2023-06-06 09:03:13 -07:00
|
|
|
import os
|
|
|
|
import pathlib
|
|
|
|
from dataclasses import dataclass
|
2023-06-07 21:22:18 -07:00
|
|
|
from typing import Any, Dict
|
2023-06-06 09:03:13 -07:00
|
|
|
|
|
|
|
import pytest
|
|
|
|
|
2023-06-07 21:22:18 -07:00
|
|
|
from unstructured.documents.elements import DataSourceMetadata
|
2023-06-06 09:03:13 -07:00
|
|
|
from unstructured.ingest.interfaces import (
|
|
|
|
BaseConnectorConfig,
|
2023-11-27 14:25:30 -05:00
|
|
|
BaseSingleIngestDoc,
|
2023-09-11 11:40:56 -04:00
|
|
|
PartitionConfig,
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
ProcessorConfig,
|
2023-09-11 11:40:56 -04:00
|
|
|
ReadConfig,
|
2023-06-06 09:03:13 -07:00
|
|
|
)
|
|
|
|
from unstructured.partition.auto import partition
|
|
|
|
from unstructured.staging.base import convert_to_dict
|
|
|
|
|
|
|
|
DIRECTORY = pathlib.Path(__file__).parent.resolve()
|
|
|
|
EXAMPLE_DOCS_DIRECTORY = os.path.join(DIRECTORY, "../..", "example-docs")
|
2023-06-20 11:19:55 -05:00
|
|
|
TEST_DOWNLOAD_DIR = "/tmp"
|
|
|
|
TEST_OUTPUT_DIR = "/tmp"
|
|
|
|
TEST_ID = "test"
|
|
|
|
TEST_FILE_PATH = os.path.join(EXAMPLE_DOCS_DIRECTORY, "book-war-and-peace-1p.txt")
|
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
|
|
|
|
@dataclass
|
2023-12-11 12:37:49 -05:00
|
|
|
class ExampleConfig(BaseConnectorConfig):
|
2023-06-06 09:03:13 -07:00
|
|
|
id: str
|
|
|
|
path: str
|
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-12-11 12:37:49 -05:00
|
|
|
TEST_CONFIG = ExampleConfig(id=TEST_ID, path=TEST_FILE_PATH)
|
2023-06-07 21:22:18 -07:00
|
|
|
TEST_SOURCE_URL = "test-source-url"
|
|
|
|
TEST_VERSION = "1.1.1"
|
|
|
|
TEST_RECORD_LOCATOR = {"id": "data-source-id"}
|
|
|
|
TEST_DATE_CREATED = "2021-01-01T00:00:00"
|
|
|
|
TEST_DATE_MODIFIED = "2021-01-02T00:00:00"
|
|
|
|
TEST_DATE_PROCESSSED = "2022-12-13T15:44:08"
|
2023-06-06 09:03:13 -07:00
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
@dataclass
|
2023-12-11 12:37:49 -05:00
|
|
|
class ExampleIngestDoc(BaseSingleIngestDoc):
|
|
|
|
connector_config: ExampleConfig
|
2023-06-06 09:03:13 -07:00
|
|
|
|
|
|
|
@property
|
|
|
|
def filename(self):
|
2023-06-07 21:22:18 -07:00
|
|
|
return TEST_FILE_PATH
|
2023-06-15 12:21:17 -04:00
|
|
|
|
2023-06-26 13:12:05 -07:00
|
|
|
@property
|
|
|
|
def _output_filename(self):
|
|
|
|
return TEST_FILE_PATH + ".json"
|
2023-07-05 11:25:11 -07:00
|
|
|
|
2023-06-07 21:22:18 -07:00
|
|
|
@property
|
|
|
|
def source_url(self) -> str:
|
|
|
|
return TEST_SOURCE_URL
|
2023-06-15 12:21:17 -04:00
|
|
|
|
|
|
|
@property
|
2023-06-07 21:22:18 -07:00
|
|
|
def version(self) -> str:
|
|
|
|
return TEST_VERSION
|
2023-06-15 12:21:17 -04:00
|
|
|
|
|
|
|
@property
|
2023-06-07 21:22:18 -07:00
|
|
|
def record_locator(self) -> Dict[str, Any]:
|
|
|
|
return TEST_RECORD_LOCATOR
|
|
|
|
|
2023-06-15 12:21:17 -04:00
|
|
|
@property
|
2023-06-07 21:22:18 -07:00
|
|
|
def date_created(self) -> str:
|
|
|
|
return TEST_DATE_CREATED
|
2023-06-15 12:21:17 -04:00
|
|
|
|
|
|
|
@property
|
2023-06-07 21:22:18 -07:00
|
|
|
def date_modified(self) -> str:
|
|
|
|
return TEST_DATE_MODIFIED
|
2023-06-15 12:21:17 -04:00
|
|
|
|
|
|
|
@property
|
2023-06-07 21:22:18 -07:00
