-
Notifications
You must be signed in to change notification settings - Fork 0
/
full_doc.py
336 lines (277 loc) · 13.5 KB
/
full_doc.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
#####################################################
### DOCUMENT PROCESSOR [FULLDOC]
#####################################################
### Jonathan Wang
# ABOUT:
# This creates an app to chat with PDFs.
# This is the FULLDOC
# which is a class that associates documents
# with their critical information
# and their tools. (keywords, summary, queryengine, etc.)
#####################################################
### TODO Board:
# Automatically determine which reader to use for each document based on the file type.
#####################################################
### PROGRAM SETTINGS
#####################################################
### PROGRAM IMPORTS
from __future__ import annotations
import asyncio
from pathlib import Path
from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, TypeVar
from uuid import UUID, uuid4
from llama_index.core import StorageContext, VectorStoreIndex
from llama_index.core.query_engine import SubQuestionQueryEngine
from llama_index.core.schema import BaseNode, TransformComponent
from llama_index.core.settings import Settings
from llama_index.core.tools import QueryEngineTool, ToolMetadata
from streamlit import session_state as ss
if TYPE_CHECKING:
from llama_index.core.base.base_query_engine import BaseQueryEngine
from llama_index.core.callbacks import CallbackManager
from llama_index.core.node_parser import NodeParser
from llama_index.core.readers.base import BaseReader
from llama_index.core.response_synthesizers import BaseSynthesizer
from llama_index.core.retrievers import BaseRetriever
# Own Modules
from engine import get_engine
from keywords import KeywordMetadataAdder
from retriever import get_retriever
from storage import get_docstore, get_vector_store
from summary import DEFAULT_ONELINE_SUMMARY_TEMPLATE, DEFAULT_TREE_SUMMARY_TEMPLATE
#####################################################
### SCRIPT
GenericNode = TypeVar("GenericNode", bound=BaseNode)
class FullDocument:
"""Bundles all the information about a document together.
Args:
name (str): The name of the document.
file_path (Path): The path to the document.
summary (str): The summary of the document.
keywords (List[str]): The keywords of the document.
entities (List[str]): The entities of the document.
vector_store (BaseDocumentStore): The vector store of the document.
"""
# Identifiers
id: UUID
name: str
file_path: Path
file_name: str
# Basic Contents
summary: str
summary_oneline: str # A one line summary of the document.
keywords: set[str] # List of keywords in document.
# entities: Set[str] # list of entities in document ## TODO: Add entities
metadata: dict[str, Any] | None
# NOTE: other metdata that might be useful:
# Document Creation / Last Date (e.g., recency important for legal/medical questions)
# Document Source and Trustworthiness
# Document Access Level (though this isn't important for us here.)
# Document Citations?
# Document Format? (text/spreadsheet/presentation/image/etc.)
# RAG Components
nodes: list[BaseNode]
storage_context: StorageContext # NOTE: current setup has single storage context per document.
vector_store_index: VectorStoreIndex
retriever: BaseRetriever # TODO(Jonathan Wang): Consider multiple retrievers for keywords vs semantic.
engine: BaseQueryEngine # TODO(Jonathan Wang): Consider mulitple engines.
subquestion_engine: SubQuestionQueryEngine
def __init__(
self,
name: str,
file_path: Path | str,
metadata: dict[str, Any] | None = None
) -> None:
self.id = uuid4()
self.name = name
if (isinstance(file_path, str)):
file_path = Path(file_path)
self.file_path = file_path
self.file_name = file_path.name
self.metadata = metadata
@classmethod
def class_name(cls) -> str:
return "FullDocument"
def add_name_to_nodes(self, nodes: list[GenericNode]) -> list[GenericNode]:
"""Add the name of the document to the nodes.
Args:
nodes (List[GenericNode]): The nodes to add the name to.
Returns:
List[GenericNode]: The nodes with the name added.
