feat(parser): expose supplemental picture data

crates/parser/src/lib.rs

@@ -197,6 +197,7 @@ pub mod contents {
     pub use crate::onenote::math_inline_object::{MathInlineObject, MathObjectType};
     pub use crate::onenote::note_tag::NoteTag;
     pub use crate::onenote::outline::{Outline, OutlineElement, OutlineGroup, OutlineItem};
+    pub use crate::onenote::picture::Picture;
     pub use crate::onenote::rich_text::{
         EmbeddedInkContainer, EmbeddedInkSpace, EmbeddedObject, ParagraphStyling, RichText,
         TextHyperlink,

crates/parser/src/one/property_set/image_node.rs

@@ -18,6 +18,7 @@ use crate::onestore::Object;
 pub(crate) struct Data {
     pub(crate) last_modified: Option<Time>,
     pub(crate) picture_container: Option<ExGuid>,
+    pub(crate) web_picture_container: Option<ExGuid>,
     pub(crate) layout_max_width: Option<f32>,
     pub(crate) layout_max_height: Option<f32>,
     pub(crate) is_layout_size_set_by_user: bool,
@@ -44,6 +45,8 @@ pub(crate) fn parse(object: &Object) -> Result<Data> {
 
     let last_modified = Time::parse(PropertyType::LastModifiedTime, object)?;
     let picture_container = ObjectReference::parse(PropertyType::PictureContainer, object)?;
+    let web_picture_container =
+        ObjectReference::parse(PropertyType::WebPictureContainer14, object)?;
     let layout_max_width = simple::parse_f32(PropertyType::LayoutMaxWidth, object)?;
     let layout_max_height = simple::parse_f32(PropertyType::LayoutMaxHeight, object)?;
     let is_layout_size_set_by_user =
@@ -73,6 +76,7 @@ pub(crate) fn parse(object: &Object) -> Result<Data> {
     let data = Data {
         last_modified,
         picture_container,
+        web_picture_container,
         layout_max_width,
         layout_max_height,
         is_layout_size_set_by_user,
@@ -96,3 +100,73 @@ pub(crate) fn parse(object: &Object) -> Result<Data> {
 
     Ok(data)
 }
+
+#[cfg(test)]
+mod tests {
+    use super::parse;
+    use crate::fsshttpb::data::cell_id::CellId;
+    use crate::fsshttpb::data::exguid::ExGuid;
+    use crate::one::property::PropertyType;
+    use crate::one::property_set::PropertySetId;
+    use crate::onestore::shared::compact_id::CompactId;
+    use crate::onestore::shared::object_prop_set::ObjectPropSet;
+    use crate::onestore::shared::prop_set::PropertySet;
+    use crate::onestore::{MappingTable, Object};
+    use crate::reader::Reader;
+    use std::rc::Rc;
+
+    struct TestMapping;
+
+    impl MappingTable for TestMapping {
+        fn resolve_id(&self, index: usize, _cid: &CompactId) -> Option<ExGuid> {
+            Some(ExGuid {
+                value: index as u32 + 1,
+                ..Default::default()
+            })
+        }
+
+        fn get_object_space(&self, _index: usize, _cid: &CompactId) -> Option<CellId> {
+            None
+        }
+    }
+
+    #[test]
+    fn parses_primary_and_web_picture_references() {
+        let property_ids = [
+            PropertyType::PictureContainer,
+            PropertyType::WebPictureContainer14,
+        ];
+        let mut bytes = Vec::new();
+        bytes.extend_from_slice(&(property_ids.len() as u16).to_le_bytes());
+        for property_id in property_ids {
+            bytes.extend_from_slice(&(property_id as u32).to_le_bytes());
+        }
+
+        let mut reader = Reader::new(&bytes);
+        let object = Object {
+            context_id: Default::default(),
+            jc_id: PropertySetId::ImageNode.as_jcid(),
+            props: ObjectPropSet {
+                object_ids: vec![
+                    CompactId {
+                        n: 0,
+                        guid_index: 0,
+                    },
+                    CompactId {
+                        n: 1,
+                        guid_index: 0,
+                    },
+                ],
+                properties: PropertySet::parse(&mut reader).unwrap(),
+                ..Default::default()
+            },
+            file_data: None,
+            mapping: Rc::new(TestMapping),
+        };
+
+        let data = parse(&object).unwrap();
+
+        assert_eq!(data.picture_container.unwrap().value, 1);
+        assert_eq!(data.web_picture_container.unwrap().value, 2);
+    }
+}

