Clean up sanitizer configs

src/config/basic.rs

@@ -1,9 +1,12 @@
 use html5ever::LocalName;
 
-use std::collections::{HashMap, HashSet};
+use crate::config::restricted::RESTRICTED_CONFIG;
+use crate::sanitizer::{Protocol, SanitizerConfig};
 
 lazy_static! {
-    pub static ref ELEMENTS: HashSet<LocalName> = hashset! {
+    pub static ref BASIC_CONFIG: SanitizerConfig = {
+        let mut config = RESTRICTED_CONFIG.clone();
+        config.allowed_elements.extend(hashset! {
             local_name!("a"),
             local_name!("abbr"),
             local_name!("blockquote"),
@@ -26,48 +29,40 @@ lazy_static! {
             local_name!("small"),
             local_name!("strike"),
             local_name!("sub"),
-        local_name!("sup"),
             local_name!("time"),
             local_name!("ul"),
             local_name!("var"),
+        });
+        config.allowed_attributes_per_element.extend(hashmap! {
+            local_name!("a") => hashset! { local_name!("href") },
+            local_name!("abbr") => hashset! { local_name!("title") },
+            local_name!("blockquote") => hashset! { local_name!("cite") },
+            local_name!("dfn") => hashset! { local_name!("title") },
+            local_name!("q") => hashset! { local_name!("cite") },
+            local_name!("time") => hashset! { local_name!("datetime"), LocalName::from("pubdate") },
+        });
+        config.add_attributes_per_element.extend(hashmap! {
+            local_name!("a") => hashmap! { local_name!("rel") => "href" },
+        });
+        config.allowed_protocols.extend(hashmap! {
+            local_name!("a") => hashmap! { local_name!("href") => hashset! {
+                Protocol::Scheme("ftp"),
+                Protocol::Scheme("http"),
+                Protocol::Scheme("https"),
+                Protocol::Scheme("mailto"),
+                Protocol::Relative,
+            }},
+            local_name!("blockquote") => hashmap! { local_name!("cite") => hashset! {
+                Protocol::Scheme("http"),
+                Protocol::Scheme("https"),
+                Protocol::Relative,
+            }},
+            local_name!("q") => hashmap! { local_name!("cite") => hashset! {
+                Protocol::Scheme("http"),
+                Protocol::Scheme("https"),
+                Protocol::Relative,
+            }},
+        });
+        config
     };
-    pub static ref ALL_ATTRIBUTES: HashSet<LocalName> = hashset! {};
-    pub static ref ATTRIBUTES: HashMap<LocalName, HashSet<LocalName>> = hashmap! {
-        local_name!("a") => hashset!{
-            local_name!("href"),
-        },
-        local_name!("abbr") => hashset!{
-            local_name!("title"),
-        },
-        local_name!("blockquote") => hashset!{
-            local_name!("cite"),
-        },
-        local_name!("dfn") => hashset!{
-            local_name!("title"),
-        },
-        local_name!("q") => hashset!{
-            local_name!("cite"),
-        },
-        local_name!("time") => hashset!{
-            local_name!("datetime"),
-            LocalName::from("pubdate"),
-        },
-    };
-    pub static ref ADD_ATTRIBUTES: HashMap<LocalName, HashMap<LocalName, &'static str>> = hashmap! {
-        local_name!("a") => hashmap! {
-            local_name!("rel") => "nofollow",
-        },
-    };
-    pub static ref PROTOCOLS: HashMap<LocalName, HashMap<LocalName, HashSet<&'static str>>> = hashmap! {
-        local_name!("a") => hashmap! {
-            local_name!("href") => hashset!{"ftp", "http", "https", "mailto"},
-        },
-        local_name!("blockquote") => hashmap! {
-            local_name!("cite") => hashset!{"http", "https"},
-        },
-        local_name!("q") => hashmap! {
-            local_name!("cite") => hashset!{"http", "https"},
-        },
-    };
-    pub static ref CSS_PROPERTIES: Vec<String> = vec![];
 }

