@graphty/graph-io 0.2.4 → 0.3.0

package/dist/chunks/{importer-CpCpfbxr.js → importer-D5pAsweZ.js}

@@ -1,7 +1,7 @@
 import { GraphFormatError, INVALID_INDEX } from "@graphty/graph-format";
-import { L as LOSS, au as joinText, ah as encodeChunks, aB as resolveExportOptions, aD as sanitizeIds, aa as checkCapabilities, aw as pairFolding, a9 as capabilities, ai as explicitWeights, al as formatF64, aj as formatDecimal, ak as formatF32, an as formatInteger, a8 as canonicalId, W as SINK_OPTION_CODE, J as OPTION_IGNORED_CODE, V as ROLE_TAKEN_CODE, S as RENAMED_CODE, n as ID_MERGED_CODE, k as DUPLICATE_NODE_CODE, Z as SYNTAX_CODE, r as INVALID_UTF8_CODE, aC as resolveImportOptions, I as ImportReportBuilder, az as reportSinkOptions, aA as reportUnusedOptions, v as LineReader, a as throwIfAborted, l as DirectionResolver, s as IdCoercer, ac as coerceIdText, af as declareResolved, at as isWeightField, ax as parseWeightText } from "./writer-DxSKC7TL.js";
-import { i as isPajekLabel, q as quotePajekLabel } from "./escape-DyI8JofU.js";
-import { W as WIDENING_UNSUPPORTED_CODE, a as isNumericText, T as TextCellWriter } from "./text-CajMdVFy.js";
+import { L as LOSS, ax as joinText, ak as encodeChunks, aE as resolveExportOptions, aG as sanitizeIds, ad as checkCapabilities, az as pairFolding, ac as capabilities, al as explicitWeights, ao as formatF64, am as formatDecimal, an as formatF32, aq as formatInteger, ab as canonicalId, Y as SINK_OPTION_CODE, P as OPTION_IGNORED_CODE, X as ROLE_TAKEN_CODE, U as RENAMED_CODE, o as ID_MERGED_CODE, k as DUPLICATE_NODE_CODE, $ as SYNTAX_CODE, A as MULTIPLE_GRAPHS_CODE, a5 as UNKNOWN_ENCODING_CODE, n as ENCODING_FALLBACK_CODE, s as INVALID_ENCODING_CODE, u as INVALID_UTF8_CODE, aF as resolveImportOptions, I as ImportReportBuilder, aD as reportUnusedOptions, x as LineReader, a as throwIfAborted, aC as reportSinkOptions, l as DirectionResolver, v as IdCoercer, af as coerceIdText, ai as declareResolved, aw as isWeightField, aA as parseWeightText } from "./writer-BdMak4_J.js";
+import { i as isPajekLabel, q as quotePajekLabel } from "./escape-CKied3Ri.js";
+import { i as inferTextDtype, W as WIDENING_UNSUPPORTED_CODE, a as isNumericText, T as TextCellWriter } from "./text-Dr0Ifpag.js";
 const LABEL_COLUMN = "label";
 const POSITION_COLUMN = "position";
 const SHAPE_COLUMN = "shape";
@@ -220,7 +220,9 @@ const PAJEK_LOSS = Object.freeze({
   /** A role-less node column named `label` reads back with the label role (its values become the vertex labels). */
   ROLE_ASSUMED: LOSS.ROLE_ASSUMED,
   /** Under sanitizeIds "mangle": an original id whose text reads back as the other type under ids "canonical". */
-  ID_TEXT_TYPE: LOSS.ID_TEXT_TYPE
+  ID_TEXT_TYPE: LOSS.ID_TEXT_TYPE,
+  /** Parameter text that looks like a number or a boolean reads back as one (parameters are untyped, design 5.1). */
+  TEXT_INFERRED: LOSS.TEXT_INFERRED
 });
 const SLOT_ROLES = /* @__PURE__ */ new Set(["label"]);
 const ROLE_NAMES = Object.freeze({ label: "label" });
@@ -477,6 +479,7 @@ function checkCells(planned, domain, notes) {
     const { name } = column.meta;
     let badText = 0;
     let nonFinite = 0;
+    let typed = 0;
     for (let r = 0; r < column.length; r++) {
       if (!column.isSet(r)) {
         continue;
@@ -491,8 +494,11 @@ function checkCells(planned, domain, notes) {
           nonFinite++;
         }
       } else if (column.dtype === "string" || column.dtype === "dict") {
-        if (!isPajekLabel(column.value(r))) {
+        const text = column.value(r);
+        if (!isPajekLabel(text)) {
           badText++;
+        } else if (slot === "param" && inferTextDtype(text) !== "string") {
+          typed++;
         }
       }
     }
@@ -506,6 +512,16 @@ function checkCells(planned, domain, notes) {
         )
       );
     }
+    if (typed > 0) {
+      notes.push(
+        note(
+          PAJEK_LOSS.TEXT_INFERRED,
+          `${typed} value(s) of ${domain} column "${name}" look like numbers or booleans; Pajek parameters are untyped and they read back as such`,
+          name,
+          typed
+        )
+      );
+    }
     if (nonFinite > 0) {
       notes.push(
         note(
@@ -947,12 +963,18 @@ const pajekExporter = Object.freeze({
 const PAJEK_ISSUE = Object.freeze({
