Spaces:
Sleeping
Sleeping
Update document_generator.py
Browse files- document_generator.py +58 -11
document_generator.py
CHANGED
@@ -39,6 +39,49 @@ FORMAT YOUR OUTPUT AS MARKDOWN ENCLOSED IN <response></response> tags
|
|
39 |
|
40 |
DOCUMENT_SECTION_PROMPT_USER = """<prompt>Output the content for the section "{section_or_subsection_title}" formatted as markdown. Follow this instruction: {content_instruction}</prompt>"""
|
41 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
# File: app.py
|
43 |
import os
|
44 |
import json
|
@@ -146,10 +189,10 @@ class DocumentGenerator:
|
|
146 |
return content.lstrip()
|
147 |
|
148 |
@log_execution
|
149 |
-
def generate_document_outline(self, query: str, max_retries: int = 3) -> Optional[Dict]:
|
150 |
messages = [
|
151 |
-
{"role": "system", "content": DOCUMENT_OUTLINE_PROMPT_SYSTEM},
|
152 |
-
{"role": "user", "content": DOCUMENT_OUTLINE_PROMPT_USER.format(query=query)}
|
153 |
]
|
154 |
|
155 |
for attempt in range(max_retries):
|
@@ -168,10 +211,11 @@ class DocumentGenerator:
|
|
168 |
return None
|
169 |
|
170 |
@log_execution
|
171 |
-
def generate_content(self, title: str, content_instruction: str, section_number: str) -> str:
|
|
|
172 |
self.content_messages.append({
|
173 |
"role": "user",
|
174 |
-
"content":
|
175 |
section_or_subsection_title=title,
|
176 |
content_instruction=content_instruction
|
177 |
)
|
@@ -240,6 +284,7 @@ router = APIRouter()
|
|
240 |
|
241 |
class DocumentRequest(BaseModel):
|
242 |
query: str
|
|
|
243 |
|
244 |
class JsonDocumentResponse(BaseModel):
|
245 |
json_document: Dict
|
@@ -247,6 +292,7 @@ class JsonDocumentResponse(BaseModel):
|
|
247 |
class MarkdownDocumentRequest(BaseModel):
|
248 |
json_document: Dict
|
249 |
query: str
|
|
|
250 |
|
251 |
MESSAGE_DELIMITER = b"\n---DELIMITER---\n"
|
252 |
|
@@ -254,16 +300,17 @@ def yield_message(message):
|
|
254 |
message_json = json.dumps(message, ensure_ascii=False).encode('utf-8')
|
255 |
return message_json + MESSAGE_DELIMITER
|
256 |
|
257 |
-
async def generate_document_stream(document_generator: DocumentGenerator, document_outline: Dict, query: str):
|
258 |
document_generator.document_outline = document_outline
|
259 |
db_manager = DatabaseManager()
|
260 |
overall_objective = query
|
261 |
document_layout = json.dumps(document_generator.document_outline, indent=2)
|
262 |
|
|
|
263 |
document_generator.content_messages = [
|
264 |
{
|
265 |
"role": "system",
|
266 |
-
"content":
|
267 |
overall_objective=overall_objective,
|
268 |
document_layout=document_layout
|
269 |
)
|
@@ -275,7 +322,7 @@ async def generate_document_stream(document_generator: DocumentGenerator, docume
|
|
275 |
section_number = section.get("SectionNumber", "")
|
276 |
content_instruction = section.get("Content", "")
|
277 |
logging.info(f"Generating content for section: {section_title}")
|
278 |
-
content = document_generator.generate_content(section_title, content_instruction, section_number)
|
279 |
section["Content"] = content
|
280 |
yield yield_message({
|
281 |
"type": "document_section",
|