src/config/mod.rs

@@ -1,4 +1,4 @@
-pub mod default;
 pub mod basic;
+pub mod default;
 pub mod relaxed;
-pub mod permissive;
+pub mod restricted;

src/config/permissive.rs

@@ -1,177 +0,0 @@
-use html5ever::LocalName;
-
-use std::collections::{HashMap, HashSet};
-
-use super::relaxed::{
-    ADD_ATTRIBUTES as RELAXED_ADD_ATTRIBUTES, ALL_ATTRIBUTES as RELAXED_ALL_ATTRIBUTES,
-    ATTRIBUTES as RELAXED_ATTRIBUTES, ELEMENTS as RELAXED_ELEMENTS,
-};
-
-lazy_static! {
-    pub static ref ELEMENTS: HashSet<LocalName> = RELAXED_ELEMENTS
-        .union(&hashset!(
-            local_name!("acronym"),
-            local_name!("basefont"),
-            local_name!("big"),
-            local_name!("blink"),
-            local_name!("center"),
-            LocalName::from("command"),
-            local_name!("dir"),
-            local_name!("font"),
-            local_name!("marquee"),
-            local_name!("strike"),
-            local_name!("tt"),
-            local_name!("form"),
-            local_name!("input"),
-            local_name!("button"),
-            LocalName::from("single"),
-            LocalName::from("double"),
-        ))
-        .into_iter()
-        .cloned()
-        .collect();
-    pub static ref ALL_ATTRIBUTES: HashSet<LocalName> = RELAXED_ALL_ATTRIBUTES
-        .union(&hashset! {
-            local_name!("bgcolor"),
-            local_name!("width"),
-            local_name!("height"),
-            local_name!("border"),
-            local_name!("color"),
-            local_name!("background"),
-        })
-        .into_iter()
-        .cloned()
-        .collect();
-    // Can't figure out how to merge HashMaps :(
-    pub static ref ATTRIBUTES: HashMap<LocalName, HashSet<LocalName>> = hashmap! {
-        local_name!("a") => hashset!{
-            local_name!("href"),
-            local_name!("hreflang"),
-            local_name!("name"),
-            local_name!("rel"),
-        },
-        local_name!("abbr") => hashset!{
-            local_name!("title"),
-        },
-        local_name!("blockquote") => hashset!{
-            local_name!("cite"),
-        },
-        local_name!("button") => hashset!{
-            local_name!("type"),
-        },
-        local_name!("col") => hashset!{
-            local_name!("span"),
-            local_name!("width"),
-        },
-        local_name!("colgroup") => hashset!{
-            local_name!("span"),
-            local_name!("width"),
-        },
-        local_name!("data") => hashset!{
-            local_name!("value"),
-        },
-        local_name!("del") => hashset!{
-            local_name!("cite"),
-            local_name!("datetime"),
-        },
-        local_name!("dfn") => hashset!{
-            local_name!("title"),
-        },
-        local_name!("img") => hashset!{
-            local_name!("align"),
-            local_name!("alt"),
-            local_name!("border"),
-            local_name!("height"),
-            local_name!("src"),
-            local_name!("srcset"),
-            local_name!("width"),
-        },
-        local_name!("input") => hashset!{
