Implement window data filtration

Cargo.lock

@@ -4,9 +4,9 @@ version = 3
 
 [[package]]
 name = "aho-corasick"
-version = "0.7.20"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cc936419f96fa211c1b9166887b38e5e40b19958e5b895be7c1f93adec7071ac"
+checksum = "67fc08ce920c31afb70f013dcce1bfc3a3195de6a228474e45e1f145b36f8d04"
 dependencies = [
  "memchr",
 ]
@@ -223,6 +223,7 @@ dependencies = [
  "fern",
  "gethostname 0.4.1",
  "log",
+ "regex",
  "serde",
  "serde_default",
  "serde_json",
@@ -1405,9 +1406,9 @@ dependencies = [
 
 [[package]]
 name = "regex"
-version = "1.7.3"
+version = "1.8.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8b1f693b24f6ac912f4893ef08244d70b6067480d2f1a46e950c9691e6749d1d"
+checksum = "af83e617f331cc6ae2da5443c602dfa5af81e517212d9d611a5b3ba1777b5370"
 dependencies = [
  "aho-corasick",
  "memchr",
@@ -1416,9 +1417,9 @@ dependencies = [
 
 [[package]]
 name = "regex-syntax"
-version = "0.6.29"
+version = "0.7.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f162c6dd7b008981e4d40210aca20b4bd0f9b60ca9271061b07f78537722f2e1"
+checksum = "a5996294f19bd3aae0453a862ad728f60e6600695733dd5df01da90c54363a3c"
 
 [[package]]
 name = "reqwest"

Cargo.toml

@@ -20,3 +20,4 @@ toml = "0.7.3"
 dirs = "5.0.0"
 serde = { version = "1.0.160", features = ["derive"] }
 serde_default = "0.1.0"
+regex = "1.8.1"

src/config.rs

@@ -1,11 +1,14 @@
 mod defaults;
 mod file_config;
+mod filters;
 
 use crate::BoxedError;
 use clap::{arg, value_parser, Command};
 use file_config::FileConfig;
 use std::{path::PathBuf, time::Duration};
 
+use self::filters::{Filter, Replacement};
+
 pub struct Config {
     pub port: u32,
     pub host: String,
@@ -14,6 +17,7 @@ pub struct Config {
     pub poll_time_window: Duration,
     pub idle_bucket_name: String,
     pub active_window_bucket_name: String,
+    filters: Vec<Filter>,
 }
 
 impl Config {
@@ -55,6 +59,17 @@ impl Config {
             poll_time_window: config.client.get_poll_time_window(),
             idle_bucket_name,
             active_window_bucket_name,
+            filters: config.client.filters,
         })
     }
+
+    pub fn window_data_replacement(&self, app_id: &str, title: &str) -> Replacement {
+        for filter in &self.filters {
+            if let Some(replacement) = filter.replacement(app_id, title) {
+                return replacement;
+            }
+        }
+
+        Replacement::default()
+    }
 }

src/config/file_config.rs

@@ -9,6 +9,8 @@ use std::{
 
 use crate::{config::defaults, BoxedError};
 
+use super::filters::Filter;
+
 #[derive(Deserialize, DefaultFromSerde)]
 pub struct ServerConfig {
     #[serde(default = "defaults::port")]
@@ -18,6 +20,7 @@ pub struct ServerConfig {
 }
 
 #[derive(Deserialize, DefaultFromSerde)]
+#[serde(rename_all = "kebab-case")]
 pub struct ClientConfig {
     #[serde(default = "defaults::idle_timeout_seconds")]
     idle_timeout_seconds: u32,
@@ -25,6 +28,8 @@ pub struct ClientConfig {
     poll_time_idle_seconds: u32,
     #[serde(default = "defaults::poll_time_window_seconds")]
     poll_time_window_seconds: u32,
+    #[serde(default)]
+    pub filters: Vec<Filter>,
 }
 
