parse yaml markdown and interpret tags

cscheid · cscheid · commit 9a0fc16db583 · 2025-10-16T21:12:59.000-05:00
diff --git a/crates/quarto-markdown-pandoc/src/pandoc/meta.rs b/crates/quarto-markdown-pandoc/src/pandoc/meta.rs
@@ -4,10 +4,12 @@
  */
 
 use crate::pandoc::block::Blocks;
-use crate::pandoc::inline::Inlines;
+use crate::pandoc::inline::{Inline, Inlines, Span, Str};
+use crate::pandoc::location::empty_source_info;
 use crate::readers;
 use crate::{pandoc::RawBlock, utils::output::VerboseOutput};
 use hashlink::LinkedHashMap;
+use std::collections::HashMap;
 use std::{io, mem};
 use yaml_rust2::parser::{Event, MarkedEventReceiver, Parser};
 
@@ -82,7 +84,29 @@ impl YamlEventHandler {
         }
     }
 
-    fn parse_scalar(&self, s: &str) -> MetaValue {
+    fn parse_scalar(&self, s: &str, tag: Option<yaml_rust2::parser::Tag>) -> MetaValue {
+        // Check if this scalar has a YAML tag (like !path, !glob, !str)
+        if let Some(t) = tag {
+            // Tagged strings bypass markdown parsing - wrap in Span immediately
+            let mut attributes = HashMap::new();
+            attributes.insert("tag".to_string(), t.suffix.clone());
+
+            let span = Span {
+                attr: (
+                    String::new(),
+                    vec!["yaml-tagged-string".to_string()],
+                    attributes,
+                ),
+                content: vec![Inline::Str(Str {
+                    text: s.to_string(),
+                    source_info: empty_source_info(),
+                })],
+                source_info: empty_source_info(),
+            };
+            return MetaValue::MetaInlines(vec![Inline::Span(span)]);
+        }
+
+        // Untagged scalars: parse as booleans or strings (will be parsed as markdown later)
         if s == "true" {
             MetaValue::MetaBool(true)
         } else if s == "false" {
@@ -116,12 +140,12 @@ impl MarkedEventReceiver for YamlEventHandler {
                     self.push_value(MetaValue::MetaList(list));
                 }
             }
-            Event::Scalar(s, ..) => match self.stack.last_mut() {
+            Event::Scalar(s, _style, _anchor, tag) => match self.stack.last_mut() {
                 Some(ContextFrame::Map(_, key_slot @ None)) => {
                     *key_slot = Some(s.to_string());
                 }
                 Some(ContextFrame::Map(_, Some(_))) | Some(ContextFrame::List(_)) => {
-                    let value = self.parse_scalar(&s);
+                    let value = self.parse_scalar(&s, tag);
                     self.push_value(value);
                 }
                 _ => {}
@@ -187,10 +211,22 @@ pub fn parse_metadata_strings(meta: MetaValue, outer_metadata: &mut Meta) -> Met
                     }
                     MetaValue::MetaBlocks(pandoc.blocks)
                 }
-                _ => panic!(
-                    "(unimplemented syntax error, this is a bug!) Failed to parse metadata string as markdown: {}",
-                    s
-                ),
+                Err(_) => {
+                    // Markdown parse failed - wrap in Span with class "yaml-markdown-syntax-error"
+                    let span = Span {
+                        attr: (
+                            String::new(),
+                            vec!["yaml-markdown-syntax-error".to_string()],
+                            HashMap::new(),
+                        ),
+                        content: vec![Inline::Str(Str {
+                            text: s.clone(),
+                            source_info: empty_source_info(),
+                        })],
+                        source_info: empty_source_info(),
+                    };
+                    MetaValue::MetaInlines(vec![Inline::Span(span)])
+                }
             }
         }
         MetaValue::MetaList(list) => {
diff --git a/crates/quarto-markdown-pandoc/src/pandoc/mod.rs b/crates/quarto-markdown-pandoc/src/pandoc/mod.rs
@@ -37,5 +37,5 @@ pub use crate::pandoc::table::{
 };
 
 pub use crate::pandoc::ast_context::ASTContext;
-pub use crate::pandoc::meta::{Meta, MetaValue, rawblock_to_meta};
+pub use crate::pandoc::meta::{Meta, MetaValue, parse_metadata_strings, rawblock_to_meta};
 pub use crate::pandoc::treesitter::treesitter_to_pandoc;
diff --git a/crates/quarto-markdown-pandoc/tests/test_meta.rs b/crates/quarto-markdown-pandoc/tests/test_meta.rs
@@ -3,8 +3,11 @@
  * Copyright (c) 2025 Posit, PBC
  */
 