"""
for node in nodes:
node.metadata["name"] = self.name
return nodes
def file_to_nodes(
self,
reader: BaseReader,
postreaders: list[Callable[[list[GenericNode]], list[GenericNode]] | TransformComponent] | None=None, # NOTE: these should be used in order. and probably all TransformComponent instead.
node_parser: NodeParser | None=None,
postparsers: list[Callable[[list[GenericNode]], list[GenericNode]] | TransformComponent] | None=None, # Stuff like chunking, adding Embeddings, etc.
) -> None:
"""Read in the file path and get the nodes.
Args:
file_path (Optional[Path], optional): The path to the file. Defaults to file_path from init.
reader (Optional[BaseReader], optional): The reader to use. Defaults to reader from init.
"""
# Use the provided reader to read in the file.
print("NEWPDF: Reading input file...")
nodes = reader.load_data(file_path=self.file_path)
# Use node postreaders to post process the nodes.
if (postreaders is not None):
for node_postreader in postreaders:
nodes = node_postreader(nodes) # type: ignore (TransformComponent allows a list of nodes)
# Use node parser to parse the nodes.
if (node_parser is None):
node_parser = Settings.node_parser
nodes = node_parser(nodes) # type: ignore (Document is a child of BaseNode)
# Use node postreaders to post process the nodes. (also add the common name to the nodes)
if (postparsers is None):
postparsers = [self.add_name_to_nodes]
else:
postparsers.append(self.add_name_to_nodes)
for node_postparser in postparsers:
nodes = node_postparser(nodes) # type: ignore (TransformComponent allows a list of nodes)
# Save nodes
self.nodes = nodes # type: ignore
def nodes_to_summary(
self,
summarizer: BaseSynthesizer, # NOTE: this is typically going to be a TreeSummarizer / SimpleSummarize for our use case
query_str: str = DEFAULT_TREE_SUMMARY_TEMPLATE,
) -> None:
"""Summarize the nodes.
Args:
summarizer (BaseSynthesizer): The summarizer to use. Takes in nodes and returns summary.
"""
if (not hasattr(self, "nodes")):
msg = "Nodes must be extracted from document using `file_to_nodes` before calling `nodes_to_summary`."
raise ValueError(msg)
text_chunks = [getattr(node, "text", "") for node in self.nodes if hasattr(node, "text")]
summary_responses = summarizer.aget_response(query_str=query_str, text_chunks=text_chunks)
loop = asyncio.get_event_loop()
summary = loop.run_until_complete(summary_responses)
if (not isinstance(summary, str)):
# TODO(Jonathan Wang): ... this should always give us a string, right? we're not doing anything fancy with TokenGen/TokenAsyncGen/Pydantic BaseModel...
msg = f"Summarizer must return a string summary. Actual type: {type(summary)}, with value {summary}."
raise TypeError(msg)
self.summary = summary
def summary_to_oneline(
self,
summarizer: BaseSynthesizer, # NOTE: this is typically going to be a SimpleSummarize / TreeSummarizer for our use case
query_str: str = DEFAULT_ONELINE_SUMMARY_TEMPLATE,
) -> None:
if (not hasattr(self, "summary")):
msg = "Summary must be extracted from document using `nodes_to_summary` before calling `summary_to_oneline`."
raise ValueError(msg)
oneline = summarizer.get_response(query_str=query_str, text_chunks=[self.summary]) # There's only one chunk.
self.summary_oneline = oneline # type: ignore | shouldn't have fancy TokenGenerators / TokenAsyncGenerators / Pydantic BaseModels
def nodes_to_document_keywords(self, keyword_extractor: Optional[KeywordMetadataAdder] = None) -> None:
"""Save the keywords from the nodes into the document.
Args:
keyword_extractor (Optional[BaseKeywordExtractor], optional): The keyword extractor to use. Defaults to None.
