Update ingest.py
Browse files
ingest.py
CHANGED
|
@@ -6,72 +6,67 @@ from langchain.document_loaders import PyPDFLoader
|
|
| 6 |
import shutil
|
| 7 |
import time
|
| 8 |
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
# Obt茅n la ruta completa del directorio actual del script
|
| 20 |
-
script_directory = os.path.dirname(os.path.abspath(__file__))
|
| 21 |
-
md_folder_path = os.path.join(script_directory,
|
| 22 |
-
mdToIngest_path = os.path.join(script_directory,
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
-
for filename in os.listdir(mdToIngest_path):
|
| 26 |
try:
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
with open(file_path, "r", encoding="utf-8") as archivo:
|
| 32 |
-
contenido = archivo.read()
|
| 33 |
-
print(f"Se ley贸 el archivo '{file_path}'.")
|
| 34 |
-
|
| 35 |
-
headersToSplitOn = [("#", "Header"), ("##", "Title")]
|
| 36 |
-
|
| 37 |
-
markdown_splitter = MarkdownHeaderTextSplitter(headers_to_split_on=headersToSplitOn)
|
| 38 |
-
md_header_splits = markdown_splitter.split_text(contenido)
|
| 39 |
-
|
| 40 |
-
for document in md_header_splits:
|
| 41 |
-
lista = []
|
| 42 |
-
|
| 43 |
-
# Extraer y mostrar los metadatos
|
| 44 |
-
metadata = document.metadata
|
| 45 |
-
page_content = document.page_content
|
| 46 |
-
for key, value in metadata.items():
|
| 47 |
-
lista.append(f"{value}{page_content}")
|
| 48 |
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
try:
|
| 52 |
-
shutil.move(file_path, ruta_destino)
|
| 53 |
-
print(f'Archivo movido a {ruta_destino} exitosamente.')
|
| 54 |
-
except shutil.Error as e:
|
| 55 |
-
print(f'Ocurri贸 un error al mover el archivo: {e}')
|
| 56 |
|
|
|
|
| 57 |
|
| 58 |
except Exception as e:
|
| 59 |
-
print(f'Ocurri贸 un error
|
| 60 |
-
print('Pasando al siguiente archivo...')
|
| 61 |
-
continue
|
| 62 |
-
|
| 63 |
-
try:
|
| 64 |
-
time.sleep(5)
|
| 65 |
-
# Eliminar la carpeta y su contenido
|
| 66 |
-
shutil.rmtree(mdToIngest_path)
|
| 67 |
-
|
| 68 |
-
# Crear la carpeta nuevamente
|
| 69 |
-
os.mkdir("mdToIngest")
|
| 70 |
-
|
| 71 |
-
print(f'Carpeta {mdToIngest_path} eliminada y recreada exitosamente.')
|
| 72 |
-
|
| 73 |
-
except Exception as e:
|
| 74 |
-
print(f'Ocurri贸 un error: {e}')
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
|
|
|
| 6 |
import shutil
|
| 7 |
import time
|
| 8 |
|
| 9 |
+
def procesar_archivos(mdToIngest_path="mdToIngest", md_folder_path="md_folder"):
|
| 10 |
+
model_name = "BAAI/bge-large-en"
|
| 11 |
+
model_kwargs = {'device': 'cpu'}
|
| 12 |
+
encode_kwargs = {'normalize_embeddings': False}
|
| 13 |
+
embeddings = HuggingFaceBgeEmbeddings(
|
| 14 |
+
model_name=model_name,
|
| 15 |
+
model_kwargs=model_kwargs,
|
| 16 |
+
encode_kwargs=encode_kwargs
|
| 17 |
+
)
|
| 18 |
+
|
| 19 |
+
# Obt茅n la ruta completa del directorio actual del script
|
| 20 |
+
script_directory = os.path.dirname(os.path.abspath(__file__))
|
| 21 |
+
md_folder_path = os.path.join(script_directory, md_folder_path)
|
| 22 |
+
mdToIngest_path = os.path.join(script_directory, mdToIngest_path)
|
| 23 |
+
|
| 24 |
+
for filename in os.listdir(mdToIngest_path):
|
| 25 |
+
try:
|
| 26 |
+
# Construye la ruta completa del archivo
|
| 27 |
+
file_path = os.path.join(mdToIngest_path, filename)
|
| 28 |
+
ruta_destino = os.path.join(md_folder_path, filename)
|
| 29 |
+
|
| 30 |
+
with open(file_path, "r", encoding="utf-8") as archivo:
|
| 31 |
+
contenido = archivo.read()
|
| 32 |
+
print(f"Se ley贸 el archivo '{file_path}'.")
|
| 33 |
+
|
| 34 |
+
headersToSplitOn = [("#", "Header"), ("##", "Title")]
|
| 35 |
+
|
| 36 |
+
markdown_splitter = MarkdownHeaderTextSplitter(headers_to_split_on=headersToSplitOn)
|
| 37 |
+
md_header_splits = markdown_splitter.split_text(contenido)
|
| 38 |
+
|
| 39 |
+
for document in md_header_splits:
|
| 40 |
+
lista = []
|
| 41 |
+
|
| 42 |
+
# Extraer y mostrar los metadatos
|
| 43 |
+
metadata = document.metadata
|
| 44 |
+
page_content = document.page_content
|
| 45 |
+
for key, value in metadata.items():
|
| 46 |
+
lista.append(f"{value}{page_content}")
|
| 47 |
+
|
| 48 |
+
vector_store = Chroma.from_documents(md_header_splits, embeddings, collection_metadata={"hnsw:space": "cosine"}, persist_directory="stores/ConserGPT")
|
| 49 |
+
|
| 50 |
+
try:
|
| 51 |
+
shutil.move(file_path, ruta_destino)
|
| 52 |
+
print(f'Archivo movido a {ruta_destino} exitosamente.')
|
| 53 |
+
except shutil.Error as e:
|
| 54 |
+
print(f'Ocurri贸 un error al mover el archivo: {e}')
|
| 55 |
+
|
| 56 |
+
except Exception as e:
|
| 57 |
+
print(f'Ocurri贸 un error al leer el archivo: {e}')
|
| 58 |
+
print('Pasando al siguiente archivo...')
|
| 59 |
+
continue
|
| 60 |
|
|
|
|
| 61 |
try:
|
| 62 |
+
time.sleep(5)
|
| 63 |
+
# Eliminar la carpeta y su contenido
|
| 64 |
+
shutil.rmtree(mdToIngest_path)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
|
| 66 |
+
# Crear la carpeta nuevamente
|
| 67 |
+
os.mkdir("mdToIngest")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
|
| 69 |
+
print(f'Carpeta {mdToIngest_path} eliminada y recreada exitosamente.')
|
| 70 |
|
| 71 |
except Exception as e:
|
| 72 |
+
print(f'Ocurri贸 un error: {e}')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|