xiaoli-cli bot
dfe21298b9
新增插件: - terminal: 持久交互式终端，纯标准库，全平台，13个测试 - operit_bridge: Operit AI工具格式桥接，47个脚本288个工具，12个测试 其他: - plugins/operit/: 47个Operit官方脚本 - 更新plugin_registry.json/plugins_config.py - 更新about.txt/CHANGELOG.md至v5.4.2
756 lines
27 KiB
TypeScript
756 lines
27 KiB
TypeScript
/* METADATA
|
|
{
|
|
"name": "pdf_vision_parser",
|
|
"display_name": {
|
|
"zh": "PDF 逐页识图解析",
|
|
"en": "PDF Vision Parser"
|
|
},
|
|
"description": {
|
|
"zh": "将 PDF 逐页渲染为图片，并调用用户当前配置的识图模型进行解析，最后按页拼接结果。",
|
|
"en": "Render a PDF page by page into images, send each image to the user's configured vision model, and merge the results."
|
|
},
|
|
"enabledByDefault": false,
|
|
"category": "File",
|
|
"tools": [
|
|
{
|
|
"name": "parse_pdf_with_vision",
|
|
"description": {
|
|
"zh": "将 PDF 逐页拆成图片，逐页调用 IMAGE_RECOGNITION 功能模型识图，并返回分页索引与输出文件路径。",
|
|
"en": "Split a PDF into page images, analyze each page with the IMAGE_RECOGNITION function model, and return per-page indexes plus output file paths."
|
|
},
|
|
"parameters": [
|
|
{
|
|
"name": "pdf_path",
|
|
"description": {
|
|
"zh": "PDF 文件路径。",
|
|
"en": "Path to the PDF file."
|
|
},
|
|
"type": "string",
|
|
"required": true
|
|
},
|
|
{
|
|
"name": "page_prompt",
|
|
"description": {
|
|
"zh": "每页识图提示词；不传则使用默认的尽量转文字提示词。",
|
|
"en": "Vision prompt for each page; defaults to a text-extraction-oriented prompt."
|
|
},
|
|
"type": "string",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "start_page",
|
|
"description": {
|
|
"zh": "起始页码，1-based，默认 1。",
|
|
"en": "Start page number, 1-based, default 1."
|
|
},
|
|
"type": "number",
|
|
"required": false
|
|
},
|
|
{
|
|
"name": "end_page",
|
|
"description": {
|
|
"zh": "结束页码，1-based，默认文档最后一页。",
|
|
"en": "End page number, 1-based, default is the last page."
|
|
},
|
|
"type": "number",
|
|
"required": false
|
|
}
|
|
]
|
|
}
|
|
]
|
|
}*/
|
|
|
|
const PdfVisionParser = (function () {
|
|
const DEFAULT_PAGE_PROMPT =
|
|
"尽量完整提取本页可见文字与结构，保留标题、列表、表格顺序，不做总结，不补充页外信息；公式和图表转成可读文本描述";
|
|
const TOOL_NAME = "parse_pdf_with_vision";
|
|
const TOOL_FUNCTION_TYPE = "IMAGE_RECOGNITION";
|
|
const RENDER_SCALE = 2;
|
|
const PAGE_CONCURRENCY = 4;
|
|
const IMAGE_FORMAT = "png";
|
|
const MERGED_OUTPUT_FILE_NAME = "parsed_output.txt";
|
|
|
|
const File = Java.type("java.io.File");
|
|
const FileOutputStream = Java.type("java.io.FileOutputStream");
|
|
const OutputStreamWriter = Java.type("java.io.OutputStreamWriter");
|
|
const BufferedWriter = Java.type("java.io.BufferedWriter");
|
|
const ParcelFileDescriptor = Java.type("android.os.ParcelFileDescriptor");
|
|
const PdfRenderer = Java.type("android.graphics.pdf.PdfRenderer");
|
|
const PdfRendererPage = Java.type("android.graphics.pdf.PdfRenderer$Page");
|
|
const Bitmap = Java.type("android.graphics.Bitmap");
|
|
const BitmapConfig = Java.type("android.graphics.Bitmap$Config");
|
|
const BitmapCompressFormat = Java.type("android.graphics.Bitmap$CompressFormat");
|
|
const Color = Java.type("android.graphics.Color");
|
|
const StandardCharsets = Java.type("java.nio.charset.StandardCharsets");
|
|
const EnhancedAIService = Java.type("com.ai.assistance.operit.api.chat.EnhancedAIService");
