Upload 2 files
Browse files- destination_connectors.json +116 -0
- source_connectors.json +86 -0
destination_connectors.json
ADDED
@@ -0,0 +1,116 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"AstraDB": {
|
3 |
+
"destination_connector": "AstraDB",
|
4 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.astradb import (AstraDBConnectionConfig, AstraDBAccessConfig, AstraDBUploadStagerConfig, AstraDBUploaderConfig)",
|
5 |
+
"configs": "destination_connection_config=AstraDBConnectionConfig(\n access_config=AstraDBAccessConfig(\n api_endpoint=os.getenv(\"ASTRA_DB_API_ENDPOINT\"),\n token=os.getenv(\"ASTRA_DB_APPLICATION_TOKEN\")\n )\n),\nstager_config=AstraDBUploadStagerConfig(),\nuploader_config=AstraDBUploaderConfig(\n namespace=os.getenv(\"ASTRA_DB_NAMESPACE\"),\n collection_name=os.getenv(\"ASTRA_DB_COLLECTION\"),\n embedding_dimension=os.getenv(\"ASTRA_DB_EMBEDDING_DIMENSIONS\")\n)",
|
6 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/astradb"
|
7 |
+
},
|
8 |
+
"AzureBlobStorage": {
|
9 |
+
"destination_connector": "AzureBlobStorage",
|
10 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.azure import (AzureConnectionConfig, AzureAccessConfig, AzureUploaderConfig)",
|
11 |
+
"configs": "destination_connection_config=AzureConnectionConfig(\n access_config=AzureAccessConfig(\n account_name=os.getenv(\"AZURE_STORAGE_ACCOUNT_NAME\"),\n account_key=os.getenv(\"AZURE_STORAGE_ACCOUNT_KEY\")\n )\n),\nuploader_config=AzureUploaderConfig(remote_url=os.getenv(\"AZURE_STORAGE_ACCOUNT_URL\"))",
|
12 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/azure"
|
13 |
+
},
|
14 |
+
"Box": {
|
15 |
+
"destination_connector": "Box",
|
16 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.box import (BoxAccessConfig, BoxConnectionConfig, BoxIndexerConfig, BoxUploaderConfig)",
|
17 |
+
"configs": "destination_connection_config=BoxConnectionConfig(\n access_config=BoxAccessConfig(\n box_app_config=os.getenv(\"BOX_APP_CONFIG_PATH\")\n )\n),\nuploader_config=BoxUploaderConfig(\n remote_url=os.getenv(\"BOX_REMOTE_URL\")\n)",
|
18 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/box"
|
19 |
+
},
|
20 |
+
"Couchbase": {
|
21 |
+
"destination_connector": "Couchbase",
|
22 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.couchbase import (CouchbaseAccessConfig, CouchbaseConnectionConfig, CouchbaseUploadStagerConfig, CouchbaseUploaderConfig)",
|
23 |
+
"configs": "destination_connection_config=CouchbaseConnectionConfig(\n access_config=CouchbaseAccessConfig(\n password=os.getenv(\"CB_PASSWORD\"),\n ),\n connection_string=os.getenv(\"CB_CONN_STR\"),\n username=os.getenv(\"CB_USERNAME\"),\n bucket=os.getenv(\"CB_BUCKET\"),\n scope=os.getenv(\"CB_SCOPE\"),\n collection=os.getenv(\"CB_COLLECTION\")\n),\nstager_config=CouchbaseUploadStagerConfig(),\nuploader_config=CouchbaseUploaderConfig(batch_size=100)",
|
24 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/couchbase"
|
25 |
+
},
|
26 |
+
"Databricks Volumes": {
|
27 |
+
"destination_connector": "Databricks Volumes",
|
28 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.databricks_volumes import (DatabricksVolumesConnectionConfig, DatabricksVolumesAccessConfig, DatabricksVolumesUploaderConfig)",
|
29 |
+
