@graphty/graph-io 0.3.17 → 0.3.19

package/src/formats/cx2/index.ts

@@ -0,0 +1,7 @@
+/**
+ * The `@graphty/graph-io/cx2` subpath (design/graph-io/cytoscape-and-obo/design.md section 1.3):
+ * the CX2 importer and exporter with their option types and code tables.
+ */
+
+export { CX2_CAPABILITIES, CX2_LOSS, cx2Exporter, type Cx2ExportOptions } from "./exporter.js";
+export { CX2_ISSUE, cx2Importer, type Cx2ImportOptions } from "./importer.js";

package/src/formats/gexf/exporter.ts

@@ -73,6 +73,8 @@ export const GEXF_LOSS = Object.freeze({
     DECLARED_TYPE: "W_DECLARED_TYPE",
     /** A node id whose text reads back as the other type under the canonical rule (design section 4.1): a non-integer number, a string of integer text. */
     ID_TEXT_TYPE: LOSS.ID_TEXT_TYPE,
+    /** A numeric edge id column: GEXF edge ids read back as strings. */
+    EDGE_ID_TEXT: "W_GEXF_EDGE_ID_TEXT",
     /** A viz role column (position, color, size, thickness) that is not f32; the importer reads viz values as f32. */
     VIZ_DTYPE: "W_GEXF_VIZ_DTYPE",
     /** A plain `weight` edge column reads back as THE weight (the importer's weightFrom default). */
@@ -267,6 +269,15 @@ function planExport(
             typeChanges,
         );
     }
+    const edgeIds = snapshot.edges.byRole("id");
+    if (edgeIds !== null && edgeIds.dtype !== "string" && edgeIds.dtype !== "dict") {
+        note(
+            GEXF_LOSS.EDGE_ID_TEXT,
+            `edge id column "${edgeIds.meta.name}" is ${edgeIds.dtype}; GEXF edge ids read back as strings`,
+            edgeIds.meta.name,
+            snapshot.edgeCount - edgeIds.nullCount,
+        );
+    }
     const tables = collectTemporalTables(snapshot, version, note);
     const nodeRoles = collectRoles(snapshot.nodes, "node", note);
     const edgeRoles = collectRoles(snapshot.edges, "edge", note);

package/src/formats/gexf/importer.ts

@@ -51,6 +51,7 @@ import {
 } from "../../common/attributes.js";
 import {
     COUNT_HINT_CODE,
+    DUPLICATE_ATTRIBUTE_CODE,
     DUPLICATE_EDGE_ID_CODE,
     DUPLICATE_NODE_CODE,
     ID_MERGED_CODE,
@@ -122,8 +123,6 @@ export const ATTRIBUTES_CLASS_CODE = "E_GEXF_ATTRIBUTES_CLASS";
 export const ATTRIBUTE_ID_CODE = "E_GEXF_ATTRIBUTE_ID";
 /** Issue code: a graph header value (`defaultedgetype`, `mode`, `timeformat`, ...) outside its set. */
 export const HEADER_VALUE_CODE = "W_GEXF_HEADER_VALUE";
-/** Issue code: a second `<attribute>` with the same id in the same class (the first is kept). */
-export const DUPLICATE_ATTRIBUTE_CODE = "W_GEXF_DUPLICATE_ATTRIBUTE";
 /** Issue code: an `<attribute>` without a type (read as string). */
 export const ATTRIBUTE_TYPE_CODE = "W_GEXF_ATTRIBUTE_TYPE";
 /** Issue code: an `<attvalue>` naming an attribute the document never declares (once per id). */
@@ -149,6 +148,7 @@ export const VIZ_VALUE_CODE = "W_GEXF_VIZ_VALUE";
 export const OPEN_BOUND_CONFLICT_CODE = "W_GEXF_OPEN_BOUND_CONFLICT";
 
 export {
+    DUPLICATE_ATTRIBUTE_CODE,
     DUPLICATE_NODE_CODE,
     ID_MERGED_CODE,
     MISSING_ENDPOINT_CODE,

package/src/formats/gexf/index.ts

@@ -11,6 +11,7 @@ import {
     COUNT_HINT_CODE,
     DIRECTION_FORCED_CODE,
     DIRECTION_REFUSED_CODE,
+    DUPLICATE_ATTRIBUTE_CODE,
     DUPLICATE_EDGE_ID_CODE,
     DUPLICATE_NODE_CODE,
     ENCODING_FALLBACK_CODE,
@@ -37,7 +38,6 @@ import {
     ATTRIBUTE_TYPE_CODE,
     ATTRIBUTES_CLASS_CODE,
     ATTVALUE_SHAPE_CODE,
-    DUPLICATE_ATTRIBUTE_CODE,
     EDGE_TYPE_CODE,
     HEADER_VALUE_CODE,
     MISSING_NODES_CODE,

package/src/formats/gml/exporter.ts

@@ -140,7 +140,7 @@ const SKIPPED_ROLES: ReadonlySet<string> = new Set([
 ]);
 
 const NODE_RESERVED: ReadonlySet<string> = new Set(["id"]);
-const EDGE_RESERVED: ReadonlySet<string> = new Set(["source", "target"]);
+const EDGE_RESERVED: ReadonlySet<string> = new Set(["source", "target", "directed"]);
 const GRAPH_RESERVED: ReadonlySet<string> = new Set(["node", "edge", "directed", "multigraph"]);
 const TOP_RESERVED: ReadonlySet<string> = new Set(["graph"]);
 const NUMBER_TEXT = /^[+-]?(?:[0-9]+(?:\.[0-9]*)?|\.[0-9]+)(?:[eE][+-]?[0-9]+)?$/;

package/src/formats/json/dialect.ts

@@ -32,17 +32,19 @@ export const JSON_DIALECTS: readonly JsonDialect[] = Object.freeze([
 ]);
 