|
|
|
|
interface ParsePdfWithVisionParams {
|
|
pdf_path: string;
|
|
page_prompt?: string;
|
|
start_page?: number;
|
|
end_page?: number;
|
|
}
|
|
|
|
interface FailedPageInfo {
|
|
page_number: number;
|
|
stage: "precheck" | "render" | "vision";
|
|
error: string;
|
|
}
|
|
|
|
interface PageResult {
|
|
page_number: number;
|
|
image_path: string;
|
|
image_link: string;
|
|
output_path: string;
|
|
}
|
|
|
|
interface InternalPageResult extends PageResult {
|
|
analysis: string;
|
|
}
|
|
|
|
interface PageProcessingFailure {
|
|
page_number: number;
|
|
stage: FailedPageInfo["stage"];
|
|
error: string;
|
|
partial_results: InternalPageResult[];
|
|
}
|
|
|
|
interface VisionModelInfo {
|
|
function_type: "IMAGE_RECOGNITION";
|
|
config_id: string;
|
|
config_name: string;
|
|
selected_model: string;
|
|
}
|
|
|
|
interface ParsePdfWithVisionData {
|
|
pdf_path: string;
|
|
image_dir: string;
|
|
output_path: string;
|
|
total_page_count: number;
|
|
processed_page_count: number;
|
|
pages: PageResult[];
|
|
vision_model: VisionModelInfo;
|
|
failed_page?: FailedPageInfo;
|
|
}
|
|
|
|
interface ToolResponse {
|
|
success: boolean;
|
|
message: string;
|
|
data?: ParsePdfWithVisionData;
|
|
}
|
|
|
|
function isBlank(value: unknown): boolean {
|
|
return String(value ?? "").trim().length === 0;
|
|
}
|
|
|
|
function toTrimmedString(value: unknown): string {
|
|
return String(value ?? "").trim();
|
|
}
|
|
|
|
function parsePositiveInteger(
|
|
value: number | undefined,
|
|
fieldName: string,
|
|
options?: { allowUndefined?: boolean }
|
|
): number | undefined {
|
|
if (value === undefined || value === null) {
|
|
if (options?.allowUndefined) {
|
|
return undefined;
|
|
}
|
|
throw new Error(`${fieldName} 不能为空。`);
|
|
}
|
|
|
|
if (!Number.isInteger(value) || value <= 0) {
|
|
throw new Error(`${fieldName} 必须是大于 0 的整数。`);
|
|
}
|
|
return value;
|
|
}
|
|
|
|
function buildMergedText(pages: InternalPageResult[]): string {
|
|
return pages
|
|
.map(page => `===== Page ${page.page_number} =====\n${page.analysis}`)
|
|
.join("\n\n");
|
|
}
|
|
|
|
function buildVisionModelInfo(binding: FunctionModelConfigResultData): VisionModelInfo {
|
|
return {
|
|
function_type: TOOL_FUNCTION_TYPE,
|
|
config_id: String(binding.configId || ""),
|
|
config_name: String(binding.configName || ""),
|
|
selected_model: String(binding.selectedModel || "")
|
|
};
|
|
}
|
|
|
|
function buildData(
|
|
pdfPath: string,
|
|
imageDir: string,
|
|
outputPath: string,
|
|
totalPageCount: number,
|
|
pages: InternalPageResult[],
|
|
visionModel: VisionModelInfo,
|
|
failedPage?: FailedPageInfo
|
|
): ParsePdfWithVisionData {
|
|
const data: ParsePdfWithVisionData = {
|
|
pdf_path: pdfPath,
|
|
image_dir: imageDir,
|
|
output_path: outputPath,
|
|
total_page_count: totalPageCount,
|
|
processed_page_count: pages.length,
|
|
pages: pages.map(page => ({
|
|
page_number: page.page_number,
|
|
image_path: page.image_path,
|
|
image_link: page.image_link,
|
|
output_path: page.output_path
|
|
})),
|
|
vision_model: visionModel
|
|
};
|
|
|
|
if (failedPage) {
