docling-haystack 0.1.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- docling_haystack-0.1.1/LICENSE +21 -0
- docling_haystack-0.1.1/PKG-INFO +54 -0
- docling_haystack-0.1.1/README.md +37 -0
- docling_haystack-0.1.1/docling_haystack/__init__.py +7 -0
- docling_haystack-0.1.1/docling_haystack/converter.py +141 -0
- docling_haystack-0.1.1/docling_haystack/py.typed +0 -0
- docling_haystack-0.1.1/pyproject.toml +80 -0
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) 2024 International Business Machines
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
|
@@ -0,0 +1,54 @@
|
|
|
1
|
+
Metadata-Version: 2.1
|
|
2
|
+
Name: docling-haystack
|
|
3
|
+
Version: 0.1.1
|
|
4
|
+
Summary: Docling Haystack converter
|
|
5
|
+
Author: Panos Vagenas
|
|
6
|
+
Author-email: pva@zurich.ibm.com
|
|
7
|
+
Requires-Python: >=3.9,<3.13
|
|
8
|
+
Classifier: Programming Language :: Python :: 3
|
|
9
|
+
Classifier: Programming Language :: Python :: 3.9
|
|
10
|
+
Classifier: Programming Language :: Python :: 3.10
|
|
11
|
+
Classifier: Programming Language :: Python :: 3.11
|
|
12
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
13
|
+
Requires-Dist: docling (>=2.9.0,<3.0.0)
|
|
14
|
+
Requires-Dist: haystack-ai (>=2.8.0,<3.0.0)
|
|
15
|
+
Description-Content-Type: text/markdown
|
|
16
|
+
|
|
17
|
+
# Docling Haystack converter
|
|
18
|
+
|
|
19
|
+
[](https://pypi.org/project/docling-haystack/)
|
|
20
|
+
[](https://pypi.org/project/docling-haystack/)
|
|
21
|
+
[](https://python-poetry.org/)
|
|
22
|
+
[](https://github.com/psf/black)
|
|
23
|
+
[](https://pycqa.github.io/isort/)
|
|
24
|
+
[](https://pydantic.dev)
|
|
25
|
+
[](https://github.com/pre-commit/pre-commit)
|
|
26
|
+
[](https://opensource.org/licenses/MIT)
|
|
27
|
+
|
|
28
|
+
A [Docling](https://github.com/DS4SD/docling) converter integration for
|
|
29
|
+
[Haystack](https://github.com/deepset-ai/haystack/).
|
|
30
|
+
|
|
31
|
+
## Installation
|
|
32
|
+
|
|
33
|
+
Simply install `docling-haystack` from your package manager, e.g. pip:
|
|
34
|
+
```bash
|
|
35
|
+
pip install docling-haystack
|
|
36
|
+
```
|
|
37
|
+
|
|
38
|
+
## Usage
|
|
39
|
+
|
|
40
|
+
Basic usage in a Haystack pipeline looks as follows:
|
|
41
|
+
|
|
42
|
+
```python
|
|
43
|
+
from haystack import Pipeline
|
|
44
|
+
from docling_haystack.converter import DoclingConverter
|
|
45
|
+
|
|
46
|
+
idx_pipe = Pipeline()
|
|
47
|
+
# ...
|
|
48
|
+
converter = DoclingConverter()
|
|
49
|
+
idx_pipe.add_component("converter", converter)
|
|
50
|
+
# ...
|
|
51
|
+
```
|
|
52
|
+
|
|
53
|
+
For end-to-end usage samples check out the [examples](examples/).
|
|
54
|
+
|
|
@@ -0,0 +1,37 @@
|
|
|
1
|
+
# Docling Haystack converter
|
|
2
|
+
|
|
3
|
+
[](https://pypi.org/project/docling-haystack/)
|
|
4
|
+
[](https://pypi.org/project/docling-haystack/)
|
|
5
|
+
[](https://python-poetry.org/)
|
|
6
|
+
[](https://github.com/psf/black)
|
|
7
|
+
[](https://pycqa.github.io/isort/)
|
|
8
|
+
[](https://pydantic.dev)
|
|
9
|
+
[](https://github.com/pre-commit/pre-commit)
|
|
10
|
+
[](https://opensource.org/licenses/MIT)
|
|
11
|
+
|
|
12
|
+
A [Docling](https://github.com/DS4SD/docling) converter integration for
|
|
13
|
+
[Haystack](https://github.com/deepset-ai/haystack/).
