feat: 实现视频的筛选规则 (#457)

crates/bili_sync_entity/src/custom_type/mod.rs

@@ -0,0 +1,2 @@
+pub mod rule;
+pub mod string_vec;

crates/bili_sync_entity/src/custom_type/rule.rs

@@ -0,0 +1,120 @@
+use std::fmt::Display;
+
+use derivative::Derivative;
+use sea_orm::FromJsonQueryResult;
+use sea_orm::prelude::DateTime;
+use serde::{Deserialize, Deserializer, Serialize, Serializer};
+
+#[derive(Clone, Debug, Serialize, Deserialize, Derivative)]
+#[derivative(PartialEq, Eq)]
+#[serde(rename_all = "camelCase", tag = "operator", content = "value")]
+pub enum Condition<T: Serialize + Display> {
+    Equals(T),
+    Contains(T),
+    #[serde(deserialize_with = "deserialize_regex", serialize_with = "serialize_regex")]
+    MatchesRegex(String, #[derivative(PartialEq = "ignore")] regex::Regex),
+    Prefix(T),
+    Suffix(T),
+    GreaterThan(T),
+    LessThan(T),
+    Between(T, T),
+}
+
+#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize, FromJsonQueryResult)]
+#[serde(rename_all = "camelCase", tag = "field", content = "rule")]
+pub enum RuleTarget {
+    Title(Condition<String>),
+    Tags(Condition<String>),
+    FavTime(Condition<DateTime>),
+    PubTime(Condition<DateTime>),
+    PageCount(Condition<usize>),
+    Not(Box<RuleTarget>),
+}
+
+pub type AndGroup = Vec<RuleTarget>;
+
+#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize, FromJsonQueryResult)]
+pub struct Rule(pub Vec<AndGroup>);
+
+impl<T: Serialize + Display> Display for Condition<T> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Condition::Equals(v) => write!(f, "等于“{}”", v),
+            Condition::Contains(v) => write!(f, "包含“{}”", v),
+            Condition::MatchesRegex(pat, _) => write!(f, "匹配“{}”", pat),
+            Condition::Prefix(v) => write!(f, "以“{}”开头", v),
+            Condition::Suffix(v) => write!(f, "以“{}”结尾", v),
+            Condition::GreaterThan(v) => write!(f, "大于“{}”", v),
+            Condition::LessThan(v) => write!(f, "小于“{}”", v),
+            Condition::Between(start, end) => write!(f, "在“{}”和“{}”之间", start, end),
+        }
+    }
+}
+
+impl Display for RuleTarget {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        fn get_field_name(rt: &RuleTarget, depth: usize) -> &'static str {
+            match rt {
+                RuleTarget::Title(_) => "标题",
+                RuleTarget::Tags(_) => "标签",
+                RuleTarget::FavTime(_) => "收藏时间",
+                RuleTarget::PubTime(_) => "发布时间",
+                RuleTarget::PageCount(_) => "视频分页数量",
+                RuleTarget::Not(inner) => {
+                    if depth == 0 {
+                        get_field_name(inner, depth + 1)
+                    } else {
+                        "格式化失败"
+                    }
+                }
+            }
+        }
+        let field_name = get_field_name(self, 0);
+        match self {
+            RuleTarget::Not(inner) => match inner.as_ref() {
+                RuleTarget::Title(cond) | RuleTarget::Tags(cond) => write!(f, "{}不{}", field_name, cond),
+                RuleTarget::FavTime(cond) | RuleTarget::PubTime(cond) => {
+                    write!(f, "{}不{}", field_name, cond)
+                }
+                RuleTarget::PageCount(cond) => write!(f, "{}不{}", field_name, cond),
+                RuleTarget::Not(_) => write!(f, "格式化失败"),
+            },
+            RuleTarget::Title(cond) | RuleTarget::Tags(cond) => write!(f, "{}{}", field_name, cond),
+            RuleTarget::FavTime(cond) | RuleTarget::PubTime(cond) => {
+                write!(f, "{}{}", field_name, cond)
+            }
+            RuleTarget::PageCount(cond) => write!(f, "{}{}", field_name, cond),
+        }
+    }
+}
+
+impl Display for Rule {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        let groups: Vec<String> = self
+            .0
+            .iter()
+            .map(|group| {
+                let conditions: Vec<String> = group.iter().map(|target| format!("({})", target)).collect();
+                format!("「{}」", conditions.join("且"))
+            })
+            .collect();
+        write!(f, "{}", groups.join("或"))
+    }
+}
+
+fn deserialize_regex<'de, D>(deserializer: D) -> Result<(String, regex::Regex), D::Error>
+where
+    D: Deserializer<'de>,
+{
+    let pattern = String::deserialize(deserializer)?;
+    // 反序列化时预编译 regex，优化性能
+    let regex = regex::Regex::new(&pattern).map_err(|e| serde::de::Error::custom(e))?;
+    Ok((pattern, regex))
+}
+
+fn serialize_regex<S>(pattern: &String, _regex: &regex::Regex, serializer: S) -> Result<S::Ok, S::Error>
+where
+    S: Serializer,
+{
+    serializer.serialize_str(pattern)
+}