|
|
data.failed_page = failedPage;
|
|
}
|
|
|
|
return data;
|
|
}
|
|
|
|
function buildPrecheckFailure(message: string): ToolResponse {
|
|
return {
|
|
success: false,
|
|
message,
|
|
data: {
|
|
pdf_path: "",
|
|
image_dir: "",
|
|
output_path: "",
|
|
total_page_count: 0,
|
|
processed_page_count: 0,
|
|
pages: [],
|
|
vision_model: {
|
|
function_type: TOOL_FUNCTION_TYPE,
|
|
config_id: "",
|
|
config_name: "",
|
|
selected_model: ""
|
|
},
|
|
failed_page: {
|
|
page_number: 0,
|
|
stage: "precheck",
|
|
error: message
|
|
}
|
|
}
|
|
};
|
|
}
|
|
|
|
function safeClose(resource: any, label: string): void {
|
|
if (!resource) {
|
|
return;
|
|
}
|
|
try {
|
|
resource.close();
|
|
} catch (error) {
|
|
console.warn(`[${TOOL_NAME}] 关闭 ${label} 失败: ${String(error)}`);
|
|
}
|
|
}
|
|
|
|
function safeRecycle(bitmap: any): void {
|
|
if (!bitmap) {
|
|
return;
|
|
}
|
|
try {
|
|
bitmap.recycle();
|
|
} catch (error) {
|
|
console.warn(`[${TOOL_NAME}] 回收 bitmap 失败: ${String(error)}`);
|
|
}
|
|
}
|
|
|
|
function ensureDirectoryExists(dir: any): void {
|
|
if (dir.exists()) {
|
|
if (!dir.isDirectory()) {
|
|
throw new Error(`目标路径不是目录: ${String(dir.getAbsolutePath())}`);
|
|
}
|
|
return;
|
|
}
|
|
const created = dir.mkdirs();
|
|
if (!created) {
|
|
throw new Error(`无法创建目录: ${String(dir.getAbsolutePath())}`);
|
|
}
|
|
}
|
|
|
|
function resolveAppContext(): JavaBridgeInstance {
|
|
if (typeof Java.getApplicationContext === "function") {
|
|
return Java.getApplicationContext();
|
|
}
|
|
if (typeof Java.getContext === "function") {
|
|
return Java.getContext();
|
|
}
|
|
throw new Error("无法获取应用 Context。");
|
|
}
|
|
|
|
function createOutputDirectory(): any {
|
|
const parserRootDir = new File(OPERIT_CLEAN_ON_EXIT_DIR, "pdf_vision_parser");
|
|
ensureDirectoryExists(parserRootDir);
|
|
|
|
const timestamp = new Date().toISOString().replace(/[:.]/g, "-");
|
|
const runDir = new File(parserRootDir, timestamp);
|
|
ensureDirectoryExists(runDir);
|
|
return runDir;
|
|
}
|
|
|
|
function writeTextFile(targetFile: any, content: string): string {
|
|
let outputStream: any = null;
|
|
let writer: any = null;
|
|
let bufferedWriter: any = null;
|
|
try {
|
|
outputStream = new FileOutputStream(targetFile);
|
|
writer = new OutputStreamWriter(outputStream, StandardCharsets.UTF_8);
|
|
bufferedWriter = new BufferedWriter(writer);
|
|
bufferedWriter.write(String(content ?? ""));
|
|
bufferedWriter.flush();
|
|
return String(targetFile.getAbsolutePath());
|
|
} finally {
|
|
safeClose(bufferedWriter, "bufferedWriter");
|
|
safeClose(writer, "writer");
|
|
safeClose(outputStream, "outputStream");
|
|
}
|
|
}
|
|
|
|
function writePageOutputFile(outputDir: any, pageNumber: number, content: string): string {
|
|
const textFile = new File(outputDir, `page-${padPageNumber(pageNumber)}.txt`);
|
|
return writeTextFile(textFile, content);
|
|
}
|
|
|
|
function writeMergedOutputFile(outputDir: any, pages: InternalPageResult[]): string {
|
|
if (!outputDir) {
|
|
return "";
|
|
}
|
|
const mergedFile = new File(outputDir, MERGED_OUTPUT_FILE_NAME);
|
|
