130 lines
3.9 KiB
Python
130 lines
3.9 KiB
Python
# Copyright (c) 2025 Patrick Motsch
|
|
# All rights reserved.
|
|
"""
|
|
Unified JSON document schema and helpers used by both generation prompts and renderers.
|
|
|
|
This defines a single canonical template and the supported section types.
|
|
"""
|
|
|
|
from typing import List, Literal, TypedDict
|
|
|
|
# Canonical list of supported section types across the system
|
|
supportedSectionTypes: List[str] = [
|
|
"table",
|
|
"bullet_list",
|
|
"heading",
|
|
"paragraph",
|
|
"code_block",
|
|
"image",
|
|
]
|
|
|
|
class InlineRun(TypedDict, total=False):
|
|
"""Single inline content run. Every paragraph/cell/list-item is a List[InlineRun]."""
|
|
type: Literal["text", "image", "link", "bold", "italic", "code"]
|
|
value: str # text content (for text/bold/italic/code/link-label)
|
|
fileId: str # for type=image: reference to FileItem
|
|
base64Data: str # for type=image: resolved base64 (post-processing)
|
|
mimeType: str # for type=image: e.g. "image/png"
|
|
widthPt: int # for type=image: optional render width
|
|
href: str # for type=link: URL target
|
|
|
|
supportedInlineRunTypes: List[str] = [
|
|
"text", "image", "link", "bold", "italic", "code",
|
|
]
|
|
|
|
|
|
# Canonical JSON template used for AI generation (documents array + sections)
|
|
# This template is used for STRUCTURE generation - sections have empty elements arrays.
|
|
# For content generation, elements arrays will be populated later.
|
|
jsonTemplateDocument: str = """{
|
|
"metadata": {
|
|
"split_strategy": "single_document",
|
|
"source_documents": [],
|
|
"extraction_method": "ai_generation",
|
|
"title": "{{DOCUMENT_TITLE}}"
|
|
},
|
|
"documents": [
|
|
{
|
|
"id": "doc_1",
|
|
"title": "{{DOCUMENT_TITLE}}",
|
|
"filename": "document.json",
|
|
"sections": [
|
|
{
|
|
"id": "section_heading_main_title",
|
|
"content_type": "heading",
|
|
"complexity": "simple",
|
|
"generation_hint": "Main document title heading",
|
|
"order": 1,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_paragraph_introduction",
|
|
"content_type": "paragraph",
|
|
"complexity": "simple",
|
|
"generation_hint": "Introduction paragraph",
|
|
"order": 2,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_heading_section_1",
|
|
"content_type": "heading",
|
|
"complexity": "simple",
|
|
"generation_hint": "Section heading for topic 1",
|
|
"order": 3,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_paragraph_section_1",
|
|
"content_type": "paragraph",
|
|
"complexity": "simple",
|
|
"generation_hint": "Content paragraph for section 1",
|
|
"order": 4,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_bullet_list_example",
|
|
"content_type": "bullet_list",
|
|
"complexity": "simple",
|
|
"generation_hint": "Bullet list items",
|
|
"order": 5,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_image_example",
|
|
"content_type": "image",
|
|
"complexity": "complex",
|
|
"generation_hint": "Illustration for document",
|
|
"image_prompt": "A detailed description for image generation",
|
|
"order": 6,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_table_example",
|
|
"content_type": "table",
|
|
"complexity": "simple",
|
|
"generation_hint": "Data table with relevant information",
|
|
"order": 7,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_code_example",
|
|
"content_type": "code_block",
|
|
"complexity": "simple",
|
|
"generation_hint": "Code example or snippet",
|
|
"order": 8,
|
|
"elements": []
|
|
},
|
|
{
|
|
"id": "section_paragraph_conclusion",
|
|
"content_type": "paragraph",
|
|
"complexity": "simple",
|
|
"generation_hint": "Conclusion paragraph",
|
|
"order": 9,
|
|
"elements": []
|
|
}
|
|
]
|
|
}
|
|
]
|
|
}"""
|
|
|
|
|