"""
if (not hasattr(self, "nodes")):
msg = "Nodes must be extracted from document using `file_to_nodes` before calling `nodes_to_keywords`."
raise ValueError(msg)
if (keyword_extractor is None):
keyword_extractor = KeywordMetadataAdder()
# Add keywords to nodes using KeywordMetadataAdder
keyword_extractor.process_nodes(self.nodes)
# Save keywords
keywords: list[str] = []
for node in self.nodes:
node_keywords = node.metadata.get("keyword_metadata", "").split(", ") # NOTE: KeywordMetadataAdder concatinates b/c required string output
keywords = keywords + node_keywords
# TODO(Jonathan Wang): handle dedupling keywords which are similar to each other (fuzzy?)
self.keywords = set(keywords)
def nodes_to_storage(self, create_new_storage: bool = True) -> None:
"""Save the nodes to storage."""
if (not hasattr(self, "nodes")):
msg = "Nodes must be extracted from document using `file_to_nodes` before calling `nodes_to_storage`."
raise ValueError(msg)
if (create_new_storage):
docstore = get_docstore(documents=self.nodes)
self.docstore = docstore
vector_store = get_vector_store()
storage_context = StorageContext.from_defaults(
docstore=docstore,
vector_store=vector_store
)
self.storage_context = storage_context
vector_store_index = VectorStoreIndex(
self.nodes, storage_context=storage_context
)
self.vector_store_index = vector_store_index
else:
### TODO(Jonathan Wang): use an existing storage instead of creating a new one.
msg = "Currently creates new storage for every document."
raise NotImplementedError(msg)
# TODO(Jonathan Wang): Create multiple different retrievers based on the question type(?)
# E.g., if the question is focused on specific keywords or phrases, use a retriever oriented towards sparse scores.
def storage_to_retriever(
self,
semantic_nodes: int = 6,
sparse_nodes: int = 3,
fusion_nodes: int = 3,
semantic_weight: float = 0.6,
merge_up_thresh: float = 0.5,
callback_manager: CallbackManager | None=None
) -> None:
"""Create retriever from storage."""
if (not hasattr(self, "vector_store_index")):
msg = "Vector store must be extracted from document using `nodes_to_storage` before calling `storage_to_retriever`."
raise ValueError(msg)
retriever = get_retriever(
_vector_store_index=self.vector_store_index,
semantic_top_k=semantic_nodes,
sparse_top_k=sparse_nodes,
fusion_similarity_top_k=fusion_nodes,
semantic_weight_fraction=semantic_weight,
merge_up_thresh=merge_up_thresh,
verbose=True,
_callback_manager=callback_manager or ss.callback_manager
)
self.retriever = retriever
def retriever_to_engine(
self,
response_synthesizer: BaseSynthesizer,
callback_manager: CallbackManager | None=None
) -> None:
"""Create query engine from retriever."""
if (not hasattr(self, "retriever")):
msg = "Retriever must be extracted from document using `storage_to_retriever` before calling `retriver_to_engine`."
raise ValueError(msg)
engine = get_engine(
retriever=self.retriever,
response_synthesizer=response_synthesizer,
callback_manager=callback_manager or ss.callback_manager
)
self.engine = engine
# TODO(Jonathan Wang): Create Summarization Index and Engine.
def engine_to_sub_question_engine(self) -> None:
"""Convert a basic query engine into a sub-question query engine for handling complex, multi-step questions.
Args:
query_engine (BaseQueryEngine): The Base Query Engine to convert.
"""
if (not hasattr(self, "summary_oneline")):
msg = "One Line Summary must be created for the document before calling `engine_to_sub_query_engine`"
raise ValueError(msg)
elif (not hasattr(self, "engine")):
msg = "Basic Query Engine must be created before calling `engine_to_sub_query_engine`"
raise ValueError(msg)
sqe_tools = [
QueryEngineTool(
query_engine=self.engine, # TODO(Jonathan Wang): handle mulitple engines?
metadata=ToolMetadata(
name=(self.name + "simple query answerer"),
description=f"""A tool that answers simple questions about the following document: {self.summary_oneline}"""
)
)
# TODO(Jonathan Wang): add more tools
]
subquestion_engine = SubQuestionQueryEngine.from_defaults(
query_engine_tools=sqe_tools,
verbose=True,
use_async=True
)
self.subquestion_engine = subquestion_engine