 /**
- * The dialects the importer reads: every JsonDialect plus two it only reads, NetworkX
- * adjacency_data (`nodes` plus one neighbour list per node under `adjacency`) and tree_data (a
- * nested `id` / `children` record). The exporter writes neither; write node-link instead.
+ * The dialects the importer reads: every JsonDialect plus three it only reads, NetworkX
+ * adjacency_data (`nodes` plus one neighbour list per node under `adjacency`), tree_data (a
+ * nested `id` / `children` record) and OBO Graphs (`graphs[]` of `sub` / `pred` / `obj` edges, the
+ * JSON form of the Gene Ontology and the OBO Foundry ontologies). The exporter writes none of them.
  */
-export type JsonImportDialect = JsonDialect | "adjacency" | "tree";
+export type JsonImportDialect = JsonDialect | "adjacency" | "tree" | "obographs";
 
 /** Every dialect the importer reads, for option checking and messages. */
 export const JSON_IMPORT_DIALECTS: readonly JsonImportDialect[] = Object.freeze([
     ...JSON_DIALECTS,
     "adjacency",
     "tree",
+    "obographs",
 ]);
 
 /** The key under `meta.extra` that holds the shape record (design section 8.5). */
@@ -116,8 +118,9 @@ export function isJsonImportDialect(value: unknown): value is JsonImportDialect
 
 /**
  * The dialect of a parsed JSON document, by the shape rules of design section 8.2 (Cytoscape:
- * `elements` or a top-level array of `{ data }` elements; JGF: `graph.nodes` / `graph.edges` or
- * `graphs[]`; graphology: `options.type` / `options.multi`, `key` nodes without `id`, edges with
+ * `elements` or a top-level array of `{ data }` elements; OBO Graphs: `graphs[]` whose first graph
+ * has an edge with `sub` (or `subj`) and `obj` or a `pred`, or a node with `lbl`, `meta` or an OWL
+ * `type`; JGF: `graph.nodes` / `graph.edges` or any other `graphs[]`; graphology: `options.type` / `options.multi`, `key` nodes without `id`, edges with
  * `undirected` or an `attributes` record; vis: edges with `from` / `to`; d3: `links` without
  * `directed` / `multigraph` / `graph`; NetworkX adjacency_data: `nodes` and `adjacency` without
  * `links` / `edges`; NetworkX tree_data: `children` without `nodes` / `links` / `edges`; else
@@ -141,7 +144,7 @@ export function sniffJsonDialect(root: unknown): JsonImportDialect | null {
         return "jgf";
     }
     if (Array.isArray(root.graphs)) {
-        return "jgf";
+        return isOboGraph(firstJsonObject(root.graphs)) ? "obographs" : "jgf";
     }
     if (!hasKey(root, "edges") && !hasKey(root, "links")) {
         if (hasKey(root, "nodes") && hasKey(root, "adjacency")) {
@@ -171,6 +174,31 @@ export function sniffJsonDialect(root: unknown): JsonImportDialect | null {
     return bare && hasKey(root, "links") ? "d3" : "node-link";
 }
 
+/** The node types of OBO Graphs (the OWL entity kinds). */
+const OBOGRAPHS_NODE_TYPES: ReadonlySet<unknown> = new Set(["CLASS", "INDIVIDUAL", "PROPERTY"]);
+
+/**
+ * Whether a graph of a `graphs[]` document is an OBO Graphs graph rather than JGF: anywhere in its
+ * nodes or edges, an edge with `sub` (or the outdated `subj`) and `obj`, or a `pred`, or a node
+ * with `lbl`, `meta` or a `type` of CLASS / INDIVIDUAL / PROPERTY (design 1.6: looking only at the
+ * first node and edge misses graphs with no edges and graphs whose first node has no `lbl`).
+ * @param graph - the first graph, or null
+ * @returns true for OBO Graphs
+ */
+function isOboGraph(graph: Record<string, unknown> | null): boolean {
+    if (graph === null) {
+        return false;
+    }
+    const isEdge = (e: unknown): boolean =>
+        isJsonObject(e) && ((hasKey(e, "obj") && (hasKey(e, "sub") || hasKey(e, "subj"))) || hasKey(e, "pred"));
+    const isNode = (n: unknown): boolean =>
+        isJsonObject(n) && (hasKey(n, "lbl") || hasKey(n, "meta") || OBOGRAPHS_NODE_TYPES.has(n.type));
+    return (
+        (Array.isArray(graph.edges) && graph.edges.some(isEdge)) ||
+        (Array.isArray(graph.nodes) && graph.nodes.some(isNode))
+    );
+}
+
 /**
  * The first object of an array, or null.
  * @param value - maybe an array
@@ -223,6 +251,8 @@ export const DIALECT_DEFAULT_DIRECTED: Readonly<Record<JsonImportDialect, boolea
     // networkx adjacency_data declares `directed`; tree_graph always builds a DiGraph
     adjacency: false,
     tree: true,
+    // OBO Graphs edges point from the subject (child) to the object (parent)
+    obographs: true,
 });
 
 /**