pixeltable 0.3.14__py3-none-any.whl → 0.5.7__py3-none-any.whl

pixeltable/utils/azure_store.py

@@ -0,0 +1,346 @@
+import datetime
+import logging
+import re
+import threading
+import uuid
+from datetime import timezone
+from pathlib import Path
+from typing import TYPE_CHECKING, Iterator
+from urllib.parse import quote
+
+from azure.core.exceptions import AzureError
+from azure.storage.blob import BlobSasPermissions, generate_blob_sas
+
+from pixeltable import env, exceptions as excs
+from pixeltable.config import Config
+from pixeltable.utils.object_stores import ObjectPath, ObjectStoreBase, StorageObjectAddress
+
+if TYPE_CHECKING:
+    from azure.storage.blob import BlobProperties, BlobServiceClient
+
+    from pixeltable.catalog import Column
+
+
+_logger = logging.getLogger('pixeltable')
+
+
+client_lock = threading.Lock()
+
+
+@env.register_client('azure_blob')
+def _() -> dict[str, 'BlobServiceClient']:
+    return {}
+
+
+class AzureBlobStore(ObjectStoreBase):
+    """Class to handle Azure Blob Storage operations."""
+
+    # TODO: This needs to be redesigned to use asyncio.
+
+    # URI of the Azure Blob Storage container
+    # Always ends with a slash
+    __base_uri: str
+
+    # Storage account name
+    __account_name: str
+
+    # Container name extracted from the URI
+    __container_name: str
+
+    # Prefix path within the container, either empty or ending with a slash
+    __prefix_name: str
+
+    # URI scheme (wasb, wasbs, abfs, abfss, https)
+    __scheme: str
+
+    soa: StorageObjectAddress
+
+    def __init__(self, soa: StorageObjectAddress):
+        self.soa = soa
+        self.__scheme = soa.scheme
+        self.__account_name = soa.account
+        self.__container_name = soa.container
+        self.__prefix_name = soa.prefix
+
+        # Reconstruct base URI in normalized format
+        self.__base_uri = self.soa.prefix_free_uri + self.__prefix_name
+        _logger.info(
+            f'Initialized AzureBlobStore with base URI: {self.__base_uri}, '
+            f'account: {self.__account_name}, container: {self.__container_name}, prefix: {self.__prefix_name}'
+        )
+
+    def client(self) -> 'BlobServiceClient':
+        """Return the Azure Blob Storage client."""
+        client_dict: dict[str, 'BlobServiceClient'] = env.Env.get().get_client('azure_blob')
+        with client_lock:
+            uri = self.soa.container_free_uri
+            if uri not in client_dict:
+                storage_account_name = Config.get().get_string_value('storage_account_name', section='azure')
+                storage_account_key = Config.get().get_string_value('storage_account_key', section='azure')
+                if (storage_account_name is None) != (storage_account_key is None):
+                    raise excs.Error(
+                        "Azure 'storage_account_name' and 'storage_account_key' must be specified together."
+                    )
+                if storage_account_name is None or storage_account_name != self.__account_name:
+                    # Attempt a connection to a public resource, with no account key
+                    client_dict[uri] = self.create_client(endpoint_url=uri)
+                else:
+                    client_dict[uri] = self.create_client(
+                        endpoint_url=uri, account_name=self.__account_name, account_key=storage_account_key
+                    )
+            return client_dict[uri]
+
+    @property
+    def account_name(self) -> str:
+        """Return the storage account name."""
+        return self.__account_name
+
+    @property
+    def container_name(self) -> str:
+        """Return the container name from the base URI."""
+        return self.__container_name
+
+    @property
+    def prefix(self) -> str:
+        """Return the prefix from the base URI."""
+        return self.__prefix_name
+
+    def validate(self, error_col_name: str) -> str | None:
+        """
+        Checks if the URI exists and is accessible.
+
+        Returns:
+            str: The base URI if the container exists and is accessible, None otherwise.
+        """
+        try:
+            container_client = self.client().get_container_client(self.container_name)
+            # Check if container exists by trying to get its properties
+            container_client.get_container_properties()
+            return self.__base_uri
+        except AzureError as e:
+            self.handle_azure_error(e, self.container_name, f'validate container {error_col_name}')
+        return None
+
+    def copy_object_to_local_file(self, src_path: str, dest_path: Path) -> None:
+        """Copies a blob to a local file. Thread safe."""
+        try:
+            blob_client = self.client().get_blob_client(container=self.container_name, blob=self.prefix + src_path)
+            with open(dest_path, 'wb') as download_file:
+                download_stream = blob_client.download_blob()
+                download_file.write(download_stream.readall())
+        except AzureError as e:
+            self.handle_azure_error(e, self.container_name, f'download file {src_path}')
+            raise
+
+    # TODO: utils package should not include back-references to `Column`
+    def copy_local_file(self, col: 'Column', src_path: Path) -> str:
+        """Copy a local file to Azure Blob Storage, and return its new URL"""
+        prefix, filename = ObjectPath.create_prefix_raw(
+            col.get_tbl().id, col.id, col.get_tbl().version, ext=src_path.suffix
+        )
+        blob_name = f'{self.prefix}{prefix}/{filename}'
+        new_file_uri = f'{self.__base_uri}{prefix}/{filename}'
+
+        try:
+            blob_client = self.client().get_blob_client(container=self.container_name, blob=blob_name)
+            with open(src_path, 'rb') as data:
+                blob_client.upload_blob(data, overwrite=True)
+            _logger.debug(f'Media Storage: copied {src_path} to {new_file_uri}')
+            return new_file_uri
+        except AzureError as e:
+            self.handle_azure_error(e, self.container_name, f'upload file {src_path}')
+            raise
+
+    def _get_filtered_blobs(
+        self, tbl_id: uuid.UUID | None, tbl_version: int | None = None
+    ) -> Iterator['BlobProperties']:
+        """Private method to get filtered blobs for a table, optionally filtered by version.
+
+        Args:
+            tbl_id: Table UUID to filter by
+            tbl_version: Optional table version to filter by
+
+        Returns:
+            Iterator over blob objects matching the criteria
+        """
+        # Use ObjectPath to construct the prefix for this table
+        if tbl_id is None:
+            prefix = self.prefix
+            assert tbl_version is None, 'tbl_version must be None if tbl_id is None'
+        else:
+            table_prefix = ObjectPath.table_prefix(tbl_id)
+            prefix = f'{self.prefix}{table_prefix}/'
+
+        try:
+            container_client = self.client().get_container_client(self.container_name)
+
+            blob_iterator: Iterator['BlobProperties']
+            if tbl_version is None:
+                # Return all blobs with the table prefix
+                blob_iterator = container_client.list_blobs(name_starts_with=prefix)
+            else:
+                # Filter by both table_id and table_version using the ObjectPath pattern
+                # Pattern: tbl_id_col_id_version_uuid
+                version_pattern = re.compile(
+                    rf'{re.escape(table_prefix)}_\d+_{re.escape(str(tbl_version))}_[0-9a-fA-F]+.*'
+                )
+                # Get all blobs with the prefix and filter by version pattern
+                all_blobs = container_client.list_blobs(name_starts_with=prefix)
+                blob_iterator = (blob for blob in all_blobs if version_pattern.match(blob.name.split('/')[-1]))
+
+            return blob_iterator
+
+        except AzureError as e:
+            self.handle_azure_error(e, self.container_name, f'setup iterator {self.prefix}')
+            raise
+
+    def count(self, tbl_id: uuid.UUID | None, tbl_version: int | None = None) -> int:
+        """Count the number of files belonging to tbl_id. If tbl_version is not None,
+        count only those files belonging to the specified tbl_version.
+
+        Args:
+            tbl_id: Table UUID to count blobs for
+            tbl_version: Optional table version to filter by
+
+        Returns:
+            Number of blobs matching the criteria
+        """
+        blob_iterator = self._get_filtered_blobs(tbl_id, tbl_version)
+        return sum(1 for _ in blob_iterator)
+
+    def delete(self, tbl_id: uuid.UUID, tbl_version: int | None = None) -> int:
+        """Delete all files belonging to tbl_id. If tbl_version is not None, delete
+        only those files belonging to the specified tbl_version.
+
+        Args:
+            tbl_id: Table UUID to delete blobs for
+            tbl_version: Optional table version to filter by
+
+        Returns:
+            Number of blobs deleted
+        """
+        assert tbl_id is not None
+        blob_iterator = self._get_filtered_blobs(tbl_id, tbl_version)
+        total_deleted = 0
+
+        try:
+            container_client = self.client().get_container_client(self.container_name)
+
+            for blob in blob_iterator:
+                # TODO: Figure out now to properly use batch method delete_blobs(), it doesn't seem to work properly
+                container_client.delete_blob(blob.name)
+                total_deleted += 1
+
+            # print(f"Deleted {total_deleted} blobs from container '{self.container_name}'.")
+            return total_deleted
+
+        except AzureError as e:
+            self.handle_azure_error(e, self.container_name, f'deleting with {self.prefix}')
+            raise
+
+    def list_objects(self, return_uri: bool, n_max: int = 10) -> list[str]:
+        """Return a list of objects found in the specified destination bucket.
+        Each returned object includes the full set of prefixes.
+        if return_uri is True, full URI's are returned; otherwise, just the object keys.