-            local_name!("type"),
-            local_name!("name"),
-            local_name!("value"),
-        },
-        local_name!("ins") => hashset!{
-            local_name!("cite"),
-            local_name!("datetime"),
-        },
-        local_name!("li") => hashset!{
-            local_name!("value"),
-        },
-        local_name!("ol") => hashset!{
-            LocalName::from("reversed"),
-            local_name!("start"),
-            local_name!("type"),
-        },
-        local_name!("q") => hashset!{
-            local_name!("cite"),
-        },
-        local_name!("style") => hashset!{
-            local_name!("media"),
-            local_name!("scoped"),
-            local_name!("type"),
-        },
-        local_name!("table") => hashset!{
-            local_name!("align"),
-            local_name!("bgcolor"),
-            local_name!("border"),
-            local_name!("cellpadding"),
-            local_name!("cellspacing"),
-            local_name!("frame"),
-            local_name!("rules"),
-            LocalName::from("sortable"),
-            local_name!("summary"),
-            local_name!("width"),
-        },
-        local_name!("td") => hashset!{
-            local_name!("abbr"),
-            local_name!("align"),
-            local_name!("axis"),
-            local_name!("colspan"),
-            local_name!("headers"),
-            local_name!("rowspan"),
-            local_name!("valign"),
-            local_name!("width"),
-        },
-        local_name!("th") => hashset!{
-            local_name!("abbr"),
-            local_name!("align"),
-            local_name!("axis"),
-            local_name!("colspan"),
-            local_name!("headers"),
-            local_name!("rowspan"),
-            local_name!("scope"),
-            LocalName::from("sorted"),
-            local_name!("valign"),
-            local_name!("width"),
-        },
-        local_name!("time") => hashset!{
-            local_name!("datetime"),
-            LocalName::from("pubdate"),
-        },
-        local_name!("ul") => hashset!{
-            local_name!("type"),
-        },
-    };
-    pub static ref ADD_ATTRIBUTES: HashMap<LocalName, HashMap<LocalName, &'static str>> = RELAXED_ADD_ATTRIBUTES.clone();
-    pub static ref PROTOCOLS: HashMap<LocalName, HashMap<LocalName, HashSet<&'static str>>> = hashmap! {
-        local_name!("a") => hashmap! {
-            local_name!("href") => hashset!{"ftp", "http", "https", "mailto"},
-        },
-        local_name!("blockquote") => hashmap! {
-            local_name!("cite") => hashset!{"http", "https"},
-        },
-        local_name!("del") => hashmap! {
-            local_name!("cite") => hashset!{"http", "https"},
-        },
-        local_name!("img") => hashmap! {
-            local_name!("src") => hashset!{"http", "https"},
-        },
-        local_name!("ins") => hashmap! {
-            local_name!("cite") => hashset!{"http", "https"},
-        },
-        local_name!("q") => hashmap! {
-            local_name!("cite") => hashset!{"http", "https"},
-        },
-    };
-}