return writeTextFile(mergedFile, buildMergedText(pages));
|
|
}
|
|
|
|
function padPageNumber(pageNumber: number): string {
|
|
return String(pageNumber).padStart(4, "0");
|
|
}
|
|
|
|
function assertVisionBinding(binding: FunctionModelConfigResultData): void {
|
|
if (!binding || isBlank(binding.configId)) {
|
|
throw new Error("IMAGE_RECOGNITION 当前没有可用绑定。");
|
|
}
|
|
if (!binding.config || binding.config.enableDirectImageProcessing !== true) {
|
|
throw new Error("IMAGE_RECOGNITION 当前绑定的模型未启用识图能力。");
|
|
}
|
|
}
|
|
|
|
function validatePdfFile(pdfPath: string): any {
|
|
if (isBlank(pdfPath)) {
|
|
throw new Error("pdf_path 不能为空。");
|
|
}
|
|
|
|
const pdfFile = new File(pdfPath);
|
|
if (!pdfFile.exists() || !pdfFile.isFile()) {
|
|
throw new Error(`PDF 文件不存在: ${pdfPath}`);
|
|
}
|
|
|
|
if (!String(pdfFile.getName()).toLowerCase().endsWith(".pdf")) {
|
|
throw new Error(`文件不是 PDF: ${pdfPath}`);
|
|
}
|
|
|
|
return pdfFile;
|
|
}
|
|
|
|
function stripThinkingContent(content: string): string {
|
|
return String(content ?? "")
|
|
.replace(/<think(?:ing)?>[\s\S]*?(<\/think(?:ing)?>|\z)/gi, "")
|
|
.replace(/<search>[\s\S]*?(<\/search>|\z)/gi, "")
|
|
.trim();
|
|
}
|
|
|
|
function sortPageResults(pages: InternalPageResult[]): InternalPageResult[] {
|
|
return [...pages].sort((left, right) => left.page_number - right.page_number);
|
|
}
|
|
|
|
function normalizeErrorMessage(error: unknown): string {
|
|
return String(error instanceof Error ? error.message : error);
|
|
}
|
|
|
|
async function analyzePageImage(
|
|
service: JavaBridgeInstance,
|
|
imagePath: string,
|
|
prompt: string
|
|
): Promise<string> {
|
|
const rawAnalysis = String(
|
|
await service.callSuspend("analyzeImageWithIntent", imagePath, prompt)
|
|
).trim();
|
|
const analysis = stripThinkingContent(rawAnalysis);
|
|
|
|
if (analysis.length === 0) {
|
|
throw new Error("识图模型返回了空结果。");
|
|
}
|
|
if (/^Image recognition failed:/i.test(rawAnalysis) || /^Image recognition failed:/i.test(analysis)) {
|
|
throw new Error(rawAnalysis || analysis);
|
|
}
|
|
|
|
return analysis;
|
|
}
|
|
|
|
async function processSinglePage(
|
|
pdfFile: any,
|
|
outputDir: any,
|
|
pageNumber: number,
|
|
prompt: string,
|
|
service: JavaBridgeInstance
|
|
): Promise<InternalPageResult> {
|
|
console.log(`[${TOOL_NAME}] 开始处理第 ${pageNumber} 页`);
|
|
|
|
const zeroBasedPageIndex = pageNumber - 1;
|
|
const imageFile = new File(
|
|
outputDir,
|
|
`page-${padPageNumber(pageNumber)}.${IMAGE_FORMAT}`
|
|
);
|
|
const imagePath = String(imageFile.getAbsolutePath());
|
|
|
|
let fileDescriptor: any = null;
|
|
let pdfRenderer: any = null;
|
|
let page: any = null;
|
|
let bitmap: any = null;
|
|
let outputStream: any = null;
|
|
let stage: FailedPageInfo["stage"] = "render";
|
|
|
|
try {
|
|
fileDescriptor = ParcelFileDescriptor.open(
|
|
pdfFile,
|
|
ParcelFileDescriptor.MODE_READ_ONLY
|
|
);
|
|
pdfRenderer = new PdfRenderer(fileDescriptor);
|
|
page = pdfRenderer.openPage(zeroBasedPageIndex);
|
|
|
|
const width = Number(page.getWidth()) * RENDER_SCALE;
|
|
const height = Number(page.getHeight()) * RENDER_SCALE;
|
|
|
|
