test / src /make_db.py
iblfe's picture
Upload folder using huggingface_hub
b585c7f verified
raw
history blame
16.2 kB
import ast
import os
import sys
from typing import Union, List
if os.path.dirname(os.path.abspath(os.path.join(__file__, '..'))) not in sys.path:
sys.path.append(os.path.dirname(os.path.abspath(os.path.join(__file__, '..'))))
from gpt_langchain import path_to_docs, get_some_dbs_from_hf, all_db_zips, some_db_zips, create_or_update_db, \
get_persist_directory, get_existing_db
from utils import H2O_Fire, makedirs, n_gpus_global
def glob_to_db(user_path, chunk=True, chunk_size=512, verbose=False,
fail_any_exception=False, n_jobs=-1, url=None,
# urls
use_unstructured=True,
use_playwright=False,
use_selenium=False,
use_scrapeplaywright=False,
use_scrapehttp=False,
# pdfs
use_pymupdf='auto',
use_unstructured_pdf='auto',
use_pypdf='auto',
enable_pdf_ocr='auto',
try_pdf_as_html='auto',
enable_pdf_doctr='auto',
# images
enable_ocr=False,
enable_doctr=False,
enable_pix2struct=False,
enable_captions=True,
enable_llava=True,
enable_transcriptions=True,
captions_model=None,
caption_loader=None,
doctr_loader=None,
llava_model=None,
llava_prompt=None,
asr_model=None,
asr_loader=None,
# json
jq_schema='.[]',
extract_frames=10,
db_type=None,
selected_file_types=None,
is_public=False):
assert db_type is not None
loaders_and_settings = dict(
# diag/error handling
verbose=verbose, fail_any_exception=fail_any_exception,
# speed
n_jobs=n_jobs,
# chunking
chunk=chunk,
chunk_size=chunk_size,
# urls
use_unstructured=use_unstructured,
use_playwright=use_playwright,
use_selenium=use_selenium,
use_scrapeplaywright=use_scrapeplaywright,
use_scrapehttp=use_scrapehttp,
# pdfs
use_pymupdf=use_pymupdf,
use_unstructured_pdf=use_unstructured_pdf,
use_pypdf=use_pypdf,
enable_pdf_ocr=enable_pdf_ocr,
try_pdf_as_html=try_pdf_as_html,
enable_pdf_doctr=enable_pdf_doctr,
# images
enable_ocr=enable_ocr,
enable_doctr=enable_doctr,
enable_pix2struct=enable_pix2struct,
enable_captions=enable_captions,
enable_llava=enable_llava,
enable_transcriptions=enable_transcriptions,
captions_model=captions_model,
caption_loader=caption_loader,
doctr_loader=doctr_loader,
llava_model=llava_model,
llava_prompt=llava_prompt,
asr_model=asr_model,
asr_loader=asr_loader,
# json
jq_schema=jq_schema,
extract_frames=extract_frames,
db_type=db_type,
is_public=is_public,
)
sources1 = path_to_docs(user_path,
url=url,
**loaders_and_settings,
selected_file_types=selected_file_types,
)
return sources1
def make_db_main(use_openai_embedding: bool = False,
hf_embedding_model: str = None,
migrate_embedding_model=False,
auto_migrate_db=False,
persist_directory: str = None,
user_path: str = 'user_path',
langchain_type: str = 'shared',
url: Union[List[str], str] = None,
add_if_exists: bool = True,
collection_name: str = 'UserData',
verbose: bool = False,
chunk: bool = True,
chunk_size: int = 512,
fail_any_exception: bool = False,
download_all: bool = False,
download_some: bool = False,
download_one: str = None,
download_dest: str = None,
n_jobs: int = -1,
# urls
use_unstructured=True,
use_playwright=False,
use_selenium=False,
use_scrapeplaywright=False,
use_scrapehttp=False,
# pdfs
use_pymupdf='auto',
use_unstructured_pdf='auto',
use_pypdf='auto',
enable_pdf_ocr='auto',
enable_pdf_doctr='auto',
try_pdf_as_html='auto',
# images
enable_ocr=False,
enable_doctr=False,
enable_pix2struct=False,
enable_captions=True,
enable_llava=True,
captions_model: str = "Salesforce/blip-image-captioning-base",
llava_model: str = None,
llava_prompt: str = None,
pre_load_image_audio_models: bool = False,
caption_gpu: bool = True,
# caption_loader=None, # set internally
# doctr_loader=None, # set internally
# asr_loader=None # set internally
enable_transcriptions: bool = True,
asr_model: str = "openai/whisper-medium",
asr_gpu: bool = True,
# json
jq_schema='.[]',
extract_frames=10,
db_type: str = 'chroma',
selected_file_types: Union[List[str], str] = None,
fail_if_no_sources: bool = True
):
"""
# To make UserData db for generate.py, put pdfs, etc. into path user_path and run:
python src/make_db.py