+use hashlink::LinkedHashMap;
 use quarto_markdown_pandoc::pandoc::location::{Location, Range, SourceInfo};
-use quarto_markdown_pandoc::pandoc::{MetaValue, RawBlock, rawblock_to_meta};
+use quarto_markdown_pandoc::pandoc::{
+    Inline, MetaValue, RawBlock, parse_metadata_strings, rawblock_to_meta,
+};
 use std::fs;
 
 #[test]
@@ -54,3 +57,183 @@ fn test_metadata_parsing() {
         Some(MetaValue::MetaList(_))
     ));
 }
+
+#[test]
+fn test_yaml_tagged_strings() {
+    // Test that YAML tags (!path, !glob, !str) prevent markdown parsing
+    let content = fs::read_to_string("tests/yaml-tagged-strings.qmd").unwrap();
+
+    let block = RawBlock {
+        format: "quarto_minus_metadata".to_string(),
+        text: content,
+        source_info: SourceInfo::with_range(Range {
+            start: Location {
+                offset: 0,
+                row: 0,
+                column: 0,
+            },
+            end: Location {
+                offset: 0,
+                row: 0,
+                column: 0,
+            },
+        }),
+    };
+
+    let mut meta = rawblock_to_meta(block);
+    let mut outer_meta = LinkedHashMap::new();
+
+    // Parse metadata strings
+    for (k, v) in meta.drain() {
+        let parsed = parse_metadata_strings(v, &mut outer_meta);
+        outer_meta.insert(k, parsed);
+    }
+
+    // Check plain_path - should be MetaInlines with Span wrapper
+    let plain_path = outer_meta.get("plain_path").expect("plain_path not found");
+    if let MetaValue::MetaInlines(inlines) = plain_path {
+        assert_eq!(inlines.len(), 1, "Expected exactly one inline");
+        if let Inline::Span(span) = &inlines[0] {
+            assert!(span.attr.1.contains(&"yaml-tagged-string".to_string()));
+            assert_eq!(span.attr.2.get("tag"), Some(&"path".to_string()));
+            // Extract the string content
+            if let Inline::Str(s) = &span.content[0] {
+                assert_eq!(s.text, "images/neovim-*.png");
+            } else {
+                panic!("Expected Str inline inside Span");
+            }
+        } else {
+            panic!("Expected Span inline, got: {:?}", inlines[0]);
+        }
+    } else {
+        panic!("Expected MetaInlines for plain_path");
+    }
+
+    // Check glob_pattern
+    let glob_pattern = outer_meta
+        .get("glob_pattern")
+        .expect("glob_pattern not found");
+    if let MetaValue::MetaInlines(inlines) = glob_pattern {
+        if let Inline::Span(span) = &inlines[0] {
+            assert_eq!(span.attr.2.get("tag"), Some(&"glob".to_string()));
+            if let Inline::Str(s) = &span.content[0] {
+                assert_eq!(s.text, "posts/*/index.qmd");
+            }
+        }
+    }
+
+    // Check literal_string
+    let literal_string = outer_meta
+        .get("literal_string")
+        .expect("literal_string not found");
+    if let MetaValue::MetaInlines(inlines) = literal_string {
+        if let Inline::Span(span) = &inlines[0] {
+            assert_eq!(span.attr.2.get("tag"), Some(&"str".to_string()));
+            if let Inline::Str(s) = &span.content[0] {
+                assert_eq!(s.text, "_foo_.py");
+            }
+        }
+    }
+
+    // Check regular_markdown - should have parsed markdown (Emph element)
+    let regular_markdown = outer_meta
+        .get("regular_markdown")
+        .expect("regular_markdown not found");
+    if let MetaValue::MetaInlines(inlines) = regular_markdown {
+        // Should contain Emph for *emphasis*
+        let has_emph = inlines
+            .iter()
+            .any(|inline| matches!(inline, Inline::Emph(_)));
+        assert!(
+            has_emph,
+            "regular_markdown should have Emph element from *emphasis*"
+        );
+    } else {
+        panic!("Expected MetaInlines for regular_markdown");
+    }
+}
+
+#[test]
+fn test_yaml_markdown_parse_failure() {
+    // Test that untagged strings that fail markdown parsing are gracefully handled
+    let content = fs::read_to_string("tests/yaml-markdown-parse-failure.qmd").unwrap();
