// A REPORT AS PLAIN MARKDOWN — `report.md`, the export that reads anywhere
// text does (publish/reportExports.ts writes it). Built from the report's page
// view (./views.ts), like the HTML export (./exportHtml.ts), with the same
// footer.
//
// The report's own markdown (summary, timeline entries, section bodies,
// findings) passes through as written, each inline citation
// `[label](cite:<id>)` becoming `label [n]` — the number of its entry in the
// numbered reference list at the end. Plain
// text fields (titles, quotes) are escaped so a `*` or `[` in a quote stays a
// character. No images: a still or a screenshot is in the HTML export and the
// evidence pack; here the quote is the text.
//
// PURE: no I/O, deterministic for its input.

import { CITE_SCHEME } from "../citations/inline";
import type { SourceArchive } from "../citations/schema";
import { dateLabel, reportExportFooterLine, siteLink, type ReportExportOptions } from "./exportHtml";
import {
  CITATION_KIND_LABELS,
  entryDateLabel,
  foundGroups,
  orderedCitations,
  reportDateParts,
  reportPagePath,
  reportRevisionLabel,
  spanLabel,
  subjectMetaParts,
  verdictTally,
  type CitationView,
  type ClaimView,
  type ReportPageView,
} from "./views";

// Plain text, safe in a Markdown paragraph: the characters Markdown reads as
// syntax are escaped, and a line can never open a heading, a list or a quote.
export function mdText(s: string): string {
  return s
    .replace(/[\\`*_[\]<>|]/g, (c) => `\\${c}`)
    .replace(/\s*\n\s*/g, " ")
    .replace(/^([#>+-]|\d+[.)])/, "\\$1");
}

// A URL as a Markdown link target: spaces and parentheses escaped.
function mdUrl(u: string): string {
  return u.replace(/ /g, "%20").replace(/\(/g, "%28").replace(/\)/g, "%29");
}

const mdLink = (text: string, url: string) => `[${mdText(text)}](${mdUrl(url)})`;

// The report's markdown with each inline citation as `label [n]`. A site-root
// link goes to the site when its URL is known, else stays its label.
export function citedMarkdownToPlain(md: string, view: Pick<ReportPageView, "citations">, siteUrl?: string): string {
  return md.replace(/\[([^\]]*)\]\(\s*([^)\s]+)(\s+"[^"]*")?\s*\)/g, (m, label: string, href: string, title?: string) => {
    if (href.startsWith(CITE_SCHEME)) {
      const c = view.citations[href.slice(CITE_SCHEME.length).trim()];
      return `${label} [${c?.number ?? "?"}]`;
    }
    if (href.startsWith("/")) {
      const abs = siteLink(siteUrl, href);
      return abs ? `[${label}](${mdUrl(abs)}${title ?? ""})` : label;
    }
    return m;
  });
}

function archiveLines(archives: readonly SourceArchive[]): string[] {
  return archives.map((a) => `- ${mdLink(a.label, a.url)} <${a.url}>${a.context ? ` — ${mdText(a.context)}` : ""}`);
}

function citationWhere(c: CitationView): string {
  switch (c.kind) {
    case "video":
    case "audio":
      return [c.record.title ? `*${mdText(c.record.title)}*` : null, c.record.channelTitle ? mdText(c.record.channelTitle) : null, spanLabel(c.start, c.end)]
        .filter(Boolean)
        .join(", ");
    case "post":
      return [c.author ? mdText(c.author) : null, c.record.channelTitle ? mdText(c.record.channelTitle) : null, "post"].filter(Boolean).join(", ");
    case "source":
      return `from *${mdText(c.sourceTitle)}*`;
    case "page":
      return c.title ? `*${mdText(c.title)}*` : "web page";
  }
}

function citationDate(c: CitationView): string | undefined {
  return dateLabel(c.date ?? (c.kind === "video" || c.kind === "audio" || c.kind === "post" ? c.record.date : undefined));
}