|
|
|
def exists(self) -> bool:
|
|
|
|
return True
|
2023-06-15 12:21:17 -04:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
def cleanup_file(self):
|
|
|
|
pass
|
|
|
|
|
|
|
|
def get_file(self):
|
|
|
|
pass
|
|
|
|
|
|
|
|
def has_output(self):
|
|
|
|
return True
|
2023-06-15 12:21:17 -04:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
def write_result(self, result):
|
|
|
|
pass
|
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
@pytest.fixture()
|
|
|
|
def partition_test_results():
|
|
|
|
# Reusable partition test results, calculated only once
|
2023-06-07 21:22:18 -07:00
|
|
|
result = partition(
|
|
|
|
filename=str(TEST_FILE_PATH),
|
|
|
|
data_source_metadata=DataSourceMetadata(
|
|
|
|
url=TEST_SOURCE_URL,
|
|
|
|
version=TEST_VERSION,
|
|
|
|
record_locator=TEST_RECORD_LOCATOR,
|
|
|
|
date_created=TEST_DATE_CREATED,
|
|
|
|
date_modified=TEST_DATE_MODIFIED,
|
|
|
|
date_processed=TEST_DATE_PROCESSSED,
|
|
|
|
),
|
|
|
|
)
|
|
|
|
return result
|
2023-06-06 09:03:13 -07:00
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
@pytest.fixture()
|
|
|
|
def partition_file_test_results(partition_test_results):
|
|
|
|
# Reusable partition_file test results, calculated only once
|
|
|
|
return convert_to_dict(partition_test_results)
|
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-07 21:22:18 -07:00
|
|
|
def test_partition_file():
|
|
|
|
"""Validate partition_file returns a list of dictionaries with the expected keys,
|
|
|
|
metadatakeys, and data source metadata values."""
|
2023-12-11 12:37:49 -05:00
|
|
|
test_ingest_doc = ExampleIngestDoc(
|
2023-09-11 11:40:56 -04:00
|
|
|
connector_config=TEST_CONFIG,
|
|
|
|
read_config=ReadConfig(download_dir=TEST_DOWNLOAD_DIR),
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
processor_config=ProcessorConfig(output_dir=TEST_OUTPUT_DIR),
|
2023-06-07 21:22:18 -07:00
|
|
|
)
|
|
|
|
test_ingest_doc._date_processed = TEST_DATE_PROCESSSED
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
isd_elems_raw = test_ingest_doc.partition_file(partition_config=PartitionConfig())
|
|
|
|
isd_elems = convert_to_dict(isd_elems_raw)
|
2023-06-07 21:22:18 -07:00
|
|
|
assert len(isd_elems)
|
2023-06-20 11:19:55 -05:00
|
|
|
expected_keys = {
|
|
|
|
"element_id",
|
|
|
|
"text",
|
|
|
|
"type",
|
|
|
|
"metadata",
|
|
|
|
}
|
2023-07-05 11:25:11 -07:00
|
|
|
# The document in TEST_FILE_PATH does not have elements with coordinates so
|
|
|
|
# partition is not expected to return coordinates metadata.
|
2023-09-11 11:40:56 -04:00
|
|
|
expected_metadata_keys = {
|
|
|
|
"data_source",
|
|
|
|
"filename",
|
|
|
|
"file_directory",
|
|
|
|
"filetype",
|
2023-09-26 14:09:27 -04:00
|
|
|
"languages",
|
2023-09-11 11:40:56 -04:00
|
|
|
"last_modified",
|
|
|
|
}
|
2023-06-07 21:22:18 -07:00
|
|
|
for elem in isd_elems:
|
Feat: Create a naive hierarchy for elements (#1268)
## **Summary**
By adding hierarchy to unstructured elements, users will have more
information for implementing vector db/LLM chunking strategies. For
example, text elements could be queried by their preceding title
element. The hierarchy is implemented by a parent_id tag in the
element's metadata.