   /** The input holds invalid UTF-8 (fatal). */
   INVALID_UTF8: INVALID_UTF8_CODE,
+  /** Invalid bytes in the encoding a BOM, a declaration or the encoding option chose (fatal). */
+  INVALID_ENCODING: INVALID_ENCODING_CODE,
+  /** Bytes that are not UTF-8 and declare no encoding were read as windows-1252. */
+  ENCODING_FALLBACK: ENCODING_FALLBACK_CODE,
+  /** A declared encoding the platform cannot decode was ignored. */
+  UNKNOWN_ENCODING: UNKNOWN_ENCODING_CODE,
   /** Fatal: no `*Vertices` section (an empty file, or not a Pajek network). */
   NO_VERTICES: "E_PAJEK_NO_VERTICES",
   /** Fatal: `*Vertices` without a vertex count, or one the sink cannot hold. */
   VERTICES_COUNT: "E_PAJEK_VERTICES_COUNT",
-  /** Aborts: a second `*Vertices` or `*Network` section; only one network per file is read. */
-  MULTIPLE_NETWORKS: "E_PAJEK_MULTIPLE_NETWORKS",
+  /** A project file holds several networks; import() reads the first, importAll() reads every one. */
+  MULTIPLE_GRAPHS: MULTIPLE_GRAPHS_CODE,
   /** A data line before the first section header. */
   OUTSIDE_SECTION: "E_PAJEK_OUTSIDE_SECTION",
   /** A section header the importer cannot parse. */
@@ -1111,6 +1133,7 @@ class PajekParser {
     this.relationHandle = INVALID_INDEX;
     this.edgeSpellsHandle = INVALID_INDEX;
     this.nodeParams = /* @__PURE__ */ new Map();
+    this.ended = false;
     this.edgeParams = /* @__PURE__ */ new Map();
     this.sink = sink;
     this.report = report;
@@ -1123,15 +1146,16 @@ class PajekParser {
    * Handle one line; every problem of the line is recorded as an issue and the line skipped.
    * @param text - the line without its terminator
    * @param line - the 1-based line number
+   * @returns true when the line is the header of the next network (not consumed: this network is complete)
    */
   line(text, line) {
     if (isCommentOrBlank(text)) {
-      return;
+      return false;
     }
     try {
       if (isSectionLine(text)) {
         this.header(text, line);
-        return;
+        return this.ended;
       }
       switch (this.section) {
         case "none":
@@ -1142,18 +1166,18 @@ class PajekParser {
           );
         case "vertices":
           this.vertexLine(text, line);
-          return;
+          return false;
         case "lines":
           this.edgeLine(text, line);
-          return;
+          return false;
         case "list":
           this.listLine(text, line);
-          return;
+          return false;
         case "matrix":
           this.matrixLine(text, line);
-          return;
+          return false;
         case "skip":
-          return;
+          return false;
         default: {
           const name = this.section;
           throw new Error(`unknown section state ${name}`);
@@ -1162,9 +1186,10 @@ class PajekParser {
     } catch (err) {
       if (err instanceof LineError) {
         this.report.error(err.category, err.code, err.message, { line });
-        return;
+        return false;
       }
       this.report.recordError(err, { line });
+      return false;
     }
   }
   /**
@@ -1224,7 +1249,8 @@ class PajekParser {
     switch (h.kind) {
       case "network":
         if (this.networkName !== null || this.vertexCount >= 0) {
-          this.refuseSecondNetwork(h.keyword, line);
+          this.ended = true;
+          return;
         }
         this.networkName = h.name ?? "";
         this.section = "none";
@@ -1234,6 +1260,10 @@ class PajekParser {
           this.section = "skip";
           return;
         }
+        if (this.vertexCount >= 0) {
+          this.ended = true;
+          return;
+        }
         this.verticesHeader(h, line);
         return;
       case "arcs":
@@ -1269,9 +1299,6 @@ class PajekParser {
    * @param line - the line number
    */
   verticesHeader(h, line) {
-    if (this.vertexCount >= 0) {
-      this.refuseSecondNetwork(h.keyword, line);
-    }
     if (h.count === null) {
       this.report.fail(PAJEK_ISSUE.VERTICES_COUNT, "*Vertices needs a vertex count", { line });
     }
@@ -1330,16 +1357,6 @@ class PajekParser {
     this.kind = kind;
     this.relation = h.relation === null ? null : h.name ?? String(h.relation);
   }
-  /**
-   * Refuse a second network: an issue, then ImportError with the report so far.
