feat: htmx-based web ui (#147)

This new web ui is written with htmx and maud for the goal of making the
ui self-contained in the rust project. Which means it requires no
additional build-step for front-end. htmx does most of the heavy lifting
so the new ui is smaller and hopefully easier to maintain than the old.

Currently, the new ui has mostly reached feature parity with the old
inspector ui. New features can be expected in the future.

The new ui is enabled by default. You can disable it the same way by
customizing `RSS_FUNNEL_INSPECTOR_UI` environment variable.

For compatibility reasons, the old ui is still available through the
`/_inspector` route. This route will be removed in a future release. Any
feedback on the new ui is welcome.

Cargo.toml

@@ -35,7 +35,7 @@ blake3 = "1.5.1"
 clap = { version = "4.5.1", features = ["derive", "env"] }
 serde = { version = "1.0.197", features = ["derive", "rc"] }
 serde_yaml = "0.9.32"
-duration-str = { version = "0.7.1", default-features = false, features = ["serde"] }
+duration-str = { version = "0.11.2", default-features = false, features = ["serde"] }
 
 # webserver
 axum-macros = "0.4.1"
@@ -87,6 +87,9 @@ glob-match = "0.2.1"
 # Logging
 tracing = { version = "0.1.40"}
 tracing-subscriber = "0.3.18"
+maud = { version = "0.26.0", features = ["axum"] }
+ammonia = "4.0.0"
+serde_with = "3.9.0"
 
 [patch.crates-io]
 ego-tree = { git = "https://github.com/shouya/ego-tree.git" }

scripts/watch-dev.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+trap "trap - SIGTERM && kill -- -$$" SIGINT SIGTERM EXIT
+
+cargo watch -x build -s 'touch /tmp/.trigger' &
+cargo watch -w /tmp/.trigger -d0 -s 'target/debug/rss-funnel -c ~/.config/rss-funnel-dev/funnel.yaml server' &
+
+wait
+

src/cli.rs

@@ -50,7 +50,7 @@ struct TestConfig {
   source: Option<Url>,
   /// Limit the first N filter steps to run
   #[clap(long, short)]
-  limit_filters: Option<usize>,
+  filter_skip: Option<usize>,
   /// Limit the number of items in the feed
   #[clap(long, short('n'))]
   limit_posts: Option<usize>,
@@ -66,7 +66,7 @@ impl TestConfig {
   fn to_endpoint_param(&self) -> server::EndpointParam {
     server::EndpointParam::new(
       self.source.as_ref().cloned(),
-      self.limit_filters,
+      self.filter_skip,
       self.limit_posts,
       self.base.clone(),
     )

src/client.rs

@@ -22,43 +22,50 @@ struct HttpFixture {
   content: String,
 }
 
+#[serde_with::skip_serializing_none]
 #[derive(
   JsonSchema, Serialize, Deserialize, Debug, Clone, Default, PartialEq, Eq, Hash,
 )]
 pub struct ClientConfig {
   /// The "user-agent" header to send with requests
-  user_agent: Option<String>,
+  #[serde(default)]
+  pub user_agent: Option<String>,
   /// The "accept" header to send with requests
-  accept: Option<String>,
+  #[serde(default)]
+  pub accept: Option<String>,
   /// The "cookie" header to send with requests (Deprecated, specify "cookie" field instead)
-  set_cookie: Option<String>,
+  #[serde(default)]
+  pub set_cookie: Option<String>,
   /// The "cookie" header to send with requests
-  cookie: Option<String>,
+  #[serde(default)]
+  pub cookie: Option<String>,
   /// The "referer" header to send with requests
-  referer: Option<String>,
-  /// The maximum number of cached responses
-  cache_size: Option<usize>,
+  #[serde(default)]
+  pub referer: Option<String>,
   /// Ignore tls error
   #[serde(default)]
-  accept_invalid_certs: bool,
+  pub accept_invalid_certs: bool,
+  /// The maximum number of cached responses
+  #[serde(default)]
+  pub cache_size: Option<usize>,
   /// The maximum time a response is kept in the cache (Format: "4s",
   /// 10m", "1h", "1d")
   #[serde(default)]
   #[serde(deserialize_with = "duration_str::deserialize_option_duration")]
   #[schemars(with = "String")]
-  cache_ttl: Option<Duration>,
+  pub cache_ttl: Option<Duration>,
   /// Request timeout (Format: "4s", "10m", "1h", "1d")
   #[serde(deserialize_with = "duration_str::deserialize_option_duration")]
   #[schemars(with = "String")]
-  timeout: Option<Duration>,
+  pub timeout: Option<Duration>,
   /// Sometimes the feed doesn't report a correct content type, so we
   /// need to override it.
   #[serde(default)]
-  assume_content_type: Option<String>,
+  pub assume_content_type: Option<String>,
   /// The proxy to use for requests
   /// (Format: "http://user:pass@host:port", "socks5://user:pass@host:port")
   #[serde(default)]
-  proxy: Option<String>,
+  pub proxy: Option<String>,
 }
 