bitmap = Bitmap.createBitmap(width, height, BitmapConfig.ARGB_8888);
|
|
bitmap.eraseColor(Color.WHITE);
|
|
page.render(
|
|
bitmap,
|
|
null,
|
|
null,
|
|
PdfRendererPage.RENDER_MODE_FOR_DISPLAY
|
|
);
|
|
|
|
outputStream = new FileOutputStream(imageFile);
|
|
const compressed = bitmap.compress(
|
|
BitmapCompressFormat.PNG,
|
|
100,
|
|
outputStream
|
|
);
|
|
if (!compressed) {
|
|
throw new Error(`页面 ${pageNumber} 图片写入失败。`);
|
|
}
|
|
|
|
safeClose(outputStream, `page-${pageNumber} output stream`);
|
|
outputStream = null;
|
|
|
|
const imageLink = String(
|
|
NativeInterface.registerImageFromPath(imagePath) || ""
|
|
).trim();
|
|
if (!imageLink) {
|
|
throw new Error(`页面 ${pageNumber} 图片链接注册失败。`);
|
|
}
|
|
|
|
stage = "vision";
|
|
const analysis = await analyzePageImage(service, imagePath, prompt);
|
|
const pageOutputPath = writePageOutputFile(outputDir, pageNumber, analysis);
|
|
|
|
console.log(`[${TOOL_NAME}] 第 ${pageNumber} 页处理完成`);
|
|
return {
|
|
page_number: pageNumber,
|
|
image_path: imagePath,
|
|
image_link: imageLink,
|
|
output_path: pageOutputPath,
|
|
analysis
|
|
};
|
|
} catch (error) {
|
|
throw {
|
|
page_number: pageNumber,
|
|
stage,
|
|
error: normalizeErrorMessage(error),
|
|
partial_results: []
|
|
} as PageProcessingFailure;
|
|
} finally {
|
|
safeClose(outputStream, `page-${pageNumber} output stream`);
|
|
safeRecycle(bitmap);
|
|
safeClose(page, `page-${pageNumber}`);
|
|
safeClose(pdfRenderer, `page-${pageNumber} pdfRenderer`);
|
|
safeClose(fileDescriptor, `page-${pageNumber} parcelFileDescriptor`);
|
|
}
|
|
}
|
|
|
|
async function processPagesWithConcurrency(
|
|
pageNumbers: number[],
|
|
limit: number,
|
|
worker: (pageNumber: number) => Promise<InternalPageResult>
|
|
): Promise<InternalPageResult[]> {
|
|
const results: Array<InternalPageResult | undefined> = new Array(pageNumbers.length);
|
|
let nextIndex = 0;
|
|
let failure: Omit<PageProcessingFailure, "partial_results"> | null = null;
|
|
|
|
async function runWorker(): Promise<void> {
|
|
while (true) {
|
|
if (failure) {
|
|
return;
|
|
}
|
|
|
|
const currentIndex = nextIndex;
|
|
nextIndex += 1;
|
|
|
|
if (currentIndex >= pageNumbers.length) {
|
|
return;
|
|
}
|
|
|
|
const pageNumber = pageNumbers[currentIndex];
|
|
try {
|
|
results[currentIndex] = await worker(pageNumber);
|
|
} catch (error) {
|
|
if (!failure) {
|
|
const errorData = error as Partial<PageProcessingFailure> | undefined;
|
|
failure = {
|
|
page_number: Number(errorData?.page_number ?? pageNumber),
|
|
stage: (errorData?.stage as FailedPageInfo["stage"] | undefined) ?? "render",
|
|
error: normalizeErrorMessage(errorData?.error ?? error)
|
|
};
|
|
}
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
const workerCount = Math.max(1, Math.min(limit, pageNumbers.length));
|
|
await Promise.all(
|
|
Array.from({ length: workerCount }, () => runWorker())
|
|
);
|
|
|
|
const partialResults = sortPageResults(
|
|
results.filter((page): page is InternalPageResult => Boolean(page))
|
|
);
|
|
if (failure) {
|
|
const failureInfo = failure as Omit<PageProcessingFailure, "partial_results">;
|
|
throw {
|
|