+        """
+        p = self.soa.prefix_free_uri if return_uri else ''
+        r: list[str] = []
+        try:
+            blob_iterator = self._get_filtered_blobs(tbl_id=None, tbl_version=None)
+            for blob in blob_iterator:
+                r.append(f'{p}{blob.name}')
+                if len(r) >= n_max:
+                    return r
+
+        except AzureError as e:
+            self.handle_azure_error(e, self.__container_name, f'list objects from {self.__base_uri}')
+        return r
+
+    @classmethod
+    def handle_azure_error(
+        cls, e: 'AzureError', container_name: str, operation: str = '', *, ignore_404: bool = False
+    ) -> None:
+        from azure.core.exceptions import ClientAuthenticationError, HttpResponseError, ResourceNotFoundError
+
+        if ignore_404 and isinstance(e, ResourceNotFoundError):
+            return
+
+        if isinstance(e, ResourceNotFoundError):
+            raise excs.Error(f'Container {container_name} or blob not found during {operation}: {str(e)!r}')
+        elif isinstance(e, ClientAuthenticationError):
+            raise excs.Error(f'Authentication failed for container {container_name} during {operation}: {str(e)!r}')
+        elif isinstance(e, HttpResponseError):
+            if e.status_code == 403:
+                raise excs.Error(f'Access denied to container {container_name} during {operation}: {str(e)!r}')
+            elif e.status_code == 412:
+                raise excs.Error(f'Precondition failed for container {container_name} during {operation}: {str(e)!r}')
+            else:
+                raise excs.Error(
+                    f'HTTP error during {operation} in container {container_name}: {e.status_code} - {str(e)!r}'
+                )
+        else:
+            raise excs.Error(f'Error during {operation} in container {container_name}: {str(e)!r}')
+
+    def create_presigned_url(self, soa: StorageObjectAddress, expiration_seconds: int) -> str:
+        """Create a presigned URL for downloading an object from Azure Blob Storage."""
+        if not soa.has_object:
+            raise excs.Error(f'StorageObjectAddress does not contain an object name: {soa}')
+
+        azure_client = self.client()
+        account_name = azure_client.account_name if azure_client.account_name else self.__account_name
+
+        # Account key cannot be extracted from client for security reasons, get from config
+        storage_account_key = Config.get().get_string_value('storage_account_key', section='azure')
+
+        if not account_name or not storage_account_key:
+            raise excs.Error(
+                'Azure storage_account_name and storage_account_key must be configured '
+                'to generate presigned URLs. Set them in the config under the [azure] section, '
+                'or include the account name in the Azure URL.'
+            )
+
+        # Use datetime.now(timezone.utc) + timedelta like in pixeltable cloud
+        expiry_time = datetime.datetime.now(timezone.utc) + datetime.timedelta(seconds=expiration_seconds)
+
+        sas_token = generate_blob_sas(
+            account_name=account_name,
+            container_name=soa.container,
+            blob_name=soa.key,
+            account_key=storage_account_key,
+            permission=BlobSasPermissions(read=True),
+            expiry=expiry_time,
+            version='2022-11-02',  # Specify API version to avoid version mismatch issues
+        )
+
+        # Build URL directly - URL encode the blob key to handle special characters
+        # Use safe='/' to preserve path separators in the blob key
+        encoded_key = quote(soa.key, safe='/')
+        blob_url = f'https://{account_name}.blob.core.windows.net/{soa.container}/{encoded_key}?{sas_token}'
+        return blob_url
+
+    @classmethod
+    def create_client(
+        cls, endpoint_url: str, account_name: str | None = None, account_key: str | None = None
+    ) -> 'BlobServiceClient':
+        from azure.core.credentials import AzureNamedKeyCredential
+        from azure.storage.blob import BlobServiceClient  # TODO: Use azure.storage.blob.aio instead
+
+        assert (account_name is None) == (account_key is None)
+        try:
+            #  e.g. endpoint_url: str = f'https://{account_name}.blob.core.windows.net'
+            assert endpoint_url is not None, 'No Azure Storage account information provided'
+
+            # Use empty SAS token for anonymous authentication
+            credential = None
+            if account_name is not None:
+                credential = AzureNamedKeyCredential(name=account_name, key=account_key)
+            return BlobServiceClient(
+                account_url=endpoint_url,
+                credential=credential,
+                max_single_get_size=(32 * 2**20),
+                max_chunk_get_size=(4 * 2**20),
+                connection_timeout=15,
+                read_timeout=30,
+            )
+        except Exception as e:
+            raise excs.Error(f'Failed to create Azure Blob Storage client: {str(e)!r}') from e