### Features
- Introduces a parent_id to ElementMetadata (The id of the parent
element, not a pointer)
- Creates a rule set for assigning hierarchies. Sensible default is
assigned, with an optional override parameter
- Sets element parent ids if there isn't an existing parent id or
matches the ruleset
### How it works
Hierarchies are assigned via a parent id field in element metadata.
Elements are read sequentially and evaluated against a ruleset. For
example take the following elements:
1. Title, "This is the Title"
2. Text, "this is the text"
And the ruleset: `{"title": ["text"]}`. When evaluated, the parent_id of
2 will be the id of 1. The algorithm for determining this is more
complex and resolves several edge cases, so please read the code for
further details.
### Schema Changes
```
@dataclass
class ElementMetadata:
coordinates: Optional[CoordinatesMetadata] = None
data_source: Optional[DataSourceMetadata] = None
filename: Optional[str] = None
file_directory: Optional[str] = None
last_modified: Optional[str] = None
filetype: Optional[str] = None
attached_to_filename: Optional[str] = None
+ parent_id: Optional[Union[str, uuid.UUID, NoID, UUID]] = None
+ category_depth: Optional[int] = None
...
```
### Testing
```
from unstructured.partition.auto import partition
from typing import List
elements = partition(filename="./unstructured/example-docs/fake-html.html", strategy="auto")
for element in elements:
print(
f"Category: {getattr(element, 'category', '')}\n"\
f"Text: {getattr(element, 'text', '')}\n"
f"ID: {element.id}\n" \
f"Parent ID: {element.metadata.parent_id}\n"\
f"Depth: {element.metadata.category_depth}\n" \
)
```
### Additional Notes
Implementing this feature revealed a possibly undesired side-effect in
how element metadata are processed. In
`unstructured/partition/common.py` the `_add_element_metadata` is
invoked as part of the `add_metadata_with_filetype` decorator for
filetype partitioning. This method is intended to add additional
information to the metadata generated with the element including
filename and filetype, however the existing metadata is merged into a
newly created metadata object rather than the other way around. Because
of the way it's structured, new metadata fields can easily be forgotten
and pose debugging challenges to developers. This likely warrants a new
issue.
I'm guessing that the implementation is done this way to avoid issues
with deserializing elements, but could be wrong.
---------
Co-authored-by: Benjamin Torres <benjats07@users.noreply.github.com>
2023-09-14 11:23:16 -04:00
|
|
|
# Parent IDs are non-deterministic - remove them from the test
|
|
|
|
elem["metadata"].pop("parent_id", None)
|
|
|
|
|
2023-06-07 21:22:18 -07:00
|
|
|
assert expected_keys == set(elem.keys())
|
|
|
|
assert expected_metadata_keys == set(elem["metadata"].keys())
|
|
|
|
data_source_metadata = elem["metadata"]["data_source"]
|
|
|
|
assert data_source_metadata["url"] == TEST_SOURCE_URL
|
|
|
|
assert data_source_metadata["version"] == TEST_VERSION
|
|
|
|
assert data_source_metadata["record_locator"] == TEST_RECORD_LOCATOR
|
|
|
|
assert data_source_metadata["date_created"] == TEST_DATE_CREATED
|
|
|
|
assert data_source_metadata["date_modified"] == TEST_DATE_MODIFIED
|
|
|
|
assert data_source_metadata["date_processed"] == TEST_DATE_PROCESSSED
|
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
def test_process_file_fields_include_default(mocker, partition_test_results):
|
|
|
|
"""Validate when metadata_include and metadata_exclude are not set, all fields:
|
|
|
|
("element_id", "text", "type", "metadata") are included"""
|
2023-06-07 21:22:18 -07:00
|
|
|
mock_partition = mocker.patch(
|
2023-06-06 09:03:13 -07:00
|
|
|
"unstructured.ingest.interfaces.partition",
|
|
|
|
return_value=partition_test_results,
|
|
|
|
)
|
2023-12-11 12:37:49 -05:00
|
|
|
test_ingest_doc = ExampleIngestDoc(
|
2023-09-11 11:40:56 -04:00
|
|
|
connector_config=TEST_CONFIG,
|
|
|
|
read_config=ReadConfig(download_dir=TEST_DOWNLOAD_DIR),
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
processor_config=ProcessorConfig(output_dir=TEST_OUTPUT_DIR),
|
2023-06-06 09:03:13 -07:00
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
isd_elems_raw = test_ingest_doc.partition_file(partition_config=PartitionConfig())
|
|
|
|
isd_elems = convert_to_dict(isd_elems_raw)
|
2023-06-06 09:03:13 -07:00
|
|
|
assert len(isd_elems)
|
2023-06-07 21:22:18 -07:00
|
|
|
assert mock_partition.call_count == 1
|
2023-06-06 09:03:13 -07:00
|
|
|
for elem in isd_elems:
|
Feat: Create a naive hierarchy for elements (#1268)
## **Summary**
By adding hierarchy to unstructured elements, users will have more
information for implementing vector db/LLM chunking strategies. For
example, text elements could be queried by their preceding title
element. The hierarchy is implemented by a parent_id tag in the
element's metadata.
