@graphty/graph-io 0.3.40 → 0.3.41

package/src/formats/xgmml/emit.ts

@@ -237,19 +237,7 @@ export function membersOf(
     const edges: EdgeRec[] = [];
     let dangling = 0;
     for (const member of graph.members) {
-        if (member.href === null) {
-            if (member.kind === "node") {
-                nodes.push(member);
-            } else {
-                edges.push(member);
-            }
-            continue;
-        }
-        const id = localRef(member.href);
-        let target: NodeRec | EdgeRec | undefined;
-        if (id !== null) {
-            target = member.kind === "node" ? nodeById.get(id) : edgeById.get(id);
-        }
+        const target = resolveMember(member, nodeById, edgeById);
         if (target === undefined) {
             dangling++;
         } else if (target.kind === "node") {
@@ -261,6 +249,29 @@ export function membersOf(
     return { nodes, edges, dangling };
 }
 
+/**
+ * The record a graph member stands for: itself when declared in place, else the node or edge its
+ * local `#id` reference names.
+ * @param member - the member
+ * @param nodeById - the first node declaration of each id
+ * @param edgeById - the first edge declaration of each id
+ * @returns the record, or undefined for a cross-file, malformed or dangling reference
+ */
+function resolveMember(
+    member: NodeRec | EdgeRec,
+    nodeById: ReadonlyMap<string, NodeRec>,
+    edgeById: ReadonlyMap<string, EdgeRec>,
+): NodeRec | EdgeRec | undefined {
+    if (member.href === null) {
+        return member;
+    }
+    const id = localRef(member.href);
+    if (id === null) {
+        return undefined;
+    }
+    return member.kind === "node" ? nodeById.get(id) : edgeById.get(id);
+}
+
 /**
  * The first declaration of each id.
  * @param records - node or edge records
@@ -276,6 +287,42 @@ function indexById<T extends NodeRec | EdgeRec>(records: readonly T[]): Map<stri
     return out;
 }
 
+/**
+ * Whether an edge repeats one already seen, remembering it otherwise: a repeat has the same id,
+ * or without one the same label and endpoints.
+ * @param seen - the keys seen so far, or null when repeats are kept
+ * @param record - the edge
+ * @returns true for a repeat
+ */
+function isRepeat(seen: Set<string> | null, record: EdgeRec): boolean {
+    if (seen === null) {
+        return false;
+    }
+    const key =
+        record.id ?? (record.label === null ? null : JSON.stringify([record.label, record.source, record.target]));
+    if (key === null) {
+        return false;
+    }
+    if (seen.has(key)) {
+        return true;
+    }
+    seen.add(key);
+    return false;
+}
+
+/**
+ * Whether a graph's att is one of its values: not RDF metadata, not the unnamed holder of a
+ * nested graph.
+ * @param att - the att
+ * @returns true to write it to the graph table
+ */
+function isGraphValueAtt(att: AttRec): boolean {
+    if (att.name !== null && META_ATTS.has(att.name)) {
+        return false;
+    }
+    return !(att.hasGraph && att.name === null);
+}
+
 /** Per-node state while emitting. */
 interface NodeRow {
     readonly id: string;
@@ -731,16 +778,7 @@ export class XgmmlEmitter {
             if (graph.owner !== null && isGroup(graph.owner)) {
                 continue;
             }
-            for (const member of graph.members) {
-                if (member.href === null) {
-                    continue;
-                }
-                const id = localRef(member.href);
-                const known = id !== null && (member.kind === "node" ? this.rows.has(id) : edgeIds.has(id));
-                if (!known) {
-                    dangling++;
-                }
-            }
+            dangling += this.danglingReferences(graph, edgeIds);
         }
         if (dangling > 0) {
             this.report.warning(
@@ -751,6 +789,27 @@ export class XgmmlEmitter {
         }
     }
 