pixeltable/utils/coco.py

@@ -50,11 +50,11 @@ def _verify_input_dict(input_dict: dict[str, Any]) -> None:
             raise excs.Error(f'Value for "category" is not a str or int: {annotation}{format_msg}')
 
 
-def write_coco_dataset(df: pxt.DataFrame, dest_path: Path) -> Path:
-    """Export a DataFrame result set as a COCO dataset in dest_path and return the path of the data.json file."""
+def write_coco_dataset(query: pxt.Query, dest_path: Path) -> Path:
+    """Export a ResultSet as a COCO dataset in dest_path and return the path of the data.json file."""
     # TODO: validate schema
-    if len(df._select_list_exprs) != 1 or not df._select_list_exprs[0].col_type.is_json_type():
-        raise excs.Error(f'Expected exactly one json-typed column in select list: {df._select_list_exprs}')
+    if len(query._select_list_exprs) != 1 or not query._select_list_exprs[0].col_type.is_json_type():
+        raise excs.Error(f'Expected exactly one json-typed column in select list: {query._select_list_exprs}')
     input_dict_slot_idx = -1  # df._select_list_exprs[0].slot_idx isn't valid until _exec()
 
     # create output dir
@@ -68,9 +68,9 @@ def write_coco_dataset(df: pxt.DataFrame, dest_path: Path) -> Path:
     annotations: list[dict[str, Any]] = []
     ann_id = -1
     categories: set[Any] = set()
-    for input_row in df._exec():
+    for input_row in query._exec():
         if input_dict_slot_idx == -1:
-            input_dict_expr = df._select_list_exprs[0]
+            input_dict_expr = query._select_list_exprs[0]
             input_dict_slot_idx = input_dict_expr.slot_idx
             input_dict = input_row[input_dict_slot_idx]
             _verify_input_dict(input_dict)

pixeltable/utils/code.py

@@ -1,12 +1,11 @@
 import types
-from typing import Optional
 
 from pixeltable.func import Function
 
 # Utilities related to the organization of the Pixeltable codebase.
 
 
-def local_public_names(mod_name: str, exclude: Optional[list[str]] = None) -> list[str]:
+def local_public_names(mod_name: str, exclude: list[str] | None = None) -> list[str]:
     """
     Returns a list of all functions and submodules that are local to the specified module and are
     publicly accessible. Intended to facilitate implementation of module __dir__() methods for
@@ -21,7 +20,8 @@ def local_public_names(mod_name: str, exclude: Optional[list[str]] = None) -> li
     for obj in mod.__dict__.values():
         if isinstance(obj, Function):
             # Pixeltable function
-            names.append(obj.name)
+            if not obj.name.startswith('_'):
+                names.append(obj.name)
         elif isinstance(obj, types.FunctionType):
             # Python function
             if obj.__module__ == mod.__name__ and not obj.__name__.startswith('_'):

pixeltable/utils/console_output.py

@@ -1,6 +1,8 @@
 import logging
 from typing import TextIO
 
+from pixeltable import exceptions as excs
+
 
 def map_level(verbosity: int) -> int:
     """
@@ -19,7 +21,8 @@ def map_level(verbosity: int) -> int:
         return logging.INFO
     if verbosity == 2:
         return logging.DEBUG
-    return logging.INFO
+
+    raise excs.Error(f'Invalid verbosity level: {verbosity}')
 
 
 class ConsoleOutputHandler(logging.StreamHandler):

pixeltable/utils/coroutine.py

@@ -1,10 +1,10 @@
 import asyncio
 import threading
-from concurrent.futures import ThreadPoolExecutor
 from typing import Any, Coroutine, TypeVar
 
-T = TypeVar('T')
+from pixeltable.env import Env
 
+T = TypeVar('T')
 