### Features
- Introduces a parent_id to ElementMetadata (The id of the parent
element, not a pointer)
- Creates a rule set for assigning hierarchies. Sensible default is
assigned, with an optional override parameter
- Sets element parent ids if there isn't an existing parent id or
matches the ruleset
### How it works
Hierarchies are assigned via a parent id field in element metadata.
Elements are read sequentially and evaluated against a ruleset. For
example take the following elements:
1. Title, "This is the Title"
2. Text, "this is the text"
And the ruleset: `{"title": ["text"]}`. When evaluated, the parent_id of
2 will be the id of 1. The algorithm for determining this is more
complex and resolves several edge cases, so please read the code for
further details.
### Schema Changes
```
@dataclass
class ElementMetadata:
coordinates: Optional[CoordinatesMetadata] = None
data_source: Optional[DataSourceMetadata] = None
filename: Optional[str] = None
file_directory: Optional[str] = None
last_modified: Optional[str] = None
filetype: Optional[str] = None
attached_to_filename: Optional[str] = None
+ parent_id: Optional[Union[str, uuid.UUID, NoID, UUID]] = None
+ category_depth: Optional[int] = None
...
```
### Testing
```
from unstructured.partition.auto import partition
from typing import List
elements = partition(filename="./unstructured/example-docs/fake-html.html", strategy="auto")
for element in elements:
print(
f"Category: {getattr(element, 'category', '')}\n"\
f"Text: {getattr(element, 'text', '')}\n"
f"ID: {element.id}\n" \
f"Parent ID: {element.metadata.parent_id}\n"\
f"Depth: {element.metadata.category_depth}\n" \
)
```
### Additional Notes
Implementing this feature revealed a possibly undesired side-effect in
how element metadata are processed. In
`unstructured/partition/common.py` the `_add_element_metadata` is
invoked as part of the `add_metadata_with_filetype` decorator for
filetype partitioning. This method is intended to add additional
information to the metadata generated with the element including
filename and filetype, however the existing metadata is merged into a
newly created metadata object rather than the other way around. Because
of the way it's structured, new metadata fields can easily be forgotten
and pose debugging challenges to developers. This likely warrants a new
issue.
I'm guessing that the implementation is done this way to avoid issues
with deserializing elements, but could be wrong.
---------
Co-authored-by: Benjamin Torres <benjats07@users.noreply.github.com>
2023-09-14 11:23:16 -04:00
|
|
|
# Parent IDs are non-deterministic - remove them from the test
|
|
|
|
elem["metadata"].pop("parent_id", None)
|
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
assert {"element_id", "text", "type", "metadata"} == set(elem.keys())
|
2023-06-07 21:22:18 -07:00
|
|
|
data_source_metadata = elem["metadata"]["data_source"]
|
|
|
|
assert data_source_metadata["url"] == TEST_SOURCE_URL
|
|
|
|
assert data_source_metadata["version"] == TEST_VERSION
|
|
|
|
assert data_source_metadata["record_locator"] == TEST_RECORD_LOCATOR
|
|
|
|
assert data_source_metadata["date_created"] == TEST_DATE_CREATED
|
|
|
|
assert data_source_metadata["date_modified"] == TEST_DATE_MODIFIED
|
|
|
|
assert data_source_metadata["date_processed"] == TEST_DATE_PROCESSSED
|
2023-06-06 09:03:13 -07:00
|
|
|
|
|
|
|
|
Feat: Create a naive hierarchy for elements (#1268)
## **Summary**
By adding hierarchy to unstructured elements, users will have more
information for implementing vector db/LLM chunking strategies. For
example, text elements could be queried by their preceding title
element. The hierarchy is implemented by a parent_id tag in the
element's metadata.
