Skip to main content

yggdrasil/api/assets/
rebuild.rs

1//! 素材索引全量重建接口。
2//!
3//! 以磁盘为准自愈 DB 与文件系统的不一致:
4//! 1. 扫 `uploads/`(跳过 `.cache` 等点目录)→ upsert assets(技术字段变化才更新,保留 alt 与原始文件名);
5//! 2. 删除文件已消失的 DB 行(refs 级联);
6//! 3. 全表扫 posts(含回收站)重建 asset_refs。
7//!
8//! 幂等:重跑结果相同(技术字段无变化时 updated 为 0,alt 与 filename 不被覆盖)。
9//! 幂等性由「手动触发」语义承载,非常态路径。Dioxus server function,仅 admin 可用。
10
11use dioxus::prelude::*;
12
13use super::types::RebuildAssetsResponse;
14
15/// 可登记的图片扩展名(与 upload.rs 的 ALLOWED_MIME_TYPES 对应)。
16#[cfg(feature = "server")]
17const IMAGE_EXTS: &[&str] = &["jpg", "jpeg", "png", "gif", "webp"];
18
19#[cfg(feature = "server")]
20/// 扫描到的磁盘文件信息(spawn_blocking 产物)。
21struct ScannedFile {
22    /// 相对路径 "2026/07/24/x.webp"。
23    rel_path: String,
24    filename: String,
25    mime: &'static str,
26    size_bytes: i64,
27    width: i32,
28    height: i32,
29}
30
31#[cfg(feature = "server")]
32/// 递归收集 dir 下的图片文件(跳过以 `.` 开头的目录/文件)。
33/// `out` 收集可读文件;`unreadable` 收集磁盘存在但尺寸读取失败的路径
34/// (保护其 DB 行不被删除步骤误删——见 issue #30)。
35fn walk_images(
36    dir: &std::path::Path,
37    base: &std::path::Path,
38    out: &mut Vec<ScannedFile>,
39    unreadable: &mut Vec<String>,
40) {
41    let Ok(entries) = std::fs::read_dir(dir) else {
42        return;
43    };
44    for entry in entries.flatten() {
45        let name = entry.file_name();
46        let Some(name_str) = name.to_str() else {
47            continue;
48        };
49        if name_str.starts_with('.') {
50            continue;
51        }
52        let path = entry.path();
53        if path.is_dir() {
54            walk_images(&path, base, out, unreadable);
55            continue;
56        }
57        let ext = name_str.rsplit('.').next().unwrap_or("");
58        if !IMAGE_EXTS.iter().any(|e| ext.eq_ignore_ascii_case(e)) {
59            continue;
60        }
61        let Ok(rel) = path.strip_prefix(base) else {
62            continue;
63        };
64        let rel_path = rel.to_string_lossy().replace('\\', "/");
65        // 尺寸读 header(命中 IMAGE_DIMENSIONS_CACHE 时零 IO);读不到说明文件损坏或
66        // 格式不支持——记录路径但不删其 DB 行(issue #30:曾因跳过而误删大 WebP)。
67        let Some((w, h)) = crate::api::image::get_image_dimensions(&rel_path) else {
68            tracing::warn!("Rebuild: skip unreadable image {}", rel_path);
69            unreadable.push(rel_path);
70            continue;
71        };
72        let size_bytes = entry.metadata().map(|m| m.len() as i64).unwrap_or(0);
73        let mime = match ext.to_ascii_lowercase().as_str() {
74            "jpg" | "jpeg" => "image/jpeg",
75            "png" => "image/png",
76            "gif" => "image/gif",
77            _ => "image/webp",
78        };
79        out.push(ScannedFile {
80            rel_path,
81            filename: name_str.to_string(),
82            mime,
83            size_bytes,
84            width: w as i32,
85            height: h as i32,
86        });
87    }
88}
89
90/// 全量重建素材索引。
91#[server(RebuildAssetsIndex, "/api")]
92pub async fn rebuild_assets_index() -> Result<RebuildAssetsResponse, ServerFnError> {
93    #[cfg(feature = "server")]
94    {
95        use crate::api::auth::get_current_admin_user;
96        use crate::api::error::AppError;
97        use crate::db::pool::get_conn;
98
99        let _admin = get_current_admin_user().await?;
100
101        // 磁盘扫描 + header 尺寸读取是 IO 密集同步操作,移到阻塞线程池。
102        let (scanned, unreadable) = tokio::task::spawn_blocking(|| {
103            let base = std::path::Path::new("uploads");
104            let mut files = Vec::new();
105            let mut unreadable = Vec::new();
106            walk_images(base, base, &mut files, &mut unreadable);
107            (files, unreadable)
108        })
109        .await
110        .map_err(|_| AppError::Internal("素材扫描任务失败"))?;
111
112        let mut client = get_conn().await.map_err(AppError::db_conn)?;
113        let tx = client.transaction().await.map_err(AppError::tx)?;
114
115        // 1. upsert assets。xmax = 0 判别新插入(PG 系统列:新行 xmax 为 0)。
116        //    ON CONFLICT 仅当技术字段(mime/size/width/height)实际变化时才更新(IS DISTINCT FROM),
117        //    保证幂等重跑 updated = 0 且不覆盖 alt 和原始 filename(issue #31)。
118        //    filename 仅在新插入时取磁盘文件名;已存在的行保留上传时记录的原始文件名。
119        let mut inserted: i64 = 0;
120        let mut updated: i64 = 0;
121        for f in &scanned {
122            // DO UPDATE 的 WHERE 不满足时不返回行(技术字段无变化),用 query_opt 区分三种结果。
123            let asset_id = uuid::Uuid::new_v4();