"configs": "destination_connection_config=DatabricksVolumesConnectionConfig(\n access_config=DatabricksVolumesAccessConfig(\n username=os.getenv(\"DATABRICKS_USERNAME\"),\n password=os.getenv(\"DATABRICKS_PASSWORD\")\n ),\n host=os.getenv(\"DATABRICKS_HOST\")\n),\nuploader_config=DatabricksVolumesUploaderConfig(\n catalog=os.getenv(\"DATABRICKS_CATALOG\"),\n volume=os.getenv(\"DATABRICKS_VOLUME\")\n)",
|
30 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/databricks-volumes"
|
31 |
+
},
|
32 |
+
"Dropbox": {
|
33 |
+
"destination_connector": "Dropbox",
|
34 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.dropbox import (DropboxAccessConfig, DropboxConnectionConfig, DropboxUploaderConfig)",
|
35 |
+
"configs": "destination_connection_config=DropboxConnectionConfig(\n access_config=DropboxAccessConfig(\n token=os.getenv(\"DROPBOX_ACCESS_TOKEN\")\n )\n),\nuploader_config=DropboxUploaderConfig(\n remote_url=os.getenv(\"DROPBOX_REMOTE_URL\")\n)",
|
36 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/dropbox"
|
37 |
+
},
|
38 |
+
"Elasticsearch": {
|
39 |
+
"destination_connector": "Elasticsearch",
|
40 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.elasticsearch import (ElasticsearchConnectionConfig, ElasticsearchAccessConfig, ElasticsearchUploadStagerConfig, ElasticsearchUploaderConfig)",
|
41 |
+
"configs": "destination_connection_config=ElasticsearchConnectionConfig(\n access_config=ElasticsearchAccessConfig(\n password=os.getenv(\"ELASTICSEARCH_PASSWORD\"),\n ssl_assert_fingerprint=os.getenv(\"ELASTICSEARCH_SSL_ASSERT_FINGERPRINT\")\n ),\n hosts=[os.getenv(\"ELASTICSEARCH_HOST\")],\n username=os.getenv(\"ELASTICSEARCH_USERNAME\"),\n ca_certs=os.getenv(\"ELASTICSEARCH_CA_CERTS\")\n),\nstager_config=ElasticsearchUploadStagerConfig(index_name=os.getenv(\"ELASTICSEARCH_INDEX_NAME\")),\nuploader_config=ElasticsearchUploaderConfig(index_name=os.getenv(\"ELASTICSEARCH_INDEX_NAME\"))",
|
42 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/elasticsearch"
|
43 |
+
},
|
44 |
+
"Google Cloud Storage": {
|
45 |
+
"destination_connector": "Google Cloud Storage",
|
46 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.gcs import (GcsConnectionConfig, GcsAccessConfig, GcsUploaderConfig)",
|
47 |
+
"configs": "destination_connection_config=GcsConnectionConfig(\n access_config=GcsAccessConfig(\n service_account_key=os.getenv(\"GCS_SERVICE_ACCOUNT_KEY\")\n )\n),\nuploader_config=GcsUploaderConfig(remote_url=os.getenv(\"GCS_REMOTE_URL\"))",
|
48 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/google-cloud-service"
|
49 |
+
},
|
50 |
+
"KDB": {
|
51 |
+
"destination_connector": "KDB",
|
52 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.kdbai import (KdbaiConnectionConfig, KdbaiAccessConfig, KdbaiUploadStagerConfig, KdbaiUploaderConfig)",
|
53 |
+
"configs": "destination_connection_config=KdbaiConnectionConfig(\n access_config=KdbaiAccessConfig(\n api_key=os.getenv(\"KDBAI_API_KEY\")\n ),\n endpoint=os.getenv(\"KDBAI_ENDPOINT\")\n),\nstager_config=KdbaiUploadStagerConfig(),\nuploader_config=KdbaiUploaderConfig(table_name=os.getenv(\"KDBAI_TABLE\"))",
|
54 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/kdbai"
|
55 |
+
},
|
56 |
+
"Local directory": {
|
57 |
+
"destination_connector": "Local directory",
|
58 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.local import LocalUploaderConfig",
|
59 |
+
"configs": "uploader_config=LocalUploaderConfig(output_dir=os.getenv(\"LOCAL_FILE_OUTPUT_DIR\"))",
|
60 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/local"
|
61 |