function referenceLines(c: CitationView, siteUrl: string | undefined): string[] {
  const meta = [
    CITATION_KIND_LABELS[c.kind],
    c.speaker ? mdText(c.speaker) : null,
    citationDate(c) ?? null,
    citationWhere(c),
    c.verification?.quoteScore !== undefined ? `quote match ${Math.round(c.verification.quoteScore * 100)}%` : null,
  ].filter(Boolean);
  const out = [`${c.number ?? "-"}. “${mdText(c.quote)}”`];
  out.push(`   ${meta.join(" · ")}`);
  for (const x of [c.label, c.note]) if (x) out.push(`   ${mdText(x)}`);
  const link = (label: string, url: string) => out.push(`   ${label}: <${url}>`);
  switch (c.kind) {
    case "video":
    case "audio": {
      if (c.record.originalUrl) link(`Original at ${spanLabel(c.start, c.end).split("–")[0]}`, c.record.originalUrl);
      const moment = siteLink(siteUrl, c.href);
      if (moment) link("Moment page", moment);
      break;
    }
    case "post": {
      if (c.record.originalUrl) link("Original", c.record.originalUrl);
      const moment = siteLink(siteUrl, c.href);
      if (moment) link("Moment page", moment);
      break;
    }
    case "source":
      if (c.href) link("Original", c.href);
      break;
    case "page":
      link("Original", c.href);
      if (c.archiveUrl) link("Archived", c.archiveUrl);
      break;
  }
  // One list item: its lines indented under the number, each ending in a
  // hard break but the last.
  return out.map((l, i, a) => (i < a.length - 1 ? `${l}  ` : l));
}

function evidenceLine(c: CitationView): string {
  return `- [${c.number ?? "?"}] ${CITATION_KIND_LABELS[c.kind]}, ${citationWhere(c)}: “${mdText(c.quote)}”`;
}

function claimLines(
  claim: ClaimView,
  view: ReportPageView,
  subjectNoun: string,
  siteUrl: string | undefined,
): string[] {
  const out: string[] = [`### ${mdText(claim.title ?? claim.text)}`, ""];
  const marks = [
    claim.verdict ? `**Verdict: ${mdText(view.verdicts[claim.verdict].label)}**` : null,
    claim.flag ? `*[${mdText(claim.flag)}]*` : null,
  ].filter(Boolean);
  if (marks.length > 0) out.push(marks.join(" "), "");
  // The document's own sentence when the claim cites one; else the claim as
  // the report states it, plain.
  const sentence = claim.sourceQuote ? view.citations[claim.sourceQuote] : undefined;
  if (sentence?.kind !== "source") {
    if (claim.title) out.push(mdText(claim.text), "");
  } else {
    out.push(`> “${mdText(sentence.quote)}” [${sentence.number ?? "?"}]`);
    if (sentence.sourceId !== view.subject) out.push(">", `> — from *${mdText(sentence.sourceTitle)}*`);
    out.push("");
    if (claim.archives?.length) out.push(...archiveLines(claim.archives), "");
  }
  if (claim.findings) out.push(citedMarkdownToPlain(claim.findings, view, siteUrl).trim(), "");
  const evidence = claim.citations
    .filter((id) => id !== claim.sourceQuote)
    .map((id) => view.citations[id])
    .filter((c): c is CitationView => !!c);
  // What the report adds first, then evidence of unknown origin, then what
  // the document under review gave itself.
  const shown = [...evidence.filter((c) => c.origin === "added"), ...evidence.filter((c) => c.origin === undefined)];
  const given = evidence.filter((c) => c.origin === "subject");
  if (shown.length > 0) out.push("Evidence:", "", ...shown.map((c) => evidenceLine(c)), "");
  if (given.length > 0) out.push(`In the ${subjectNoun}:`, "", ...given.map((c) => evidenceLine(c)), "");
  return out;
}

export function reportExportMarkdown(view: ReportPageView, opts: ReportExportOptions): string {
  const out: string[] = [];
  const subject = view.subject ? view.sources[view.subject] : undefined;
  const subjectNoun = subject?.kind === "article" ? "article" : "source";
  const pageUrl = siteLink(opts.siteUrl, reportPagePath(view.id));
  const isFactcheck = view.kind === "factcheck";