src/config/relaxed.rs

@@ -1,20 +1,12 @@
 use html5ever::LocalName;
 
-use std::collections::{HashMap, HashSet};
+use crate::config::basic::BASIC_CONFIG;
-use std::iter::FromIterator;
+use crate::sanitizer::{Protocol, SanitizerConfig};
-
-use super::basic::{
-    ADD_ATTRIBUTES as BASIC_ADD_ATTRIBUTES, ALL_ATTRIBUTES as BASIC_ALL_ATTRIBUTES,
-    ATTRIBUTES as BASIC_ATTRIBUTES, CSS_PROPERTIES as BASIC_CSS_PROPERTIES,
-    ELEMENTS as BASIC_ELEMENTS,
-};
-
-use crate::css_property::CssProperty;
-use crate::css_at_rule::CssAtRule;
 
 lazy_static! {
-    pub static ref ELEMENTS: HashSet<LocalName> = BASIC_ELEMENTS
+    pub static ref RELAXED_CONFIG: SanitizerConfig = {
-        .union(&hashset!(
+        let mut config = BASIC_CONFIG.clone();
+        config.allowed_elements.extend(hashset! {
             local_name!("address"),
             local_name!("article"),
             local_name!("aside"),
@@ -62,11 +54,9 @@ lazy_static! {
             local_name!("title"),
             local_name!("tr"),
             local_name!("wbr"),
-        ))
+        });
-        .into_iter()
+        config.allow_doctype = true;
-        .cloned()
+        config.allowed_attributes.extend(hashset! {
-        .collect();
-    pub static ref ALL_ATTRIBUTES: HashSet<LocalName> = BASIC_ALL_ATTRIBUTES.union(&hashset! {
             local_name!("class"),
             local_name!("dir"),
             local_name!("hidden"),
@@ -76,21 +66,14 @@ lazy_static! {
             local_name!("tabindex"),
             local_name!("title"),
             LocalName::from("translate"),
-    }).into_iter().cloned().collect();
+        });
-    // Can't figure out how to merge HashMaps :(
+        config.allowed_attributes_per_element.extend(hashmap! {
-    pub static ref ATTRIBUTES: HashMap<LocalName, HashSet<LocalName>> = hashmap! {
             local_name!("a") => hashset!{
                 local_name!("href"),
                 local_name!("hreflang"),
                 local_name!("name"),
                 local_name!("rel"),
             },
-        local_name!("abbr") => hashset!{
-            local_name!("title"),
-        },
-        local_name!("blockquote") => hashset!{
-            local_name!("cite"),
-        },
             local_name!("col") => hashset!{
                 local_name!("span"),
                 local_name!("width"),
@@ -106,9 +89,6 @@ lazy_static! {
                 local_name!("cite"),
                 local_name!("datetime"),
             },
-        local_name!("dfn") => hashset!{
-            local_name!("title"),
-        },
             local_name!("img") => hashset!{
                 local_name!("align"),
                 local_name!("alt"),
@@ -130,9 +110,6 @@ lazy_static! {
                 local_name!("start"),
                 local_name!("type"),
             },
-        local_name!("q") => hashset!{
-            local_name!("cite"),
-        },
             local_name!("style") => hashset!{
                 local_name!("media"),
                 local_name!("scoped"),
@@ -172,36 +149,56 @@ lazy_static! {
                 local_name!("valign"),
                 local_name!("width"),
             },
-        local_name!("time") => hashset!{
-            local_name!("datetime"),
-            LocalName::from("pubdate"),
-        },
             local_name!("ul") => hashset!{
                 local_name!("type"),
             },
-    };
+        });
-    pub static ref ADD_ATTRIBUTES: HashMap<LocalName, HashMap<LocalName, &'static str>> = BASIC_ADD_ATTRIBUTES.clone();
+        config.allowed_protocols.extend(hashmap! {
-    pub static ref PROTOCOLS: HashMap<LocalName, HashMap<LocalName, HashSet<&'static str>>> = hashmap! {
+            local_name!("del") => hashmap! { local_name!("cite") => hashset! {
-        local_name!("a") => hashmap! {
+                Protocol::Scheme("http"),
-            local_name!("href") => hashset!{"ftp", "http", "https", "mailto"},
+                Protocol::Scheme("https"),
-        },
+                Protocol::Relative,
-        local_name!("blockquote") => hashmap! {
+            }},
-            local_name!("cite") => hashset!{"http", "https"},
+            local_name!("img") => hashmap! { local_name!("src") => hashset! {
-        },
+                Protocol::Scheme("http"),
-        local_name!("del") => hashmap! {
+                Protocol::Scheme("https"),
-            local_name!("cite") => hashset!{"http", "https"},