crates/bili_sync_entity/src/custom_type/string_vec.rs

@@ -0,0 +1,20 @@
+use sea_orm::FromJsonQueryResult;
+use serde::{Deserialize, Serialize};
+
+// reference: https://www.sea-ql.org/SeaORM/docs/generate-entity/column-types/#json-column
+// 在 entity 中使用裸 Vec 仅在 postgres 中支持，sea-orm 会将其映射为 postgres array
+// 如果需要实现跨数据库的 array，必须将其包裹在 wrapper type 中
+#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize, FromJsonQueryResult)]
+pub struct StringVec(pub Vec<String>);
+
+impl From<Vec<String>> for StringVec {
+    fn from(value: Vec<String>) -> Self {
+        Self(value)
+    }
+}
+
+impl From<StringVec> for Vec<String> {
+    fn from(value: StringVec) -> Self {
+        value.0
+    }
+}

crates/bili_sync_entity/src/entities/collection.rs

@@ -2,6 +2,8 @@
 
 use sea_orm::entity::prelude::*;
 
+use crate::rule::Rule;
+
 #[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq)]
 #[sea_orm(table_name = "collection")]
 pub struct Model {
@@ -14,6 +16,7 @@ pub struct Model {
     pub path: String,
     pub created_at: String,
     pub latest_row_at: DateTime,
+    pub rule: Option<Rule>,
     pub enabled: bool,
 }
 

crates/bili_sync_entity/src/entities/favorite.rs

@@ -2,6 +2,8 @@
 
 use sea_orm::entity::prelude::*;
 
+use crate::rule::Rule;
+
 #[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq)]
 #[sea_orm(table_name = "favorite")]
 pub struct Model {
@@ -13,6 +15,7 @@ pub struct Model {
     pub path: String,
     pub created_at: String,
     pub latest_row_at: DateTime,
+    pub rule: Option<Rule>,
     pub enabled: bool,
 }
 

crates/bili_sync_entity/src/entities/submission.rs

@@ -2,6 +2,8 @@
 
 use sea_orm::entity::prelude::*;
 
+use crate::rule::Rule;
+
 #[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq)]
 #[sea_orm(table_name = "submission")]
 pub struct Model {
@@ -12,6 +14,7 @@ pub struct Model {
     pub path: String,
     pub created_at: String,
     pub latest_row_at: DateTime,
+    pub rule: Option<Rule>,
     pub enabled: bool,
 }
 

crates/bili_sync_entity/src/entities/video.rs

@@ -2,6 +2,8 @@
 
 use sea_orm::entity::prelude::*;
 
+use crate::string_vec::StringVec;
+
 #[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq, Default)]
 #[sea_orm(table_name = "video")]
 pub struct Model {
@@ -25,7 +27,8 @@ pub struct Model {
     pub favtime: DateTime,
     pub download_status: u32,
     pub valid: bool,
-    pub tags: Option<serde_json::Value>,
+    pub should_download: bool,
+    pub tags: Option<StringVec>,
     pub single_page: Option<bool>,
     pub created_at: String,
 }

crates/bili_sync_entity/src/entities/watch_later.rs

@@ -2,6 +2,8 @@
 
 use sea_orm::entity::prelude::*;
 
+use crate::rule::Rule;
+
 #[derive(Clone, Debug, PartialEq, DeriveEntityModel, Eq)]
 #[sea_orm(table_name = "watch_later")]
 pub struct Model {
@@ -10,6 +12,7 @@ pub struct Model {
     pub path: String,
     pub created_at: String,
     pub latest_row_at: DateTime,
+    pub rule: Option<Rule>,
     pub enabled: bool,
 }
 

crates/bili_sync_entity/src/lib.rs

@@ -1,2 +1,5 @@
+mod custom_type;
 mod entities;
+
+pub use custom_type::*;
 pub use entities::*;