+
},
|
62 |
+
"Milvus": {
|
63 |
+
"destination_connector": "Milvus",
|
64 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.milvus import (MilvusConnectionConfig, MilvusAccessConfig, MilvusUploadStagerConfig, MilvusUploaderConfig)",
|
65 |
+
"configs": "destination_connection_config=MilvusConnectionConfig(\n access_config=MilvusAccessConfig(\n password=os.getenv(\"MILVUS_PASSWORD\")\n ),\n uri=os.getenv(\"MILVUS_URI\"),\n user=os.getenv(\"MILVUS_USER\"),\n db_name=os.getenv(\"MILVUS_DB\")\n),\nstager_config=MilvusUploadStagerConfig(),\nuploader_config=MilvusUploaderConfig(collection_name=os.getenv(\"MILVUS_COLLECTION\"))",
|
66 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/milvus"
|
67 |
+
},
|
68 |
+
"MongoDB": {
|
69 |
+
"destination_connector": "MongoDB",
|
70 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.mongodb import (MongoDBAccessConfig, MongoDBConnectionConfig, MongoDBUploadStagerConfig, MongoDBUploaderConfig)",
|
71 |
+
"configs": "destination_connection_config=MongoDBConnectionConfig(\n access_config=MongoDBAccessConfig(\n uri=os.getenv(\"MONGODB_URI\")\n ),\n database=os.getenv(\"MONGODB_DATABASE\"),\n collection=os.getenv(\"MONGODB_COLLECTION\")\n),\nstager_config=MongoDBUploadStagerConfig(),\nuploader_config=MongoDBUploaderConfig(batch_size=100)",
|
72 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/mongodb"
|
73 |
+
},
|
74 |
+
"OpenSearch": {
|
75 |
+
"destination_connector": "OpenSearch",
|
76 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.opensearch import (OpenSearchConnectionConfig, OpenSearchAccessConfig, OpensearchUploadStagerConfig, OpensearchUploaderConfig)",
|
77 |
+
"configs": "destination_connection_config=OpenSearchConnectionConfig(\n access_config=OpenSearchAccessConfig(\n password=os.getenv(\"OPENSEARCH_PASSWORD\"),\n use_ssl=True\n ),\n hosts=[os.getenv(\"OPENSEARCH_HOST\")],\n username=os.getenv(\"OPENSEARCH_USERNAME\")\n),\nstager_config=OpensearchUploadStagerConfig(\n index_name=os.getenv(\"OPENSEARCH_INDEX_NAME\")\n),\nuploader_config=OpensearchUploaderConfig(\n index_name=os.getenv(\"OPENSEARCH_INDEX_NAME\")\n)",
|
78 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/opensearch"
|
79 |
+
},
|
80 |
+
"Pinecone": {
|
81 |
+
"destination_connector": "Pinecone",
|
82 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.pinecone import (PineconeConnectionConfig, PineconeAccessConfig, PineconeUploaderConfig, PineconeUploadStagerConfig)",
|
83 |
+
"configs": "destination_connection_config=PineconeConnectionConfig(\n access_config=PineconeAccessConfig(\n api_key=os.getenv(\"PINECONE_API_KEY\")\n ),\n index_name=os.getenv(\"PINECONE_INDEX_NAME\")\n),\nstager_config=PineconeUploadStagerConfig(),\nuploader_config=PineconeUploaderConfig()",
|
84 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/pinecone"
|
85 |
+
},
|
86 |
+
"S3": {
|
87 |
+
"destination_connector": "S3",
|
88 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.s3 import (S3ConnectionConfig, S3AccessConfig, S3UploaderConfig)",
|
89 |
+
"configs": "destination_connection_config=S3ConnectionConfig(\n access_config=S3AccessConfig(\n key=os.getenv(\"AWS_ACCESS_KEY_ID\"),\n secret=os.getenv(\"AWS_SECRET_ACCESS_KEY\")\n )\n),\nuploader_config=S3UploaderConfig(remote_url=os.getenv(\"AWS_S3_URL\"))",
|
90 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/s3"
|
91 |
+
},
|
92 |
+
"SFTP": {
|
93 |
+
