@botpress/adk-cli/dist/chunk-kkk13rcb.js

Command-line interface for the Botpress Agent Development Kit (ADK)

// @bun
import {
  getTokenizer
} from "./chunk-na956zz3.js";

// ../../node_modules/.bun/llmz@0.0.79+b49d396f5ed96e7f/node_modules/llmz/dist/chunk-M4WTTARE.js
var DEFAULT_REMOVE_CHUNK = 250;
var WRAP_OPEN_TAG_1 = "\u3010TRUNCATE";
var WRAP_OPEN_TAG_2 = "\u3011";
var WRAP_CLOSE_TAG = "\u3010/TRUNCATE\u3011";
var REGEXP = `(${WRAP_OPEN_TAG_1}(?:\\s+[\\w:]+)*\\s*${WRAP_OPEN_TAG_2})([\\s\\S]*?)(${WRAP_CLOSE_TAG})`;
var DEFAULT_TRUNCATE_OPTIONS = {
  preserve: "top",
  flex: 1,
  minTokens: 0
};
function wrapContent(content, options) {
  const preserve = (options == null ? undefined : options.preserve) ?? DEFAULT_TRUNCATE_OPTIONS.preserve;
  const flex = (options == null ? undefined : options.flex) ?? DEFAULT_TRUNCATE_OPTIONS.flex;
  const minTokens = (options == null ? undefined : options.minTokens) ?? DEFAULT_TRUNCATE_OPTIONS.minTokens;
  return `${WRAP_OPEN_TAG_1} preserve:${preserve} flex:${flex} min:${minTokens} ${WRAP_OPEN_TAG_2}${content}${WRAP_CLOSE_TAG}`;
}
function truncateWrappedContent({
  messages,
  tokenLimit,
  throwOnFailure = true
}) {
  var _a, _b, _c;
  const tokenizer = getTokenizer();
  const parts = [];
  for (const msg of messages) {
    const current = [];
    const content = typeof msg.content === "string" ? msg.content : "";
    let match;
    const parser = new _MessageContentParser;
    while ((match = parser.parse(content)) !== null) {
      const { attributes, nonTruncatableContent, wrappedContent } = match;
      if (nonTruncatableContent) {
        current.push({
          content: nonTruncatableContent,
          tokens: tokenizer.count(nonTruncatableContent),
          truncatable: false
        });
      }
      current.push({
        content: wrappedContent,
        tokens: tokenizer.count(wrappedContent),
        truncatable: true,
        attributes: {
          preserve: attributes.preserve,
          flex: Number(attributes.flex) || DEFAULT_TRUNCATE_OPTIONS.flex,
          minTokens: Number(attributes.min) || DEFAULT_TRUNCATE_OPTIONS.minTokens
        }
      });
    }
    const remainingContent = parser.getRemainingContent(content);
    if (remainingContent) {
      current.push({
        content: remainingContent,
        tokens: tokenizer.count(remainingContent),
        truncatable: false
      });
    }
    parts.push(current);
  }
  let currentCount = _countTotalTokens(parts);
  while (currentCount > tokenLimit) {
    const { biggest, secondBiggest } = _getTwoBiggestTruncables(parts);
    if (!biggest || !biggest.truncatable || biggest.tokens <= 0) {
      if (throwOnFailure) {
        throw new Error(`Cannot truncate further, current count: ${currentCount}`);
      } else {
        break;
      }
    }
    const delta = Math.max(biggest.tokens - ((secondBiggest == null ? undefined : secondBiggest.tokens) ?? 0), DEFAULT_REMOVE_CHUNK);
    const room = Math.min(delta, biggest.tokens);
    let toRemove = Math.min(room, currentCount - tokenLimit);
    if (toRemove <= 0) {
      if (throwOnFailure) {
        throw new Error(`Cannot truncate further, current count: ${currentCount}`);
      } else {
        break;
      }
    }
    if (biggest.tokens - toRemove < (((_a = biggest.attributes) == null ? undefined : _a.minTokens) ?? 0)) {
      toRemove = biggest.tokens - (((_b = biggest.attributes) == null ? undefined : _b.minTokens) ?? 0);
    }
    const preserve = ((_c = biggest.attributes) == null ? undefined : _c.preserve) ?? DEFAULT_TRUNCATE_OPTIONS.preserve;
    const split = tokenizer.splitAndSlice(biggest.content);
    if (preserve === "bottom") {
      biggest.content = split.slice(toRemove).join("");
    } else if (preserve === "top") {
      biggest.content = split.slice(0, -toRemove).join("");
    } else {
      const anchor = Math.ceil(split.length / 2);
      const radius = Math.ceil(toRemove / 2);
      const left = anchor - radius;
      const right = anchor + radius;
      biggest.content = split.slice(0, left).join("") + split.slice(right).join("");
    }
    biggest.tokens -= toRemove;
    currentCount -= toRemove;
  }
  return messages.map((msg, i) => {
    const p = parts[i];
    return {
      ...msg,
      content: typeof msg.content === "string" ? _renderRemainingWrappers(p.map((part) => part.content).join("")) : msg.content
    };
  });
}
var _MessageContentParser = class {
  _regex;
  _lastIndex = 0;
  constructor() {
    this._regex = _createRegex();
  }
  parse(content) {
    const match = this._regex.exec(content);
    if (!match) {
      return null;
    }
    const attributes = match[1].split(/\s+/).slice(1).filter((x) => x !== WRAP_OPEN_TAG_2).map((x) => x.split(":")).reduce((acc, [key, value]) => ({ ...acc, [key]: value }), {});
    let nonTruncatableContent = undefined;
    if (match.index > this._lastIndex) {
      nonTruncatableContent = content.slice(this._lastIndex, match.index);
    }
    const wrappedContent = match[2];
    this._lastIndex = this._regex.lastIndex;
    return { attributes, nonTruncatableContent, wrappedContent };
  }
  getRemainingContent(content) {
    if (this._lastIndex < content.length) {
      const remainingContent = content.slice(this._lastIndex);
      return remainingContent;
    }
    return null;
  }
};
var _createRegex = () => new RegExp(REGEXP, "g");
var _renderRemainingWrappers = (content) => content.replace(_createRegex(), "$2");
var _countTotalTokens = (parts) => parts.reduce((acc, x) => acc + x.reduce((acc2, y) => acc2 + y.tokens, 0), 0);
var _getTwoBiggestTruncables = (parts) => {
  var _a, _b;
  let biggest = null;
  let secondBiggest = null;
  for (const part of parts.flat()) {
    if (part.truncatable) {
      if (part.tokens <= (((_a = part.attributes) == null ? undefined : _a.minTokens) ?? 0)) {
        continue;
      }
      const flex = ((_b = part.attributes) == null ? undefined : _b.flex) ?? DEFAULT_TRUNCATE_OPTIONS.flex;
      const tokens = part.tokens * flex;
      if (!biggest || tokens > biggest.tokens) {
        secondBiggest = biggest;
        biggest = part;
      } else if (!secondBiggest || tokens > secondBiggest.tokens) {
        secondBiggest = part;
      }
    }
  }
  return { biggest, secondBiggest };
};

export { wrapContent, truncateWrappedContent };