crates/parser/src/onenote/embedded_file.rs

@@ -4,6 +4,7 @@ use crate::one::property::file_type::FileType;
 use crate::one::property_set::{embedded_file_container, embedded_file_node};
 use crate::onenote::ParserContext;
 use crate::onenote::note_tag::{NoteTag, parse_note_tags};
+use crate::onenote::picture::Picture;
 use crate::onestore::ObjectSpace;
 use crate::onestore::shared::file_blob::{FileBlob, FileDataStatus};
 
@@ -17,6 +18,7 @@ pub struct EmbeddedFile {
     pub(crate) filename: String,
     pub(crate) file_type: FileType,
     pub(crate) data: FileBlob,
+    pub(crate) icon: Option<Picture>,
 
     pub(crate) layout_max_width: Option<f32>,
     pub(crate) layout_max_height: Option<f32>,
@@ -86,6 +88,11 @@ impl EmbeddedFile {
         self.data.status()
     }
 
+    /// The icon stored by OneNote for this embedded file, when available.
+    pub fn icon(&self) -> Option<&Picture> {
+        self.icon.as_ref()
+    }
+
     /// The max width of the embedded file's icon in half-inch increments.
     ///
     /// See [\[MS-ONE\] 2.3.21].
@@ -138,12 +145,31 @@ pub(crate) fn parse_embedded_file(
         .ok_or_else(|| ErrorKind::MalformedOneNoteData("embedded file is missing".into()))?;
     let node = embedded_file_node::parse(node_object)?;
 
+    let icon = node.picture_container.and_then(|container_id| {
+        let result = space
+            .get_object(container_id)
+            .ok_or_else(|| {
+                ErrorKind::MalformedOneNoteData("embedded file icon container is missing".into())
+                    .into()
+            })
+            .and_then(crate::one::property_set::picture_container::parse);
+
+        match result {
+            Ok(data) => Some(Picture::new(data.data, data.extension)),
+            Err(err) => {
+                warn!(ctx, "could not parse embedded file icon: {err}");
+                None
+            }
+        }
+    });
+
     // Helper function to create a fallback for corrupted files
     let create_fallback = |filename: String| -> Result<EmbeddedFile> {
         Ok(EmbeddedFile {
             filename,
             file_type: node.file_type,
             data: FileBlob::missing(),
+            icon: icon.clone(),
             layout_max_width: node.layout_max_width,
             layout_max_height: node.layout_max_height,
             offset_horizontal: node.offset_from_parent_horiz,
@@ -184,6 +210,7 @@ pub(crate) fn parse_embedded_file(
         filename: embedded_filename,
         file_type: node.file_type,
         data: container.into_value(),
+        icon,
         layout_max_width: node.layout_max_width,
         layout_max_height: node.layout_max_height,
         offset_horizontal: node.offset_from_parent_horiz,

crates/parser/src/onenote/image.rs

@@ -5,6 +5,7 @@ use crate::one::property_set::{image_node, picture_container};
 use crate::onenote::ParserContext;
 use crate::onenote::iframe::{IFrame, parse_iframe};
 use crate::onenote::note_tag::{NoteTag, parse_note_tags};
+use crate::onenote::picture::Picture;
 use crate::onestore::ObjectSpace;
 use crate::onestore::shared::file_blob::{FileBlob, FileDataStatus};
 
