update
This commit is contained in:
@@ -4,16 +4,15 @@ use crate::db::Database;
|
||||
use crate::engine::title_parser;
|
||||
use crate::engine::tag_extractor;
|
||||
use crate::engine::person_merger;
|
||||
use std::collections::HashMap;
|
||||
|
||||
/// 处理所有相册的元数据:提取人名 + 标签
|
||||
/// 两阶段:先用正则提取人名入库,再用匹配精炼,最后提取标签
|
||||
/// 处理所有相册的元数据:提取人名 + 标签(单次 jieba 分词)
|
||||
/// 一阶段:一次调用 extract_person_and_tags 同时获取人名和标签
|
||||
pub async fn process_all_albums(db: &Database) -> (usize, usize) {
|
||||
let albums = db.get_all_albums_full().await;
|
||||
let mut person_count = 0;
|
||||
let mut tag_count = 0;
|
||||
|
||||
// Phase 1: 先用正则从标题直接提取人名(冷启动,不需要已有 persons)
|
||||
// 已有人物库时优先用匹配(更准确)
|
||||
let patterns = person_merger::collect_person_patterns(db).await;
|
||||
let use_patterns = !patterns.is_empty();
|
||||
|
||||
@@ -21,14 +20,26 @@ pub async fn process_all_albums(db: &Database) -> (usize, usize) {
|
||||
let title = album.title.as_deref().unwrap_or("");
|
||||
if title.is_empty() { continue; }
|
||||
|
||||
// 提取人名
|
||||
if album.person_name_original.is_none() || album.person_name_original.as_deref() == Some("") {
|
||||
// 判断是否需要处理
|
||||
let need_person = album.person_name_original.is_none()
|
||||
|| album.person_name_original.as_deref() == Some("");
|
||||
let need_tags = album.tags.as_deref().unwrap_or("").is_empty()
|
||||
|| album.tags.as_deref() == Some("[]");
|
||||
|
||||
if !need_person && !need_tags { continue; }
|
||||
|
||||
// 一次 jieba 分词,同时获取人名 + 标签
|
||||
let (person_from_jieba, tags_from_jieba) = title_parser::extract_person_and_tags(title);
|
||||
|
||||
// ── 处理人名 ──
|
||||
if need_person {
|
||||
let person_name = if use_patterns {
|
||||
// 已有人物库,用匹配
|
||||
person_merger::match_title(title, &patterns).map(|(_, name, _)| name)
|
||||
// 已有人物库,优先用匹配;匹配不到再回退到 jieba
|
||||
person_merger::match_title(title, &patterns)
|
||||
.map(|(_, name, _)| name)
|
||||
.or(person_from_jieba)
|
||||
} else {
|
||||
// 冷启动,直接用正则提取
|
||||
title_parser::extract_person_name_from_title(title)
|
||||
person_from_jieba
|
||||
};
|
||||
|
||||
if let Some(person_name) = person_name {
|
||||
@@ -52,27 +63,17 @@ pub async fn process_all_albums(db: &Database) -> (usize, usize) {
|
||||
person_count += 1;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Phase 2: 提取标签(使用已填充的人名来排除)
|
||||
// 重新获取 albums(因为 person_name_cleaned 已更新)
|
||||
let albums_refreshed = db.get_all_albums_full().await;
|
||||
let mut seen_albums: HashMap<String, String> = HashMap::new();
|
||||
for album in &albums {
|
||||
if let Some(pn) = &album.person_name_cleaned {
|
||||
seen_albums.insert(album.id.clone(), pn.clone());
|
||||
}
|
||||
}
|
||||
|
||||
for album in &albums_refreshed {
|
||||
let title = album.title.as_deref().unwrap_or("");
|
||||
if title.is_empty() { continue; }
|
||||
|
||||
let existing_tags = album.tags.as_deref().unwrap_or("");
|
||||
if existing_tags.is_empty() || existing_tags == "[]" {
|
||||
// ── 处理标签 ──
|
||||
if need_tags {
|
||||
// 用当前 album 的人名(优先已有的,否则用刚提取的)
|
||||
let person = album.person_name_cleaned.as_deref()
|
||||
.or_else(|| seen_albums.get(&album.id).map(|s| s.as_str()));
|
||||
let tags = tag_extractor::extract_tags_from_title(title, person);
|
||||
.or_else(|| album.person_name_original.as_deref());
|
||||
|
||||
// 结合分类词库匹配(高质量标签)+ jieba 名词提取
|
||||
let mut tags = tag_extractor::extract_tags_from_title(title, person);
|
||||
tags.extend(tags_from_jieba);
|
||||
|
||||
let unique = tag_extractor::unique_tags(tags);
|
||||
if !unique.is_empty() {
|
||||
let tag_names: Vec<String> = unique.iter().map(|(t, _)| t.clone()).collect();
|
||||
@@ -87,4 +88,4 @@ pub async fn process_all_albums(db: &Database) -> (usize, usize) {
|
||||
|
||||
println!("[管线] 处理完成:{} 个人名,{} 个标签", person_count, tag_count);
|
||||
(person_count, tag_count)
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user