+    /**
+     * How many `xlink:href` members of a graph name no node or edge of the document.
+     * @param graph - the graph
+     * @param edgeIds - the id of every edge
+     * @returns the count
+     */
+    private danglingReferences(graph: GraphRec, edgeIds: ReadonlySet<string | null>): number {
+        let dangling = 0;
+        for (const member of graph.members) {
+            if (member.href === null) {
+                continue;
+            }
+            const id = localRef(member.href);
+            const known = id !== null && (member.kind === "node" ? this.rows.has(id) : edgeIds.has(id));
+            if (!known) {
+                dangling++;
+            }
+        }
+        return dangling;
+    }
+
     /**
      * Write the per-node values: label, attributes, XML attributes, graphics, positions, nested
      * network pointers and subgraph memberships.
@@ -966,17 +1025,10 @@ export class XgmmlEmitter {
         this.decideNameLabel("edge", records);
         for (const record of records) {
             const where = { line: record.line, element: record.id ?? record.label };
-            // a repeat has the same id, or without one the same label and endpoints
-            const key =
-                record.id ??
-                (record.label === null ? null : JSON.stringify([record.label, record.source, record.target]));
-            if (dedupe !== null && key !== null) {
-                if (dedupe.has(key)) {
-                    repeats++;
-                    firstRepeat ??= where;
-                    continue;
-                }
-                dedupe.add(key);
+            if (isRepeat(dedupe, record)) {
+                repeats++;
+                firstRepeat ??= where;
+                continue;
             }
             const index = this.addEdge(record, header, ids, where);
             this.checkAbort();
@@ -1244,13 +1296,9 @@ export class XgmmlEmitter {
             }
         }
         for (const att of g.atts) {
-            if (att.name !== null && META_ATTS.has(att.name)) {
-                continue;
-            }
-            if (att.hasGraph && att.name === null) {
-                continue;
+            if (isGraphValueAtt(att)) {
+                this.graphColumns.addAtt(0, att, where);
             }
-            this.graphColumns.addAtt(0, att, where);
         }
         this.graphColumns.write(this.sink, () => 0);
         const json: Record<string, unknown> = { ...this.extras.graphJson };

package/src/formats/xgmml/exporter.ts

@@ -394,6 +394,17 @@ function slotOf(column: Column, domain: "graph" | "node" | "edge", hasLabelRole:
     if (domain !== "node") {
         return "att";
     }
+    return nodeSlotOf(column);
+}
+
+/**
+ * Where a node column goes when it is no XML attribute, label or graphics: one of the node-only
+ * slots, else a typed att.
+ * @param column - the column
+ * @returns the slot
+ */
+function nodeSlotOf(column: Column): ColumnWrite["kind"] {
+    const { meta } = column;
     if (meta.name === Z_COLUMN && meta.dtype === "f64" && meta.role === null) {
         return "z";
     }
@@ -1004,18 +1015,7 @@ function cellOf(column: Column | null, row: number): unknown {
  * @returns the coordinates as XML attributes
  */
 function coordsOf(plan: Plan, zColumn: Column | null, node: number): [string, string][] {
-    const out: [string, string][] = [];
-    const xyz = cellOf(plan.position, node);
-    if (xyz !== null) {
-        const [x, y, z] = Array.from(xyz as ArrayLike<number>);
-        const fmt = plan.position?.dtype === "f32" ? formatF32 : formatF64;
-        if (Number.isFinite(x) && Number.isFinite(y)) {
-            out.push(["x", fmt(x)], ["y", fmt(y === 0 ? 0 : -y)]);
-            if (zColumn === null && z !== undefined && z !== 0 && Number.isFinite(z)) {
-                out.push(["z", fmt(z)]);
-            }
-        }
-    }
+    const out = positionCoordsOf(plan, zColumn, node);
     const z = cellOf(zColumn, node);
     if (typeof z === "number") {
         out.push(["z", formatF64(z)]);
@@ -1023,6 +1023,34 @@ function coordsOf(plan: Plan, zColumn: Column | null, node: number): [string, st
     return out;
 }
 