 impl ClientConfig {
@@ -124,6 +131,10 @@ impl ClientConfig {
     );
     Ok(client)
   }
+
+  pub fn to_yaml(&self) -> Result<String, ConfigError> {
+    Ok(serde_yaml::to_string(self)?)
+  }
 }
 
 pub struct Client {

src/feed.rs

@@ -40,6 +40,15 @@ pub enum FeedFormat {
   Atom,
 }
 
+impl FeedFormat {
+  pub fn as_str(&self) -> &'static str {
+    match self {
+      FeedFormat::Rss => "rss",
+      FeedFormat::Atom => "atom",
+    }
+  }
+}
+
 impl Feed {
   pub fn format(&self) -> FeedFormat {
     match self {

src/filter.rs

@@ -1,37 +1,57 @@
-mod convert;
-mod full_text;
-mod highlight;
-mod html;
-mod image_proxy;
-mod js;
-mod limit;
-mod magnet;
-mod merge;
-mod note;
-mod sanitize;
-mod select;
-mod simplify_html;
-
-use std::collections::HashMap;
+pub(crate) mod convert;
+pub(crate) mod full_text;
+pub(crate) mod highlight;
+pub(crate) mod html;
+pub(crate) mod image_proxy;
+pub(crate) mod js;
+pub(crate) mod limit;
+pub(crate) mod magnet;
+pub(crate) mod merge;
+pub(crate) mod note;
+pub(crate) mod sanitize;
+pub(crate) mod select;
+pub(crate) mod simplify_html;
+
+use std::collections::{HashMap, HashSet};
 use std::sync::Arc;
 
 use schemars::JsonSchema;
 use serde::{Deserialize, Serialize};
+use serde_with::{formats::CommaSeparator, serde_as, StringWithSeparator};
 use url::Url;
 
 use crate::{
   feed::Feed,
   util::{ConfigError, Error, Result},
 };
 
+#[serde_as]
+#[derive(Clone, Debug, Deserialize)]
+#[serde(transparent)]
+pub struct FilterSkip {
+  #[serde_as(as = "StringWithSeparator::<CommaSeparator, usize>")]
+  indices: HashSet<usize>,
+}
+
+impl FilterSkip {
+  pub(crate) fn upto(n: usize) -> Self {
+    let indices = (0..n).collect::<HashSet<usize>>();
+    Self { indices }
+  }
+
+  pub fn allows_filter(&self, index: usize) -> bool {
+    !self.indices.contains(&index)
+  }
+}
+
 #[derive(Clone)]
 pub struct FilterContext {
   /// The base URL of the application. Used to construct absolute URLs
   /// from a relative path.
   base: Option<Url>,
 
   /// The maximum number of filters to run on this pipeline
-  limit_filters: Option<usize>,
+  filter_skip: Option<FilterSkip>,
 
   /// The extra query parameters passed to the endpoint
   extra_queries: HashMap<String, String>,
@@ -42,15 +62,11 @@ impl FilterContext {
   pub fn new() -> Self {
     Self {
       base: None,
-      limit_filters: None,
+      filter_skip: None,
       extra_queries: HashMap::new(),
     }
   }
 
-  pub fn limit_filters(&self) -> Option<usize> {
-    self.limit_filters
-  }
-
   pub fn base(&self) -> Option<&Url> {
     self.base.as_ref()
   }
@@ -63,8 +79,8 @@ impl FilterContext {
     &self.extra_queries
   }
 
-  pub fn set_limit_filters(&mut self, limit: usize) {
-    self.limit_filters = Some(limit);
+  pub fn set_filter_skip(&mut self, filter_skip: FilterSkip) {
+    self.filter_skip = Some(filter_skip);
   }
 
   pub fn set_base(&mut self, base: Url) {
@@ -74,18 +90,26 @@ impl FilterContext {
   pub fn subcontext(&self) -> Self {
     Self {
       base: self.base.clone(),
-      limit_filters: None,
+      filter_skip: None,
       extra_queries: self.extra_queries.clone(),
     }
   }
 
   pub fn from_param(param: &crate::server::EndpointParam) -> Self {
     Self {
       base: param.base().cloned(),
-      limit_filters: param.limit_filters(),
+      filter_skip: param.filter_skip().cloned(),
       extra_queries: param.extra_queries().clone(),
     }
   }
+
+  pub fn allows_filter(&self, index: usize) -> bool {
+    if let Some(f) = &self.filter_skip {
+      f.allows_filter(index)
+    } else {
+      true
+    }
+  }
 }
 