"destination_connector": "SFTP",
|
94 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.sftp import (SftpConnectionConfig, SftpAccessConfig, SftpUploaderConfig)",
|
95 |
+
"configs": "destination_connection_config=SftpConnectionConfig(\n access_config=SftpAccessConfig(password=os.getenv(\"SFTP_PASSWORD\")),\n host=os.getenv(\"SFTP_HOST\"),\n port=os.getenv(\"SFTP_PORT\"),\n username=os.getenv(\"SFTP_USERNAME\")\n),\nuploader_config=SftpUploaderConfig(remote_url=os.getenv(\"SFTP_REMOTE_URL\"))",
|
96 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/sftp"
|
97 |
+
},
|
98 |
+
"SingleStore": {
|
99 |
+
"destination_connector": "SingleStore",
|
100 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.singlestore import (SingleStoreConnectionConfig, SingleStoreAccessConfig, SingleStoreUploadStagerConfig, SingleStoreUploaderConfig)",
|
101 |
+
"configs": "destination_connection_config=SingleStoreConnectionConfig(\n host=os.getenv(\"SINGLESTORE_HOST\"),\n port=os.getenv(\"SINGLESTORE_PORT\"),\n user=os.getenv(\"SINGLESTORE_USER\"),\n database=os.getenv(\"SINGLESTORE_DB\"),\n access_config=SingleStoreAccessConfig(\n password=os.getenv(\"SINGLESTORE_PASSWORD\")\n )\n),\nstager_config=SingleStoreUploadStagerConfig(drop_empty_cols=True),\nuploader_config=SingleStoreUploaderConfig(table_name=os.getenv(\"SINGLESTORE_TABLE\"))",
|
102 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/singlestore"
|
103 |
+
},
|
104 |
+
"SQL": {
|
105 |
+
"destination_connector": "SQL",
|
106 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.sql import (SQLConnectionConfig, SQLAccessConfig, SQLUploaderConfig, SQLUploadStagerConfig)",
|
107 |
+
"configs": "destination_connection_config=SQLConnectionConfig(\n access_config=SQLAccessConfig(\n username=os.getenv(\"PGUSER\"),\n password=os.getenv(\"PGPASSWORD\")\n ),\n db_type=os.getenv(\"SQL_DB_TYPE\"),\n host=os.getenv(\"PGHOST\"),\n port=os.getenv(\"PGPORT\"),\n database=os.getenv(\"PGDATABASE\")\n),\nstager_config=SQLUploadStagerConfig(),\nuploader_config=SQLUploaderConfig()",
|
108 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/sql"
|
109 |
+
},
|
110 |
+
"Weaviate": {
|
111 |
+
"destination_connector": "Weaviate",
|
112 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.weaviate import (WeaviateConnectionConfig, WeaviateAccessConfig, WeaviateUploaderConfig, WeaviateUploadStagerConfig)",
|
113 |
+
"configs": "destination_connection_config=WeaviateConnectionConfig(\n access_config=WeaviateAccessConfig(\n api_key=os.getenv(\"WEAVIATE_API_KEY\")\n ),\n host_url=os.getenv(\"WEAVIATE_URL\"),\n class_name=os.getenv(\"WEAVIATE_COLLECTION_CLASS_NAME\")\n),\nstager_config=WeaviateUploadStagerConfig(),\nuploader_config=WeaviateUploaderConfig()",
|
114 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/destination-connector/weaviate"
|
115 |
+
}
|
116 |
+
}
|
source_connectors.json
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Azure": {
|
3 |
+
"source_connector": "AzureBlobStorage",
|
4 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.azure import (AzureBlobStorageIndexerConfig, AzureBlobStorageDownloaderConfig, AzureBlobStorageConnectionConfig)",
|
5 |
+
"configs": "indexer_config=AzureBlobStorageIndexerConfig(container_name=os.getenv(\"AZURE_CONTAINER_NAME\")),\ndownloader_config=AzureBlobStorageDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=AzureBlobStorageConnectionConfig(\n connection_string=os.getenv(\"AZURE_CONNECTION_STRING\")\n),",