+/**
+ * The graphics coordinates a node's position gives: x, y negated back to screen coordinates, and
+ * z when no z column carries it.
+ * @param plan - the plan
+ * @param zColumn - the z column, or null
+ * @param node - the node
+ * @returns the coordinates as XML attributes (none for an unset or non-finite position)
+ */
+function positionCoordsOf(plan: Plan, zColumn: Column | null, node: number): [string, string][] {
+    const xyz = cellOf(plan.position, node);
+    if (xyz === null) {
+        return [];
+    }
+    const [x, y, z] = Array.from(xyz as ArrayLike<number>);
+    if (!Number.isFinite(x) || !Number.isFinite(y)) {
+        return [];
+    }
+    const fmt = plan.position?.dtype === "f32" ? formatF32 : formatF64;
+    const out: [string, string][] = [
+        ["x", fmt(x)],
+        ["y", fmt(y === 0 ? 0 : -y)],
+    ];
+    if (zColumn === null && z !== undefined && z !== 0 && Number.isFinite(z)) {
+        out.push(["z", fmt(z)]);
+    }
+    return out;
+}
+
 /**
  * The group marker and the nested `<graph>` of a group node: its own id, label and atts, and an
  * `xlink:href` to each child.

package/src/formats/xgmml/importer.ts

@@ -174,23 +174,7 @@ export async function parseXgmml(
     common: ResolvedImportOptions,
     options: XgmmlImportOptions | undefined,
 ): Promise<XgmmlDocument> {
-    const repairs: XmlRepairs = {
-        bareAmpersand: booleanOption("repairBareAmpersands", options?.repairBareAmpersands, false)
-            ? (line): void => {
-                  report.warning("parse-error", XGMML_ISSUE.AMPERSAND_REPAIRED, "a bare & was read as &amp;", { line });
-              }
-            : undefined,
-        surrogatePair: booleanOption("pairSurrogateReferences", options?.pairSurrogateReferences, false)
-            ? (line): void => {
-                  report.warning(
-                      "parse-error",
-                      XGMML_ISSUE.SURROGATE_PAIRED,
-                      "two surrogate character references were joined into one character",
-                      { line },
-                  );
-              }
-            : undefined,
-    };
+    const repairs = repairsOf(options, report);
     const parser = new XgmmlParser(report);
     const seen = { content: false };
     try {
@@ -200,12 +184,7 @@ export async function parseXgmml(
             repairs,
         );
     } catch (err) {
-        if (err instanceof XmlSyntaxError) {
-            if (!seen.content) {
-                report.fail(XGMML_ISSUE.EMPTY_INPUT, "the input is empty");
-            }
-            report.fail(XGMML_ISSUE.XML_SYNTAX, err.message, { line: err.line });
-        }
+        failOnSyntax(err, seen.content, report);
         throw err;
     }
     const doc = parser.document();
@@ -220,6 +199,49 @@ export async function parseXgmml(
     return doc;
 }
 
+/**
+ * The XML repairs the options switch on, each reporting where it repaired.
+ * @param options - the XGMML options
+ * @param report - the report
+ * @returns the repairs
+ */
+function repairsOf(options: XgmmlImportOptions | undefined, report: ImportReportBuilder): XmlRepairs {
+    return {
+        bareAmpersand: booleanOption("repairBareAmpersands", options?.repairBareAmpersands, false)
+            ? (line): void => {
+                  report.warning("parse-error", XGMML_ISSUE.AMPERSAND_REPAIRED, "a bare & was read as &amp;", { line });
+              }
+            : undefined,
+        surrogatePair: booleanOption("pairSurrogateReferences", options?.pairSurrogateReferences, false)
+            ? (line): void => {
+                  report.warning(
+                      "parse-error",
+                      XGMML_ISSUE.SURROGATE_PAIRED,
+                      "two surrogate character references were joined into one character",
+                      { line },
+                  );
+              }
+            : undefined,
+    };
+}
+
+/**
+ * Fail the report on an XML syntax error: an input with no content is empty, any other is
+ * malformed. Any other error is left to the caller.
+ * @param err - the error the tokenizer threw
+ * @param content - whether non-whitespace text was seen
+ * @param report - the report
+ */
+function failOnSyntax(err: unknown, content: boolean, report: ImportReportBuilder): void {
+    if (!(err instanceof XmlSyntaxError)) {
+        return;
+    }
+    if (!content) {
+        report.fail(XGMML_ISSUE.EMPTY_INPUT, "the input is empty");
+    }
+    report.fail(XGMML_ISSUE.XML_SYNTAX, err.message, { line: err.line });
+}
+
 /**
  * Pass text chunks through while noting non-whitespace content.
  * @param chunks - the chunks