 # TODO This is a temporary hack to be able to run async UDFs in contexts that are not properly handled by the existing
 #   scheduler logic (e.g., as an embedding function as part of a similarity lookup). Once the scheduler is fully
@@ -15,27 +15,10 @@ def run_coroutine_synchronously(coroutine: Coroutine[Any, Any, T], timeout: floa
     """
     Runs the given coroutine synchronously, even if called in the context of a running event loop.
     """
-
-    def run_in_new_loop() -> T:
-        new_loop = asyncio.new_event_loop()
-        asyncio.set_event_loop(new_loop)
-        try:
-            return new_loop.run_until_complete(coroutine)
-        finally:
-            new_loop.close()
-
-    try:
-        loop = asyncio.get_running_loop()
-    except RuntimeError:
-        # No event loop; just call `asyncio.run()`
-        return asyncio.run(coroutine)
+    loop = Env.get().event_loop
 
     if threading.current_thread() is threading.main_thread():
-        if not loop.is_running():
-            return loop.run_until_complete(coroutine)
-        else:
-            with ThreadPoolExecutor() as pool:
-                future = pool.submit(run_in_new_loop)
-                return future.result(timeout=timeout)
+        return loop.run_until_complete(coroutine)
     else:
-        return asyncio.run_coroutine_threadsafe(coroutine, loop).result()
+        # Not in main thread, use run_coroutine_threadsafe
+        return asyncio.run_coroutine_threadsafe(coroutine, loop).result(timeout)

pixeltable/utils/dbms.py

@@ -1,6 +1,6 @@
 import abc
 
-from sqlalchemy import URL
+import sqlalchemy as sql
 
 
 class Dbms(abc.ABC):
@@ -11,9 +11,9 @@ class Dbms(abc.ABC):
     name: str
     transaction_isolation_level: str
     version_index_type: str
-    db_url: URL
+    db_url: sql.URL
 
-    def __init__(self, name: str, transaction_isolation_level: str, version_index_type: str, db_url: URL) -> None:
+    def __init__(self, name: str, transaction_isolation_level: str, version_index_type: str, db_url: sql.URL) -> None:
         self.name = name
         self.transaction_isolation_level = transaction_isolation_level
         self.version_index_type = version_index_type
@@ -28,14 +28,19 @@ class Dbms(abc.ABC):
     @abc.abstractmethod
     def default_system_db_url(self) -> str: ...
 
+    @abc.abstractmethod
+    def create_vector_index_stmt(
+        self, store_index_name: str, sa_value_col: sql.Column, metric: str
+    ) -> sql.Compiled: ...
+
 
 class PostgresqlDbms(Dbms):
     """
     Implements utilities to interact with Postgres database.
     """
 
-    def __init__(self, db_url: URL):
-        super().__init__('postgresql', 'REPEATABLE READ', 'brin', db_url)
+    def __init__(self, db_url: sql.URL):
+        super().__init__('postgresql', 'SERIALIZABLE', 'brin', db_url)
 
     def drop_db_stmt(self, database: str) -> str:
         return f'DROP DATABASE {database}'
@@ -47,13 +52,25 @@ class PostgresqlDbms(Dbms):
         a = self.db_url.set(database='postgres').render_as_string(hide_password=False)
         return a
 
+    def create_vector_index_stmt(self, store_index_name: str, sa_value_col: sql.Column, metric: str) -> sql.Compiled:
+        from sqlalchemy.dialects import postgresql
+
+        sa_idx = sql.Index(
+            store_index_name,
+            sa_value_col,
+            postgresql_using='hnsw',
+            postgresql_with={'m': 16, 'ef_construction': 64},
+            postgresql_ops={sa_value_col.name: metric},
+        )
+        return sql.schema.CreateIndex(sa_idx, if_not_exists=True).compile(dialect=postgresql.dialect())
+
 
 class CockroachDbms(Dbms):
     """
     Implements utilities to interact with CockroachDb database.
     """
 