|
6 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/astradb"
|
7 |
+
},
|
8 |
+
"Box": {
|
9 |
+
"source_connector": "Box",
|
10 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.box import (BoxAccessConfig, BoxConnectionConfig, BoxIndexerConfig, BoxDownloaderConfig)",
|
11 |
+
"configs": "indexer_config=BoxIndexerConfig(remote_url=os.getenv(\"BOX_REMOTE_URL\")),\n downloader_config=BoxDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=BoxConnectionConfig(\n access_config=BoxAccessConfig(\n box_app_config=os.getenv(\"BOX_APP_CONFIG_PATH\")\n )\n),",
|
12 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/box"
|
13 |
+
},
|
14 |
+
"Couchbase": {
|
15 |
+
"source_connector": "Couchbase",
|
16 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.couchbase import (CouchbaseIndexerConfig, CouchbaseDownloaderConfig, CouchbaseConnectionConfig, CouchbaseAccessConfig)",
|
17 |
+
"configs": "indexer_config=CouchbaseIndexerConfig(batch_size=100),\n downloader_config=CouchbaseDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=CouchbaseConnectionConfig(\n access_config=CouchbaseAccessConfig(\n password=os.getenv(\"CB_PASSWORD\")\n ),\n username=os.getenv(\"CB_USERNAME\"),\n connection_string=os.getenv(\"CB_CONN_STR\"),\n bucket=os.getenv(\"CB_BUCKET\"),\n scope=os.getenv(\"CB_SCOPE\"),\n collection=os.getenv(\"CB_COLLECTION\")\n),",
|
18 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/couchbase"
|
19 |
+
},
|
20 |
+
"Dropbox": {
|
21 |
+
"source_connector": "Dropbox",
|
22 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.dropbox import (DropboxIndexerConfig, DropboxDownloaderConfig, DropboxAccessConfig, DropboxConnectionConfig)",
|
23 |
+
"configs": "indexer_config=DropboxIndexerConfig(remote_url=os.getenv(\"DROPBOX_REMOTE_URL\")),\ndownloader_config=DropboxDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=DropboxConnectionConfig(\n access_config=DropboxAccessConfig(\n token=os.getenv(\"DROPBOX_ACCESS_TOKEN\")\n )\n),",
|
24 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/dropbox"
|
25 |
+
},
|
26 |
+
"Elasticsearch": {
|
27 |
+
"source_connector": "Elasticsearch",
|
28 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.elasticsearch import (ElasticsearchIndexerConfig, ElasticsearchDownloaderConfig, ElasticsearchConnectionConfig, ElasticsearchAccessConfig, ElasticsearchUploadStagerConfig, ElasticsearchUploaderConfig)",
|
29 |
+
"configs": "indexer_config=ElasticsearchIndexerConfig(index_name=os.getenv(\"ELASTICSEARCH_INDEX_NAME\")),\ndownloader_config=ElasticsearchDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=ElasticsearchConnectionConfig(\n access_config=ElasticsearchAccessConfig(\n password=os.getenv(\"ELASTICSEARCH_PASSWORD\"),\n ssl_assert_fingerprint=os.getenv(\"ELASTICSEARCH_SSL_ASSERT_FINGERPRINT\")\n ),\n hosts=[os.getenv(\"ELASTICSEARCH_HOST\")],\n username=os.getenv(\"ELASTICSEARCH_USERNAME\"),\n ca_certs=os.getenv(\"ELASTICSEARCH_CA_CERTS\")\n),",
|
30 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/elastic-search"
|
31 |
+
},
|
32 |
+
"Google Cloud Storage": {
|
33 |
+
"source_connector": "Google Cloud Storage",
|
34 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.gcs import (GCSIndexerConfig, GCSDownloaderConfig, GCSConnectionConfig)",
|
35 |
+