@@ -291,7 +338,7 @@ async def generate_document_stream(document_generator: DocumentGenerator, docume
|
|
291 |
subsection_number = subsection.get("SectionNumber", "")
|
292 |
subsection_content_instruction = subsection.get("Content", "")
|
293 |
logging.info(f"Generating content for subsection: {subsection_title}")
|
294 |
-
content = document_generator.generate_content(subsection_title, subsection_content_instruction, subsection_number)
|
295 |
subsection["Content"] = content
|
296 |
yield yield_message({
|
297 |
"type": "document_section",
|
@@ -322,7 +369,7 @@ async def generate_document_outline_endpoint(request: DocumentRequest):
|
|
322 |
document_generator = DocumentGenerator(ai_client)
|
323 |
|
324 |
try:
|
325 |
-
json_document = document_generator.generate_document_outline(request.query)
|
326 |
|
327 |
if json_document is None:
|
328 |
raise HTTPException(status_code=500, detail="Failed to generate a valid document outline")
|
@@ -338,7 +385,7 @@ async def generate_markdown_document_stream_endpoint(request: MarkdownDocumentRe
|
|
338 |
|
339 |
async def stream_generator():
|
340 |
try:
|
341 |
-
async for chunk in generate_document_stream(document_generator, request.json_document, request.query):
|
342 |
yield chunk
|
343 |
except Exception as e:
|
344 |
yield yield_message({
|
|
|
39 |
|
40 |
DOCUMENT_SECTION_PROMPT_USER = """<prompt>Output the content for the section "{section_or_subsection_title}" formatted as markdown. Follow this instruction: {content_instruction}</prompt>"""
|
41 |
|
42 |
+
##########################################
|
43 |
+
|
44 |
+
DOCUMENT_TEMPLATE_OUTLINE_PROMPT_SYSTEM = """You are a document template generator. Provide the outline of the document requested in <prompt></prompt> in JSON format.
|
45 |
+
Include sections and subsections if required. Use the "Content" field to provide a specific prompt or instruction for generating template (with empty input fields) that particular section or subsection. Specify that input fields should be empty in each prompt.
|
46 |
+
make sure the Sections follow a logical flow and each prompt's content does not overlap with other sections.
|
47 |
+
OUTPUT IN FOLLOWING JSON FORMAT enclosed in <output> tags
|
48 |
+
<output>
|
49 |
+
{
|
50 |
+
"Document": {
|
51 |
+
"Title": "Document Title",
|
52 |
+
"Author": "Author Name",
|
53 |
+
"Date": "YYYY-MM-DD",
|
54 |
+
"Version": "1.0",
|
55 |
+
|
56 |
+
"Sections": [
|
57 |
+
{
|
58 |
+
"SectionNumber": "1",
|
59 |
+
"Title": "Section Title",
|
60 |
+
"Content": "Specific prompt or instruction for generating empty template for this section",
|
61 |
+
"Subsections": [
|
62 |
+
{
|
63 |
+
"SectionNumber": "1.1",
|
64 |
+
"Title": "Subsection Title",
|
65 |
+
"Content": "Specific prompt or instruction for generating empty template for this subsection"
|
66 |
+
}
|
67 |
+
]
|
68 |
+
}
|
69 |
+
]
|
70 |
+
}
|
71 |
+
}
|
72 |
+
</output>"""
|
73 |
+
|
74 |
+
DOCUMENT_TEMPLATE_PROMPT_USER = """<prompt>{query}</prompt>"""
|
75 |
+
|
76 |
+
DOCUMENT_TEMPLATE_SECTION_PROMPT_SYSTEM = """You are a document template generator, You need to output only the content requested in the section in the prompt.
