@graphty/graph-io 0.3.18 → 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/importer.ts

@@ -66,6 +66,7 @@ import {
 import { DirectionResolver, type EdgeKind } from "../../common/direction.js";
 import { ID_MERGED_CODE, IdCoercer } from "../../common/ids.js";
 import { readText, textChunks, throwIfAborted } from "../../common/input.js";
+import { MAYBE_UNSAFE_INTEGER, reviveNonstandard, rewriteNumbers } from "../../common/json-elements.js";
 import {
     chooseGraph,
     type ImportFormatDefaults,
@@ -77,6 +78,7 @@ import {
 } from "../../common/options.js";
 import { ImportReportBuilder } from "../../common/report.js";
 import { weightFromValue } from "../../common/weights.js";
+import { sniffJsonDialectHead } from "../../sniff.js";
 import {
     type CommonImportOptions,
     type GraphChoiceOptions,
@@ -1272,113 +1274,9 @@ function parseDocument(text: string, report: ImportReportBuilder): unknown {
     return root;
 }
 
-/** A run of 16 digits not inside a fraction: the shortest integer literal that can exceed 2^53 (9007199254740992). */
-const MAYBE_UNSAFE_INTEGER = /(?<![0-9.])[0-9]{16}/;
-
 /** How many of the integers kept as text the warning lists. */
 const BIG_INTEGERS_SHOWN = 10;
 
-/**
- * The prefix of the string a non-standard token is rewritten to (a NUL character first, which no
- * sensible attribute value starts with); reviveNonstandard() turns it back into the number.
- */
-const NONSTANDARD_SENTINEL = `${String.fromCharCode(0)}graph-io:`;
-
-/** The non-standard tokens Python's json module writes, longest first so -Infinity wins over a bare minus. */
-const NONSTANDARD_TOKENS: readonly (readonly [string, number])[] = [
-    ["-Infinity", -Infinity],
-    ["Infinity", Infinity],
-    ["NaN", NaN],
-];
-
-/** A JSON integer literal (no fraction, no exponent, no leading zero), as CANONICAL_INTEGER in common/ids.ts. */
-const INTEGER_LITERAL = /^-?(0|[1-9][0-9]*)$/;
-
-/**
- * Rewrite the numbers JSON.parse cannot read exactly, outside strings and in value positions only:
- * NaN / Infinity / -Infinity become sentinel strings, an integer literal that is not a safe
- * integer becomes a string of its digits. A container stack tells a value position (after `:`,
- * `[`, or `,` inside an array) from a key position, so `{NaN: 1}` stays invalid.
- * @param text - the document text
- * @returns the rewritten text, the non-standard tokens seen and the integer literals quoted
- */
-function rewriteNumbers(text: string): { text: string; tokens: Set<string>; bigIntegers: string[] } {
-    const parts: string[] = [];
-    const tokens = new Set<string>();
-    const bigIntegers: string[] = [];
-    const arrays: boolean[] = [];
-    let expectValue = true;
-    let copied = 0;
-    let i = 0;
-    const n = text.length;
-    while (i < n) {
-        const ch = text[i];
-        if (ch === '"') {
-            i++;
-            while (i < n && text[i] !== '"') {
-                i += text[i] === "\\" ? 2 : 1;
-            }
-            i++;
-            expectValue = false;
-            continue;
-        }
-        if (ch === "{" || ch === "[") {
-            arrays.push(ch === "[");
-            expectValue = ch === "[";
-        } else if (ch === "}" || ch === "]") {
-            arrays.pop();
-            expectValue = false;
-        } else if (ch === ":") {
-            expectValue = true;
-        } else if (ch === ",") {
-            expectValue = arrays.length > 0 && arrays[arrays.length - 1];
-        } else if (expectValue && ch !== " " && ch !== "\t" && ch !== "\n" && ch !== "\r") {
-            expectValue = false;
-            const token = NONSTANDARD_TOKENS.find(([word]) => text.startsWith(word, i));
-            let end = i;
-            let replacement: string | null = null;
-            if (token !== undefined) {
-                end = i + token[0].length;
-                tokens.add(token[0]);
-                replacement = JSON.stringify(`${NONSTANDARD_SENTINEL}${token[0]}`);
-            } else if (ch === "-" || (ch >= "0" && ch <= "9")) {
-                end = i + 1;
-                while (end < n && "0123456789+-.eE".includes(text[end])) {
-                    end++;
-                }
-                const literal = text.slice(i, end);
-                if (INTEGER_LITERAL.test(literal) && !Number.isSafeInteger(Number(literal))) {
-                    bigIntegers.push(literal);
-                    replacement = `"${literal}"`;
-                }
-            }
-            if (replacement !== null) {
-                parts.push(text.slice(copied, i), replacement);
-                copied = end;
-            }
-            i = Math.max(end, i + 1);
-            continue;
-        }
-        i++;
-    }
-    parts.push(text.slice(copied));
-    return { text: parts.join(""), tokens, bigIntegers };
-}
-
-/**
- * The JSON.parse reviver that turns the sentinel strings of rewriteNumbers() back into numbers.
- * @param _key - the member key (unused)
- * @param value - the parsed value
- * @returns the number for a sentinel string, the value otherwise
- */
-function reviveNonstandard(_key: string, value: unknown): unknown {
-    if (typeof value === "string" && value.startsWith(NONSTANDARD_SENTINEL)) {
-        const found = NONSTANDARD_TOKENS.find(([word]) => word === value.slice(NONSTANDARD_SENTINEL.length));
-        return found === undefined ? value : found[1];
-    }
-    return value;
-}
-
 /**
  * The dialect to read: the forced one, else the shape rule of sniffJsonDialect(); a document
  * that matches no dialect is a fatal E_JSON_DIALECT.
@@ -2979,7 +2877,10 @@ export const jsonImporter: GraphImporter<JsonImportOptions> = Object.freeze({
         if (!trimmed.startsWith("{") && !trimmed.startsWith("[")) {
             return 0;
         }
-        return SNIFF_KEYS.some((key) => trimmed.includes(key)) ? 0.9 : 0.5;
+        const score = SNIFF_KEYS.some((key) => trimmed.includes(key)) ? 0.9 : 0.5;
+        // a top-level array is a graph only as Cytoscape.js elements (`[{"data": ...}]`): a CX or
+        // CX2 document (`[{"metaData": ...}]`, `[{"CXVersion": ...}]`) is another format's
+        return trimmed.startsWith("[") && sniffJsonDialectHead(trimmed) !== "cytoscape" ? Math.min(score, 0.3) : score;
     },
 