124            let row = tx
125                .query_opt(
126                    "INSERT INTO assets (id, path, filename, mime, size_bytes, width, height) \
127                     VALUES ($1, $2, $3, $4, $5, $6, $7) \
128                    ON CONFLICT (path) DO UPDATE SET \
129                        mime = EXCLUDED.mime, \
130                        size_bytes = EXCLUDED.size_bytes, \
131                        width = EXCLUDED.width, \
132                        height = EXCLUDED.height, \
133                        updated_at = NOW() \
134                    WHERE assets.size_bytes IS DISTINCT FROM EXCLUDED.size_bytes \
135                       OR assets.width IS DISTINCT FROM EXCLUDED.width \
136                       OR assets.height IS DISTINCT FROM EXCLUDED.height \
137                       OR assets.mime IS DISTINCT FROM EXCLUDED.mime \
138                    RETURNING (xmax = 0) AS was_inserted",
139                    &[
140                        &asset_id,
141                        &f.rel_path,
142                        &f.filename,
143                        &f.mime,
144                        &f.size_bytes,
145                        &f.width,
146                        &f.height,
147                    ],
148                )
149                .await
150                .map_err(AppError::tx)?;
151            match row {
152                Some(r) if r.get::<_, bool>("was_inserted") => inserted += 1,
153                Some(_) => updated += 1,
154                None => {} // 技术字段无变化,幂等跳过
155            }
156        }
157
158        // 2. 删除文件已消失的 DB 行(refs 级联删)。
159        //    排除集 = 可读文件 + 尺寸读取失败的文件(后者仍在磁盘上,
160        //    不应误删——issue #30 的根因就是此处曾遗漏无法读尺寸的大 WebP)。
161        let mut keep_paths: Vec<String> = scanned.iter().map(|f| f.rel_path.clone()).collect();
162        keep_paths.extend(unreadable.iter().cloned());
163        let removed = tx
164            .execute(
165                "DELETE FROM assets WHERE NOT (path = ANY($1))",
166                &[&keep_paths],
167            )
168            .await
169            .map_err(AppError::tx)?;
170
171        // 3. 重建 asset_refs:全表扫 posts(含回收站——回收站文章的引用同样阻止删除)。
172        let post_rows = tx
173            .query("SELECT id, content_html, cover_image FROM posts", &[])
174            .await
175            .map_err(AppError::query)?;
176        tx.execute("DELETE FROM asset_refs", &[])
177            .await
178            .map_err(AppError::tx)?;
179        let mut ref_count: i64 = 0;
180        for pr in &post_rows {
181            let post_id: i32 = pr.get("id");
182            let content_html: Option<String> = pr.get("content_html");
183            let cover_image: Option<String> = pr.get("cover_image");
184            let found = crate::api::posts::helpers::extract_asset_paths(
185                content_html.as_deref().unwrap_or(""),
186                cover_image.as_deref(),
187            );
188            if found.is_empty() {
189                continue;
190            }
191            let n = tx
192                .execute(
193                    "INSERT INTO asset_refs (asset_id, post_id) \
194                     SELECT id, $1 FROM assets WHERE path = ANY($2) \
195                     ON CONFLICT DO NOTHING",
196                    &[&post_id, &found],
197                )
198                .await
199                .map_err(AppError::tx)?;
200            ref_count += n as i64;
201        }
202
203        tx.commit().await.map_err(AppError::tx)?;
204
205        let scanned_count = scanned.len() as i64;
206        let skipped_count = unreadable.len() as i64;
207        let message = if skipped_count > 0 {
208            format!(
209                "重建完成:扫描 {} 个文件,新增 {},更新 {},移除 {},跳过 {} 个无法读取的文件(已保留)",
210                scanned_count, inserted, updated, removed, skipped_count
211            )
212        } else {
213            format!(
214                "重建完成:扫描 {} 个文件,新增 {},更新 {},移除 {}",
215                scanned_count, inserted, updated, removed
216            )
217        };
218        Ok(RebuildAssetsResponse {
219            success: true,
220            message,
221            scanned: scanned_count,
222            inserted,
223            updated,
224            removed: removed as i64,
225            ref_count,
226            skipped: skipped_count,
227        })
228    }
229    #[cfg(not(feature = "server"))]
230    unreachable!()
231}