package/src/sniff.ts

@@ -356,40 +356,57 @@ function skeletonOf(text: string): unknown {
     }
     const root: Record<string, unknown> = {};
     for (const key of rootKeys) {
-        switch (key) {
-            case "graph":
-            case "options":
-                root[key] = objectOf(keys.get(key));
-                break;
-            case "graphs": {
-                // the keys of any node or edge of the first graph: OBO Graphs or JGF (design 1.6)
-                const nodes = keys.get("graphs[0].nodes[]");
-                const edges = keys.get("graphs[0].edges[]");
-                const graph: Record<string, unknown> = {};
-                if (nodes !== undefined) {
-                    graph.nodes = [objectOf(nodes)];
-                }
-                if (edges !== undefined) {
-                    graph.edges = [objectOf(edges)];
-                }
-                root[key] = nodes === undefined && edges === undefined ? [] : [graph];
-                break;
-            }
-            case "nodes":
-            case "edges":
-            case "links": {
-                const first = keys.get(`${key}[0]`);
-                root[key] = first === undefined ? [] : [objectOf(first)];
-                break;
-            }
-            default:
-                root[key] = true;
-                break;
-        }
+        root[key] = placeholderOf(key, keys);
     }
     return root;
 }
 
+/**
+ * The placeholder value of a root key: the shape the dialect rule tests that key for.
+ * @param key - the root key
+ * @param keys - the key sets by path
+ * @returns the placeholder
+ */
+function placeholderOf(key: string, keys: ReadonlyMap<KeyPath, Set<string>>): unknown {
+    switch (key) {
+        case "graph":
+        case "options":
+            return objectOf(keys.get(key));
+        case "graphs":
+            return graphsPlaceholder(keys);
+        case "nodes":
+        case "edges":
+        case "links": {
+            const first = keys.get(`${key}[0]`);
+            return first === undefined ? [] : [objectOf(first)];
+        }
+        default:
+            return true;
+    }
+}
+
+/**
+ * The placeholder of `graphs`: the keys of any node or edge of the first graph, which tell OBO
+ * Graphs from JGF (design 1.6).
+ * @param keys - the key sets by path
+ * @returns `[]` when neither was seen, else one graph holding them
+ */
+function graphsPlaceholder(keys: ReadonlyMap<KeyPath, Set<string>>): unknown[] {
+    const nodes = keys.get("graphs[0].nodes[]");
+    const edges = keys.get("graphs[0].edges[]");
+    if (nodes === undefined && edges === undefined) {
+        return [];
+    }
+    const graph: Record<string, unknown> = {};
+    if (nodes !== undefined) {
+        graph.nodes = [objectOf(nodes)];
+    }
+    if (edges !== undefined) {
+        graph.edges = [objectOf(edges)];
+    }
+    return [graph];
+}
+
 /**
  * An object holding placeholder values for a key set: `{}` for the keys the rule tests with
  * isJsonObject (`attributes`, `data`), `true` otherwise.
@@ -424,34 +441,6 @@ function scanKeys(text: string): Map<KeyPath, Set<string>> {
     const counts: number[] = [];
     let pendingKey: string | null = null;
     let expectingKey = false;
-    const pathOf = (): KeyPath | null => {
-        const depth = kinds.length;
-        if (depth === 1) {
-            return kinds[0] === "object" ? "" : null;
-        }
-        if (depth === 2 && kinds[0] === "array" && kinds[1] === "object" && labels[1] === "0") {
-            return "nodes[0]";
-        }
-        if (depth === 2 && kinds[0] === "object" && kinds[1] === "object") {
-            return labels[1] === "graph" || labels[1] === "options" ? labels[1] : null;
-        }
-        if (
-            depth === 5 &&
-            kinds.join() === "object,array,object,array,object" &&
-            labels[1] === "graphs" &&
-            labels[2] === "0" &&
-            (labels[3] === "nodes" || labels[3] === "edges")
-        ) {
-            return labels[3] === "nodes" ? "graphs[0].nodes[]" : "graphs[0].edges[]";
-        }
-        if (depth === 3 && kinds[0] === "object" && kinds[1] === "array" && kinds[2] === "object") {
-            const section = labels[1];
-            if ((section === "nodes" || section === "edges" || section === "links") && labels[2] === "0") {
-                return `${section}[0]`;
-            }
-        }
-        return null;
-    };
     let i = 0;
     const n = text.length;
     while (i < n) {
@@ -487,7 +476,7 @@ function scanKeys(text: string): Map<KeyPath, Set<string>> {
             pendingKey = null;
         } else if (ch === ":") {
             if (pendingKey !== null) {
-                const path = pathOf();
+                const path = keyPathOf(kinds, labels);
                 if (path !== null) {
                     let set = found.get(path);
                     if (set === undefined) {
@@ -510,6 +499,73 @@ function scanKeys(text: string): Map<KeyPath, Set<string>> {
     return found;
 }
 
