更新

difyPlugin/pdf/tools/pdf_summary.yaml

@@ -0,0 +1,99 @@
+identity:
+  name: "pdf_summary"
+  author: "yslg"
+  label:
+    en_US: "PDF Page Summary"
+    zh_Hans: "PDF页面概述"
+    pt_BR: "Resumo de Página PDF"
+    ja_JP: "PDFページ概要"
+description:
+  human:
+    en_US: "Extract core elements (text, image, table, path) from PDF pages with coordinates, then summarize via LLM"
+    zh_Hans: "提取PDF页面核心元素（文本、图片、表格、路径）及坐标，并通过LLM进行概述"
+    pt_BR: "Extrair elementos principais (texto, imagem, tabela, caminho) de páginas PDF com coordenadas e resumir via LLM"
+    ja_JP: "PDFページからコア要素（テキスト、画像、テーブル、パス）を座標付きで抽出し、LLMで要約"
+  llm: "Extract core elements (text, image, table, drawing path) with coordinates from specified PDF page range, then use LLM to summarize the content"
+parameters:
+  - name: file
+    type: file
+    required: true
+    label:
+      en_US: PDF File
+      zh_Hans: PDF 文件
+      pt_BR: Arquivo PDF
+      ja_JP: PDFファイル
+    human_description:
+      en_US: "PDF file to process"
+      zh_Hans: "要处理的 PDF 文件"
+      pt_BR: "Arquivo PDF para processar"
+      ja_JP: "処理するPDFファイル"
+    llm_description: "PDF file to extract elements from and summarize"
+    form: llm
+    fileTypes:
+      - "pdf"
+  - name: pdf_start_page
+    type: number
+    required: true
+    label:
+      en_US: Start Page
+      zh_Hans: 起始页码
+      pt_BR: Página Inicial
+      ja_JP: 開始ページ
+    human_description:
+      en_US: "Start page index (0-based)"
+      zh_Hans: "起始页码（从0开始）"
+      pt_BR: "Índice da página inicial (base 0)"
+      ja_JP: "開始ページ番号（0始まり）"
+    llm_description: "Start page index (0-based) for element extraction"
+    form: llm
+    default: 0
+  - name: pdf_end_page
+    type: number
+    required: true
+    label:
+      en_US: End Page
+      zh_Hans: 结束页码
+      pt_BR: Página Final
+      ja_JP: 終了ページ
+    human_description:
+      en_US: "End page index (0-based, inclusive)"
+      zh_Hans: "结束页码（从0开始，包含该页）"
+      pt_BR: "Índice da página final (base 0, inclusivo)"
+      ja_JP: "終了ページ番号（0始まり、含む）"
+    llm_description: "End page index (0-based, inclusive) for element extraction"
+    form: llm
+    default: 0
+  - name: model
+    type: model-selector
+    scope: llm
+    required: true
+    label:
+      en_US: LLM Model
+      zh_Hans: LLM 模型
+      pt_BR: Modelo LLM
+      ja_JP: LLMモデル
+    human_description:
+      en_US: "LLM model used for summarizing extracted content"
+      zh_Hans: "用于概述提取内容的 LLM 模型"
+      pt_BR: "Modelo LLM usado para resumir o conteúdo extraído"
+      ja_JP: "抽出内容の要約に使用するLLMモデル"
+    form: form
+  - name: llm_prompt
+    type: string
+    required: false
+    label:
+      en_US: LLM Prompt
+      zh_Hans: LLM 提示词
+      pt_BR: Prompt do LLM
+      ja_JP: LLMプロンプト
+    human_description:
+      en_US: "System prompt for LLM summarization"
+      zh_Hans: "LLM 概述的系统提示词"
+      pt_BR: "Prompt do sistema para resumo LLM"
+      ja_JP: "LLM要約用のシステムプロンプト"
+    llm_description: "System prompt guiding LLM on how to summarize the extracted PDF content"
+    form: form
+    default: "你是一个专业的文档分析助手。请根据以下从PDF页面中提取的结构化内容（包含文本、图片信息、表格和矢量图形），对每页内容进行准确、简洁的概述。"
+extra:
+  python:
+    source: tools/pdf_summary.py