From 7d0db3f8ec910c5975b076fbbc70b36f9b20535d Mon Sep 17 00:00:00 2001
From: yggverse <yggverse@project>
Date: Sun, 8 Jun 2025 19:48:10 +0300
Subject: [PATCH] implement `match_time` argument, make some optimizations

---
 Cargo.toml      |  1 +
 README.md       | 16 ++++++++++------
 src/argument.rs |  4 ++++
 src/debug.rs    |  2 +-
 src/main.rs     | 25 ++++++++++++++++++++-----
 5 files changed, 36 insertions(+), 12 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 5101505..3d2fbfb 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -12,4 +12,5 @@ repository = "https://github.com/YGGverse/htcount"
 
 [dependencies]
 anyhow = "1.0"
+chrono = "0.4"
 clap = { version = "4.5", features = ["derive"] }
diff --git a/README.md b/README.md
index d0be592..e720e5c 100644
--- a/README.md
+++ b/README.md
@@ -50,19 +50,22 @@ htcount --source      /var/log/nginx/access.log\
 
         [default: nginx]
 
-    --export-json <EXPORT_JSON>
+--export-json <EXPORT_JSON>
         Export results to JSON file (e.g. `/path/to/stats.json`)
 
-    --export-svg <EXPORT_SVG>
+--export-svg <EXPORT_SVG>
         Export results to SVG file (e.g. `/path/to/badge.svg`)
 
         * use `{hits}` / `{hosts}` pattern to replace parsed values
 
-    --template-svg <TEMPLATE_SVG>
+--template-svg <TEMPLATE_SVG>
         Use custom SVG file template with `{hits}` / `{hosts}` placeholders
 
         [default: default/counter.svg]
 
+-m, --match-time <MATCH_TIME>
+        Filter records match time pattern (e.g. `%d/%b/%Y`)
+
 -c, --capacity <CAPACITY>
         Expected memory index capacity
 
@@ -86,7 +89,6 @@ htcount --source      /var/log/nginx/access.log\
         Print version
 ```
 
-
 ### systemd
 
 ``` /etc/systemd/system/htcount.service
@@ -101,8 +103,9 @@ Type=simple
 ExecStart=/usr/local/bin/htcount --source /var/log/nginx/access.log\
                                  --export-svg /var/www/htcount/visitors.svg\
                                  --template-svg /path/to/default/template.svg\
-                                 --ignore-host 127.0.0.1
-                                 --ignore-host 127.0.0.2
+                                 --ignore-host 127.0.0.1\
+                                 --ignore-host 127.0.0.2\
+                                 --match-time %d/%b/%Y\
                                  --update 3600\
                                  --debug n
 StandardOutput=null
@@ -114,6 +117,7 @@ WantedBy=multi-user.target
 * make sure `/var/www/htcount` exists
 * replace `/path/to/default/template.svg` with your value
 * use `ignore-host` to skip local host requests
+* `match-time %d/%b/%Y` today records only
 
 * `systemctl daemon-reload` - update configuration
 * `systemctl enable` - launch on system startup
diff --git a/src/argument.rs b/src/argument.rs
index bea400c..ac7e383 100644
--- a/src/argument.rs
+++ b/src/argument.rs
@@ -30,6 +30,10 @@ pub struct Argument {
     #[arg(long, default_value_t = String::from("default/counter.svg"))]
     pub template_svg: String,
 
+    /// Filter records match time pattern (e.g. `%d/%b/%Y`)
+    #[arg(short, long)]
+    pub match_time: Option<String>,
+
     /// Expected memory index capacity
     #[arg(short, long, default_value_t = 100)]
     pub capacity: usize,
diff --git a/src/debug.rs b/src/debug.rs
index 69cf562..43847c8 100644
--- a/src/debug.rs
+++ b/src/debug.rs
@@ -1,4 +1,4 @@
-pub fn info(message: String) {
+pub fn info(message: &str) {
     println!("[{}] [info] {message}", now())
 }
 
diff --git a/src/main.rs b/src/main.rs
index f434539..8aa36e8 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -15,6 +15,10 @@ fn main() -> anyhow::Result<()> {
         argument::Argument::parse()
     };
 
+    let match_time = argument
+        .match_time
+        .map(|ref t| chrono::Local::now().format(t).to_string());
+
     // parse some arguments once
     let is_debug_i = argument.debug.contains("i");
     let is_debug_d = argument.debug.contains("d");
@@ -24,12 +28,12 @@ fn main() -> anyhow::Result<()> {
     }
 
     if is_debug_i {
-        debug::info("Crawler started".into());
+        debug::info("Crawler started");
     }
 
     loop {
         if is_debug_i {
-            debug::info("Index queue begin...".into());
+            debug::info("Index queue begin...");
         }
 
         let file = File::open(&argument.source)?;
@@ -38,7 +42,18 @@ fn main() -> anyhow::Result<()> {
         let mut index: HashMap<String, usize> = HashMap::with_capacity(argument.capacity);
 
         'l: for line in reader.lines() {
-            let host = line?
+            let l = line?;
+
+            if let Some(ref t) = match_time {
+                if !l.contains(t) {
+                    if is_debug_d {
+                        debug::info(&format!("Record time mismatch time filter {t}"))
+                    }
+                    continue;
+                }
+            }
+
+            let host = l
                 .split_whitespace()
                 .next()
                 .map(|s| s.into())
@@ -47,7 +62,7 @@ fn main() -> anyhow::Result<()> {
             for h in &argument.ignore_host {
                 if h == &host {
                     if is_debug_d {
-                        debug::info(format!("Host `{h}` ignored by settings"))
+                        debug::info(&format!("Host `{h}` ignored by settings"))
                     }
                     continue 'l;
                 }
@@ -60,7 +75,7 @@ fn main() -> anyhow::Result<()> {
         let hits: usize = index.values().sum();
 
         if is_debug_i {
-            debug::info(format!(
+            debug::info(&format!(
                 "Index queue completed:\n{}\n\thosts: {} / hits: {}, await {} seconds to continue...",
                 if is_debug_d {
                     let mut b = Vec::with_capacity(hosts);