@@ -17,6 +18,7 @@ use crate::onestore::shared::file_blob::{FileBlob, FileDataStatus};
 pub struct Image {
     pub(crate) data: Option<FileBlob>,
     pub(crate) extension: Option<String>,
+    pub(crate) web_picture: Option<Picture>,
 
     pub(crate) layout_max_width: Option<f32>,
     pub(crate) layout_max_height: Option<f32>,
@@ -57,12 +59,12 @@ impl Image {
     /// [`data_status`](Self::data_status) before reading to distinguish
     /// invalid data from a valid zero-byte payload.
     pub fn read(&self) -> Option<Box<dyn std::io::Read>> {
-        self.data.as_ref().map(|blob| blob.read())
+        self.data.as_ref().map(FileBlob::read)
     }
 
     /// The size of the image in bytes, or `None` if not yet uploaded.
     pub fn size(&self) -> Option<u64> {
-        self.data.as_ref().map(|blob| blob.size())
+        self.data.as_ref().map(FileBlob::size)
     }
 
     /// Availability of the image's binary data.
@@ -81,6 +83,14 @@ impl Image {
         self.extension.as_deref()
     }
 
+    /// A browser-compatible picture stored as a fallback for the primary image.
+    ///
+    /// OneNote may provide this when the primary representation uses a format
+    /// that browsers and other consumers cannot display directly.
+    pub fn web_picture(&self) -> Option<&Picture> {
+        self.web_picture.as_ref()
+    }
+
     /// The maximum width to display the image in half-inch increments.
     ///
     /// See [\[MS-ONE\] 2.3.21].
@@ -256,6 +266,23 @@ pub(crate) fn parse_image(
         (None, None)
     };
 
+    let web_picture = node.web_picture_container.and_then(|container_id| {
+        let result = space
+            .get_object(container_id)
+            .ok_or_else(|| {
+                ErrorKind::MalformedOneNoteData("web picture container is missing".into()).into()
+            })
+            .and_then(picture_container::parse);
+
+        match result {
+            Ok(data) => Some(Picture::new(data.data, data.extension)),
+            Err(err) => {
+                warn!(ctx, "could not parse web picture fallback: {err}");
+                None
+            }
+        }
+    });
+
     let embed = node
         .iframe
         .into_iter()
@@ -265,6 +292,7 @@ pub(crate) fn parse_image(
     let image = Image {
         data,
         extension,
+        web_picture,
         layout_max_width: node.layout_max_width,
         layout_max_height: node.layout_max_height,
         alt_text: node.alt_text,

crates/parser/src/onenote/mod.rs

@@ -40,6 +40,7 @@ pub(crate) mod outline;
 pub(crate) mod page;
 pub(crate) mod page_content;
 pub(crate) mod page_series;
+pub(crate) mod picture;
 pub(crate) mod rich_text;
 pub(crate) mod section;
 pub(crate) mod table;

crates/parser/src/onenote/picture.rs

@@ -0,0 +1,62 @@
+use crate::onestore::shared::file_blob::{FileBlob, FileDataStatus};
+
+/// A stored picture representation associated with a OneNote object.
+///
+/// Some objects carry supplemental pictures, such as a browser-compatible
+/// fallback for an image or the icon for an embedded file. The bytes remain
+/// backed by the parser's lazy file source and are read only on demand.
+#[derive(Clone, PartialEq, Debug)]
+pub struct Picture {
+    pub(crate) data: FileBlob,
+    pub(crate) extension: Option<String>,
+}
+
+impl Picture {
+    pub(crate) fn new(data: FileBlob, extension: Option<String>) -> Self {
+        Self { data, extension }
+    }
+
+    /// A [`std::io::Read`] over the picture's binary data.
+    pub fn read(&self) -> Box<dyn std::io::Read> {
+        self.data.read()
+    }
+
+    /// The size of the picture in bytes.
+    pub fn size(&self) -> u64 {
+        self.data.size()
+    }
+
+    /// Availability of the picture's binary data.
+    pub fn data_status(&self) -> FileDataStatus {
+        self.data.status()
+    }
+
+    /// The picture's file extension, when stored by OneNote.
+    pub fn extension(&self) -> Option<&str> {
+        self.extension.as_deref()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::Picture;
+    use crate::fs::file_source::BytesSource;
+    use crate::onestore::shared::file_blob::{FileBlob, FileDataStatus};
+    use bytes::Bytes;
+    use std::io::Read;
+    use std::sync::Arc;
+
+    #[test]
+    fn exposes_lazy_picture_data_and_metadata() {
+        let source = Arc::new(BytesSource::new(Bytes::from_static(b"picture")));
+        let picture = Picture::new(FileBlob::from_source(source, 0, 7), Some("png".to_owned()));
+
+        let mut bytes = Vec::new();
+        picture.read().read_to_end(&mut bytes).unwrap();
+
+        assert_eq!(bytes, b"picture");
+        assert_eq!(picture.size(), 7);
+        assert_eq!(picture.data_status(), FileDataStatus::Available);
+        assert_eq!(picture.extension(), Some("png"));
+    }
+}