 #[async_trait::async_trait]
@@ -187,6 +211,10 @@ macro_rules! define_filters {
         }
       }
 
+      pub fn to_yaml(&self) -> Result<String, ConfigError> {
+        Ok(serde_yaml::to_string(self)?)
+      }
+
       pub fn name(&self) -> &'static str {
         match self {
           $(FilterConfig::$variant(_) => paste::paste! {stringify!([<$variant:snake>])},)*

src/filter/js.rs

@@ -17,7 +17,7 @@ use super::{FeedFilter, FeedFilterConfig, FilterContext};
 /// See <a href="https://github.com/shouya/rss-funnel/wiki/JavaScript-API" target="_blank">JavaScript API</a>.
 
 pub struct JsConfig {
-  code: String,
+  pub code: String,
 }
 
 #[derive(
@@ -30,7 +30,7 @@ pub struct JsConfig {
 /// <br><br>
 /// Example: <code>modify_post: post.title += " (modified)";</code>
 pub struct ModifyPostConfig {
-  code: String,
+  pub code: String,
 }
 
 #[derive(
@@ -43,7 +43,7 @@ pub struct ModifyPostConfig {
 /// <br><br>
 /// Example: <code>modify_feed: feed.title.value = "Modified Feed";</code>
 pub struct ModifyFeedConfig {
-  code: String,
+  pub code: String,
 }
 
 pub struct JsFilter {
@@ -83,8 +83,8 @@ impl FeedFilterConfig for JsConfig {
 
   async fn build(self) -> Result<Self::Filter, ConfigError> {
     let runtime = Runtime::new().await?;
-    runtime.eval(&self.code).await?;
-    runtime.eval(MODIFY_POSTS_CODE).await?;
+    let () = runtime.eval(&self.code).await?;
+    let () = runtime.eval(MODIFY_POSTS_CODE).await?;
 
     Ok(Self::Filter { runtime })
   }

src/filter/merge.rs

@@ -7,7 +7,7 @@ use serde::{Deserialize, Serialize};
 use crate::client::{Client, ClientConfig};
 use crate::feed::Feed;
 use crate::filter_pipeline::{FilterPipeline, FilterPipelineConfig};
-use crate::source::{Source, SourceConfig};
+use crate::source::{SimpleSourceConfig, Source};
 use crate::util::{ConfigError, Result, SingleOrVec};
 
 use super::{FeedFilter, FeedFilterConfig, FilterContext};
@@ -28,24 +28,24 @@ pub enum MergeConfig {
 )]
 #[serde(transparent)]
 pub struct MergeSimpleConfig {
-  source: SingleOrVec<SourceConfig>,
+  pub source: SingleOrVec<SimpleSourceConfig>,
 }
 
 #[derive(
   JsonSchema, Serialize, Deserialize, Clone, Debug, PartialEq, Eq, Hash,
 )]
 pub struct MergeFullConfig {
   /// Source configuration
-  source: SingleOrVec<SourceConfig>,
+  pub source: SingleOrVec<SimpleSourceConfig>,
   /// Number of concurrent requests to make for fetching multiple sources (default: 20)
   #[serde(default)]
-  parallelism: Option<usize>,
+  pub parallelism: Option<usize>,
   /// Client configuration
   #[serde(default)]
-  client: Option<ClientConfig>,
+  pub client: Option<ClientConfig>,
   /// Filters to apply to the merged feed
   #[serde(default)]
-  filters: Option<FilterPipelineConfig>,
+  pub filters: Option<FilterPipelineConfig>,
 }
 
 impl From<MergeSimpleConfig> for MergeFullConfig {

src/filter_pipeline.rs

@@ -14,7 +14,7 @@ use crate::{
 )]
 #[serde(transparent)]
 pub struct FilterPipelineConfig {
-  filters: Vec<FilterConfig>,
+  pub filters: Vec<FilterConfig>,
 }
 
 impl From<Vec<FilterConfig>> for FilterPipelineConfig {
@@ -88,16 +88,11 @@ impl Inner {
     mut context: FilterContext,
     mut feed: Feed,
   ) -> Result<Feed> {
-    let limit_filters = context
-      .limit_filters()
-      .unwrap_or_else(|| self.num_filters());
-    for filter in self.filters.iter().take(limit_filters) {
-      feed = filter.run(&mut context, feed).await?;
+    for (i, filter) in self.filters.iter().enumerate() {
+      if context.allows_filter(i) {
+        feed = filter.run(&mut context, feed).await?;
+      }
     }
     Ok(feed)
   }
-
-  fn num_filters(&self) -> usize {
-    self.filters.len()
-  }
 }