### Features
- Introduces a parent_id to ElementMetadata (The id of the parent
element, not a pointer)
- Creates a rule set for assigning hierarchies. Sensible default is
assigned, with an optional override parameter
- Sets element parent ids if there isn't an existing parent id or
matches the ruleset
### How it works
Hierarchies are assigned via a parent id field in element metadata.
Elements are read sequentially and evaluated against a ruleset. For
example take the following elements:
1. Title, "This is the Title"
2. Text, "this is the text"
And the ruleset: `{"title": ["text"]}`. When evaluated, the parent_id of
2 will be the id of 1. The algorithm for determining this is more
complex and resolves several edge cases, so please read the code for
further details.
### Schema Changes
```
@dataclass
class ElementMetadata:
coordinates: Optional[CoordinatesMetadata] = None
data_source: Optional[DataSourceMetadata] = None
filename: Optional[str] = None
file_directory: Optional[str] = None
last_modified: Optional[str] = None
filetype: Optional[str] = None
attached_to_filename: Optional[str] = None
+ parent_id: Optional[Union[str, uuid.UUID, NoID, UUID]] = None
+ category_depth: Optional[int] = None
...
```
### Testing
```
from unstructured.partition.auto import partition
from typing import List
elements = partition(filename="./unstructured/example-docs/fake-html.html", strategy="auto")
for element in elements:
print(
f"Category: {getattr(element, 'category', '')}\n"\
f"Text: {getattr(element, 'text', '')}\n"
f"ID: {element.id}\n" \
f"Parent ID: {element.metadata.parent_id}\n"\
f"Depth: {element.metadata.category_depth}\n" \
)
```
### Additional Notes
Implementing this feature revealed a possibly undesired side-effect in
how element metadata are processed. In
`unstructured/partition/common.py` the `_add_element_metadata` is
invoked as part of the `add_metadata_with_filetype` decorator for
filetype partitioning. This method is intended to add additional
information to the metadata generated with the element including
filename and filetype, however the existing metadata is merged into a
newly created metadata object rather than the other way around. Because
of the way it's structured, new metadata fields can easily be forgotten
and pose debugging challenges to developers. This likely warrants a new
issue.
I'm guessing that the implementation is done this way to avoid issues
with deserializing elements, but could be wrong.
---------
Co-authored-by: Benjamin Torres <benjats07@users.noreply.github.com>
2023-09-14 11:23:16 -04:00
|
|
|
def test_process_file_metadata_includes_filename_and_filetype(
|
|
|
|
mocker,
|
|
|
|
partition_test_results,
|
|
|
|
):
|
2023-06-15 12:21:17 -04:00
|
|
|
"""Validate when metadata_include is set to "filename,filetype",
|
2023-06-06 09:03:13 -07:00
|
|
|
only filename is included in metadata"""
|
|
|
|
mocker.patch(
|
|
|
|
"unstructured.ingest.interfaces.partition",
|
|
|
|
return_value=partition_test_results,
|
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
partition_config = PartitionConfig(
|
|
|
|
metadata_include=["filename", "filetype"],
|
|
|
|
)
|
2023-12-11 12:37:49 -05:00
|
|
|
test_ingest_doc = ExampleIngestDoc(
|
2023-09-11 11:40:56 -04:00
|
|
|
connector_config=TEST_CONFIG,
|
|
|
|
read_config=ReadConfig(download_dir=TEST_DOWNLOAD_DIR),
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
processor_config=ProcessorConfig(output_dir=TEST_OUTPUT_DIR),
|
2023-06-06 09:03:13 -07:00
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
isd_elems = test_ingest_doc.process_file(partition_config=partition_config)
|
2023-06-06 09:03:13 -07:00
|
|
|
assert len(isd_elems)
|
|
|
|
for elem in isd_elems:
|
Feat: Create a naive hierarchy for elements (#1268)
## **Summary**
By adding hierarchy to unstructured elements, users will have more
information for implementing vector db/LLM chunking strategies. For
example, text elements could be queried by their preceding title
element. The hierarchy is implemented by a parent_id tag in the
element's metadata.