page_number: failureInfo.page_number,
|
|
stage: failureInfo.stage,
|
|
error: failureInfo.error,
|
|
partial_results: partialResults
|
|
} as PageProcessingFailure;
|
|
}
|
|
|
|
return partialResults;
|
|
}
|
|
|
|
async function parsePdfWithVisionInternal(
|
|
params: ParsePdfWithVisionParams
|
|
): Promise<ToolResponse> {
|
|
let binding: FunctionModelConfigResultData;
|
|
try {
|
|
binding = await Tools.SoftwareSettings.getFunctionModelConfig(TOOL_FUNCTION_TYPE);
|
|
assertVisionBinding(binding);
|
|
} catch (error) {
|
|
const message = String(error instanceof Error ? error.message : error);
|
|
return buildPrecheckFailure(message);
|
|
}
|
|
|
|
const visionModel = buildVisionModelInfo(binding);
|
|
const rawPdfPath = toTrimmedString(params?.pdf_path);
|
|
|
|
let pdfFile: any;
|
|
let startPage: number | undefined;
|
|
let endPageInput: number | undefined;
|
|
const resolvedPrompt =
|
|
toTrimmedString(params?.page_prompt) || DEFAULT_PAGE_PROMPT;
|
|
|
|
try {
|
|
pdfFile = validatePdfFile(rawPdfPath);
|
|
startPage = parsePositiveInteger(params?.start_page, "start_page", {
|
|
allowUndefined: true
|
|
}) ?? 1;
|
|
endPageInput = parsePositiveInteger(params?.end_page, "end_page", {
|
|
allowUndefined: true
|
|
});
|
|
} catch (error) {
|
|
const message = String(error instanceof Error ? error.message : error);
|
|
return {
|
|
success: false,
|
|
message,
|
|
data: {
|
|
pdf_path: rawPdfPath,
|
|
image_dir: "",
|
|
output_path: "",
|
|
total_page_count: 0,
|
|
processed_page_count: 0,
|
|
pages: [],
|
|
vision_model: visionModel,
|
|
failed_page: {
|
|
page_number: 0,
|
|
stage: "precheck",
|
|
error: message
|
|
}
|
|
}
|
|
};
|
|
}
|
|
|
|
const pages: InternalPageResult[] = [];
|
|
const context = resolveAppContext();
|
|
const enhancedAiService = EnhancedAIService.getInstance(context) as JavaBridgeInstance;
|
|
let outputDir: any = null;
|
|
let imageDir = "";
|
|
let outputPath = "";
|
|
|
|
let fileDescriptor: any = null;
|
|
let pdfRenderer: any = null;
|
|
let totalPageCount = 0;
|
|
|
|
try {
|
|
fileDescriptor = ParcelFileDescriptor.open(
|
|
pdfFile,
|
|
ParcelFileDescriptor.MODE_READ_ONLY
|
|
);
|
|
pdfRenderer = new PdfRenderer(fileDescriptor);
|
|
totalPageCount = Number(pdfRenderer.getPageCount());
|
|
|
|
if (!Number.isInteger(totalPageCount) || totalPageCount <= 0) {
|
|
const error = "PDF 没有可处理的页面。";
|
|
return {
|
|
success: false,
|
|
message: error,
|
|
data: buildData(rawPdfPath, imageDir, outputPath, 0, pages, visionModel, {
|
|
page_number: 0,
|
|
stage: "render",
|
|
error
|
|
})
|
|
};
|
|
}
|
|
|
|
const resolvedEndPage = Math.min(endPageInput ?? totalPageCount, totalPageCount);
|
|
if ((startPage ?? 1) > resolvedEndPage) {
|
|
const error =
|
|
(startPage ?? 1) > totalPageCount
|
|
? `start_page 超出了总页数，文档总页数为 ${totalPageCount}。`
|
|
: "start_page 不能大于 end_page。";
|
|
return {
|
|
success: false,
|
|
message: error,
|
|
data: buildData(rawPdfPath, imageDir, outputPath, totalPageCount, pages, visionModel, {
|
|
page_number: 0,
|
|
stage: "precheck",
|
|
error
|
|
})
|
|
};
|
|
}
|
|
|
|
outputDir = createOutputDirectory();
|
|
imageDir = String(outputDir.getAbsolutePath());
