doc/shape: add top-level contentMediaType to all shapes

crates/assemble/src/lib.rs

@@ -55,6 +55,12 @@ pub fn inference(shape: &Shape, exists: Exists) -> flow::Inference {
         shape::Redact::Unset => flow::inference::Redact::Unset,
     };
 
+    let content_media_type = shape
+        .content_media_type
+        .as_deref()
+        .unwrap_or_default()
+        .to_string();
+
     flow::Inference {
         types: shape.type_.to_vec(),
         exists: exists as i32,
@@ -66,14 +72,12 @@ pub fn inference(shape: &Shape, exists: Exists) -> flow::Inference {
             .unwrap_or_default(),
         default_json,
         secret: shape.secret.unwrap_or_default(),
+        // `content_media_type` lives at the top level. For back-compat,
+        // `Inference.String.content_type` (below) carries the same value
+        // whenever the projection's type includes "string".
         string: if shape.type_.overlaps(types::STRING) {
             Some(flow::inference::String {
-                content_type: shape
-                    .string
-                    .content_type
-                    .clone()
-                    .map(Into::into)
-                    .unwrap_or_default(),
+                content_type: content_media_type.clone(),
                 format: shape
                     .string
                     .format
@@ -135,6 +139,7 @@ pub fn inference(shape: &Shape, exists: Exists) -> flow::Inference {
         enum_json_vec,
         reduce: reduce as i32,
         redact: redact as i32,
+        content_media_type,
     }
 }
 
@@ -601,10 +606,10 @@ mod test {
             description: Some("the description".into()),
             title: Some("the title".into()),
             secret: Some(true),
+            content_media_type: Some("a/type".into()),
             string: StringShape {
                 content_encoding: Some("BaSE64".into()),
                 format: Some(json::schema::formats::Format::DateTime),
-                content_type: Some("a/type".into()),
                 min_length: 10,
                 max_length: Some(123),
             },

crates/assemble/src/snapshots/assemble__test__inference.snap

@@ -30,6 +30,7 @@ expression: "&[out1, out2, out3, out4]"
         ],
         reduce: Sum,
         redact: Sha256,
+        content_media_type: "a/type",
     },
     Inference {
         types: [
@@ -50,6 +51,7 @@ expression: "&[out1, out2, out3, out4]"
         ],
         reduce: Sum,
         redact: Sha256,
+        content_media_type: "a/type",
     },
     Inference {
         types: [
@@ -85,6 +87,7 @@ expression: "&[out1, out2, out3, out4]"
         ],
         reduce: Sum,
         redact: Sha256,
+        content_media_type: "a/type",
     },
     Inference {
         types: [
@@ -116,5 +119,6 @@ expression: "&[out1, out2, out3, out4]"
         ],
         reduce: Sum,
         redact: Sha256,
+        content_media_type: "a/type",
     },
 ]

crates/doc/src/shape/inference.rs

@@ -219,7 +219,7 @@ impl Shape {
                         shape.string.content_encoding = Some((&**enc).into());
                     }
                     Annotation::Core(CoreAnnotation::ContentMediaType(mt)) => {
-                        shape.string.content_type = Some((&**mt).into());
+                        shape.content_media_type = Some((&**mt).into());
                     }
                     Annotation::Core(CoreAnnotation::Format(format)) => {
                         shape.string.format = Some(*format);
@@ -476,6 +476,9 @@ mod test {
                 - reduce: {strategy: firstWriteWins}
                 - redact: {strategy: sha256}
                 - default: john.doe@gmail.com
+                - oneOf:
+                  - contentMediaType: some/thing
+                  - contentMediaType: some/thing
                 anyOf:
                 - contentEncoding: base64
                 - type: object # Elided (impossible).
@@ -547,9 +550,9 @@ mod test {
                     None,
                 ))),
                 secret: Some(true),
+                content_media_type: Some("some/thing".into()),
                 string: StringShape {
                     content_encoding: Some("base64".into()),
-                    content_type: Some("some/thing".into()),
                     format: Some(Format::Email),
                     max_length: None,
                     min_length: 0,
@@ -559,6 +562,25 @@ mod test {
         );
     }
 