-   * @param keyword - the header keyword
-   * @param line - the line number
-   */
-  refuseSecondNetwork(keyword, line) {
-    const message = `*${keyword} starts a second network; the importer reads one network per file`;
-    this.report.error("unsupported", PAJEK_ISSUE.MULTIPLE_NETWORKS, message, { line });
-    throw this.report.abort(message, { code: PAJEK_ISSUE.MULTIPLE_NETWORKS, line });
-  }
   /**
    * Checks that run when a section ends.
    */
@@ -1993,23 +2010,111 @@ const pajekImporter = Object.freeze({
     reportUnusedOptions(options, report, USED_OPTIONS);
     const parser = new PajekParser(sink, report, resolved, firstVertex);
     const reader = new LineReader(input, report, resolved);
+    let rest = null;
+    let restLine = 0;
     let sinceCheck = 0;
     for await (const text of reader) {
-      parser.line(text, reader.line);
+      if (rest !== null) {
+        rest.line(text);
+      } else if (parser.line(text, reader.line)) {
+        rest = new NetworkCounter();
+        rest.line(text);
+        restLine = reader.line;
+      }
       if (++sinceCheck >= ABORT_CHECK_INTERVAL) {
         sinceCheck = 0;
         throwIfAborted(resolved.signal);
       }
     }
     parser.finish();
+    if (rest !== null) {
+      report.warning(
+        "unsupported",
+        PAJEK_ISSUE.MULTIPLE_GRAPHS,
+        `the project file holds ${rest.count} more network(s) after the first; import() reads the first, importAll() reads every one`,
+        { line: restLine }
+      );
+    }
     throwIfAborted(resolved.signal);
     return report.finish();
+  },
+  /**
+   * Read every network of a Pajek project file (`.paj`), each into its own sink. A network starts
+   * at `*Network` or at a `*Vertices` that does not belong to a `*Partition` / `*Vector` section.
+   * @param input - the text, bytes or stream
+   * @param sinkFor - the sink of the network with this index, called before its first push
+   * @param options - common and Pajek options
+   * @returns one report per network
+   */
+  async importAll(input, sinkFor, options) {
+    const resolved = resolveImportOptions(options, DEFAULTS);
+    const firstVertex = firstVertexOption(options?.firstVertex);
+    const reports = [new ImportReportBuilder("pajek", resolved.errorLimit)];
+    reportUnusedOptions(options, reports[0], USED_OPTIONS);
+    const open = () => {
+      const report = reports[reports.length - 1];
+      const sink = sinkFor(reports.length - 1);
+      reportSinkOptions(sink, options, report);
+      return new PajekParser(sink, report, resolved, firstVertex);
+    };
+    let parser = open();
+    const reader = new LineReader(input, reports[0], resolved);
+    let sinceCheck = 0;
+    for await (const text of reader) {
+      if (parser.line(text, reader.line)) {
+        parser.finish();
+        reports.push(new ImportReportBuilder("pajek", resolved.errorLimit));
+        parser = open();
+        parser.line(text, reader.line);
+      }
+      if (++sinceCheck >= ABORT_CHECK_INTERVAL) {
+        sinceCheck = 0;
+        throwIfAborted(resolved.signal);
+      }
+    }
+    parser.finish();
+    throwIfAborted(resolved.signal);
+    return reports.map((r) => r.finish());
   }
 });
+class NetworkCounter {
+  constructor() {
+    this.count = 0;
+    this.named = false;
+    this.populated = false;
+    this.skipping = false;
+  }
+  /**
+   * Feed one line.
+   * @param text - the line
+   */
+  line(text) {
+    if (!isSectionLine(text)) {
+      return;
+    }
+    const h = parseSectionHeader(text);
+    const kind = h === null ? "unsupported" : h.kind;
+    const { skipping } = this;
+    this.skipping = kind === "unsupported" || kind === "vertices" && skipping && this.populated;
+    if (kind === "network") {
+      if (this.count === 0 || this.named || this.populated) {
+        this.count++;
+        this.populated = false;
+      }
+      this.named = true;
+    } else if (kind === "vertices" && !this.skipping) {
+      if (this.count === 0 || this.populated) {
+        this.count++;
+        this.named = false;
+      }
+      this.populated = true;
+    }
+  }
+}
 export {
   PAJEK_ISSUE as P,
   PAJEK_LOSS as a,
   pajekImporter as b,
   pajekExporter as p
 };
-//# sourceMappingURL=importer-CpCpfbxr.js.map
+//# sourceMappingURL=importer-D5pAsweZ.js.map