|
|
const pageNumbers: number[] = [];
|
|
for (let pageNumber = startPage ?? 1; pageNumber <= resolvedEndPage; pageNumber += 1) {
|
|
pageNumbers.push(pageNumber);
|
|
}
|
|
|
|
try {
|
|
const processedPages = await processPagesWithConcurrency(
|
|
pageNumbers,
|
|
PAGE_CONCURRENCY,
|
|
currentPageNumber =>
|
|
processSinglePage(
|
|
pdfFile,
|
|
outputDir,
|
|
currentPageNumber,
|
|
resolvedPrompt,
|
|
enhancedAiService
|
|
)
|
|
);
|
|
pages.push(...processedPages);
|
|
} catch (error) {
|
|
const failedPage = error as PageProcessingFailure;
|
|
pages.push(...sortPageResults(failedPage.partial_results ?? []));
|
|
outputPath = writeMergedOutputFile(outputDir, pages);
|
|
|
|
return {
|
|
success: false,
|
|
message: `第 ${failedPage.page_number} 页处理失败: ${failedPage.error}`,
|
|
data: buildData(rawPdfPath, imageDir, outputPath, totalPageCount, pages, visionModel, {
|
|
page_number: failedPage.page_number,
|
|
stage: failedPage.stage,
|
|
error: failedPage.error
|
|
})
|
|
};
|
|
}
|
|
|
|
outputPath = writeMergedOutputFile(outputDir, pages);
|
|
return {
|
|
success: true,
|
|
message: `成功解析 ${pages.length} 页 PDF 并完成识图，结果已写入 ${outputPath}。`,
|
|
data: buildData(rawPdfPath, imageDir, outputPath, totalPageCount, pages, visionModel)
|
|
};
|
|
} catch (error) {
|
|
const message = String(error instanceof Error ? error.message : error);
|
|
outputPath = writeMergedOutputFile(outputDir, pages);
|
|
return {
|
|
success: false,
|
|
message,
|
|
data: buildData(rawPdfPath, imageDir, outputPath, totalPageCount, pages, visionModel, {
|
|
page_number: 0,
|
|
stage: "render",
|
|
error: message
|
|
})
|
|
};
|
|
} finally {
|
|
safeClose(pdfRenderer, "pdfRenderer");
|
|
safeClose(fileDescriptor, "parcelFileDescriptor");
|
|
}
|
|
}
|
|
|
|
async function wrapToolExecution(
|
|
func: (params: ParsePdfWithVisionParams) => Promise<ToolResponse>,
|
|
params: ParsePdfWithVisionParams
|
|
): Promise<void> {
|
|
try {
|
|
const result = await func(params || ({} as ParsePdfWithVisionParams));
|
|
complete(result);
|
|
} catch (error) {
|
|
const message = String(error instanceof Error ? error.message : error);
|
|
complete({
|
|
success: false,
|
|
message,
|
|
data: {
|
|
pdf_path: toTrimmedString(params?.pdf_path),
|
|
image_dir: "",
|
|
output_path: "",
|
|
total_page_count: 0,
|
|
processed_page_count: 0,
|
|
pages: [],
|
|
vision_model: {
|
|
function_type: TOOL_FUNCTION_TYPE,
|
|
config_id: "",
|
|
config_name: "",
|
|
selected_model: ""
|
|
},
|
|
failed_page: {
|
|
page_number: 0,
|
|
stage: "precheck",
|
|
error: message
|
|
}
|
|
}
|
|
});
|
|
}
|
|
}
|
|
|
|
async function parse_pdf_with_vision(
|
|
params: ParsePdfWithVisionParams
|
|
): Promise<void> {
|
|
await wrapToolExecution(parsePdfWithVisionInternal, params);
|
|
}
|
|
|
|
async function main(params: ParsePdfWithVisionParams): Promise<void> {
|
|
await wrapToolExecution(parsePdfWithVisionInternal, params);
|
|
}
|
|
|
|
return {
|
|
parse_pdf_with_vision,
|
|
main
|
|
};
|
|
})();
|
|
|
|
exports.parse_pdf_with_vision = PdfVisionParser.parse_pdf_with_vision;
|
|
exports.main = PdfVisionParser.main;
|