### Features
- Introduces a parent_id to ElementMetadata (The id of the parent
element, not a pointer)
- Creates a rule set for assigning hierarchies. Sensible default is
assigned, with an optional override parameter
- Sets element parent ids if there isn't an existing parent id or
matches the ruleset
### How it works
Hierarchies are assigned via a parent id field in element metadata.
Elements are read sequentially and evaluated against a ruleset. For
example take the following elements:
1. Title, "This is the Title"
2. Text, "this is the text"
And the ruleset: `{"title": ["text"]}`. When evaluated, the parent_id of
2 will be the id of 1. The algorithm for determining this is more
complex and resolves several edge cases, so please read the code for
further details.
### Schema Changes
```
@dataclass
class ElementMetadata:
coordinates: Optional[CoordinatesMetadata] = None
data_source: Optional[DataSourceMetadata] = None
filename: Optional[str] = None
file_directory: Optional[str] = None
last_modified: Optional[str] = None
filetype: Optional[str] = None
attached_to_filename: Optional[str] = None
+ parent_id: Optional[Union[str, uuid.UUID, NoID, UUID]] = None
+ category_depth: Optional[int] = None
...
```
### Testing
```
from unstructured.partition.auto import partition
from typing import List
elements = partition(filename="./unstructured/example-docs/fake-html.html", strategy="auto")
for element in elements:
print(
f"Category: {getattr(element, 'category', '')}\n"\
f"Text: {getattr(element, 'text', '')}\n"
f"ID: {element.id}\n" \
f"Parent ID: {element.metadata.parent_id}\n"\
f"Depth: {element.metadata.category_depth}\n" \
)
```
### Additional Notes
Implementing this feature revealed a possibly undesired side-effect in
how element metadata are processed. In
`unstructured/partition/common.py` the `_add_element_metadata` is
invoked as part of the `add_metadata_with_filetype` decorator for
filetype partitioning. This method is intended to add additional
information to the metadata generated with the element including
filename and filetype, however the existing metadata is merged into a
newly created metadata object rather than the other way around. Because
of the way it's structured, new metadata fields can easily be forgotten
and pose debugging challenges to developers. This likely warrants a new
issue.
I'm guessing that the implementation is done this way to avoid issues
with deserializing elements, but could be wrong.
---------
Co-authored-by: Benjamin Torres <benjats07@users.noreply.github.com>
2023-09-14 11:23:16 -04:00
|
|
|
# Parent IDs are non-deterministic - remove them from the test
|
|
|
|
elem["metadata"].pop("parent_id", None)
|
|
|
|
|
2023-06-15 12:21:17 -04:00
|
|
|
assert set(elem["metadata"].keys()) == {"filename", "filetype"}
|
2023-06-06 09:03:13 -07:00
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
def test_process_file_metadata_exclude_filename_pagenum(mocker, partition_test_results):
|
|
|
|
"""Validate when metadata_exclude is set to "filename,page_number",
|
|
|
|
neither filename nor page_number are included in metadata"""
|
|
|
|
mocker.patch(
|
|
|
|
"unstructured.ingest.interfaces.partition",
|
|
|
|
return_value=partition_test_results,
|
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
partition_config = PartitionConfig(
|
|
|
|
metadata_exclude=["filename", "page_number"],
|
|
|
|
)
|
2023-12-11 12:37:49 -05:00
|
|
|
test_ingest_doc = ExampleIngestDoc(
|
2023-09-11 11:40:56 -04:00
|
|
|
connector_config=TEST_CONFIG,
|
|
|
|
read_config=ReadConfig(download_dir=TEST_DOWNLOAD_DIR),
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
processor_config=ProcessorConfig(
|
2023-06-06 09:03:13 -07:00
|
|
|
output_dir=TEST_OUTPUT_DIR,
|
|
|
|
),
|
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
isd_elems = test_ingest_doc.process_file(partition_config=partition_config)
|
2023-06-06 09:03:13 -07:00
|
|
|
assert len(isd_elems)
|
|
|
|
for elem in isd_elems:
|
2023-08-01 11:31:13 -04:00
|
|
|
assert "filename" not in elem["metadata"]
|
|
|
|
assert "page_number" not in elem["metadata"]
|
2023-06-06 09:03:13 -07:00
|
|
|
|
2023-06-20 11:19:55 -05:00
|
|
|
|
2023-06-06 09:03:13 -07:00
|
|
|
def test_process_file_flatten_metadata(mocker, partition_test_results):
|
|
|
|
mocker.patch(
|
|
|
|
"unstructured.ingest.interfaces.partition",
|
|
|
|
return_value=partition_test_results,
|
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
partition_config = PartitionConfig(
|
2023-11-03 12:47:21 -04:00
|
|
|
metadata_include=["filename", "file_directory", "filetype"],