|
|
14
|
+
|
|
15
|
+
## Installation
|
|
16
|
+
|
|
17
|
+
Simply install `docling-haystack` from your package manager, e.g. pip:
|
|
18
|
+
```bash
|
|
19
|
+
pip install docling-haystack
|
|
20
|
+
```
|
|
21
|
+
|
|
22
|
+
## Usage
|
|
23
|
+
|
|
24
|
+
Basic usage in a Haystack pipeline looks as follows:
|
|
25
|
+
|
|
26
|
+
```python
|
|
27
|
+
from haystack import Pipeline
|
|
28
|
+
from docling_haystack.converter import DoclingConverter
|
|
29
|
+
|
|
30
|
+
idx_pipe = Pipeline()
|
|
31
|
+
# ...
|
|
32
|
+
converter = DoclingConverter()
|
|
33
|
+
idx_pipe.add_component("converter", converter)
|
|
34
|
+
# ...
|
|
35
|
+
```
|
|
36
|
+
|
|
37
|
+
For end-to-end usage samples check out the [examples](examples/).
|
|
@@ -0,0 +1,141 @@
|
|
|
1
|
+
#
|
|
2
|
+
# Copyright IBM Corp. 2024 - 2024
|
|
3
|
+
# SPDX-License-Identifier: MIT
|
|
4
|
+
#
|
|
5
|
+
|
|
6
|
+
"""Docling Haystack converter module."""
|
|
7
|
+
|
|
8
|
+
from abc import ABC, abstractmethod
|
|
9
|
+
from enum import Enum
|
|
10
|
+
from pathlib import Path
|
|
11
|
+
from typing import Any, Iterable, Optional, Union
|
|
12
|
+
|
|
13
|
+
from docling.chunking import BaseChunk, BaseChunker, HybridChunker
|
|
14
|
+
from docling.datamodel.document import DoclingDocument
|
|
15
|
+
from docling.document_converter import DocumentConverter
|
|
16
|
+
from haystack import Document, component
|
|
17
|
+
|
|
18
|
+
|
|
19
|
+
class ExportType(str, Enum):
|
|
20
|
+
"""Enumeration of available export types."""
|
|
21
|
+
|
|
22
|
+
MARKDOWN = "markdown"
|
|
23
|
+
DOC_CHUNKS = "doc_chunks"
|
|
24
|
+
|
|
25
|
+
|
|
26
|
+
class BaseMetaExtractor(ABC):
|
|
27
|
+
"""BaseMetaExtractor."""
|
|
28
|
+
|
|
29
|
+
@abstractmethod
|
|
30
|
+
def extract_chunk_meta(self, chunk: BaseChunk) -> dict[str, Any]:
|
|
31
|
+
"""Extract chunk meta."""
|
|
32
|
+
raise NotImplementedError()
|
|
33
|
+
|
|
34
|
+
@abstractmethod
|
|
35
|
+
def extract_dl_doc_meta(self, dl_doc: DoclingDocument) -> dict[str, Any]:
|
|
36
|
+
"""Extract Docling document meta."""
|
|
37
|
+
raise NotImplementedError()
|
|
38
|
+
|
|
39
|
+
|
|
40
|
+
class MetaExtractor(BaseMetaExtractor):
|
|
41
|
+
"""MetaExtractor."""
|
|
42
|
+
|
|
43
|
+
def extract_chunk_meta(self, chunk: BaseChunk) -> dict[str, Any]:
|
|
44
|
+
"""Extract chunk meta."""
|
|
45
|
+
return {"dl_meta": chunk.export_json_dict()}
|
|
46
|
+
|
|
47
|
+
def extract_dl_doc_meta(self, dl_doc: DoclingDocument) -> dict[str, Any]:
|
|
48
|
+
"""Extract Docling document meta."""
|
|
49
|
+
return (
|
|
50
|
+
{"dl_meta": {"origin": dl_doc.origin.model_dump(exclude_none=True)}}
|
|
51
|
+
if dl_doc.origin
|
|
52
|
+
else {}
|
|
53
|
+
)
|
|
54
|
+
|
|
55
|
+
|
|
56
|
+
@component
|
|
57
|
+
class DoclingConverter:
|
|
58
|
+
"""Docling Haystack converter."""
|
|
59
|
+
|
|
60
|
+
def __init__(
|
|
61
|
+
self,
|
|
62
|
+
converter: Optional[DocumentConverter] = None,
|
|
63
|
+
convert_kwargs: Optional[dict[str, Any]] = None,
|
|
64
|
+
export_type: ExportType = ExportType.DOC_CHUNKS,
|
|
65
|
+
md_export_kwargs: Optional[dict[str, Any]] = None,
|
|
66
|
+
chunker: Optional[BaseChunker] = None,
|
|
67
|
+
meta_extractor: Optional[BaseMetaExtractor] = None,
|
|
68
|
+
):
|
|
69
|
+
"""Create a Docling Haystack converter.
