melib: don't exclude whitespaces in mime encoded words

1 files changed, 60 insertions, 10 deletions

diff --git a/melib/src/email/compose/mime.rs b/melib/src/email/compose/mime.rs
index fa6b8517..7eddb8c0 100644
--- a/melib/src/email/compose/mime.rs
+++ b/melib/src/email/compose/mime.rs
@@ -1,18 +1,68 @@
 use super::*;
+use crate::grapheme_clusters::Graphemes;
 
 pub fn encode_header(value: &str) -> String {
-    eprintln!("encoding \"{}\"", value);
     let mut ret = String::with_capacity(value.len());
-    for word in value.split_whitespace() {
-        if word.is_ascii() {
-            ret.push_str(word);
-        } else {
-            ret.push_str(
-                format!("=?UTF-8?B?{}?=", BASE64_MIME.encode(word.trim().as_bytes())).trim(),
-            );
+    let graphemes = value.graphemes_indices();
+    let mut is_current_window_ascii = true;
+    let mut current_window_start = 0;
+    for (idx, g) in graphemes {
+        match (g.is_ascii(), is_current_window_ascii) {
+            (true, true) => {
+                ret.push_str(g);
+            }
+            (false, true) => {
+                current_window_start = idx;
+                is_current_window_ascii = false;
+            }
+            (true, false) => {
+                /* If !g.is_whitespace()
+                 *
+                 * Whitespaces inside encoded tokens must be greedily taken,
+                 * instead of splitting each non-ascii word into separate encoded tokens. */
+                if !g.split_whitespace().collect::<Vec<&str>>().is_empty() {
+                    ret.push_str(&format!(
+                        "=?UTF-8?B?{}?=",
+                        BASE64_MIME
+                            .encode(value[current_window_start..idx].as_bytes())
+                            .trim()
+                    ));
+                    if idx != value.len() - 1 {
+                        ret.push(' ');
+                    }
+                    is_current_window_ascii = true;
+                }
+            }
+            /* RFC2047 recommends:
+             * 'While there is no limit to the length of a multiple-line header field, each line of
+             * a header field that contains one or more 'encoded-word's is limited to 76
+             * characters.'
+             * This is a rough compliance.
+             */
+            (false, false) if (((4 * (idx - current_window_start) / 3) + 3) & !3) > 33 => {
+                ret.push_str(&format!(
+                    "=?UTF-8?B?{}?=",
+                    BASE64_MIME
+                        .encode(value[current_window_start..idx].as_bytes())
+                        .trim()
+                ));
+                if idx != value.len() - 1 {
+                    ret.push(' ');
+                }
+                current_window_start = idx;
+            }
+            (false, false) => {}
         }
-        ret.push(' ');
     }
-    ret.pop();
+    /* If the last part of the header value is encoded, it won't be pushed inside the previous for
+     * block */
+    if !is_current_window_ascii {
+        ret.push_str(&format!(
+            "=?UTF-8?B?{}?=",
+            BASE64_MIME
+                .encode(value[current_window_start..].as_bytes())
+                .trim()
+        ));
+    }
     ret
 }