# once db is made, can use in generate.py like:
python generate.py --base_model=h2oai/h2ogpt-oig-oasst1-512-6_9b --langchain_mode=UserData
or zip-up the db_dir_UserData and share:
zip -r db_dir_UserData.zip db_dir_UserData
# To get all db files (except large wiki_full) do:
python src/make_db.py --download_some=True
# To get a single db file from HF:
python src/make_db.py --download_one=db_dir_DriverlessAI_docs.zip
:param use_openai_embedding: Whether to use OpenAI embedding
:param hf_embedding_model: HF embedding model to use. Like generate.py, uses 'hkunlp/instructor-large' if have GPUs, else "sentence-transformers/all-MiniLM-L6-v2"
:param migrate_embedding_model: whether to migrate to newly chosen hf_embedding_model or stick with one in db
:param auto_migrate_db: whether to migrate database for chroma<0.4 -> >0.4
:param persist_directory: where to persist db (note generate.py always uses db_dir_<collection name>
If making personal database for user, set persistent_directory to users/<username>/db_dir_<collection name>
and pass --langchain_type=personal
:param user_path: where to pull documents from (None means url is not None. If url is not None, this is ignored.)
:param langchain_type: type of database, i.e.. 'shared' or 'personal'
:param url: url (or urls) to generate documents from (None means user_path is not None)
:param add_if_exists: Add to db if already exists, but will not add duplicate sources
:param collection_name: Collection name for new db if not adding
Normally same as langchain_mode
:param verbose: whether to show verbose messages
:param chunk: whether to chunk data
:param chunk_size: chunk size for chunking
:param fail_any_exception: whether to fail if any exception hit during ingestion of files
:param download_all: whether to download all (including 23GB Wikipedia) example databases from h2o.ai HF
:param download_some: whether to download some small example databases from h2o.ai HF
:param download_one: whether to download one chosen example databases from h2o.ai HF
:param download_dest: Destination for downloads
:param n_jobs: Number of cores to use for ingesting multiple files
:param use_unstructured: see gen.py
:param use_playwright: see gen.py
:param use_selenium: see gen.py
:param use_scrapeplaywright: see gen.py
:param use_scrapehttp: see gen.py
:param use_pymupdf: see gen.py
:param use_unstructured_pdf: see gen.py
:param use_pypdf: see gen.py
:param enable_pdf_ocr: see gen.py
:param try_pdf_as_html: see gen.py
:param enable_pdf_doctr: see gen.py
:param enable_ocr: see gen.py
:param enable_doctr: see gen.py
:param enable_pix2struct: see gen.py
:param enable_captions: Whether to enable captions on images
:param enable_llava: See gen.py
:param captions_model: See gen.py
:param llava_model: See gen.py
:param llava_prompt: See gen.py
:param pre_load_image_audio_models: See generate.py
:param caption_gpu: Caption images on GPU if present
:param db_type: 'faiss' for in-memory
'chroma' (for chroma >= 0.4)
'chroma_old' (for chroma < 0.4) -- recommended for large collections
'weaviate' for persisted on disk
:param selected_file_types: File types (by extension) to include if passing user_path
For a list of possible values, see:
https://github.com/h2oai/h2ogpt/blob/main/docs/README_LangChain.md#shoosing-document-types
e.g. --selected_file_types="['pdf', 'html', 'htm']"
:return: None
"""
db = None
if isinstance(selected_file_types, str):
selected_file_types = ast.literal_eval(selected_file_types)
if persist_directory is None:
persist_directory, langchain_type = get_persist_directory(collection_name, langchain_type=langchain_type)
if download_dest is None:
download_dest = makedirs('./', use_base=True)
# match behavior of main() in generate.py for non-HF case
n_gpus = n_gpus_global
if n_gpus == 0:
if hf_embedding_model is None:
# if no GPUs, use simpler embedding model to avoid cost in time
hf_embedding_model = "sentence-transformers/all-MiniLM-L6-v2"
else:
if hf_embedding_model is None:
# if still None, then set default
hf_embedding_model = 'hkunlp/instructor-large'
