From d9b0eef4dfc3ad8d7a1e8c7b5f1e99b57803520c Mon Sep 17 00:00:00 2001
From: delorean <jackdelmar@protonmail.com>
Date: Thu, 23 May 2024 19:41:09 -0500
Subject: [PATCH] fixed body text boundary truncation

---
 src/common/conf.rs    |  2 +-
 src/common/modules.rs | 19 ++++++++++---------
 2 files changed, 11 insertions(+), 10 deletions(-)

diff --git a/src/common/conf.rs b/src/common/conf.rs
index c596494..189ea09 100644
--- a/src/common/conf.rs
+++ b/src/common/conf.rs
@@ -38,7 +38,7 @@ pub struct Config {
     /// follow redirects
     pub follow: bool,
 
-    #[clap(long = "body")]
+    #[clap(default_value_t = 0, long = "body")]
     /// read n bytes of the response document body
     pub bodysize: usize,
 }
diff --git a/src/common/modules.rs b/src/common/modules.rs
index 886bbd2..1496c4a 100644
--- a/src/common/modules.rs
+++ b/src/common/modules.rs
@@ -17,15 +17,16 @@ pub fn get_title(body: &String) -> String {
 pub fn read_body(body: &String, lim: usize) -> String {
     let document = Document::from(body.as_str());
 
-    let mut text = String::new();
-    for tag in document.find(Name("body")) {
-        text.push_str(&tag.text());
+    let mut bodytext = document.find(Name("body")).next().map(|n| n.text()).unwrap_or_else(|| "".to_string());
+
+    bodytext = fmtwhitespace(bodytext);
+    if bodytext.len() > lim {
+        bodytext = bodytext
+            .char_indices()
+            .take_while(|(i, _)| *i < lim)
+            .map(|(_, c)| c)
+            .collect();
     }
 
-    text = fmtwhitespace(text);
-    if text.len() > lim {
-        text.truncate(lim);
-    }
-
-    parsebody(text)
+    parsebody(bodytext)
 }
\ No newline at end of file