Skip to main content

LinkedInExportIntegration

What it is​

  • A small integration that works with a LinkedIn data export ZIP file:
    • Extracts the ZIP into a sibling folder.
    • Lists extracted files/folders.
    • Reads CSV files from the extracted export into a pandas.DataFrame.
  • Includes a helper to expose the integration as LangChain StructuredTools.

Public API​

LinkedInExportIntegrationConfiguration​

  • Dataclass configuration (extends IntegrationConfiguration).
  • Fields
    • export_file_path: str — path to the LinkedIn export ZIP file.

LinkedInExportIntegration​

Integration class (extends Integration).

  • __init__(configuration: LinkedInExportIntegrationConfiguration)

    • Stores configuration.

  • unzip_export() -> Dict[str, Any]

    • Validates the configured ZIP path and extracts it to a directory next to the ZIP.
    • Returns a dict with:
      • extracted_directory: str
      • files_count: int, folders_count: int
      • files: List[str] (relative paths)
      • folders: List[str] (relative paths)
      • file_created_at: datetime, file_modified_at: datetime

  • list_files_and_folders(recursive: bool = True) -> Dict

    • Calls unzip_export() and lists the extracted directory contents.
    • Returns:
      • files: List[str], folders: List[str] (relative paths)
      • total_files: int, total_folders: int
      • path: str (extracted directory path)

  • list_files() -> List[str]

    • Convenience wrapper returning list_files_and_folders()["files"].

  • read_csv(csv_file_name: str, sep: str = ",", encodings: List[str] = ["utf-8", "latin-1"], header: Optional[int] = 0, skiprows: Optional[int] = None, nrows: Optional[int] = None) -> pd.DataFrame

    • Calls unzip_export(), then reads a CSV inside the extracted directory.
    • Tries multiple encodings; attempts to detect the header row by scanning for the first “header-like” line containing sep.
    • Returns a pandas.DataFrame.

as_tools(configuration: LinkedInExportIntegrationConfiguration) -> list​

  • Returns a list of LangChain StructuredTools:
    • linkedin_export_unzip
    • linkedin_export_list_files_and_folders
    • linkedin_export_list_files
    • linkedin_export_read_csv (args: csv_file_name, sep)

Configuration/Dependencies​

  • Requires:
    • pandas
    • Standard library: os, zipfile, dataclasses, datetime, pathlib
    • naas_abi_core.integration.integration (Integration, IntegrationConfiguration)
  • Optional (only if using as_tools):
    • langchain_core.tools.StructuredTool
    • pydantic (BaseModel, Field)

Usage​

from naas_abi_marketplace.applications.linkedin.integrations.LinkedInExportIntegration import (
    LinkedInExportIntegration,
    LinkedInExportIntegrationConfiguration,
)

cfg = LinkedInExportIntegrationConfiguration(
    export_file_path="/path/to/LinkedInDataExport.zip"
)

integration = LinkedInExportIntegration(cfg)

info = integration.unzip_export()
print(info["extracted_directory"])

files = integration.list_files()
print(files[:5])

df = integration.read_csv("Connections.csv", sep=",")
print(df.head())

Caveats​

  • list_files_and_folders(), list_files(), and read_csv() call unzip_export() internally; repeated calls may re-extract the ZIP into the same directory.
  • Extraction directory naming is derived from export_file_path (<zip parent>/<zip stem>), and the directory is created if missing.
  • read_csv() header detection is heuristic; it scans for the first line containing the separator and assumes it is a header when the first column name length is below a fixed threshold.