Fixes and reorganization. Release 0.1.0 maybe soon

Scripter17 · Feb 26, 2024 · 63bb0a5 · 63bb0a5
1 parent 0feb2ce
commit 63bb0a5
Show file tree

Hide file tree

Showing 16 changed files with 336 additions and 275 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/default-config.json b/default-config.json
@@ -1,4 +1,16 @@
 {
+	"docs": {
+		"flags": {
+			"no-unmangle": "...",
+			"...": ["..."]
+		},
+		"vars": {
+			"...": [
+				"...",
+				"...2"
+			]
+		}
+	},
 	"rules": [
 		{
 			"RepeatUntilNonePass": {
@@ -108,19 +120,55 @@
 
 
 		{
-			"condition": {"PartIs": {"part": "Scheme", "value": "http"}},
-			"mapper": {"SetPart": {"part": "Scheme", "value": "https"}}
+			"comment": "https://abc.com/https://abc.com/user",
+			"condition": {"All": [
+				{"Not": {"FlagIsSet": "no-unmangle"}},
+				{"Any": [
+					{"PartContains": {"part": "Path", "where": "Start", "value": "/http:"}},
+					{"PartContains": {"part": "Path", "where": "Start", "value": "/https:"}}
+				]}
+			]},
+			"mapper": {"SetPart": {
+				"part": "Whole",
+				"value": {"Modified": {"source": {"Part": "Path"}, "modification": {"StripPrefix": "/"}}}
+			}}
+		},
+		{
+			"comment": "https://abc.com/profile/https://abc.com/profile/user",
+			"comment2": "The general case requires conditions and mappers to talk to each other and would break the WayBack machine.",
+			"condition": {"All": [
+				{"Not": {"FlagIsSet": "no-unmangle"}},
+				{"Any": [
+					{"PartIs": {"part": {"PathSegment": 1}, "value": "http:"}},
+					{"PartIs": {"part": {"PathSegment": 1}, "value": "https:"}}
+				]}
+			]},
+			"mapper": {"SetPart": {
+				"part": "Whole",
+				"value": {"Part": {"PartSegments": {"part": "Path", "split": "/", "start": 2, "end": null}}}
+			}}
 		},
 
+		{
+			"comment": "https://profile.abc.com.abc.com",
+			"condition": {"All": [
+				{"Not": {"FlagIsSet": "no-unmangle"}},
+				{"TreatErrorAsFail": {"PartContains": {
+					"part": {"PartSegments": {"part": "Domain", "split": ".", "start": 1, "end": null}},
+					"where": "Start",
+					"value": {"Join": {"sources": [{"Part": "NotSubdomain"}, {"String": "."}]}}
+				}}}
+			]},
+			"mapper": {"SetPart": {"part": "NotSubdomain", "value": null}}
+		},
 
 
 		{
-			"condition": {"TreatErrorAsFail": {"PartContains": {
-				"part": {"PartSegments": {"part": "Domain", "split": ".", "start": 1, "end": null}},
-				"where": "Start",
-				"value": {"Join": {"sources": [{"Part": "NotSubdomain"}, {"String": "."}]}}
-			}}},
-			"mapper": {"SetPart": {"part": "NotSubdomain", "value": null}}
+			"condition": {"All": [
+				{"Not": {"FlagIsSet": "no-https"}},
+				{"PartIs": {"part": "Scheme", "value": "http"}}
+			]},
+			"mapper": {"SetPart": {"part": "Scheme", "value": "https"}}
 		},
 
 
@@ -132,7 +180,7 @@
 			]},
 			"mapper": {"All": [
 				{"SetPart": {"part": "Path", "value": "/users"}},
-				{"CopyPart": {"from": {"QueryParam": "id"}, "none_to_empty_string": false, "to": "NextPathSegment"}},
+				{"CopyPart": {"from": {"QueryParam": "id"}, "from_none_to_empty_string": false, "to": "NextPathSegment"}},
 				"RemoveQuery"
 			]}
 		},

diff --git a/src/lib.rs b/src/lib.rs
@@ -10,10 +10,9 @@ pub mod rules;
 pub mod glue;
 /// Types that don't fit in the other modules.
 pub mod types;
-/// Deserializing and handling configuration.
-pub mod config;
+pub(crate) mod util;
 
