feat: use "inherit left, wildcard right" behavior (#44)

crowlKats · web-flow · commit 218f408d7bea · 2024-07-26T09:34:28.000-07:00
diff --git a/src/canonicalize_and_process.rs b/src/canonicalize_and_process.rs
@@ -270,3 +270,25 @@ pub fn special_scheme_default_port(scheme: &str) -> Option<&'static str> {
     _ => None,
   }
 }
+
+// Ref: https://urlpattern.spec.whatwg.org/#process-a-base-url-string
+pub fn process_base_url(input: &str, kind: &ProcessType) -> String {
+  if kind != &ProcessType::Pattern {
+    input.to_string()
+  } else {
+    escape_pattern_string(input)
+  }
+}
+
+// Ref: https://wicg.github.io/urlpattern/#escape-a-pattern-string
+pub fn escape_pattern_string(input: &str) -> String {
+  assert!(input.is_ascii());
+  let mut result = String::new();
+  for char in input.chars() {
+    if matches!(char, '+' | '*' | '?' | ':' | '{' | '}' | '(' | ')' | '\\') {
+      result.push('\\');
+    }
+    result.push(char);
+  }
+  result
+}
diff --git a/src/component.rs b/src/component.rs
@@ -1,5 +1,6 @@
 // Copyright 2018-2021 the Deno authors. All rights reserved. MIT license.
 
+use crate::canonicalize_and_process::escape_pattern_string;
 use crate::matcher::InnerMatcher;
 use crate::matcher::Matcher;
 use crate::parser::Options;
@@ -262,19 +263,6 @@ fn generate_pattern_string(part_list: &[&Part], options: &Options) -> String {
   result
 }
 
-// Ref: https://wicg.github.io/urlpattern/#escape-a-pattern-string
-fn escape_pattern_string(input: &str) -> String {
-  assert!(input.is_ascii());
-  let mut result = String::new();
-  for char in input.chars() {
-    if matches!(char, '+' | '*' | '?' | ':' | '{' | '}' | '(' | ')' | '\\') {
-      result.push('\\');
-    }
-    result.push(char);
-  }
-  result
-}
-
 /// This function generates a matcher for a given parts list.
 fn generate_matcher<R: RegExp>(
   mut part_list: &[&Part],
diff --git a/src/constructor_parser.rs b/src/constructor_parser.rs
@@ -127,7 +127,11 @@ impl<'a> ConstructorStringParser<'a> {
   }
 
   // Ref: https://wicg.github.io/urlpattern/#change-state
-  fn change_state(&mut self, state: ConstructorStringParserState, skip: usize) {
+  fn change_state(
+    &mut self,
+    new_state: ConstructorStringParserState,
+    skip: usize,
+  ) {
     match self.state {
       ConstructorStringParserState::Protocol => {
         self.result.protocol = Some(self.make_component_string())
@@ -153,10 +157,69 @@ impl<'a> ConstructorStringParser<'a> {
       ConstructorStringParserState::Hash => {
         self.result.hash = Some(self.make_component_string())
       }
-      _ => {}
+      ConstructorStringParserState::Init
+      | ConstructorStringParserState::Authority
+      | ConstructorStringParserState::Done => {}
     }
 