|
77 |
+
FORMAT YOUR OUTPUT AS MARKDOWN ENCLOSED IN <response></response> tags
|
78 |
+
<overall_objective>{overall_objective}</overall_objective>
|
79 |
+
<document_layout>{document_layout}</document_layout>"""
|
80 |
+
|
81 |
+
DOCUMENT_TEMPLATE_SECTION_PROMPT_USER = """<prompt>Output the content for the section "{section_or_subsection_title}" formatted as markdown. Follow this instruction: {content_instruction}</prompt>"""
|
82 |
+
|
83 |
+
|
84 |
+
|
85 |
# File: app.py
|
86 |
import os
|
87 |
import json
|
|
|
189 |
return content.lstrip()
|
190 |
|
191 |
@log_execution
|
192 |
+
def generate_document_outline(self, query: str, template: bool = False, max_retries: int = 3) -> Optional[Dict]:
|
193 |
messages = [
|
194 |
+
{"role": "system", "content": DOCUMENT_OUTLINE_PROMPT_SYSTEM if not template else DOCUMENT_TEMPLATE_OUTLINE_PROMPT_SYSTEM},
|
195 |
+
{"role": "user", "content": DOCUMENT_OUTLINE_PROMPT_USER.format(query=query) if not template else DOCUMENT_TEMPLATE_PROMPT_USER.format(query=query)}
|
196 |
]
|
197 |
|
198 |
for attempt in range(max_retries):
|
|
|
211 |
return None
|
212 |
|
213 |
@log_execution
|
214 |
+
def generate_content(self, title: str, content_instruction: str, section_number: str, template: bool = False) -> str:
|
215 |
+
SECTION_PROMPT_USER = DOCUMENT_SECTION_PROMPT_USER if not template else DOCUMENT_TEMPLATE_SECTION_PROMPT_USER
|
216 |
self.content_messages.append({
|
217 |
"role": "user",
|
218 |
+
"content": SECTION_PROMPT_USER.format(
|
219 |
section_or_subsection_title=title,
|
220 |
content_instruction=content_instruction
|
221 |
)
|
|
|
284 |
|
285 |
class DocumentRequest(BaseModel):
|
286 |
query: str
|
287 |
+
template: bool = False
|
288 |
|
289 |
class JsonDocumentResponse(BaseModel):
|
290 |
json_document: Dict
|
|
|
292 |
class MarkdownDocumentRequest(BaseModel):
|
293 |
json_document: Dict
|
294 |
query: str
|
295 |
+
template: bool = False
|
296 |
|
297 |
MESSAGE_DELIMITER = b"\n---DELIMITER---\n"
|
298 |
|
|
|
300 |
message_json = json.dumps(message, ensure_ascii=False).encode('utf-8')
|
301 |
return message_json + MESSAGE_DELIMITER
|
302 |
|
303 |
+
async def generate_document_stream(document_generator: DocumentGenerator, document_outline: Dict, query: str, template: bool = False):
|
304 |
document_generator.document_outline = document_outline
|
305 |
db_manager = DatabaseManager()
|
306 |
overall_objective = query
|
307 |
document_layout = json.dumps(document_generator.document_outline, indent=2)
|
308 |
|
309 |
+
SECTION_PROMPT_SYSTEM = DOCUMENT_SECTION_PROMPT_SYSTEM if not template else DOCUMENT_TEMPLATE_SECTION_PROMPT_SYSTEM
|
310 |
document_generator.content_messages = [
|
311 |
{
|
312 |
"role": "system",
|
313 |
+
"content": SECTION_PROMPT_SYSTEM.format(
|
314 |
overall_objective=overall_objective,
|
315 |
document_layout=document_layout
|
316 |
)
|
|
|
322 |
section_number = section.get("SectionNumber", "")
|
323 |
content_instruction = section.get("Content", "")
|
324 |
logging.info(f"Generating content for section: {section_title}")
|
325 |
+
content = document_generator.generate_content(section_title, content_instruction, section_number, template)
|
326 |
section["Content"] = content
|
327 |
yield yield_message({
|
328 |
"type": "document_section",
|
|
|
338 |
subsection_number = subsection.get("SectionNumber", "")
|
339 |
subsection_content_instruction = subsection.get("Content", "")
|
340 |
logging.info(f"Generating content for subsection: {subsection_title}")
|
341 |
+
content = document_generator.generate_content(subsection_title, subsection_content_instruction, subsection_number, template)
|
342 |
subsection["Content"] = content
|
343 |
yield yield_message({
|
344 |
"type": "document_section",
|
|
|
369 |
document_generator = DocumentGenerator(ai_client)
|
370 |
|
371 |
try:
|
372 |
+
json_document = document_generator.generate_document_outline(request.query, request.template)
|
373 |
|
374 |
if json_document is None:
|
375 |
raise HTTPException(status_code=500, detail="Failed to generate a valid document outline")
|
|
|
385 |
|
386 |
async def stream_generator():
|
387 |
try:
|
388 |
+
async for chunk in generate_document_stream(document_generator, request.json_document, request.query, request.template):
|
389 |
yield chunk
|
390 |
except Exception as e:
|
391 |
yield yield_message({
|