@promptbook/pdf 0.84.0-16 → 0.84.0-18

package/esm/typings/src/_packages/core.index.d.ts

@@ -14,6 +14,7 @@ import { DEFAULT_BOOK_TITLE } from '../config';
 import { DEFAULT_TASK_TITLE } from '../config';
 import { DEFAULT_PROMPT_TASK_TITLE } from '../config';
 import { DEFAULT_BOOK_OUTPUT_PARAMETER_NAME } from '../config';
+import { DEFAULT_MAX_FILE_SIZE } from '../config';
 import { MAX_FILENAME_LENGTH } from '../config';
 import { DEFAULT_INTERMEDIATE_FILES_STRATEGY } from '../config';
 import { DEFAULT_MAX_PARALLEL_COUNT } from '../config';
@@ -144,6 +145,7 @@ export { DEFAULT_BOOK_TITLE };
 export { DEFAULT_TASK_TITLE };
 export { DEFAULT_PROMPT_TASK_TITLE };
 export { DEFAULT_BOOK_OUTPUT_PARAMETER_NAME };
+export { DEFAULT_MAX_FILE_SIZE };
 export { MAX_FILENAME_LENGTH };
 export { DEFAULT_INTERMEDIATE_FILES_STRATEGY };
 export { DEFAULT_MAX_PARALLEL_COUNT };

package/esm/typings/src/config.d.ts

@@ -77,6 +77,12 @@ export declare const DEFAULT_PROMPT_TASK_TITLE = "Prompt";
  * @public exported from `@promptbook/core`
  */
 export declare const DEFAULT_BOOK_OUTPUT_PARAMETER_NAME = "result";
+/**
+ * Maximum file size limit
+ *
+ * @public exported from `@promptbook/core`
+ */
+export declare const DEFAULT_MAX_FILE_SIZE: number;
 /**
  * Warning message for the generated sections and files files
  *

package/esm/typings/src/conversion/compilePipelineOnRemoteServer.d.ts

@@ -17,5 +17,5 @@ import type { RemoteClientOptions } from '../remote-server/types/RemoteClientOpt
  */
 export declare function compilePipelineOnRemoteServer<TCustomOptions = undefined>(pipelineString: PipelineString, options: RemoteClientOptions<TCustomOptions>): Promise<PipelineJson>;
 /**
- * TODO: !!!! Do not return Promise<PipelineJson> But PreparePipelineTask
+ * TODO: [🐚] Do not return Promise<PipelineJson> But PreparePipelineTask
  */

package/esm/typings/src/prepare/preparePipelineOnRemoteServer.d.ts

@@ -13,5 +13,5 @@ import type { RemoteClientOptions } from '../remote-server/types/RemoteClientOpt
  */
 export declare function preparePipelineOnRemoteServer<TCustomOptions = undefined>(pipeline: PipelineJson, options: RemoteClientOptions<TCustomOptions>): Promise<PipelineJson>;
 /**
- * TODO: !!!! Do not return Promise<PipelineJson> But PreparePipelineTask
+ * TODO: [🐚] Do not return Promise<PipelineJson> But PreparePipelineTask
  */

package/esm/typings/src/remote-server/socket-types/_subtypes/PromptbookServer_Identification.d.ts

@@ -9,7 +9,7 @@ import type { ApplicationRemoteServerClientOptions } from '../../types/RemoteSer
  */
 export type PromptbookServer_Identification<TCustomOptions> = PromptbookServer_ApplicationIdentification<TCustomOptions> | PromptbookServer_AnonymousIdentification;
 /**
- * Application mode is @@@!!!
+ * Application mode is situation when you run known and well-defined books with your own api keys
  *
  * @public exported from `@promptbook/remote-server`
  * @public exported from `@promptbook/remote-client`
@@ -21,7 +21,10 @@ export type PromptbookServer_ApplicationIdentification<TCustomOptions> = Applica
     readonly isAnonymous: false;
 };
 /**
- * Anonymous mode is @@@!!!
+ * Anonymous mode is when you run arbitrary user books without api keys from user
+ *
+ * Note: This is useful in situations when the LLM provider does not allow to call the API requests from the client side
+ *       It is kind of a proxy mode
  *
  * @public exported from `@promptbook/remote-server`
  * @public exported from `@promptbook/remote-client`