-    self.state = state;
+    if self.state != ConstructorStringParserState::Init
+      && new_state != ConstructorStringParserState::Done
+    {
+      if matches!(
+        self.state,
+        ConstructorStringParserState::Protocol
+          | ConstructorStringParserState::Authority
+          | ConstructorStringParserState::Username
+          | ConstructorStringParserState::Password
+      ) && matches!(
+        new_state,
+        ConstructorStringParserState::Port
+          | ConstructorStringParserState::Pathname
+          | ConstructorStringParserState::Search
+          | ConstructorStringParserState::Hash
+      ) && self.result.hostname.is_none()
+      {
+        self.result.hostname = Some(String::new());
+      }
+
+      if matches!(
+        self.state,
+        ConstructorStringParserState::Protocol
+          | ConstructorStringParserState::Authority
+          | ConstructorStringParserState::Username
+          | ConstructorStringParserState::Password
+          | ConstructorStringParserState::Hostname
+          | ConstructorStringParserState::Port
+      ) && matches!(
+        new_state,
+        ConstructorStringParserState::Search
+          | ConstructorStringParserState::Hash
+      ) && self.result.pathname.is_none()
+      {
+        if self.protocol_matches_special_scheme {
+          self.result.pathname = Some(String::from("/"));
+        } else {
+          self.result.pathname = Some(String::new());
+        }
+      }
+
+      if matches!(
+        self.state,
+        ConstructorStringParserState::Protocol
+          | ConstructorStringParserState::Authority
+          | ConstructorStringParserState::Username
+          | ConstructorStringParserState::Password
+          | ConstructorStringParserState::Hostname
+          | ConstructorStringParserState::Port
+          | ConstructorStringParserState::Pathname
+      ) && new_state == ConstructorStringParserState::Hash
+        && self.result.search.is_none()
+      {
+        self.result.search = Some(String::new());
+      }
+    }
+
+    self.state = new_state;
     self.token_index += skip;
     self.component_start = self.token_index;
     self.token_increment = 0;
@@ -273,11 +336,8 @@ pub(crate) fn parse_constructor_string<R: RegExp>(
           parser.change_state(ConstructorStringParserState::Hash, 1);
         } else if parser.is_search_prefix() {
           parser.change_state(ConstructorStringParserState::Search, 1);
-          parser.result.hash = Some(String::new());
         } else {
           parser.change_state(ConstructorStringParserState::Pathname, 0);
-          parser.result.search = Some(String::new());
-          parser.result.hash = Some(String::new());
         }
         parser.token_index += parser.token_increment;
         continue;
@@ -306,22 +366,12 @@ pub(crate) fn parse_constructor_string<R: RegExp>(
     match parser.state {
       ConstructorStringParserState::Init => {
         if parser.is_protocol_suffix() {
-          parser.result.username = Some(String::new());
-          parser.result.password = Some(String::new());
-          parser.result.hostname = Some(String::new());
-          parser.result.port = Some(String::new());
-          parser.result.pathname = Some(String::new());
-          parser.result.search = Some(String::new());
-          parser.result.hash = Some(String::new());
           parser.rewind_and_set_state(ConstructorStringParserState::Protocol);
         }
       }
       ConstructorStringParserState::Protocol => {
         if parser.is_protocol_suffix() {
           parser.compute_protocol_matches_special_scheme::<R>()?;
-          if parser.protocol_matches_special_scheme {
-            parser.result.pathname = Some(String::from("/"));
-          }
           let mut next_state = ConstructorStringParserState::Pathname;
           let mut skip = 1;
           if parser.next_is_authority_slashes() {
@@ -398,5 +448,10 @@ pub(crate) fn parse_constructor_string<R: RegExp>(
     }
     parser.token_index += parser.token_increment;
   }
+
+  if parser.result.hostname.is_some() && parser.result.port.is_none() {
+    parser.result.port = Some(String::new());
+  }
+
   Ok(parser.result)
 }
diff --git a/src/lib.rs b/src/lib.rs
@@ -18,8 +18,9 @@ mod tokenizer;
 pub use error::Error;
 use url::Url;
 
-use crate::canonicalize_and_process::is_special_scheme;
 use crate::canonicalize_and_process::special_scheme_default_port;
+use crate::canonicalize_and_process::ProcessType;
+use crate::canonicalize_and_process::{is_special_scheme, process_base_url};
 use crate::component::Component;
 use crate::regexp::RegExp;
 
@@ -55,7 +56,7 @@ impl UrlPatternInit {
   #[allow(clippy::too_many_arguments)]
   fn process(
     &self,
-    kind: canonicalize_and_process::ProcessType,
+    kind: ProcessType,
     protocol: Option<String>,
     username: Option<String>,
     password: Option<String>,
@@ -78,18 +79,84 @@ impl UrlPatternInit {
     };
 