  // The series on its own line and the title as the heading (or the title
  // alone), one line of dates and this export's revision, the document under
  // review as a cited line, the subtitle.
  if (view.series) out.push(`**${mdText(view.series)}**`, "");
  out.push(`# ${mdText(view.title)}`, "");
  const historyUrl = view.history ? siteLink(opts.siteUrl, view.history.href) : undefined;
  const revision =
    opts.footer.revision !== undefined
      ? historyUrl
        ? mdLink(reportRevisionLabel(opts.footer.revision), historyUrl)
        : mdText(reportRevisionLabel(opts.footer.revision))
      : undefined;
  const dateLine = [...reportDateParts(view).map(mdText), ...(revision ? [revision] : [])];
  if (dateLine.length > 0) out.push(dateLine.join(" · "), "");
  if (subject) {
    const meta = subjectMetaParts(subject).map(mdText);
    out.push(`${subject.url ? mdLink(subject.title, subject.url) : mdText(subject.title)}${meta.length ? ` — ${meta.join(" · ")}` : ""}`, "");
  }
  if (view.subtitle) out.push(`*${mdText(view.subtitle)}*`, "");

  // Tier 1: the quick take, after a rule (each tier opens with one).
  out.push("---", "");
  const tally = isFactcheck ? verdictTally(view) : [];
  if (tally.length > 0) {
    const claims = view.sections.reduce((n, s) => n + s.claims.length, 0);
    out.push(
      `${claims} claim${claims === 1 ? "" : "s"} checked: ${tally.map((t) => `${mdText(view.verdicts[t.verdict].label)} ${t.count}`).join(" · ")}`,
      "",
    );
  }
  if (view.summary) out.push(citedMarkdownToPlain(view.summary, view, opts.siteUrl).trim(), "");

  // The timeline: its dated entries, newest first.
  const entries = view.entries ?? [];
  if (entries.length > 0) {
    out.push("## Timeline", "");
    for (const e of entries) {
      out.push(`### ${entryDateLabel(e.date)} — ${mdText(e.title)}`, "");
      if (e.updated) out.push(`*updated ${entryDateLabel(e.updated)}*`, "");
      out.push(citedMarkdownToPlain(e.body, view, opts.siteUrl).trim(), "");
    }
  }

  // Tier 2: what the check found (a fact-check with verdicts).
  const groups = isFactcheck ? foundGroups(view) : [];
  if (groups.length > 0) {
    out.push("---", "", "## What the check found", "");
    for (const g of groups) {
      out.push(`### ${mdText(view.verdicts[g.verdict].label)} (${g.claims.length})`, "");
      for (const c of g.claims) {
        out.push(`- ${mdText(c.title ?? c.text)}${c.gist ? ` — ${mdText(c.gist)}` : ""}${c.flag ? ` *[${mdText(c.flag)}]*` : ""}`);
      }
      out.push("");
    }
  }

  // Tier 3: every claim, with its evidence.
  out.push("---", "", "## Every claim, with its evidence", "");
  if (view.method) out.push("**How it was checked**", "", citedMarkdownToPlain(view.method, view, opts.siteUrl).trim(), "");
  for (const s of view.sections) {
    out.push(`## ${mdText(s.title)}`, "");
    if (s.body) out.push(citedMarkdownToPlain(s.body, view, opts.siteUrl).trim(), "");
    for (const c of s.claims) out.push(...claimLines(c, view, subjectNoun, opts.siteUrl));
  }

  const sources = Object.values(view.sources);
  if (sources.length > 0) {
    out.push("## Sources", "");
    for (const s of sources) {
      out.push(`### ${mdText(s.title)}${s.id === view.subject ? " (under review)" : ""}`, "");
      if (s.url) out.push(`<${s.url}>`, "");
      const byline = [s.publisher, s.author, dateLabel(s.date)].filter(Boolean).map((x) => mdText(x!));
      if (byline.length > 0) out.push(byline.join(" · "), "");
      if (s.note) out.push(mdText(s.note), "");
      if (s.archives.length > 0) out.push(`${s.archives.length} archive link${s.archives.length === 1 ? "" : "s"} in context:`, "", ...archiveLines(s.archives), "");
    }
  }

  const refs = orderedCitations(view);
  if (refs.length > 0) {
    out.push("## References", "");
    for (const c of refs) out.push(...referenceLines(c, opts.siteUrl), "");
  }

  out.push("---", "", reportExportFooterLine(opts.footer));
  if (pageUrl) out.push("", `Published at <${pageUrl}>`);
  return `${out.join("\n").replace(/\n{3,}/g, "\n\n").trimEnd()}\n`;
}