-    def __init__(self, db_url: URL):
+    def __init__(self, db_url: sql.URL):
         super().__init__('cockroachdb', 'SERIALIZABLE', 'btree', db_url)
 
     def drop_db_stmt(self, database: str) -> str:
@@ -64,3 +81,12 @@ class CockroachDbms(Dbms):
 
     def default_system_db_url(self) -> str:
         return self.db_url.set(database='defaultdb').render_as_string(hide_password=False)
+
+    def sa_vector_index(self, store_index_name: str, sa_value_col: sql.schema.Column, metric: str) -> sql.Index | None:
+        return None
+
+    def create_vector_index_stmt(self, store_index_name: str, sa_value_col: sql.Column, metric: str) -> sql.Compiled:
+        return sql.text(
+            f'CREATE VECTOR INDEX IF NOT EXISTS {store_index_name} ON {sa_value_col.table.name}'
+            f'({sa_value_col.name} {metric})'
+        ).compile()

pixeltable/utils/description_helper.py

@@ -1,5 +1,4 @@
 import dataclasses
-from typing import Optional, Union
 
 import pandas as pd
 from pandas.io.formats.style import Styler
@@ -7,11 +6,11 @@ from pandas.io.formats.style import Styler
 
 @dataclasses.dataclass
 class _Descriptor:
-    body: Union[str, pd.DataFrame]
+    body: str | pd.DataFrame
     # The remaining fields only affect the behavior if `body` is a pd.DataFrame.
     show_index: bool
     show_header: bool
-    styler: Optional[Styler] = None
+    styler: Styler | None = None
 
 
 class DescriptionHelper:
@@ -33,10 +32,10 @@ class DescriptionHelper:
 
     def append(
         self,
-        descriptor: Union[str, pd.DataFrame],
+        descriptor: str | pd.DataFrame,
         show_index: bool = False,
         show_header: bool = True,
-        styler: Optional[Styler] = None,
+        styler: Styler | None = None,
     ) -> None:
         self.__descriptors.append(_Descriptor(descriptor, show_index, show_header, styler))
 

pixeltable/utils/documents.py

@@ -1,10 +1,9 @@
 import dataclasses
 import os
-from typing import Optional
 
 import bs4
-import fitz  # type: ignore[import-untyped]
 import puremagic
+from pypdfium2 import PdfDocument  # type: ignore[import-untyped]
 
 from pixeltable import exceptions as excs, type_system as ts
 from pixeltable.env import Env
@@ -13,10 +12,10 @@ from pixeltable.env import Env
 @dataclasses.dataclass
 class DocumentHandle:
     format: ts.DocumentType.DocumentFormat
-    bs_doc: Optional[bs4.BeautifulSoup] = None
-    md_ast: Optional[dict] = None
-    pdf_doc: Optional[fitz.Document] = None
-    txt_doc: Optional[str] = None
+    bs_doc: bs4.BeautifulSoup | None = None
+    md_ast: dict | None = None
+    pdf_doc: PdfDocument | None = None
+    txt_doc: str | None = None
 
 
 def get_document_handle(path: str) -> DocumentHandle:
@@ -34,7 +33,7 @@ def get_document_handle(path: str) -> DocumentHandle:
     raise excs.Error(f'Unrecognized document format: {path}')
 
 
-def get_handle_by_extension(path: str, extension: str) -> Optional[DocumentHandle]:
+def get_handle_by_extension(path: str, extension: str) -> DocumentHandle | None:
     doc_format = ts.DocumentType.DocumentFormat.from_extension(extension)
 
     try:
@@ -43,7 +42,7 @@ def get_handle_by_extension(path: str, extension: str) -> Optional[DocumentHandl
         if doc_format == ts.DocumentType.DocumentFormat.MD:
             return DocumentHandle(doc_format, md_ast=get_markdown_handle(path))
         if doc_format == ts.DocumentType.DocumentFormat.PDF:
-            return DocumentHandle(doc_format, pdf_doc=get_pdf_handle(path))
+            return DocumentHandle(doc_format, pdf_doc=PdfDocument(path))
         if doc_format == ts.DocumentType.DocumentFormat.XML:
             return DocumentHandle(doc_format, bs_doc=get_xml_handle(path))
         if doc_format == ts.DocumentType.DocumentFormat.TXT:
@@ -72,16 +71,6 @@ def get_markdown_handle(path: str) -> dict:
     return md_ast(text)
 
 
-def get_pdf_handle(path: str) -> fitz.Document:
-    doc = fitz.open(path)
-    # check pdf (bc it will work for images)
-    if not doc.is_pdf:
-        raise excs.Error(f'Not a valid PDF document: {path}')
-    # try to read one page
-    next(page for page in doc)
-    return doc
-
-
 def get_xml_handle(path: str) -> bs4.BeautifulSoup:
     with open(path, 'r', encoding='utf8') as fp:
         doc = bs4.BeautifulSoup(fp, 'xml')