"configs": "indexer_config=GCSIndexerConfig(bucket_name=os.getenv(\"GCS_BUCKET_NAME\")),\ndownloader_config=GCSDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=GCSConnectionConfig(\n project_id=os.getenv(\"GCS_PROJECT_ID\"),\n credentials_path=os.getenv(\"GCS_CREDENTIALS_PATH\")\n),",
|
36 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/google-cloud-storage"
|
37 |
+
},
|
38 |
+
"Google Drive": {
|
39 |
+
"source_connector": "Google Drive",
|
40 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.google_drive import (GoogleDriveConnectionConfig, GoogleDriveAccessConfig, GoogleDriveIndexerConfig, GoogleDriveDownloaderConfig)",
|
41 |
+
"configs": "indexer_config=GoogleDriveIndexerConfig(),\ndownloader_config=GoogleDriveDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=GoogleDriveConnectionConfig(\n access_config=GoogleDriveAccessConfig(\n service_account_key_path=os.getenv(\"GCP_SERVICE_ACCOUNT_KEY_FILEPATH\"), # Or\n service_account_key=os.getenv(\"GCP_SERVICE_ACCOUNT_KEY_STRING\")\n ),\n drive_id=os.getenv(\"GOOGLE_DRIVE_FOLDER_ID\"),),",
|
42 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/google-drive"
|
43 |
+
},
|
44 |
+
"Local directory": {
|
45 |
+
"source_connector": "Local directory",
|
46 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.local import (LocalIndexerConfig, LocalDownloaderConfig, LocalConnectionConfig, LocalUploaderConfig)",
|
47 |
+
"configs": "indexer_config=LocalIndexerConfig(input_path=os.getenv(\"LOCAL_FILE_INPUT_DIR\")),\ndownloader_config=LocalDownloaderConfig(),\nsource_connection_config=LocalConnectionConfig(),",
|
48 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/local"
|
49 |
+
},
|
50 |
+
"OneDrive": {
|
51 |
+
"source_connector": "OneDrive",
|
52 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.onedrive import (OnedriveIndexerConfig, OnedriveDownloaderConfig, OnedriveConnectionConfig, OnedriveAccessConfig)",
|
53 |
+
"configs": "indexer_config=OnedriveIndexerConfig(path=os.getenv(\"ONEDRIVE_PATH\")),\ndownloader_config=OnedriveDownloaderConfig(\n download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")\n),\nsource_connection_config=OnedriveConnectionConfig(\n access_config=OnedriveAccessConfig(\n client_cred=os.getenv(\"ONEDRIVE_CLIENT_CRED\")\n ),\n client_id=os.getenv(\"ONEDRIVE_CLIENT_ID\"),\n tenant=os.getenv(\"ONEDRIVE_TENANT\"),\n user_pname=os.getenv(\"ONEDRIVE_USER_PNAME\"),\n authority_url=os.getenv(\"ONEDRIVE_AUTHORITY_URL\")\n),",
|
54 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/one-drive"
|
55 |
+
},
|
56 |
+
"OpenSearch": {
|
57 |
+
"source_connector": "OpenSearch",
|
58 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.opensearch import (OpensearchIndexerConfig, OpensearchDownloaderConfig, OpenSearchConnectionConfig, OpenSearchAccessConfig)",
|
59 |
+
"configs": "indexer_config=OpensearchIndexerConfig(index_name=os.getenv(\"OPENSEARCH_INDEX_NAME\")),\ndownloader_config=OpensearchDownloaderConfig(\n download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\"),\n fields=[\"director\", \"plot\"]\n),\nsource_connection_config=OpenSearchConnectionConfig(\n access_config=OpenSearchAccessConfig(\n password=os.getenv(\"OPENSEARCH_PASSWORD\"),\n use_ssl=True\n ),\n hosts=[os.getenv(\"OPENSEARCH_HOST\")],\n username=os.getenv(\"OPENSEARCH_USERNAME\")\n),",
|
60 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/opensearch"
|
61 |
+
},
|
62 |
+
"S3": {
|
63 |
+
"source_connector": "S3",