     /**

package/src/index.ts

@@ -76,6 +76,16 @@ export {
     csvImporter,
     type CsvImportOptions,
 } from "./formats/csv/index.js";
+export { CX_ISSUE, cxImporter, type CxImportOptions } from "./formats/cx/index.js";
+export {
+    CX2_CAPABILITIES,
+    CX2_ISSUE,
+    CX2_LOSS,
+    cx2Exporter,
+    type Cx2ExportOptions,
+    cx2Importer,
+    type Cx2ImportOptions,
+} from "./formats/cx2/index.js";
 export {
     DOT_ISSUE,
     DOT_LOSS,

package/src/registry.ts

@@ -19,6 +19,8 @@ import {
 import { throwIfAborted } from "./common/input.js";
 import { ImportReportBuilder } from "./common/report.js";
 import { csvExporter, csvImporter } from "./formats/csv/index.js";
+import { cxImporter } from "./formats/cx/index.js";
+import { cx2Exporter, cx2Importer } from "./formats/cx2/index.js";
 import { dotExporter, dotImporter } from "./formats/dot/index.js";
 import { gexfExporter, gexfImporter } from "./formats/gexf/index.js";
 import { gmlExporter, gmlImporter } from "./formats/gml/index.js";
@@ -374,6 +376,9 @@ export function createRegistry(): FormatRegistry {
         .registerExporter(pajekExporter)
         .registerImporter(neo4jImporter)
         .registerExporter(neo4jExporter)
+        .registerImporter(cx2Importer)
+        .registerExporter(cx2Exporter)
+        .registerImporter(cxImporter)
         .registerImporter(oboImporter);
 }
 

package/src/sniff.ts

@@ -25,7 +25,18 @@ import { type JsonImportDialect, sniffJsonDialect } from "./formats/json/dialect
 import { type GraphImporter } from "./types.js";
 
 /** The format names of the eight built-in importers and exporters. */
-export type GraphFormatName = "gexf" | "graphml" | "gml" | "dot" | "pajek" | "csv" | "json" | "neo4j" | "obo";
+export type GraphFormatName =
+    | "gexf"
+    | "graphml"
+    | "gml"
+    | "dot"
+    | "pajek"
+    | "csv"
+    | "json"
+    | "neo4j"
+    | "cx2"
+    | "cx"
+    | "obo";
 
 /**
  * The built-in format names in the default registry's order, which is also the tie-break order of
@@ -41,6 +52,8 @@ export const GRAPH_FORMATS: readonly GraphFormatName[] = Object.freeze([
     "dot",
     "pajek",
     "neo4j",
+    "cx2",
+    "cx",
     "obo",
 ]);