Enable conditional splitting for html files (#184)

* enable html header

Signed-off-by: XuhuiRen <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Update utils.py

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: XuhuiRen <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

comps/dataprep/milvus/prepare_doc_milvus.py

@@ -8,6 +8,7 @@
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_community.embeddings import HuggingFaceBgeEmbeddings, HuggingFaceEmbeddings, HuggingFaceHubEmbeddings
 from langchain_milvus.vectorstores import Milvus
+from langchain_text_splitters import HTMLHeaderTextSplitter
 
 from comps.cores.mega.micro_service import opea_microservices, register_microservice
 from comps.cores.proto.docarray import DocPath
@@ -34,9 +35,18 @@ def ingest_documents(doc_path: DocPath):
     path = doc_path.path
     print(f"Parsing document {path}.")
 
-    text_splitter = RecursiveCharacterTextSplitter(
-        chunk_size=doc_path.chunk_size, chunk_overlap=doc_path.chunk_size, add_start_index=True
-    )
+    if path.endswith(".html"):
+        headers_to_split_on = [
+            ("h1", "Header 1"),
+            ("h2", "Header 2"),
+            ("h3", "Header 3"),
+        ]
+        text_splitter = HTMLHeaderTextSplitter(headers_to_split_on=headers_to_split_on)
+    else:
+        text_splitter = RecursiveCharacterTextSplitter(
+            chunk_size=doc_path.chunk_size, chunk_overlap=100, add_start_index=True
+        )
+
     content = document_loader(path)
     chunks = text_splitter.split_text(content)
 

comps/dataprep/milvus/requirements.txt

@@ -6,6 +6,7 @@ frontend==0.0.3
 huggingface_hub
 langchain
 langchain-community
+langchain-text-splitters
 langchain_milvus
 numpy
 opentelemetry-api

comps/dataprep/qdrant/prepare_doc_qdrant.py

@@ -7,6 +7,7 @@
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_community.embeddings import HuggingFaceBgeEmbeddings, HuggingFaceEmbeddings, HuggingFaceHubEmbeddings
 from langchain_community.vectorstores import Qdrant
+from langchain_text_splitters import HTMLHeaderTextSplitter
 
 from comps import DocPath, opea_microservices, opea_telemetry, register_microservice
 from comps.dataprep.utils import document_loader
@@ -28,9 +29,18 @@ def ingest_documents(doc_path: DocPath):
     path = doc_path.path
     print(f"Parsing document {path}.")
 
-    text_splitter = RecursiveCharacterTextSplitter(
-        chunk_size=doc_path.chunk_size, chunk_overlap=doc_path.chunk_size, add_start_index=True
-    )
+    if path.endswith(".html"):
+        headers_to_split_on = [
+            ("h1", "Header 1"),
+            ("h2", "Header 2"),
+            ("h3", "Header 3"),
+        ]
+        text_splitter = HTMLHeaderTextSplitter(headers_to_split_on=headers_to_split_on)
+    else:
+        text_splitter = RecursiveCharacterTextSplitter(
+            chunk_size=doc_path.chunk_size, chunk_overlap=100, add_start_index=True
+        )
+
     content = document_loader(path)
     chunks = text_splitter.split_text(content)
 

comps/dataprep/qdrant/requirements.txt

@@ -5,6 +5,7 @@ fastapi
 huggingface_hub
 langchain
 langchain-community
+langchain-text-splitters
 numpy
 opentelemetry-api
 opentelemetry-exporter-otlp

comps/dataprep/redis/langchain/prepare_doc_redis.py

@@ -12,6 +12,7 @@
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_community.embeddings import HuggingFaceBgeEmbeddings, HuggingFaceHubEmbeddings
 from langchain_community.vectorstores import Redis
+from langchain_text_splitters import HTMLHeaderTextSplitter
 from langsmith import traceable
 
 from comps import DocPath, opea_microservices, register_microservice
@@ -36,10 +37,20 @@ def ingest_data_to_redis(doc_path: DocPath):
     path = doc_path.path
     print(f"Parsing document {path}.")
 
-    text_splitter = RecursiveCharacterTextSplitter(
-        chunk_size=doc_path.chunk_size, chunk_overlap=doc_path.chunk_size, add_start_index=True
-    )
+    if path.endswith(".html"):
+        headers_to_split_on = [
+            ("h1", "Header 1"),
+            ("h2", "Header 2"),
+            ("h3", "Header 3"),
+        ]
+        text_splitter = HTMLHeaderTextSplitter(headers_to_split_on=headers_to_split_on)
+    else:
+        text_splitter = RecursiveCharacterTextSplitter(
+            chunk_size=doc_path.chunk_size, chunk_overlap=100, add_start_index=True
+        )
+
     content = document_loader(path)
+
     chunks = text_splitter.split_text(content)
     print("Done preprocessing. Created ", len(chunks), " chunks of the original pdf")
 

comps/dataprep/redis/langchain/requirements.txt

@@ -5,6 +5,7 @@ fastapi
 huggingface_hub
 langchain
 langchain-community
+langchain-text-splitters
 langsmith
 numpy
 opentelemetry-api

comps/dataprep/utils.py

@@ -23,6 +23,7 @@
 from bs4 import BeautifulSoup
 from docx import Document as DDocument
 from langchain_community.document_loaders import (
+    UnstructuredHTMLLoader,
     UnstructuredImageLoader,
     UnstructuredMarkdownLoader,
     UnstructuredPowerPointLoader,
@@ -112,11 +113,11 @@ def load_pdf(pdf_path):
 
 def load_html(html_path):
     """Load the html file."""
-    with open(html_path, "r", encoding="utf-8") as file:
-        html = file.read()
-    soup = BeautifulSoup(html, "html.parser")
-    text = soup.get_text(strip=True)
-    return text
+    data_html = UnstructuredHTMLLoader(html_path).load()
+    content = ""
+    for ins in data_html:
+        content += ins.page_content
+    return content
 
 
 def load_txt(txt_path):