+    #[test]
+    fn test_one_of_branch_elided_by_outer_type_constraint() {
+        // A `oneOf` branch whose type is incompatible with the outer
+        // `$ref`-imposed type set must not leak that type into the final
+        // shape. Here the `null` branch is impossible given the $ref
+        // restricts to [string, array].
+        let shape = shape_from(
+            r#"
+            $defs:
+              aDef: {type: [string, array]}
+            oneOf:
+            - type: array
+            - type: 'null'
+            $ref: '#/$defs/aDef'
+            "#,
+        );
+        assert_eq!(shape.type_, types::ARRAY);
+    }
+
     #[test]
     fn test_multiple_reduce_and_redact() {
         infer_test(
@@ -742,6 +764,7 @@ mod test {
                 provenance: Inline,
                 default: None,
                 secret: None,
+                content_media_type: None,
                 annotations: {},
                 array: ArrayShape {
                     additional_items: None,
@@ -760,7 +783,6 @@ mod test {
                 },
                 string: StringShape {
                     content_encoding: None,
-                    content_type: None,
                     format: None,
                     max_length: None,
                     min_length: 0,
@@ -888,6 +910,7 @@ mod test {
                         shape: enum_fixture(json!(["b"])),
                     }],
                     additional_properties: None,
+                    ..ObjShape::new()
                 },
                 ..Shape::anything()
             },
@@ -933,6 +956,7 @@ mod test {
                     }],
                     pattern_properties: Vec::new(),
                     additional_properties: Some(Box::new(enum_fixture(json!(["a", "b"])))),
+                    ..ObjShape::new()
                 },
                 ..Shape::anything()
             },
@@ -1133,6 +1157,7 @@ mod test {
                 numeric: NumericShape {
                     minimum: Some(5u64.into()),
                     maximum: Some(10u64.into()),
+                    ..NumericShape::new()
                 },
                 ..Shape::anything()
             },
@@ -1244,6 +1269,7 @@ mod test {
                         shape: enum_fixture(json!(["c", "d"])),
                     }],
                     additional_properties: Some(Box::new(enum_fixture(json!([1, 2])))),
+                    ..ObjShape::new()
                 },
                 ..Shape::anything()
             },
@@ -1276,6 +1302,7 @@ mod test {
                         shape: enum_fixture(json!(["c", "d"])),
                     }],
                     additional_properties: Some(Box::new(Shape::nothing())),
+                    ..ObjShape::new()
                 },
                 ..Shape::anything()
             },

crates/doc/src/shape/intersect.rs

@@ -39,7 +39,6 @@ impl StringShape {
         };
         StringShape {
             content_encoding: lhs.content_encoding.or(rhs.content_encoding),
-            content_type: lhs.content_type.or(rhs.content_type),
             format: lhs.format.or(rhs.format),
             min_length: lhs.min_length.max(rhs.min_length),
             max_length,
@@ -256,6 +255,8 @@ impl Shape {
         let default = lhs.default.or(rhs.default);
         let secret = lhs.secret.or(rhs.secret);
 
+        let content_media_type = lhs.content_media_type.or(rhs.content_media_type);
+
         let mut annotations = rhs.annotations;
         annotations.extend(lhs.annotations.into_iter());
 
@@ -298,6 +299,7 @@ impl Shape {
             provenance,
             default,
             secret,
+            content_media_type,
             annotations,
             string,
             array,

crates/doc/src/shape/mod.rs

@@ -36,6 +36,11 @@ pub struct Shape {
     pub default: Option<Box<(Value, Option<super::FailedValidation>)>>,
     /// Is this location sensitive? For example, a password or credential.
     pub secret: Option<bool>,
+    /// Annotated `contentMediaType`. The JSON-Schema specification defines
+    /// this annotation only for strings; Flow extends it to apply to any
+    /// type, so it lives at the top level rather than nested in a typed
+    /// sub-shape.
+    pub content_media_type: Option<Box<str>>,
     /// Annotations are any keywords starting with `X-` or `x-`.
     /// Their keys and values are collected here, without performing any
     /// normalization of prefix case. Technically both `x-foo` and `X-foo` may be
@@ -52,7 +57,6 @@ pub struct Shape {
 #[derive(Clone, Debug, Eq, PartialEq)]
 pub struct StringShape {
     pub content_encoding: Option<Box<str>>,
-    pub content_type: Option<Box<str>>,
     pub format: Option<Format>,
     pub max_length: Option<u32>,
     pub min_length: u32,
@@ -135,7 +139,6 @@ impl StringShape {
     pub const fn new() -> Self {
         Self {
             content_encoding: None,
-            content_type: None,
             format: None,
             max_length: None,
             min_length: 0,
@@ -187,6 +190,7 @@ impl Shape {
             provenance: Provenance::Unset,
             default: None,
             secret: None,
+            content_media_type: None,
             annotations: BTreeMap::new(),
             array: ArrayShape::new(),
             numeric: NumericShape::new(),
@@ -208,6 +212,7 @@ impl Shape {
             provenance: Provenance::Inline,
             default: None,
             secret: None,
+            content_media_type: None,
             annotations: BTreeMap::new(),
             array: ArrayShape::new(),
             numeric: NumericShape::new(),
@@ -286,7 +291,7 @@ mod test {
     fn shape_size_regression() {
         use super::{ArrayShape, ObjShape, Shape, StringShape};
         assert_eq!(std::mem::size_of::<ObjShape>(), 56);
-        assert_eq!(std::mem::size_of::<StringShape>(), 48);
+        assert_eq!(std::mem::size_of::<StringShape>(), 32);
         assert_eq!(std::mem::size_of::<ArrayShape>(), 48);
         assert_eq!(std::mem::size_of::<Shape>(), 328);
     }

