@ -59,6 +59,7 @@ pub struct CodeBlock {
     pub attr: Attr,
     pub text: String,
     pub source_info: quarto_source_map::SourceInfo,
+    pub attr_source: AttrSourceInfo,
 }
 
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
@@ -99,6 +100,7 @@ pub struct Header {
     pub attr: Attr,
     pub content: Inlines,
     pub source_info: quarto_source_map::SourceInfo,
+    pub attr_source: AttrSourceInfo,
 }
 
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
@@ -112,13 +114,15 @@ pub struct Figure {
     pub caption: Caption,
     pub content: Blocks,
     pub source_info: quarto_source_map::SourceInfo,
+    pub attr_source: AttrSourceInfo,
 }
 
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
 pub struct Div {
     pub attr: Attr,
     pub content: Blocks,
     pub source_info: quarto_source_map::SourceInfo,
+    pub attr_source: AttrSourceInfo,
 }
 
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]

crates/quarto-markdown-pandoc/src/pandoc/caption.rs

@@ -11,4 +11,5 @@ use serde::{Deserialize, Serialize};
 pub struct Caption {
     pub short: Option<Inlines>,
     pub long: Option<Blocks>,
+    pub source_info: quarto_source_map::SourceInfo,
 }