existing_db = False
if download_all:
print("Downloading all (and unzipping): %s" % all_db_zips, flush=True)
get_some_dbs_from_hf(download_dest, db_zips=all_db_zips)
if verbose:
print("DONE", flush=True)
existing_db = True
elif download_some:
print("Downloading some (and unzipping): %s" % some_db_zips, flush=True)
get_some_dbs_from_hf(download_dest, db_zips=some_db_zips)
if verbose:
print("DONE", flush=True)
existing_db = True
elif download_one:
print("Downloading %s (and unzipping)" % download_one, flush=True)
get_some_dbs_from_hf(download_dest, db_zips=[[download_one, '', 'Unknown License']])
if verbose:
print("DONE", flush=True)
existing_db = True
if existing_db:
load_db_if_exists = True
langchain_mode = collection_name
langchain_mode_paths = dict(langchain_mode=None)
langchain_mode_types = dict(langchain_mode='shared')
db, use_openai_embedding, hf_embedding_model = \
get_existing_db(None, persist_directory, load_db_if_exists, db_type,
use_openai_embedding,
langchain_mode, langchain_mode_paths, langchain_mode_types,
hf_embedding_model, migrate_embedding_model, auto_migrate_db,
verbose=False,
n_jobs=n_jobs)
return db, collection_name
if enable_captions and pre_load_image_audio_models:
# preload, else can be too slow or if on GPU have cuda context issues
# Inside ingestion, this will disable parallel loading of multiple other kinds of docs
# However, if have many images, all those images will be handled more quickly by preloaded model on GPU
from image_captions import H2OImageCaptionLoader
caption_loader = H2OImageCaptionLoader(None,
blip_model=captions_model,
blip_processor=captions_model,
caption_gpu=caption_gpu,
).load_model()
else:
if enable_captions:
caption_loader = 'gpu' if n_gpus > 0 and caption_gpu else 'cpu'
else:
caption_loader = False
if enable_doctr or enable_pdf_ocr in [True, 'auto', 'on']:
doctr_loader = 'gpu' if n_gpus > 0 and caption_gpu else 'cpu'
else:
doctr_loader = False
if enable_transcriptions:
asr_loader = 'gpu' if n_gpus > 0 and asr_gpu else 'cpu'
else:
asr_loader = False
if verbose:
print("Getting sources", flush=True)
assert user_path is not None or url is not None, "Can't have both user_path and url as None"
if not url:
assert os.path.isdir(user_path), "user_path=%s does not exist" % user_path
sources = glob_to_db(user_path, chunk=chunk, chunk_size=chunk_size, verbose=verbose,
fail_any_exception=fail_any_exception, n_jobs=n_jobs, url=url,
# urls
use_unstructured=use_unstructured,
use_playwright=use_playwright,
use_selenium=use_selenium,
use_scrapeplaywright=use_scrapeplaywright,
use_scrapehttp=use_scrapehttp,
# pdfs
use_pymupdf=use_pymupdf,
use_unstructured_pdf=use_unstructured_pdf,
use_pypdf=use_pypdf,
enable_pdf_ocr=enable_pdf_ocr,
try_pdf_as_html=try_pdf_as_html,
enable_pdf_doctr=enable_pdf_doctr,
# images
enable_ocr=enable_ocr,
enable_doctr=enable_doctr,
enable_pix2struct=enable_pix2struct,
enable_captions=enable_captions,
enable_llava=enable_llava,
enable_transcriptions=enable_transcriptions,
captions_model=captions_model,
caption_loader=caption_loader,
doctr_loader=doctr_loader,
llava_model=llava_model,
llava_prompt=llava_prompt,
# Note: we don't reload doctr model
asr_loader=asr_loader,
asr_model=asr_model,
# json
jq_schema=jq_schema,
extract_frames=extract_frames,
db_type=db_type,
selected_file_types=selected_file_types,
is_public=False,
)
exceptions = [x for x in sources if x.metadata.get('exception')]
print("Exceptions: %s/%s %s" % (len(exceptions), len(sources), exceptions), flush=True)
sources = [x for x in sources if 'exception' not in x.metadata]
assert len(sources) > 0 or not fail_if_no_sources, "No sources found"
db = create_or_update_db(db_type, persist_directory,
collection_name, user_path, langchain_type,
sources, use_openai_embedding, add_if_exists, verbose,
hf_embedding_model, migrate_embedding_model, auto_migrate_db,
n_jobs=n_jobs)
assert db is not None or not fail_if_no_sources
if verbose:
print("DONE", flush=True)
return db, collection_name
if __name__ == "__main__":
H2O_Fire(make_db_main)