crates/doc/src/shape/schema.rs

@@ -18,6 +18,7 @@ fn to_sub_schema(shape: Shape) -> Schema {
         provenance: _, // Not mapped to a schema.
         default,
         secret,
+        content_media_type,
         annotations,
         array,
         numeric,
@@ -53,6 +54,16 @@ fn to_sub_schema(shape: Shape) -> Schema {
         out.insert("default".to_string(), d.0);
     }
 
+    // The JSON-Schema spec defines `contentMediaType` only for strings;
+    // Flow extends it to apply to any type, so it lives at the top level
+    // of Shape.
+    if let Some(ct) = content_media_type {
+        out.insert(
+            keywords::CONTENT_MEDIA_TYPE.to_string(),
+            serde_json::json!(ct),
+        );
+    }
+
     // Object keywords.
     if type_.overlaps(types::OBJECT) {
         let ObjShape {
@@ -136,7 +147,6 @@ fn to_sub_schema(shape: Shape) -> Schema {
     if type_.overlaps(types::STRING) {
         let StringShape {
             content_encoding,
-            content_type,
             format,
             max_length,
             min_length,
@@ -148,12 +158,6 @@ fn to_sub_schema(shape: Shape) -> Schema {
                 serde_json::json!(encoding),
             );
         }
-        if let Some(content_type) = content_type {
-            out.insert(
-                keywords::CONTENT_MEDIA_TYPE.to_string(),
-                serde_json::json!(content_type),
-            );
-        }
         if let Some(f) = format {
             out.insert("format".to_string(), serde_json::json!(f.to_string()));
         }

crates/doc/src/shape/snapshots/doc__shape__inference__test__annotation_collection.snap

@@ -12,6 +12,7 @@ Shape {
     provenance: Inline,
     default: None,
     secret: None,
+    content_media_type: None,
     annotations: {
         "x-test-top-level": Bool(true),
     },
@@ -43,6 +44,7 @@ Shape {
                     provenance: Inline,
                     default: None,
                     secret: None,
+                    content_media_type: None,
                     annotations: {
                         "X-bar-top-level": Bool(true),
                         "x-bar-two": String("twoVal"),
@@ -64,7 +66,6 @@ Shape {
                     },
                     string: StringShape {
                         content_encoding: None,
-                        content_type: None,
                         format: None,
                         max_length: None,
                         min_length: 0,
@@ -87,6 +88,7 @@ Shape {
                     provenance: Inline,
                     default: None,
                     secret: None,
+                    content_media_type: None,
                     annotations: {
                         "x-conflicting-ann": String("yes please"),
                     },
@@ -107,7 +109,6 @@ Shape {
                     },
                     string: StringShape {
                         content_encoding: None,
-                        content_type: None,
                         format: None,
                         max_length: None,
                         min_length: 0,
@@ -128,6 +129,7 @@ Shape {
                     provenance: Inline,
                     default: None,
                     secret: None,
+                    content_media_type: None,
                     annotations: {
                         "X-foo-top-level": Bool(false),
                         "x-foo-one": String("oneVal"),
@@ -150,7 +152,6 @@ Shape {
                     },
                     string: StringShape {
                         content_encoding: None,
-                        content_type: None,
                         format: None,
                         max_length: None,
                         min_length: 0,
@@ -161,7 +162,6 @@ Shape {
     },
     string: StringShape {
         content_encoding: None,
-        content_type: None,
         format: None,
         max_length: None,
         min_length: 0,