     let base_url = if let Some(parsed_base_url) = &self.base_url {
-      // TODO: check if these are correct
-      result.protocol = Some(parsed_base_url.scheme().to_string());
-      result.username = Some(parsed_base_url.username().to_string());
-      result.password =
-        Some(parsed_base_url.password().unwrap_or_default().to_string());
-      result.hostname =
-        Some(parsed_base_url.host_str().unwrap_or_default().to_string());
-      result.port = Some(url::quirks::port(parsed_base_url).to_string());
-      result.pathname =
-        Some(url::quirks::pathname(parsed_base_url).to_string());
-      result.search = Some(parsed_base_url.query().unwrap_or("").to_string());
-      result.hash = Some(parsed_base_url.fragment().unwrap_or("").to_string());
+      if self.protocol.is_none() {
+        result.protocol =
+          Some(process_base_url(parsed_base_url.scheme(), &kind));
+      }
+
+      if kind != ProcessType::Pattern
+        && (self.protocol.is_none()
+          && self.hostname.is_none()
+          && self.port.is_none()
+          && self.username.is_none())
+      {
+        result.username =
+          Some(process_base_url(parsed_base_url.username(), &kind));
+      }
+
+      if kind != ProcessType::Pattern
+        && (self.protocol.is_none()
+          && self.hostname.is_none()
+          && self.port.is_none()
+          && self.username.is_none()
+          && self.password.is_none())
+      {
+        result.password = Some(process_base_url(
+          parsed_base_url.password().unwrap_or_default(),
+          &kind,
+        ));
+      }
+
+      if self.protocol.is_none() && self.hostname.is_none() {
+        result.hostname = Some(process_base_url(
+          parsed_base_url.host_str().unwrap_or_default(),
+          &kind,
+        ));
+      }
+
+      if self.protocol.is_none()
+        && self.hostname.is_none()
+        && self.port.is_none()
+      {
+        result.port =
+          Some(process_base_url(url::quirks::port(parsed_base_url), &kind));
+      }
+
+      if self.protocol.is_none()
+        && self.hostname.is_none()
+        && self.port.is_none()
+        && self.pathname.is_none()
+      {
+        result.pathname = Some(process_base_url(
+          url::quirks::pathname(parsed_base_url),
+          &kind,
+        ));
+      }
+
+      if self.protocol.is_none()
+        && self.hostname.is_none()
+        && self.port.is_none()
+        && self.pathname.is_none()
+        && self.search.is_none()
+      {
+        result.search = Some(process_base_url(
+          parsed_base_url.query().unwrap_or_default(),
+          &kind,
+        ));
+      }
+
+      if self.protocol.is_none()
+        && self.hostname.is_none()
+        && self.port.is_none()
+        && self.pathname.is_none()
+        && self.search.is_none()
+        && self.hash.is_none()
+      {
+        result.hash = Some(process_base_url(
+          parsed_base_url.fragment().unwrap_or_default(),
+          &kind,
+        ));
+      }
 
       Some(parsed_base_url)
     } else {
@@ -235,7 +302,7 @@ impl<R: RegExp> UrlPattern<R> {
     report_regex_errors: bool,
   ) -> Result<Self, Error> {
     let mut processed_init = init.process(
-      canonicalize_and_process::ProcessType::Pattern,
+      ProcessType::Pattern,
       None,
       None,
       None,
@@ -413,7 +480,7 @@ impl<R: RegExp> UrlPattern<R> {
     &self,
     input: UrlPatternMatchInput,
   ) -> Result<Option<UrlPatternResult>, Error> {
-    let input = match crate::quirks::parse_match_input(input) {
+    let input = match quirks::parse_match_input(input) {
       Some(input) => input,
       None => return Ok(None),
     };
@@ -591,9 +658,9 @@ mod tests {
 
   fn test_case(case: TestCase) {
     let input = case.pattern.first().cloned();
-    let mut base_url = case.pattern.get(1).map(|input| match input {
-      StringOrInit::String(str) => str.clone(),
-      StringOrInit::Init(_) => unreachable!(),
+    let mut base_url = case.pattern.get(1).and_then(|input| match input {
+      StringOrInit::String(str) => Some(str.clone()),
+      StringOrInit::Init(_) => None,
     });
 
     println!("\n=====");
@@ -664,7 +731,48 @@ mod tests {
           }) = &input
           {
             expected = Some($field.to_owned())
-          } else if let Some(base_url) = &base_url {
+          } else if {
+            if let StringOrInit::Init(init) = &input {
+              match stringify!($field) {
+                "protocol" => false,
+                "hostname" => init.protocol.is_some(),
+                "port" => init.protocol.is_some() || init.hostname.is_some(),
+                "username" => false,
+                "password" => false,
+                "pathname" => {
+                  init.protocol.is_some()
+                    || init.hostname.is_some()
+                    || init.port.is_some()
+                }
+                "search" => {
+                  init.protocol.is_some()
+                    || init.hostname.is_some()
+                    || init.port.is_some()
+                    || init.pathname.is_some()
+                }
+                "hash" => {
+                  init.protocol.is_some()
+                    || init.hostname.is_some()
+                    || init.port.is_some()
+                    || init.pathname.is_some()
+                    || init.search.is_some()
+                }
+                _ => unreachable!(),
+              }
+            } else {
+              false
+            }
+          } {
+            expected = Some("*".to_owned())
+          } else if let Some(base_url) =
+            base_url.as_ref().and_then(|base_url| {
+              if !matches!(stringify!($field), "username" | "password") {
+                Some(base_url)
+              } else {
+                None
+              }
+            })
+          {
             let base_url = Url::parse(base_url).unwrap();
             let field = url::quirks::$field(&base_url);
             let field: String = match stringify!($field) {
@@ -684,8 +792,8 @@ mod tests {
         let pattern = &pattern.$field.pattern_string;
 
         assert_eq!(
-          pattern,
           &expected,
+          pattern,
           "pattern for {} does not match",
           stringify!($field)
         );
diff --git a/src/testdata/urlpatterntestdata.json b/src/testdata/urlpatterntestdata.json