-/// Takes a URL, an optional [`config::Config`], an optional [`config::Params`], and returns the result of applying the config and params to the URL.
+/// Takes a URL, an optional [`types::Config`], an optional [`types::Params`], and returns the result of applying the config and params to the URL.
 /// This function's name is set to `clean_url` in WASM for API simplicity.
 /// # Errors
 /// If the config or params can't be parsed, returns the parsing error.
@@ -25,30 +24,30 @@ pub fn wasm_clean_url(url: &str, config: wasm_bindgen::JsValue, params: wasm_bin
     Ok(JsValue::from_str(url.as_str()))
 }
 
-/// Takes a URL, an optional [`config::Config`], an optional [`config::Params`], and returns the result of applying the config and params to the URL.
+/// Takes a URL, an optional [`types::Config`], an optional [`types::Params`], and returns the result of applying the config and params to the URL.
 /// # Errors
 /// If applying the rules returns an error, that error is returned.
-pub fn clean_url(url: &mut Url, config: Option<&config::Config>, params: Option<&config::Params>) -> Result<(), types::CleaningError> {
+pub fn clean_url(url: &mut Url, config: Option<&types::Config>, params: Option<&types::Params>) -> Result<(), types::CleaningError> {
     let mut config=match config {
         Some(config) => config.clone(),
-        None => config::Config::get_default()?.clone()
+        None => types::Config::get_default()?.clone()
     };
     if let Some(params) = params {config.params.merge(params.clone());}
     config.apply(url)?;
     Ok(())
 }
 
-fn js_value_to_config(config: wasm_bindgen::JsValue) -> Result<Cow<'static, config::Config>, JsError> {
+fn js_value_to_config(config: wasm_bindgen::JsValue) -> Result<Cow<'static, types::Config>, JsError> {
     Ok(if config.is_null() {
-        Cow::Borrowed(config::Config::get_default()?)
+        Cow::Borrowed(types::Config::get_default()?)
     } else {
         Cow::Owned(serde_wasm_bindgen::from_value(config)?)
     })
 }
 
-fn js_value_to_params(params: wasm_bindgen::JsValue) -> Result<config::Params, JsError> {
+fn js_value_to_params(params: wasm_bindgen::JsValue) -> Result<types::Params, JsError> {
     Ok(if params.is_null() {
-        config::Params::default()
+        types::Params::default()
     } else {
         serde_wasm_bindgen::from_value(params)?
     })

diff --git a/src/main.rs b/src/main.rs
@@ -11,7 +11,7 @@ use url::Url;
 mod rules;
 mod glue;
 mod types;
-mod config;
+mod util;
 
 #[derive(Parser)]
 struct Args {
@@ -24,28 +24,28 @@ struct Args {
     flag: Vec<String>
 }
 
-impl TryFrom<Args> for (Vec<Url>, config::Config) {
-    type Error=config::GetConfigError;
+impl TryFrom<Args> for (Vec<Url>, types::Config) {
+    type Error=types::GetConfigError;
 
     fn try_from(args: Args) -> Result<Self, Self::Error> {
-        let mut config=config::Config::get_default_or_load(args.config.as_deref())?.into_owned();
+        let mut config=types::Config::get_default_or_load(args.config.as_deref())?.into_owned();
         config.params.merge(
             #[allow(clippy::needless_update)]
-            config::Params {
+            types::Params {
                 vars: args.var
                     .into_iter()
                     .filter_map(|mut kev| kev.find('=').map(|e| {let mut v=kev.split_off(e); v.drain(..1); kev.shrink_to_fit(); (kev, v)}))
                     .collect(),
                 flags: args.flag.into_iter().collect(),
-                ..config::Params::default()
+                ..types::Params::default()
             }
         );
         Ok((args.urls, config))
     }
 }
 