crates/parser/tests/snapshots/lib__onenote_2016_parse_notebook.snap

@@ -1,6 +1,6 @@
 ---
 source: crates/parser/tests/lib.rs
-assertion_line: 70
+assertion_line: 84
 expression: parser.parse_section(path).unwrap()
 ---
 Section {
@@ -639,6 +639,12 @@ Section {
                                                         filename: "testing.docx",
                                                         file_type: Unknown,
                                                         data: FileBlob [ 504b030414000600080000002100cd625ce778010000a5060000130008025b43 ... 637573746f6d2e786d6c504b0506000000000e000e0081030000ad3500000000; 14 KiB ],
+                                                        icon: Some(
+                                                            Picture {
+                                                                data: FileBlob [ 89504e470d0a1a0a0000000d4948445200000020000000200806000000737a7a ... 978e886a15139b10a9a10e00fadb902d8aa71aea0000000049454e44ae426082; 0 KiB ],
+                                                                extension: None,
+                                                            },
+                                                        ),
                                                         layout_max_width: None,
                                                         layout_max_height: None,
                                                         offset_horizontal: None,

crates/parser/tests/snapshots/lib__parse_notebook.snap

@@ -5933,6 +5933,7 @@ Notebook {
                                                                     extension: Some(
                                                                         ".jpg",
                                                                     ),
+                                                                    web_picture: None,
                                                                     layout_max_width: Some(
                                                                         12.5,
                                                                     ),

crates/parser/tests/snapshots/lib__parse_notebook_new.snap

@@ -5933,6 +5933,7 @@ Notebook {
                                                                     extension: Some(
                                                                         ".jpg",
                                                                     ),
+                                                                    web_picture: None,
                                                                     layout_max_width: Some(
                                                                         12.5,
                                                                     ),

crates/parser/tests/snapshots/lib__parse_section.snap

@@ -5930,6 +5930,7 @@ Section {
                                                         extension: Some(
                                                             ".jpg",
                                                         ),
+                                                        web_picture: None,
                                                         layout_max_width: Some(
                                                             12.5,
                                                         ),

crates/parser/tests/snapshots/lib__parse_section_handwriting_recognition.snap

@@ -5928,6 +5928,7 @@ Section {
                                                             FileBlob [ ffd8ffe000104a46494600010101004800480000ffdb00430003020203020203 ... f43c7f77dde2727c5dff0041dbc3cd9cfe99cdce2f4ce2f57c4feafb3e0fffd9; 88 KiB ],
                                                         ),
                                                         extension: None,
+                                                        web_picture: None,
                                                         layout_max_width: Some(
                                                             12.5,
                                                         ),

crates/parser/tests/snapshots/lib__parse_section_with_image_missing_last_modified.snap

@@ -8438,6 +8438,7 @@ Section {
                                                         extension: Some(
                                                             ".png",
                                                         ),
+                                                        web_picture: None,
                                                         layout_max_width: Some(
                                                             10.0,
                                                         ),