ai-chatbot/lib/canvas/markdown-image-streaming.ts

/** A half-open span containing a complete Markdown link or image. */
export type MarkdownImageRange = { start: number; end: number };

type Parsed = { kind: "complete"; end: number } | { kind: "pending" } | { kind: "invalid" };

// This scanner recognizes syntax only; rendering and URL safety remain the
// responsibility of the existing Markdown renderer and image splitter.
function readDestination(text: string, start: number, allowWhitespace = false): Parsed {
  if (start >= text.length) return { kind: "pending" };
  if (text[start] !== "(") return { kind: "invalid" };
  let depth = 1;
  for (let i = start + 1; i < text.length; i += 1) {
    const char = text[i];
    if (char === "\\") {
      if (i + 1 >= text.length) return { kind: "pending" };
      i += 1;
      continue;
    }
    if (char === "\n" || char === "\r" || (!allowWhitespace && /\s/.test(char))) return { kind: "invalid" };
    if (char === "(") depth += 1;
    if (char === ")" && --depth === 0) return { kind: "complete", end: i + 1 };
  }
  return { kind: "pending" };
}

function readImage(text: string, start: number, wrapped: boolean): Parsed {
  const imageStart = start + (wrapped ? 1 : 0);
  if (imageStart + 1 >= text.length) return { kind: "pending" };
  if (text[imageStart] !== "!" || text[imageStart + 1] !== "[") return { kind: "invalid" };
  let depth = 1;
  for (let i = imageStart + 2; i < text.length; i += 1) {
    const char = text[i];
    if (char === "\n" || char === "\r") return { kind: "invalid" };
    if (char === "\\") {
      if (i + 1 >= text.length) return { kind: "pending" };
      i += 1;
      continue;
    }
    if (char === "[") depth += 1;
    if (char !== "]" || --depth !== 0) continue;
    const inner = readDestination(text, i + 1);
    if (inner.kind !== "complete" || !wrapped) return inner;
    if (inner.end >= text.length) return { kind: "pending" };
    if (text[inner.end] !== "]") return { kind: "invalid" };
    return readDestination(text, inner.end + 1);
  }
  return { kind: "pending" };
}

function readLink(text: string, start: number): Parsed {
  let depth = 1;
  for (let i = start + 1; i < text.length; i += 1) {
    const char = text[i];
    // An unfinished bracket on an earlier line must not hide later output.
    if (char === "\n" || char === "\r") return { kind: "invalid" };
    if (char === "\\") {
      if (i + 1 >= text.length) return { kind: "pending" };
      i += 1;
      continue;
    }
    if (char === "[") depth += 1;
    if (char !== "]" || --depth !== 0) continue;
    if (text[i + 1] === "[") {
      for (let j = i + 2; j < text.length; j += 1) {
        if (text[j] === "\n" || text[j] === "\r") return { kind: "invalid" };
        if (text[j] === "\\") {
          if (j + 1 >= text.length) return { kind: "pending" };
          j += 1;
          continue;
        }
        if (text[j] === "]") return { kind: "complete", end: j + 1 };
      }
      return { kind: "pending" };
    }
    return readDestination(text, i + 1, true);
  }
  return { kind: "pending" };
}

/** Locate complete Markdown links/images and an incomplete trailing construct.
 * Code spans/fences and escaped markers remain normal text. */
export function scanMarkdownImageStreaming(text: string): {
  ranges: MarkdownImageRange[];
  pendingStart: number | null;
} {
  const ranges: MarkdownImageRange[] = [];
  let cursor = 0;
  while (cursor < text.length) {
    if (text[cursor] === "\\") { cursor += 2; continue; }
    if (text[cursor] === "`") {
      let end = cursor + 1;
      while (text[end] === "`") end += 1;
      const marker = text.slice(cursor, end);
      const close = text.indexOf(marker, end);
      cursor = close < 0 ? text.length : close + marker.length;
      continue;
    }
    if (cursor === 0 || text[cursor - 1] === "\n") {
      const fence = text.slice(cursor).match(/^ {0,3}(~{3,})[^\n]*(?:\n|$)/);
      if (fence) {
        const closing = new RegExp(`^ {0,3}~{${fence[1].length},}\\s*$`);
        let line = cursor + fence[0].length;
        while (line < text.length) {
          const next = text.indexOf("\n", line);
          const end = next < 0 ? text.length : next;
          if (closing.test(text.slice(line, end))) { line = next < 0 ? end : end + 1; break; }
          line = next < 0 ? end : end + 1;
        }
        cursor = line;
        continue;
      }
    }
    const char = text[cursor];
    if (char !== "[" && char !== "!") { cursor += 1; continue; }
    const wrapped = char === "[";
    const parsed = wrapped && text[cursor + 1] !== "!"
      ? readLink(text, cursor)
      : readImage(text, cursor, wrapped);
    if (parsed.kind === "pending") return { ranges, pendingStart: cursor };
    if (parsed.kind === "complete") { ranges.push({ start: cursor, end: parsed.end }); cursor = parsed.end; }
    else cursor += 1;
  }
  return { ranges, pendingStart: null };
}

/** During active delivery, withhold only unfinished link/image syntax at the tail. */
export function stableStreamingMarkdown(text: string, streaming: boolean): string {
  if (!streaming) return text;
  const { pendingStart } = scanMarkdownImageStreaming(text);
  return pendingStart === null ? text : text.slice(0, pendingStart);
}

export function advancePastAtomicRanges(length: number, target: number, ranges: MarkdownImageRange[]) {
  let position = Math.min(length, target);
  // Overlapping table/link ranges can require more than one jump.
  for (let changed = true; changed;) {
    changed = false;
    for (const range of ranges) {
      if (position > range.start && position < range.end) {
        position = Math.min(length, range.end);
        changed = true;
      }
    }
  }
  return position;
}