|
64 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.s3 import (S3IndexerConfig, S3DownloaderConfig, S3ConnectionConfig, S3AccessConfig)",
|
65 |
+
"configs": "indexer_config=S3IndexerConfig(remote_url=os.getenv(\"AWS_S3_URL\")),\ndownloader_config=S3DownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=S3ConnectionConfig(\n access_config=S3AccessConfig(\n key=os.getenv(\"AWS_ACCESS_KEY_ID\"),\n secret=os.getenv(\"AWS_SECRET_ACCESS_KEY\")\n )\n),",
|
66 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/s3"
|
67 |
+
},
|
68 |
+
"Salesforce": {
|
69 |
+
"source_connector": "Salesforce",
|
70 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.salesforce import (SalesforceIndexerConfig, SalesforceDownloaderConfig, SalesforceConnectionConfig, SalesforceAccessConfig)",
|
71 |
+
"configs": "indexer_config=SalesforceIndexerConfig(categories=[\"EmailMessage\", \"Account\", \"Lead\", \"Case\", \"Campaign\"]),\ndownloader_config=SalesforceDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=SalesforceConnectionConfig(\n access_config=SalesforceAccessConfig(\n consumer_key=os.getenv(\"SALESFORCE_CONSUMER_KEY\"),\n private_key=os.getenv(\"SALESFORCE_PRIVATE_KEY\")\n ),\n username=os.getenv(\"SALESFORCE_USERNAME\")\n),",
|
72 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/salesforce"
|
73 |
+
},
|
74 |
+
"SFTP": {
|
75 |
+
"source_connector": "SFTP",
|
76 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.fsspec.sftp import (SftpIndexerConfig, SftpDownloaderConfig, SftpConnectionConfig, SftpAccessConfig)",
|
77 |
+
"configs": "indexer_config=SftpIndexerConfig(remote_url=os.getenv(\"SFTP_REMOTE_URL\")),\ndownloader_config=SftpDownloaderConfig(\n download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\"),\n remote_url=os.getenv(\"SFTP_REMOTE_URL\")\n),\nsource_connection_config=SftpConnectionConfig(\n access_config=SftpAccessConfig(password=os.getenv(\"SFTP_PASSWORD\")),\n host=os.getenv(\"SFTP_HOST\"),\n port=os.getenv(\"SFTP_PORT\"),\n username=os.getenv(\"SFTP_USERNAME\"),\n),",
|
78 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/sftp"
|
79 |
+
},
|
80 |
+
"Sharepoint": {
|
81 |
+
"source_connector": "Sharepoint",
|
82 |
+
"imports": "from unstructured_ingest.v2.processes.connectors.sharepoint import (SharepointIndexerConfig, SharepointDownloaderConfig, SharepointConnectionConfig, SharepointAccessConfig, SharepointPermissionsConfig)",
|
83 |
+
"configs": "indexer_config=SharepointIndexerConfig(\n path=os.getenv(\"SHAREPOINT_PATH\"),\n recursive=False,\n omit_lists=True,\n omit_pages=True,\n omit_files=False\n),\ndownloader_config=SharepointDownloaderConfig(download_dir=os.getenv(\"LOCAL_FILE_DOWNLOAD_DIR\")),\nsource_connection_config=SharepointConnectionConfig(\n access_config=SharepointAccessConfig(client_cred=os.getenv(\"SHAREPOINT_APP_CLIENT_SECRET\")),\n client_id=os.getenv(\"SHAREPOINT_APP_CLIENT_ID\"),\n site=os.getenv(\"SHAREPOINT_SITE\"),\n permissions_config=SharepointPermissionsConfig(\n permissions_application_id=os.getenv(\"SHAREPOINT_APP_PERMISSIONS_CLIENT_ID\"),\n permissions_tenant=os.getenv(\"SHAREPOINT_APP_TENANT_ID\"),\n permissions_client_cred=os.getenv(\"SHAREPOINT_APP_PERMISSIONS_CLIENT_SECRET\"),\n authority_url=\"https://login.microsoftonline.com\"\n )\n),",
|
84 |
+
"docs": "https://docs.unstructured.io/api-reference/ingest/source-connectors/sharepoint"
|
85 |
+
}
|
86 |
+
}
|