+/**
+ * The watched path of the innermost open object, from the container stack.
+ * @param kinds - the kind of each open container, outermost first
+ * @param labels - the key or index each open container sits under
+ * @returns the path, or null when keys there are not watched
+ */
+function keyPathOf(kinds: readonly ("object" | "array")[], labels: readonly string[]): KeyPath | null {
+    switch (kinds.length) {
+        case 1:
+            return kinds[0] === "object" ? "" : null;
+        case 2:
+            return nestedObjectPath(kinds, labels);
+        case 3:
+            return sectionElementPath(kinds, labels);
+        case 5:
+            return graphsMemberPath(kinds, labels);
+        default:
+            return null;
+    }
+}
+
+/**
+ * At depth 2: the first element of a top-level array, or the `graph` / `options` object.
+ * @param kinds - the kind of each open container, outermost first
+ * @param labels - the key or index each open container sits under
+ * @returns the path, or null
+ */
+function nestedObjectPath(kinds: readonly ("object" | "array")[], labels: readonly string[]): KeyPath | null {
+    if (kinds[0] === "array" && kinds[1] === "object" && labels[1] === "0") {
+        return "nodes[0]";
+    }
+    if (kinds[0] === "object" && kinds[1] === "object") {
+        return labels[1] === "graph" || labels[1] === "options" ? labels[1] : null;
+    }
+    return null;
+}
+
+/**
+ * At depth 3: the first element of the root's `nodes`, `edges` or `links` array.
+ * @param kinds - the kind of each open container, outermost first
+ * @param labels - the key or index each open container sits under
+ * @returns the path, or null
+ */
+function sectionElementPath(kinds: readonly ("object" | "array")[], labels: readonly string[]): KeyPath | null {
+    if (kinds[0] !== "object" || kinds[1] !== "array" || kinds[2] !== "object" || labels[2] !== "0") {
+        return null;
+    }
+    const section = labels[1];
+    return section === "nodes" || section === "edges" || section === "links" ? `${section}[0]` : null;
+}
+
+/**
+ * At depth 5: any node or edge of the first graph of the root's `graphs` array.
+ * @param kinds - the kind of each open container, outermost first
+ * @param labels - the key or index each open container sits under
+ * @returns the path, or null
+ */
+function graphsMemberPath(kinds: readonly ("object" | "array")[], labels: readonly string[]): KeyPath | null {
+    if (kinds.join() !== "object,array,object,array,object" || labels[1] !== "graphs" || labels[2] !== "0") {
+        return null;
+    }
+    if (labels[3] === "nodes") {
+        return "graphs[0].nodes[]";
+    }
+    return labels[3] === "edges" ? "graphs[0].edges[]" : null;
+}
+
 /**
  * The index of the closing quote of a JSON string starting after its opening quote.
  * @param text - the text