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@promptbook/pdf",
-    "version": "0.84.0-16",
+    "version": "0.84.0-18",
     "description": "It's time for a paradigm shift. The future of software in plain English, French or Latin",
     "private": false,
     "sideEffects": false,
@@ -8,8 +8,10 @@
         "type": "git",
         "url": "https://github.com/webgptorg/promptbook"
     },
+    "author": "Pavol Hejný <pavol@ptbk.io> (https://www.pavolhejny.com/)",
     "contributors": [
-        "Pavol Hejný <pavol@ptbk.io> (https://www.pavolhejny.com/)"
+        "Pavol Hejný <pavol@ptbk.io> (https://www.pavolhejny.com/)",
+        "Jiří Jahn <jiri@ptbk.io> (https://www.ptbk.io/)"
     ],
     "keywords": [
         "ai",
@@ -23,6 +25,9 @@
         "natural-language-processing",
         "nlp",
         "openai",
+        "o3",
+        "o3-mini",
+        "deepseek",
         "gpt-3",
         "gpt-4",
         "gpt-4o",
@@ -33,7 +38,7 @@
         "anthropic",
         "LLMOps"
     ],
-    "license": "FSL-1.1-Apache-2.0",
+    "license": "UNLICENSED",
     "bugs": {
         "url": "https://github.com/webgptorg/promptbook/issues"
     },
@@ -42,7 +47,7 @@
     "module": "./esm/index.es.js",
     "typings": "./esm/typings/src/_packages/pdf.index.d.ts",
     "peerDependencies": {
-        "@promptbook/core": "0.84.0-16"
+        "@promptbook/core": "0.84.0-18"
     },
     "dependencies": {
         "crypto-js": "4.2.0",

