Jyotisha/frontend/tests/reader-appendix-language-parity.test.ts

import assert from "node:assert/strict";
import { spawnSync } from "node:child_process";
import { readFileSync } from "node:fs";
import test from "node:test";
import { fileURLToPath } from "node:url";
import { cleanReaderAppendixMarkdown } from "../src/lib/reader-appendix-language.ts";

const golden = JSON.parse(readFileSync(new URL("./fixtures/report-density-fictional-reader.json", import.meta.url), "utf8"));
// Vocabulary-only cases, not invented calculation fixtures.
const pageCases = [
  "PL9 第 42 页", "PL9第42页", "PL9.pdf 第 42 页", "PL9.pdf第42页后文",
  "参见PL9 第 42 页。", "PL9 第 43–44 页", "PL9 第 40 / 48 页",
  "PL9 pages 99-100", "PL9 p-42", "PL9第42页中的原始字段",
];
const chartFence = '  ```jyotish-chart\r\n{"title":"PL9 第 42 页 parameter_sensitive","id":"D1"}\r\n  ```\r\n';
const copyCases = [
  ...pageCases,
  "中文parameter_sensitive中文 PyJHora中文 cmd_full_reading后文",
  "some_parameter_sensitive_field PL99 XPL9 PL9_suffix",
  "The blocked / executed / available / computed result is not a claim.",
  "| blocked | `executed` | available | computed | sign_cn | 12.50 |\n",
  `parameter_sensitive\r\n${chartFence}| parameter_sensitive |\r\n`,
];

function pythonClean(inputs: string[]): string[] {
  const script = [
    "import json, sys",
    "from scripts.reader_appendix_language import clean_reader_appendix_markdown",
    "sys.stdout.write(json.dumps([clean_reader_appendix_markdown(value) for value in json.load(sys.stdin)], ensure_ascii=False))",
  ].join("\n");
  const result = spawnSync(process.env.PYTHON ?? (process.platform === "win32" ? "python" : "python3"), ["-c", script], {
    cwd: fileURLToPath(new URL("../../", import.meta.url)),
    encoding: "utf8", input: JSON.stringify(inputs), maxBuffer: 16 * 1024 * 1024,
    env: { ...process.env, PYTHONIOENCODING: "utf-8", PYTHONDONTWRITEBYTECODE: "1" },
  });
  assert.equal(result.status, 0, `${result.error ?? ""}\n${result.stderr}`);
  return JSON.parse(result.stdout);
}

test("appendix Python and TS share byte-identical golden and Chinese-boundary output", () => {
  assert.equal(golden.fixtureProvenance.fictional, true);
  const inputs = [golden.markdown, golden.reader_dasha_applicability, ...copyCases] as string[];
  const python = pythonClean(inputs);
  for (const [index, input] of inputs.entries()) {
    const ts = cleanReaderAppendixMarkdown(input);
    assert.deepEqual(Buffer.from(python[index], "utf8"), Buffer.from(ts, "utf8"), `case ${index}`);
    assert.equal(cleanReaderAppendixMarkdown(ts), ts, `idempotent case ${index}`);
  }
  assert.doesNotMatch(python[0], /第\s*\d+(?:\s*[–—/-]\s*\d+)*\s*页/);
  assert.doesNotMatch(cleanReaderAppendixMarkdown(golden.markdown), /第\s*\d+(?:\s*[–—/-]\s*\d+)*\s*页/);
});

test("appendix page references disappear without swallowing adjacent Chinese prose", () => {
  for (const source of pageCases) {
    const output = cleanReaderAppendixMarkdown(source);
    assert.doesNotMatch(output, /PL9|第\s*\d+|pages?\s*\d+|p-42/);
  }
  assert.equal(cleanReaderAppendixMarkdown("参见PL9第42页中的原始字段"), "参见外部参照资料中的原始字段");
  assert.equal(cleanReaderAppendixMarkdown("PL9.pdf第42页后文"), "外部参照资料后文");
  assert.equal(cleanReaderAppendixMarkdown("some_parameter_sensitive_field PL99 XPL9"), "some_parameter_sensitive_field PL99 外部参照资料");
});

test("appendix golden preserves line and table shape, field names, numeric cells and chart bytes", () => {
  const raw = golden.markdown as string;
  const clean = cleanReaderAppendixMarkdown(raw);
  const sourceLines = raw.split("\n");
  const cleanLines = clean.split("\n");
  assert.equal(cleanLines.length, sourceLines.length);
  for (const [index, line] of sourceLines.entries()) {
    if (!line.startsWith("|")) continue;
    const cells = line.split("|");
    const outputCells = cleanLines[index].split("|");
    assert.equal(outputCells.length, cells.length, `table columns at line ${index}`);
    cells.forEach((cell, column) => {
      if (/^\s*-?\d+(?:\.\d+)?\s*$/.test(cell)) assert.equal(outputCells[column], cell);
    });
  }
  for (const field of ["sign_cn", "degree_in_sign", "nakshatra_lord"]) {
    assert.equal(clean.split(field).length, raw.split(field).length);
  }
  const fences = (text: string) => [...text.matchAll(/^[ \t]*```jyotish-chart[^\n]*\n[\s\S]*?^[ \t]*```[^\n]*(?:\n|$)/gm)].map(match => Buffer.from(match[0]));
  assert.deepEqual(fences(clean), fences(raw));
  assert.deepEqual(fences(cleanReaderAppendixMarkdown(copyCases.at(-1)!)), [Buffer.from(chartFence)]);
});