Spaces:
Running
Running
Update utils/ingestion.py
Browse files- utils/ingestion.py +26 -17
utils/ingestion.py
CHANGED
@@ -23,7 +23,7 @@ class DocumentProcessor:
|
|
23 |
"""Initialize document processor with necessary components"""
|
24 |
self.setup_document_converter()
|
25 |
self.embed_model = FastEmbedEmbeddings()
|
26 |
-
self.client = chromadb.PersistentClient(path="chroma_db") #
|
27 |
|
28 |
def setup_document_converter(self):
|
29 |
"""Configure document converter with advanced processing capabilities"""
|
@@ -33,9 +33,17 @@ class DocumentProcessor:
|
|
33 |
pipeline_options.table_structure_options.do_cell_matching = True
|
34 |
pipeline_options.ocr_options.lang = ["en"]
|
35 |
pipeline_options.table_structure_options.mode = TableFormerMode.ACCURATE
|
36 |
-
|
37 |
-
|
38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
|
40 |
self.converter = DocumentConverter(
|
41 |
format_options={
|
@@ -49,7 +57,7 @@ class DocumentProcessor:
|
|
49 |
def extract_chunk_metadata(self, chunk) -> Dict[str, Any]:
|
50 |
"""Extract essential metadata from a chunk"""
|
51 |
metadata = {
|
52 |
-
"text": chunk.text,
|
53 |
"headings": [],
|
54 |
"page_info": None,
|
55 |
"content_type": None
|
@@ -73,7 +81,7 @@ class DocumentProcessor:
|
|
73 |
|
74 |
def process_document(self, pdf_path: str):
|
75 |
"""Process document and create searchable index with metadata"""
|
76 |
-
print(f"Processing document: {pdf_path}")
|
77 |
start_time = time.time()
|
78 |
|
79 |
result = self.converter.convert(pdf_path)
|
@@ -87,7 +95,7 @@ class DocumentProcessor:
|
|
87 |
metadata = self.extract_chunk_metadata(chunk)
|
88 |
processed_chunks.append(metadata)
|
89 |
|
90 |
-
print("
|
91 |
collection = self.client.get_or_create_collection(name="document_chunks")
|
92 |
|
93 |
documents = []
|
@@ -98,10 +106,10 @@ class DocumentProcessor:
|
|
98 |
for idx, chunk in enumerate(processed_chunks):
|
99 |
text = chunk.get('text', '').strip()
|
100 |
if not text:
|
101 |
-
print(f"Skipping empty chunk at index {idx}")
|
102 |
continue # Skip empty chunks
|
103 |
|
104 |
-
embedding = self.embed_model.embed_documents([text])[0] # β
|
105 |
documents.append(text)
|
106 |
embeddings.append(embedding)
|
107 |
metadata_list.append({
|
@@ -111,14 +119,15 @@ class DocumentProcessor:
|
|
111 |
})
|
112 |
ids.append(str(idx))
|
113 |
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
|
|
|
121 |
|
122 |
processing_time = time.time() - start_time
|
123 |
-
print(f"
|
124 |
return collection
|
|
|
23 |
"""Initialize document processor with necessary components"""
|
24 |
self.setup_document_converter()
|
25 |
self.embed_model = FastEmbedEmbeddings()
|
26 |
+
self.client = chromadb.PersistentClient(path="chroma_db") # Persistent Storage
|
27 |
|
28 |
def setup_document_converter(self):
|
29 |
"""Configure document converter with advanced processing capabilities"""
|
|
|
33 |
pipeline_options.table_structure_options.do_cell_matching = True
|
34 |
pipeline_options.ocr_options.lang = ["en"]
|
35 |
pipeline_options.table_structure_options.mode = TableFormerMode.ACCURATE
|
36 |
+
|
37 |
+
# β
Automatically handle CPU fallback
|
38 |
+
try:
|
39 |
+
pipeline_options.accelerator_options = AcceleratorOptions(
|
40 |
+
num_threads=8, device=AcceleratorDevice.MPS
|
41 |
+
)
|
42 |
+
except Exception as e:
|
43 |
+
print("β οΈ MPS is not available. Falling back to CPU.")
|
44 |
+
pipeline_options.accelerator_options = AcceleratorOptions(
|
45 |
+
num_threads=8, device=AcceleratorDevice.CPU
|
46 |
+
)
|
47 |
|
48 |
self.converter = DocumentConverter(
|
49 |
format_options={
|
|
|
57 |
def extract_chunk_metadata(self, chunk) -> Dict[str, Any]:
|
58 |
"""Extract essential metadata from a chunk"""
|
59 |
metadata = {
|
60 |
+
"text": chunk.text.strip(),
|
61 |
"headings": [],
|
62 |
"page_info": None,
|
63 |
"content_type": None
|
|
|
81 |
|
82 |
def process_document(self, pdf_path: str):
|
83 |
"""Process document and create searchable index with metadata"""
|
84 |
+
print(f"π Processing document: {pdf_path}")
|
85 |
start_time = time.time()
|
86 |
|
87 |
result = self.converter.convert(pdf_path)
|
|
|
95 |
metadata = self.extract_chunk_metadata(chunk)
|
96 |
processed_chunks.append(metadata)
|
97 |
|
98 |
+
print("β
Chunking completed. Creating vector database...")
|
99 |
collection = self.client.get_or_create_collection(name="document_chunks")
|
100 |
|
101 |
documents = []
|
|
|
106 |
for idx, chunk in enumerate(processed_chunks):
|
107 |
text = chunk.get('text', '').strip()
|
108 |
if not text:
|
109 |
+
print(f"β οΈ Skipping empty chunk at index {idx}")
|
110 |
continue # Skip empty chunks
|
111 |
|
112 |
+
embedding = self.embed_model.embed_documents([text])[0] # β
Corrected method
|
113 |
documents.append(text)
|
114 |
embeddings.append(embedding)
|
115 |
metadata_list.append({
|
|
|
119 |
})
|
120 |
ids.append(str(idx))
|
121 |
|
122 |
+
if documents:
|
123 |
+
collection.add(
|
124 |
+
ids=ids,
|
125 |
+
embeddings=embeddings,
|
126 |
+
documents=documents,
|
127 |
+
metadatas=metadata_list
|
128 |
+
)
|
129 |
+
print(f"β
Successfully added {len(documents)} chunks to the database.")
|
130 |
|
131 |
processing_time = time.time() - start_time
|
132 |
+
print(f"β
Document processing completed in {processing_time:.2f} seconds")
|
133 |
return collection
|