+
+    let block = RawBlock {
+        format: "quarto_minus_metadata".to_string(),
+        text: content,
+        source_info: SourceInfo::with_range(Range {
+            start: Location {
+                offset: 0,
+                row: 0,
+                column: 0,
+            },
+            end: Location {
+                offset: 0,
+                row: 0,
+                column: 0,
+            },
+        }),
+    };
+
+    let mut meta = rawblock_to_meta(block);
+    let mut outer_meta = LinkedHashMap::new();
+
+    // Parse metadata strings - this should not panic
+    for (k, v) in meta.drain() {
+        let parsed = parse_metadata_strings(v, &mut outer_meta);
+        outer_meta.insert(k, parsed);
+    }
+
+    // Check untagged_path - should be wrapped in error span
+    let untagged_path = outer_meta
+        .get("untagged_path")
+        .expect("untagged_path not found");
+    if let MetaValue::MetaInlines(inlines) = untagged_path {
+        if let Inline::Span(span) = &inlines[0] {
+            assert!(
+                span.attr
+                    .1
+                    .contains(&"yaml-markdown-syntax-error".to_string())
+            );
+            if let Inline::Str(s) = &span.content[0] {
+                assert_eq!(s.text, "posts/*/index.qmd");
+            }
+        } else {
+            panic!("Expected Span inline for failed parse");
+        }
+    } else {
+        panic!("Expected MetaInlines for untagged_path");
+    }
+
+    // Check another_glob - should also be wrapped in error span
+    let another_glob = outer_meta
+        .get("another_glob")
+        .expect("another_glob not found");
+    if let MetaValue::MetaInlines(inlines) = another_glob {
+        if let Inline::Span(span) = &inlines[0] {
+            assert!(
+                span.attr
+                    .1
+                    .contains(&"yaml-markdown-syntax-error".to_string())
+            );
+            if let Inline::Str(s) = &span.content[0] {
+                assert_eq!(s.text, "images/*.png");
+            }
+        }
+    }
+
+    // Check underscore_file - this one should successfully parse as markdown with Emph
+    let underscore_file = outer_meta
+        .get("underscore_file")
+        .expect("underscore_file not found");
+    if let MetaValue::MetaInlines(inlines) = underscore_file {
+        // _foo_ should become Emph element
+        let has_emph = inlines
+            .iter()
+            .any(|inline| matches!(inline, Inline::Emph(_)));
+        assert!(
+            has_emph,
+            "underscore_file should have Emph element from _foo_"
+        );
+    }
+}
diff --git a/crates/quarto-markdown-pandoc/tests/yaml-markdown-parse-failure.qmd b/crates/quarto-markdown-pandoc/tests/yaml-markdown-parse-failure.qmd
@@ -0,0 +1,8 @@
+---
+title: Test Markdown Parse Failure Fallback
+untagged_path: posts/*/index.qmd
+another_glob: images/*.png
+underscore_file: _foo_.py
+---
+
+Test document for graceful handling of markdown parse failures in untagged strings.
diff --git a/crates/quarto-markdown-pandoc/tests/yaml-tagged-strings.qmd b/crates/quarto-markdown-pandoc/tests/yaml-tagged-strings.qmd
@@ -0,0 +1,9 @@
+---
+title: Test YAML Tagged Strings
+plain_path: !path images/neovim-*.png
+glob_pattern: !glob posts/*/index.qmd
+literal_string: !str _foo_.py
+regular_markdown: This has *emphasis*
+---
+
+Test document for YAML tag support.
diff --git a/docs/syntax/index.qmd b/docs/syntax/index.qmd
@@ -13,3 +13,4 @@ The features documented here are currently under development. The syntax and beh
 - [Definition Lists](definition-lists.qmd) - Create definition lists using an embedded markdown DSL
 - [Editorial Marks](editorial-marks.qmd) - Annotate text with highlights, insertions, deletions, and comments
 - [Footnotes](footnotes.qmd) - Add footnotes with inline or fenced block syntax
+- [YAML Metadata](yaml-metadata.qmd) - Control markdown parsing in metadata with YAML tags
diff --git a/docs/syntax/yaml-metadata.qmd b/docs/syntax/yaml-metadata.qmd
