file connectors¶
File-system and file-protocol integrations: local FS, network FS (NFS, SMB), transfer protocols (FTP, SFTP, SCP, AS2), file formats (CSV, JSON, XML, YAML, Parquet, PDF, Excel), and supplementary content extraction (ocr-extract, pdf-extract). They cover both transport ("get this file from there") and processing ("read this CSV file into records") roles.
Connectors in this category¶
| Connector | URI / type | Auth | Native SDK | One-line description |
|---|---|---|---|---|
as2_file |
connector://as2-file/send |
Certificate signing | No (HTTP) | AS2 file transfer with signed MDN receipts. |
csv |
connector://csv/send |
None | No | CSV (RFC 4180) read / write with headers + dialect options. |
file-watch |
connector://fwatch/send |
None | No | File-system change watcher (fsnotify-style). |
ftp |
connector://ftp/send |
Username / password | No (HTTP) | FTP transfer (active / passive) via net/textproto. |
http_connector |
connector://http-conn/send |
Bearer / Basic / mTLS | No (HTTP) | HTTP file download + upload (generic). |
json-file |
connector://json-file/send |
None | No | JSON file read / write (single-doc or stream). |
ldap_connector |
connector://ldap-conn/send |
Bind DN + password | No (HTTP) | LDAP directory operations as file-style records. |
local-file |
connector://local-file/send |
OS perms | No | Local file system read / write / list / delete. |
nfs_connector |
connector://nfs/send |
OS mount | No | NFS network file system (mount-mediated). |
ocr-extract |
connector://ocr/send |
None | No | OCR text extraction (Tesseract-backed). |
parquet-file |
connector://parquet-file/send |
None | No | Apache Parquet column-store file read / write. |
pdf-extract |
connector://pdf-extract/send |
None | No | PDF text + table extraction (pdf-lib). |
pdf_lib |
(internal helper) | n/a | n/a | Shared PDF library used by pdf-extract. |
scp_connector |
connector://scp/send |
SSH key / password | No (HTTP) | SCP secure copy via golang.org/x/crypto/ssh. |
sftp |
connector://sftp/send |
SSH key / password | No (HTTP) | SFTP secure file transfer via pkg/sftp. |
smb |
connector://smb/send |
NTLM / Kerberos | No (HTTP) | SMB / CIFS file share operations. |
socket_connector |
connector://socket/send |
None / TLS | No | Raw TCP / UDP socket I/O. |
tar-archive |
connector://tar/send |
None | No | TAR archive create / extract / list. |
websocket_connector |
connector://ws/send |
Bearer / token | No (HTTP) | WebSocket bidirectional file / message streaming. |
xml-file |
connector://xml-file/send |
None | No | XML file read / write with namespace + XPath support. |
yaml-file |
connector://yaml-file/send |
None | No | YAML file read / write (yaml.v3). |
zip-archive |
connector://zip/send |
None | No | ZIP archive create / extract / list. |
pdf_lib.go is an internal helper; the other 22 register via init().
See also¶
- Connector SDK guide — adding a new connector
file://,ftp://,s3://,websocket://components — for streaming or polling primary use- storage — cloud-object stores (Box, Dropbox, Backblaze, MinIO, Cloudflare R2, Wasabi)
pkg/rag/ingest/— file readers (PDF, DOCX, CSV, HTML, Markdown, XML, image, audio) used inside the RAG ingest pipeline