package/umd/index.umd.js

@@ -25,7 +25,7 @@
      * @generated
      * @see https://github.com/webgptorg/promptbook
      */
-    var PROMPTBOOK_ENGINE_VERSION = '0.84.0-15';
+    var PROMPTBOOK_ENGINE_VERSION = '0.84.0-17';
     /**
      * TODO: string_promptbook_version should be constrained to the all versions of Promptbook engine
      * Note: [💞] Ignore a discrepancy between file name and entity name
@@ -196,6 +196,12 @@
      * @public exported from `@promptbook/core`
      */
     var DEFAULT_BOOK_TITLE = "\u2728 Untitled Book";
+    /**
+     * Maximum file size limit
+     *
+     * @public exported from `@promptbook/core`
+     */
+    var DEFAULT_MAX_FILE_SIZE = 100 * 1024 * 1024; // 100MB
     // <- TODO: [🧠] Better system for generator warnings - not always "code" and "by `@promptbook/cli`"
     /**
      * The maximum number of iterations for a loops
@@ -3627,11 +3633,10 @@
     function makeKnowledgeSourceHandler(knowledgeSource, tools, options) {
         var _a;
         return __awaiter(this, void 0, void 0, function () {
-            var _b, fetch, knowledgeSourceContent, name, _c, _d, rootDirname, url, response_1, mimeType, basename, hash, rootDirname_1, filepath, _f, _g, _h, _j, _k, filename_1, fileExtension, mimeType;
-            return __generator(this, function (_l) {
-                switch (_l.label) {
+            var _b, fetch, knowledgeSourceContent, name, _c, _d, rootDirname, url, response_1, mimeType, basename, hash, rootDirname_1, filepath, fileContent, _f, _g, filename_1, fileExtension, mimeType;
+            return __generator(this, function (_h) {
+                switch (_h.label) {
                     case 0:
-                        console.log('!!! makeKnowledgeSourceHandler', knowledgeSource);
                         _b = tools.fetch, fetch = _b === void 0 ? scraperFetch : _b;
                         knowledgeSourceContent = knowledgeSource.knowledgeSourceContent;
                         name = knowledgeSource.name;
@@ -3643,7 +3648,7 @@
                         url = knowledgeSourceContent;
                         return [4 /*yield*/, fetch(url)];
                     case 1:
-                        response_1 = _l.sent();
+                        response_1 = _h.sent();
                         mimeType = ((_a = response_1.headers.get('content-type')) === null || _a === void 0 ? void 0 : _a.split(';')[0]) || 'text/html';
                         if (tools.fs === undefined || !url.endsWith('.pdf' /* <- TODO: [💵] */)) {
                             return [2 /*return*/, {
@@ -3693,17 +3698,19 @@
                         filepath = path.join.apply(void 0, __spreadArray(__spreadArray([], __read(nameToSubfolderPath(hash /* <- TODO: [🎎] Maybe add some SHA256 prefix */)), false), ["".concat(basename.substring(0, MAX_FILENAME_LENGTH), ".").concat(mimeTypeToExtension(mimeType))], false));
                         return [4 /*yield*/, tools.fs.mkdir(path.dirname(path.join(rootDirname_1, filepath)), { recursive: true })];
                     case 2:
-                        _l.sent();
-                        _g = (_f = tools.fs).writeFile;
-                        _h = [path.join(rootDirname_1, filepath)];
-                        _k = (_j = Buffer).from;
+                        _h.sent();
+                        _g = (_f = Buffer).from;
                         return [4 /*yield*/, response_1.arrayBuffer()];
-                    case 3: return [4 /*yield*/, _g.apply(_f, _h.concat([_k.apply(_j, [_l.sent()])]))];
+                    case 3:
+                        fileContent = _g.apply(_f, [_h.sent()]);
+                        if (fileContent.length > DEFAULT_MAX_FILE_SIZE /* <- TODO: Allow to pass different value to remote server */) {
+                            throw new LimitReachedError("File is too large (".concat(Math.round(fileContent.length / 1024 / 1024), "MB). Maximum allowed size is ").concat(Math.round(DEFAULT_MAX_FILE_SIZE / 1024 / 1024), "MB."));
+                        }
+                        return [4 /*yield*/, tools.fs.writeFile(path.join(rootDirname_1, filepath), fileContent)];
                     case 4:
-                        _l.sent();
+                        _h.sent();
                         // TODO: [💵] Check the file security
-                        // TODO: !!!!!!!! Check the file size (if it is not too big)
-                        // TODO: !!!!!!!! Delete the file after the scraping is done
+                        // TODO: [🧹][🧠] Delete the file after the scraping is done
                         return [2 /*return*/, makeKnowledgeSourceHandler({ name: name, knowledgeSourceContent: filepath }, tools, __assign(__assign({}, options), { rootDirname: rootDirname_1 }))];
                     case 5:
                         if (!isValidFilePath(knowledgeSourceContent)) return [3 /*break*/, 7];
@@ -3720,7 +3727,7 @@
                         mimeType = extensionToMimeType(fileExtension || '');
                         return [4 /*yield*/, isFileExisting(filename_1, tools.fs)];
                     case 6:
-                        if (!(_l.sent())) {
+                        if (!(_h.sent())) {
                             throw new NotFoundError(spaceTrim__default["default"](function (block) { return "\n                          Can not make source handler for file which does not exist:\n\n                          File:\n                          ".concat(block(knowledgeSourceContent), "\n\n                          Full file path:\n                          ").concat(block(filename_1), "\n                      "); }));
                         }
                         // TODO: [🧠][😿] Test security file - file is scoped to the project (BUT maybe do this in `filesystemTools`)
@@ -6361,7 +6368,7 @@
             // Note: Module `markitdown-ts` has no types available, so it is imported using `require`
             // eslint-disable-next-line @typescript-eslint/no-var-requires
             var MarkItDown = require('markitdown-ts').MarkItDown;
-            // <- TODO: 'Use Markitdown directly not through this package
+            // <- TODO: [🍀] Use Markitdown directly not through this package
             this.markitdown = new MarkItDown();
         }
         Object.defineProperty(MarkitdownScraper.prototype, "metadata", {
@@ -6406,7 +6413,7 @@
                         case 2:
                             if (!!(_f.sent())) return [3 /*break*/, 5];
                             src = source.filename || source.url || null;
-                            // console.log('!!!', { src, source, cacheFilehandler });
+                            // console.log('!!', { src, source, cacheFilehandler });
                             if (src === null) {
                                 throw new UnexpectedError('Source has no filename or url');
                             }
@@ -6419,12 +6426,12 @@
                             result = _f.sent();
                             if (result === null || result === undefined) {
                                 throw new Error("Markitdown could not convert the \"".concat(source.source, "\""));
-                                // <- TODO: !!! Make MarkitdownError
+                                // <- TODO: [🍀] Make MarkitdownError
                             }
-                            // console.log('!!!', { result, cacheFilehandler });
+                            // console.log('!!', { result, cacheFilehandler });
                             return [4 /*yield*/, this.tools.fs.writeFile(cacheFilehandler.filename, result.text_content)];
                         case 4:
-                            // console.log('!!!', { result, cacheFilehandler });
+                            // console.log('!!', { result, cacheFilehandler });
                             _f.sent();
                             _f.label = 5;
                         case 5: return [2 /*return*/, cacheFilehandler];