|
|
70
|
+
|
|
71
|
+
Args:
|
|
72
|
+
converter: The Docling `DocumentConverter` to use; if not set, a system
|
|
73
|
+
default is used.
|
|
74
|
+
convert_kwargs: Any parameters to pass to Docling conversion; if not set, a
|
|
75
|
+
system default is used.
|
|
76
|
+
export_type: The export mode to use: set to `ExportType.MARKDOWN` if you
|
|
77
|
+
want to capture each input document as a separate Haystack document, or
|
|
78
|
+
`ExportType.DOC_CHUNKS` (default), if you want to first have each input
|
|
79
|
+
document chunked and to then capture each individual chunk as a separate
|
|
80
|
+
Haystack document downstream.
|
|
81
|
+
md_export_kwargs: Any parameters to pass to Markdown export (applicable in
|
|
82
|
+
case of `ExportType.MARKDOWN`).
|
|
83
|
+
chunker: The Docling chunker instance to use; if not set, a system default
|
|
84
|
+
is used.
|
|
85
|
+
meta_extractor: The extractor instance to use for populating the output
|
|
86
|
+
document metadata; if not set, a system default is used.
|
|
87
|
+
"""
|
|
88
|
+
self._converter = converter or DocumentConverter()
|
|
89
|
+
self._convert_kwargs = convert_kwargs if convert_kwargs is not None else {}
|
|
90
|
+
self._export_type = export_type
|
|
91
|
+
self._md_export_kwargs = (
|
|
92
|
+
md_export_kwargs
|
|
93
|
+
if md_export_kwargs is not None
|
|
94
|
+
else {"image_placeholder": ""}
|
|
95
|
+
)
|
|
96
|
+
if self._export_type == ExportType.DOC_CHUNKS:
|
|
97
|
+
# TODO remove tokenizer once docling-core ^2.10.0 guaranteed via docling:
|
|
98
|
+
self._chunker = chunker or HybridChunker(
|
|
99
|
+
tokenizer="sentence-transformers/all-MiniLM-L6-v2"
|
|
100
|
+
)
|
|
101
|
+
self._meta_extractor = meta_extractor or MetaExtractor()
|
|
102
|
+
|
|
103
|
+
@component.output_types(documents=list[Document])
|
|
104
|
+
def run(
|
|
105
|
+
self,
|
|
106
|
+
paths: Iterable[Union[Path, str]],
|
|
107
|
+
):
|
|
108
|
+
"""Run the DoclingConverter.
|
|
109
|
+
|
|
110
|
+
Args:
|
|
111
|
+
paths: The input document locations, either as local paths or URLs.
|
|
112
|
+
|
|
113
|
+
Returns:
|
|
114
|
+
list[Document]: The output Haystack Documents.
|
|
115
|
+
"""
|
|
116
|
+
documents: list[Document] = []
|
|
117
|
+
for filepath in paths:
|
|
118
|
+
dl_doc = self._converter.convert(
|
|
119
|
+
source=filepath,
|
|
120
|
+
**self._convert_kwargs,
|
|
121
|
+
).document
|
|
122
|
+
|
|
123
|
+
if self._export_type == ExportType.DOC_CHUNKS:
|
|
124
|
+
chunk_iter = self._chunker.chunk(dl_doc=dl_doc)
|
|
125
|
+
hs_docs = [
|
|
126
|
+
Document(
|
|
127
|
+
content=self._chunker.serialize(chunk=chunk),
|
|
128
|
+
meta=self._meta_extractor.extract_chunk_meta(chunk=chunk),
|
|
129
|
+
)
|
|
130
|
+
for chunk in chunk_iter
|
|
131
|
+
]
|
|
132
|
+
documents.extend(hs_docs)
|
|
133
|
+
elif self._export_type == ExportType.MARKDOWN:
|
|
134
|
+
hs_doc = Document(
|
|
135
|
+
content=dl_doc.export_to_markdown(**self._md_export_kwargs),
|
|
136
|
+
meta=self._meta_extractor.extract_dl_doc_meta(dl_doc=dl_doc),
|
|
137
|
+
)
|
|
