Module multimodal

Structs§

LocalImageCache: Per-path cache for resolved local image data URIs. Keyed by absolute path; stores (len, mtime) for freshness checks ((0, 0) sentinel = immutable upload). LRU evicts by both entry count and total bytes.
PreparedMessages

contains_image_markers
count_image_markers
count_user_image_markers: Count image markers that originate from genuine user messages (i.e. inbound attachments), excluding tool-result carriers (role == "tool" and [Tool results] user messages).
extract_ollama_image_payload
parse_image_markers
prepare_messages_for_provider
prepare_messages_for_provider_cached: Like prepare_messages_for_provider but reuses a LocalImageCache across calls so each unique local image file is read from disk at most once per session (or once per modification for mutable files).
strip_media_markers: Replace media markers ([IMAGE:...], [PHOTO:...], [DOCUMENT:...], [FILE:...], [VIDEO:...], [VOICE:...], [AUDIO:...]) with [media attachment]. Match is case-insensitive to align with the channel attachment parsers, which all uppercase the kind before comparing (crates/zeroclaw-channels/src/util.rs::ATTACHMENT_KINDS, telegram.rs, discord.rs, qq.rs, whatsapp_web.rs).