Cleans/shortens a filename or directory name component using simple replacement.

Builds CivitAI image API URLs with an explicit max page size. The API default is smaller than the documented maximum. Using 200 items

Handles pagination and processing for an identifier, checking for 'Not Found' errors. Args: deep_scan_pass: If set, this is a con

Lazily creates and returns the shared httpx.AsyncClient instance.

Helper to get the correct dictionary entry in run_results to update stats.

Sorts downloaded images and metadata files within a directory (base_dir) into subfolders based on the 'Model' field found in the meta

Extracts the actual metadata dict from an API item, handling nested structure. CivitAI API changed structure from: item["meta"] = {"p

Creates mode-specific subfolder within output directory.

Gets the disable_prompt_check option (True/False) from args or interactive prompt.

Robust file moving with retries and logging, checking actual outcome.

Validates a pagination URL from the API to prevent SSRF attacks. Ensures the URL uses HTTPS, belongs to an allowed CivitAI API host (

Issue #58: Clear all tracked images for a specific target. Args: target_type: Type of target ('username', 'model', 'tag', 'modelV

Inserts a batch of image data.

Inserts a batch of tag data.

Main entry point to run the downloader based on initialized configuration. Collects identifiers, initializes results, dispatches task

Logs warnings if CLI arguments conflict with the selected mode.

Fetches a single page from the Civitai images API, with retries.

Fetches one model record so modelId can be resolved to modelVersionIds.

Gets filter tags for Mode 4 from args or interactive prompt. Returns None if not specified.

Get maximum images limit, prompting in interactive mode if not set via CLI.

Gets model IDs from args or interactive prompt, validating numeric.

Gets model version IDs from args or interactive prompt, validating numeric.

Returns (modelVersionId, version name) pairs for a model ID.

Gets tags from args or interactive prompt, replacing spaces with underscores.

Gets the timeout value from args or interactive prompt.

Gets usernames from args or interactive prompt.

Handles download, meta write, tracking, stats update for one item.

Initializes the SQLite DB connection and creates tables if needed.

Issue #58: Add target_type and target_value columns for per-target DB management.

Return seconds from a Retry-After header, capped to avoid unbounded sleeps.

Prints the final download statistics summary, including per-identifier results and warnings.

Processes image items: checks skip conditions, creates download tasks.

Performs deep scan passes to retrieve images beyond the 50K API pagination cap. Strategy: 1. sort=Oldest pass: Paginates from the old

Downloads all images for a Model ID by resolving its modelVersionIds first.

Compares downloaded image count against known website counts for verification. Loads website_image_counts.json and prints a completeness repo

Searches for model version IDs by tag, handling pagination and adding validation based on first page results and a maximum page limit

Helper: Fetches one page of model search results, with retries.

Performs basic client-side validation. Returns (is_valid, reason_if_invalid).

Writes metadata to a .txt file. Args: meta: Metadata dictionary extracted from API response base_output_path_no_ext:

Writes CSV summaries for tags using SQLite data.

Checks if an image exists in the tracking database. Args: image_id: The image ID quality: Image quality (SD/HD)

Adjust logger and handler levels after CLI arguments are known.

Detects file extension from magic numbers.

Downloads a single image or video, detects extension, saves, with retries.

Connects to DB and ensures schema exists.

Marks image as downloaded in DB (INSERT OR REPLACE image, DELETE/INSERT tags). Args: image_id: The image ID image_pat

Parses command-line arguments.

Initializes the downloader with configuration and database connection.