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
flatten_metadata=True,
|
|
|
|
)
|
2023-12-11 12:37:49 -05:00
|
|
|
test_ingest_doc = ExampleIngestDoc(
|
2023-09-11 11:40:56 -04:00
|
|
|
connector_config=TEST_CONFIG,
|
|
|
|
read_config=ReadConfig(download_dir=TEST_DOWNLOAD_DIR),
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
processor_config=ProcessorConfig(
|
2023-06-06 09:03:13 -07:00
|
|
|
output_dir=TEST_OUTPUT_DIR,
|
|
|
|
),
|
|
|
|
)
|
refactor: unstructured ingest as a pipeline (#1551)
### Description
As we add more and more steps to the pipeline (i.e. chunking, embedding,
table manipulation), it would help seperate the responsibility of each
of these into their own processes, running each in parallel using json
files to share data across. This will also help guarantee data is
serializable if this code was used in an actual pipeline. Following is a
flow diagram of the proposed changes. As part of this change:
* A parent pipeline class will be responsible for running each `node`,
which can optionally be run via multiprocessing if it supports it, or
not. Possible nodes at this moment:
* Doc factory: creates all the ingest docs via the source connector
* Source: reads/downloads all of the content to process to the local
filesystem to the location set by the `download_dir` parameter.
* Partition: runs partition on all of the downloaded content in json
format.
* Any number of reformat nodes that modify the partitioned content. This
can include chunking, embedding, etc.
* Write: push the final json into the destination via the destination
connector
* This pipeline relies on the information of the ingest docs to be
available via their serialization. An optimization was introduced with
the `IngestDocJsonMixin` which adds in all the `@property` fields to the
serialized json already being created via the `DataClassJsonMixin`
* For all intermediate steps (partitioning, reformatting), the content
is saved to a dedicated location on the local filesystem. Right now it's
set to `$HOME/.cache/unstructured/ingest/pipeline/STEP_NAME/`.
* Minor changes: made sense to move some of the config parameters
between the read and partition configs when I explicitly divided the
responsibility to download vs partition the content in the pipeline.
* The pipeline class only makes the doc factory, source and partition
nodes required, keeping with the logic that has been supported so far.
All reformatting nodes and write node are optional.
* Long term, there should also be some changes to the base configs
supported by the CLI to support pipeline specific configs, but for now
what exists was used to minimize changes in this PR.
* Final step to copy the final output to the location designated by the
`_output_filename` value of the ingest doc.
* Hashing occurs at each step by hashing the parameters of that step
(i.e. partition configs) along with the previous step via the filename
used. This allows each step to be the same _if_ all the parameters for
it have not changed and the content so far is the same.
* The only data that is shared and has writes to across processes is the
dictionary of ingest json data. This dict is created using the
`multiprocessing.manager.DictProxy` to make sure any interaction with it
is behind a lock.
### Minor refactors included:
* Utility methods added to extract configs from the click options
* Utility method to add common options to click commands.
* All writers moved to using the class approach which extracts a lot of
the common code so there's less copy-paste when new runners are added.
* Use `@property` for source metadata on base ingest doc to add logic to
call `update_source_metadata` if it's still `None` at the time it's
fetched.
### Additional bug fixes included
* Fsspec connectors were not serializable due to the `ingest_doc_cls`.
This was removed from the fields captured by the `@dataclass` decorator
and added in a `__post_init__` method.
* Various reddit connector params were missing. This doesn't have an
explicit ingest test at the moment so was never caught.
* Fsspec connector had the parent `update_source_metadata` misnamed as
`update_source_metadata_metadata` so it was never being called.
### Flow Diagram

2023-10-06 14:49:29 -04:00
|
|
|
isd_elems = test_ingest_doc.process_file(partition_config=partition_config)
|
2023-11-03 12:47:21 -04:00
|
|
|
expected_keys = {"element_id", "text", "type", "filename", "file_directory", "filetype"}
|
2023-06-06 09:03:13 -07:00
|
|
|
for elem in isd_elems:
|
2023-06-07 21:22:18 -07:00
|
|
|
assert expected_keys == set(elem.keys())
|