+                Protocol::Relative,
-        },
+            }},
-        local_name!("img") => hashmap! {
+            local_name!("ins") => hashmap! { local_name!("cite") => hashset! {
-            local_name!("src") => hashset!{"http", "https"},
+                Protocol::Scheme("http"),
-        },
+                Protocol::Scheme("https"),
-        local_name!("ins") => hashmap! {
+                Protocol::Relative,
-            local_name!("cite") => hashset!{"http", "https"},
+            }},
-        },
+        });
-        local_name!("q") => hashmap! {
+        config.allow_css_comments = true;
-            local_name!("cite") => hashset!{"http", "https"},
+        config.allowed_css_at_rules.extend(hashset! {
-        },
+            css_at_rule!("bottom-center"),
-    };
+            css_at_rule!("bottom-left"),
-    pub static ref CSS_PROPERTIES: HashSet<CssProperty> = vec![
+            css_at_rule!("bottom-left-corner"),
+            css_at_rule!("bottom-right"),
+            css_at_rule!("bottom-right-corner"),
+            css_at_rule!("font-face"),
+            css_at_rule!("left-bottom"),
+            css_at_rule!("left-middle"),
+            css_at_rule!("left-top"),
+            css_at_rule!("page"),
+            css_at_rule!("right-bottom"),
+            css_at_rule!("right-middle"),
+            css_at_rule!("right-top"),
+            css_at_rule!("top-center"),
+            css_at_rule!("top-left"),
+            css_at_rule!("top-left-corner"),
+            css_at_rule!("top-right"),
+            css_at_rule!("top-right-corner"),
+            css_at_rule!("-moz-keyframes"),
+            css_at_rule!("-o-keyframes"),
+            css_at_rule!("-webkit-keyframes"),
+            css_at_rule!("document"),
+            css_at_rule!("keyframes"),
+            css_at_rule!("media"),
+            css_at_rule!("supports"),
+        });
+        config.allowed_css_properties.extend(hashset! {
             css_property!("-moz-appearance"),
             css_property!("-moz-background-inline-policy"),
             css_property!("-moz-box-sizing"),
@@ -845,32 +842,7 @@ lazy_static! {
             css_property!("wrap-through"),
             css_property!("writing-mode"),
             css_property!("z-index"),
-    ].into_iter().collect();
+        });
-    pub static ref CSS_AT_RULES: HashSet<CssAtRule> = vec![
+        config
-        css_at_rule!("bottom-center"),
+    };
-        css_at_rule!("bottom-left"),
-        css_at_rule!("bottom-left-corner"),
-        css_at_rule!("bottom-right"),
-        css_at_rule!("bottom-right-corner"),
-        css_at_rule!("font-face"),
-        css_at_rule!("left-bottom"),
-        css_at_rule!("left-middle"),
-        css_at_rule!("left-top"),
-        css_at_rule!("page"),
-        css_at_rule!("right-bottom"),
-        css_at_rule!("right-middle"),
-        css_at_rule!("right-top"),
-        css_at_rule!("top-center"),
-        css_at_rule!("top-left"),
-        css_at_rule!("top-left-corner"),
-        css_at_rule!("top-right"),
-        css_at_rule!("top-right-corner"),
-        css_at_rule!("-moz-keyframes"),
-        css_at_rule!("-o-keyframes"),
-        css_at_rule!("-webkit-keyframes"),
-        css_at_rule!("document"),
-        css_at_rule!("keyframes"),
-        css_at_rule!("media"),
-        css_at_rule!("supports"),
-    ].into_iter().collect();
 }

src/config/restricted.rs

@@ -0,0 +1,16 @@
+use crate::config::default::DEFAULT_CONFIG;
+use crate::sanitizer::SanitizerConfig;
+
+lazy_static! {
+    pub static ref RESTRICTED_CONFIG: SanitizerConfig = {
+        let mut config = DEFAULT_CONFIG.clone();
+        config.allowed_elements.extend(hashset! {
+            local_name!("b"),
+            local_name!("em"),
+            local_name!("i"),
+            local_name!("strong"),
+            local_name!("u"),
+        });
+        config
+    };
+}

src/main.rs

@@ -27,11 +27,11 @@ mod css_parser;
 mod sanitizer;
 
 use arena_dom::{create_element, Arena, NodeData, Ref};
-use config::default::DEFAULT_CONFIG;
+use config::basic::BASIC_CONFIG;
 use sanitizer::Sanitizer;
 
 fn main() {
-    let sanitizer = Sanitizer::new(&DEFAULT_CONFIG, vec![&add_spacer_elements_around_ul]);
+    let sanitizer = Sanitizer::new(&BASIC_CONFIG, vec![&add_spacer_elements_around_ul]);
     sanitizer
         .sanitize_fragment(&mut io::stdin(), &mut io::stdout())
         .unwrap();