 impl ClientConfig {
@@ -83,6 +88,19 @@ impl FileConfig {
 # idle-timeout-seconds={}
 # poll-time-idle-seconds={}
 # poll-time-window-seconds={}
+
+# Add as many filters as needed. The first matching filter stops the replacement.
+# There should be at least 1 match field, and at least 1 replace field.
+# Matches are case sensitive regular expressions between implici ^ and $, e.g.
+# - "." matches 1 any character
+# - ".*" matches any number of any characters
+# - ".+" matches 1 or more any characters.
+# - "word" is an exact match.
+# [[awatcher.filters]]
+# match-app-id = "navigator"
+# match-title = ".*Firefox.*"
+# replace-app-id = "firefox"
+# replace-title = "Unknown"
 "#,
                 defaults::port(),
                 defaults::host(),

src/config/filters.rs

@@ -0,0 +1,79 @@
+use regex::Regex;
+use serde::de::Error;
+use serde::{Deserialize, Deserializer};
+
+#[derive(Debug, Deserialize)]
+#[serde(rename_all = "kebab-case")]
+pub struct Filter {
+    #[serde(default)]
+    #[serde(deserialize_with = "string_to_regex")]
+    match_app_id: Option<Regex>,
+    #[serde(default)]
+    #[serde(deserialize_with = "string_to_regex")]
+    match_title: Option<Regex>,
+    replace_app_id: Option<String>,
+    replace_title: Option<String>,
+}
+
+fn string_to_regex<'de, D>(d: D) -> Result<Option<Regex>, D::Error>
+where
+    D: Deserializer<'de>,
+{
+    let s = <Option<String>>::deserialize(d)?;
+
+    if let Some(s) = s {
+        match format!("^{s}$").parse() {
+            Ok(regex) => Ok(Some(regex)),
+            Err(err) => Err(D::Error::custom(err)),
+        }
+    } else {
+        Ok(None)
+    }
+}
+
+#[derive(Default)]
+pub struct Replacement {
+    pub replace_app_id: Option<String>,
+    pub replace_title: Option<String>,
+}
+
+impl Filter {
+    fn is_valid(&self) -> bool {
+        (self.match_app_id.is_some() || self.match_title.is_some())
+            && (self.replace_app_id.is_some() || self.replace_title.is_some())
+    }
+
+    fn is_match(&self, app_id: &str, title: &str) -> bool {
+        if let Some(match_app_id) = &self.match_app_id {
+            if !match_app_id.is_match(app_id) {
+                return false;
+            };
+        };
+        if let Some(match_title) = &self.match_title {
+            if !match_title.is_match(title) {
+                return false;
+            };
+        };
+
+        true
+    }
+
+    pub fn replacement(&self, app_id: &str, title: &str) -> Option<Replacement> {
+        if !self.is_valid() {
+            return None;
+        }
+
+        if self.is_match(app_id, title) {
+            let mut replacement = Replacement::default();
+            if let Some(new_app_id) = &self.replace_app_id {
+                replacement.replace_app_id = Some(new_app_id.to_string());
+            }
+            if let Some(new_title) = &self.replace_title {
+                replacement.replace_title = Some(new_title.to_string());
+            }
+            Some(replacement)
+        } else {
+            None
+        }
+    }
+}

src/report_client.rs

@@ -49,8 +49,23 @@ impl ReportClient {
 
     pub fn send_active_window(&self, app_id: &str, title: &str) -> Result<(), BoxedError> {
         let mut data = Map::new();
-        data.insert("app".to_string(), Value::String(app_id.to_string()));
+        let mut data_insert = |k: &str, v: String| data.insert(k.to_string(), Value::String(v));
-        data.insert("title".to_string(), Value::String(title.to_string()));
+
+        let replacement = self.config.window_data_replacement(app_id, title);
+        let inserted_app_id = if let Some(new_app_id) = replacement.replace_app_id {
+            trace!("Replacing app_id by {new_app_id}");
+            new_app_id
+        } else {
+            app_id.to_string()
+        };
+        let inserted_title = if let Some(new_title) = replacement.replace_title {
+            trace!("Replacing title of {inserted_app_id} by {new_title}");
+            new_title
+        } else {
+            title.to_string()
+        };
+        data_insert("app", inserted_app_id);
+        data_insert("title", inserted_title);
         let event = AwEvent {
             id: None,
             timestamp: Utc::now(),