 fn main() -> Result<(), types::CleaningError> {
-    let (urls, config): (Vec<Url>, config::Config)=Args::parse().try_into()?;
+    let (urls, config): (Vec<Url>, types::Config)=Args::parse().try_into()?;
 
     for mut url in urls {
         match config.apply(&mut url) {

diff --git a/src/rules.rs b/src/rules.rs
@@ -11,7 +11,8 @@ mod conditions;
 pub use conditions::*;
 mod mappers;
 pub use mappers::*;
-use crate::config;
+
+pub use crate::types::*;
 
 /// The core unit describing when and how URLs are modified.
 #[derive(Debug, Serialize, Deserialize, Clone, PartialEq)]
@@ -21,7 +22,7 @@ pub enum Rule {
     /// # Examples
     /// ```
     /// # use url_cleaner::rules::{Rule, Mapper};
-    /// # use url_cleaner::config::Params;
+    /// # use url_cleaner::types::Params;
     /// # use url::Url;
     /// # use std::collections::HashMap;
     /// let rule=Rule::HostMap(HashMap::from_iter([
@@ -43,7 +44,7 @@ pub enum Rule {
     /// # Examples
     /// ```
     /// # use url_cleaner::rules::{Rule, Condition, Mapper};
-    /// # use url_cleaner::config::Params;
+    /// # use url_cleaner::types::Params;
     /// # use url_cleaner::types::UrlPart;
     /// # use url::Url;
     /// # use std::str::FromStr;
@@ -54,8 +55,8 @@ pub enum Rule {
     ///             condition: Condition::Always,
     ///             mapper: Mapper::SetPart {
     ///                 part: UrlPart::NextPathSegment,
-    ///                 none_to_empty_string: false,
-    ///                 value: Some(FromStr::from_str("a").unwrap())
+    ///                 value: Some(FromStr::from_str("a").unwrap()),
+    ///                 value_none_to_empty_string: false
     ///             }
     ///         }
     ///     ],
@@ -77,7 +78,7 @@ pub enum Rule {
     /// # Examples
     /// ```
     /// # use url_cleaner::rules::{Rule, Condition, Mapper};
-    /// # use url_cleaner::config::Params;
+    /// # use url_cleaner::types::Params;
     /// # use url::Url;
     /// assert!(Rule::Normal{condition: Condition::Never, mapper: Mapper::None}.apply(&mut Url::parse("https://example.com").unwrap(), &Params::default()).is_err());
     /// ```
@@ -119,7 +120,7 @@ impl Rule {
     /// If the rule is a [`Self::Normal`] and the contained condition or mapper returns an error, that error is returned.
     /// If the rule is a [`Self::HostMap`] and the provided URL doesn't have a host, returns the error [`RuleError::UrlHasNoHost`].
     /// If the rule is a [`Self::HostMap`] and the provided URL's host isn't in the rule's map, returns the error [`RuleError::HostNotInMap`].
-    pub fn apply(&self, url: &mut Url, params: &config::Params) -> Result<(), RuleError> {
+    pub fn apply(&self, url: &mut Url, params: &Params) -> Result<(), RuleError> {
         match self {
             Self::Normal{condition, mapper} => if condition.satisfied_by(url, params)? {
                 mapper.apply(url, params)?;
@@ -175,7 +176,7 @@ impl Rules {
     /// If an error is returned, `url` is left unmodified.
     /// # Errors
     /// If the error [`RuleError::FailedCondition`], [`RuleError::UrlHasNoHost`], or [`RuleError::HostNotInMap`] is encountered, it is ignored.
-    pub fn apply(&self, url: &mut Url, params: &config::Params) -> Result<(), RuleError> {
+    pub fn apply(&self, url: &mut Url, params: &Params) -> Result<(), RuleError> {
         let mut temp_url=url.clone();
         for rule in &**self {
             match rule.apply(&mut temp_url, params) {