138
|
+
documents.append(hs_doc)
|
|
139
|
+
else:
|
|
140
|
+
raise RuntimeError(f"Unexpected export type: {self._export_type}")
|
|
141
|
+
return {"documents": documents}
|
|
File without changes
|
|
@@ -0,0 +1,80 @@
|
|
|
1
|
+
[tool.poetry]
|
|
2
|
+
name = "docling-haystack"
|
|
3
|
+
version = "0.1.1" # DO NOT EDIT, updated automatically
|
|
4
|
+
description = "Docling Haystack converter"
|
|
5
|
+
authors = ["Panos Vagenas <pva@zurich.ibm.com>"]
|
|
6
|
+
readme = "README.md"
|
|
7
|
+
|
|
8
|
+
packages = [{ include = "docling_haystack" }]
|
|
9
|
+
|
|
10
|
+
[tool.poetry.dependencies]
|
|
11
|
+
python = ">=3.9,<3.13" # constraining below 3.13 due to haystack-ai 2.8.0
|
|
12
|
+
haystack-ai = "^2.8.0"
|
|
13
|
+
docling = "^2.9.0"
|
|
14
|
+
|
|
15
|
+
[tool.poetry.group.dev.dependencies]
|
|
16
|
+
ipykernel = "^6.29.5"
|
|
17
|
+
black = "^24.4.2"
|
|
18
|
+
pytest = "^7.1.2"
|
|
19
|
+
mypy = "^1.6.0"
|
|
20
|
+
isort = "^5.10.1"
|
|
21
|
+
pre-commit = "^3.7.1"
|
|
22
|
+
autoflake = "^2.0.0"
|
|
23
|
+
flake8 = "^7.1.0"
|
|
24
|
+
pycodestyle = "^2.10.0"
|
|
25
|
+
flake8-docstrings = "^1.6.0"
|
|
26
|
+
pep8-naming = "^0.13.2"
|
|
27
|
+
nbqa = "^1.9.0"
|
|
28
|
+
python-semantic-release = "^7.32.2"
|
|
29
|
+
|
|
30
|
+
[build-system]
|
|
31
|
+
requires = ["poetry-core"]
|
|
32
|
+
build-backend = "poetry.core.masonry.api"
|
|
33
|
+
|
|
34
|
+
[tool.black]
|
|
35
|
+
line-length = 88
|
|
36
|
+
target-version = ["py39", "py310"]
|
|
37
|
+
include = '\.pyi?$'
|
|
38
|
+
preview = true
|
|
39
|
+
|
|
40
|
+
[tool.isort]
|
|
41
|
+
profile = "black"
|
|
42
|
+
line_length = 88
|
|
43
|
+
py_version = 39
|
|
44
|
+
multi_line_output = 3
|
|
45
|
+
include_trailing_comma = true
|
|
46
|
+
|
|
47
|
+
[tool.autoflake]
|
|
48
|
+
in-place = true
|
|
49
|
+
ignore-init-module-imports = true
|
|
50
|
+
remove-all-unused-imports = true
|
|
51
|
+
remove-unused-variables = true
|
|
52
|
+
expand-star-imports = true
|
|
53
|
+
recursive = true
|
|
54
|
+
|
|
55
|
+
[tool.mypy]
|
|
56
|
+
pretty = true
|
|
57
|
+
no_implicit_optional = true
|
|
58
|
+
namespace_packages = true
|
|
59
|
+
show_error_codes = true
|
|
60
|
+
python_version = "3.9"
|
|
61
|
+
plugins = ["pydantic.mypy"]
|
|
62
|
+
|
|
63
|
+
[[tool.mypy.overrides]]
|
|
64
|
+
module = [
|
|
65
|
+
"haystack.*",
|
|
66
|
+
]
|
|
67
|
+
ignore_missing_imports = true
|
|
68
|
+
|
|
69
|
+
[tool.semantic_release]
|
|
70
|
+
# for default values check:
|
|
71
|
+
# https://github.com/python-semantic-release/python-semantic-release/blob/v7.32.2/semantic_release/defaults.cfg
|
|
72
|
+
|
|
73
|
+
version_source = "tag_only"
|
|
74
|
+
branch = "main"
|
|
75
|
+
|
|
76
|
+
# configure types which should trigger minor and patch version bumps respectively
|
|
77
|
+
# (note that they must be a subset of the configured allowed types):
|
|
78
|
+
parser_angular_allowed_types = "build,chore,ci,docs,feat,fix,perf,style,refactor,test"
|
|
79
|
+
parser_angular_minor_types = "feat"
|
|
80
|
+
parser_angular_patch_types = "fix,perf"
|