Trang chủ Khám phá Trợ giúp
Đăng nhập
reda
/
crabmail
3
2
Fork 0
Các tập tin Các vấn đề 0 Yêu cầu kéo về 0 Wiki
Browse Source

fix xml converter

Yurii Sokolovskyi 3 tháng trước cách đây
mục cha
327dd39e6b
commit
d534f16d82
5 tập tin đã thay đổi với 69 bổ sung và 58 xóa
Split View Hiển thị tình trạng sai khác
  1. 2 0
      .gitignore
  2. 0 1
      src/main.rs
  3. 3 28
      src/templates/html.rs
  4. 28 0
      src/templates/util.rs
  5. 36 29
      src/templates/xml.rs

+ 2 - 0
.gitignore
Xem Tập Tin 

@@ -1,3 +1,5 @@
 
 /target
 
 site/
 
 
+/.idea/.gitignore
 
+/.idea/vcs.xml

+ 0 - 1
src/main.rs
Xem Tập Tin 

@@ -87,7 +87,6 @@ impl List {
 
         if Config::global().include_html {
 
             write(self.to_html(), &self.out_dir, "html");
 
         }
 
-        write_if_changed(&self.out_dir.join("atom.xml"), self.to_xml());
 
     }
 
 
     // Used with atom

+ 3 - 28
src/templates/html.rs
Xem Tập Tin 

@@ -1,5 +1,5 @@
 
 use std::error::Error;
 
-use super::util::xml_safe as x;
 
+use super::util::{parse_email, xml_safe as x};
 
 use crate::models::*;
 
 use crate::templates::PAGE_SIZE;
 
 use crate::util::*;
 
@@ -208,18 +208,6 @@ impl MailAddress {
 
 }
 
 
 impl Thread {
 
-    fn extract_html_content(&self, part: &ParsedMail) -> Option<String> {
 
-        if part.ctype.mimetype == "text/html" {
 
-            return part.get_body().ok();
 
-        }
 
-        for subpart in &part.subparts {
 
-            if let Some(html) = self.extract_html_content(subpart) {
 
-                return Some(html);
 
-            }
 
-        }
 
-        None
 
-    }
 
-
 
     fn retrieve_attachments(parsed_mail: ParsedMail, dir_parent: PathBuf, file_name: String) -> io::Result<Vec<PathBuf>> {
 
         // lists of all paths to attachments
 
         let mut paths = Vec::new();
 
@@ -256,19 +244,6 @@ impl Thread {
 
         Ok(paths)
 
     }
 
 
-    fn parse_email(&self, parsed_mail: &ParsedMail) -> Result<String, Box<dyn Error>> {
 
-        if let Some(html_content) = self.extract_html_content(&parsed_mail) {
 
-            return Ok(html_content.replace("\r\n", ""));
 
-        } else if parsed_mail.ctype.mimetype.starts_with("text/plain") {
 
-            // Convert plain text to simple HTML
 
-            let plain_text = parsed_mail.get_body()?.replace("\r\n", "<br>");
 
-            let html_content = format!("<html><body>{}</body></html>", plain_text);
 
-            return Ok(html_content);
 
-        }
 
-
 
-        Err(From::from("No HTML or text content found"))
 
-    }
 
-
 
     pub fn to_html(&self, out_dir: PathBuf) -> String {
 
         fn extract_styles(html: &str) -> String {
 
             let style_pattern = Regex::new(r"(?is)<style[^>]*>(.*?)</style>").unwrap();
 
@@ -379,7 +354,7 @@ impl Thread {
 
                 }
 
             );
 
 
-            let data = match fs::read_to_string(&msg.original_path) {
 
+            let data = match fs::read_to_string(&msg.original_path.clone()) {
 
                 Ok(content) => content,
 
                 Err(e) => {
 
                     eprintln!("Error reading file: {}", e);
 
@@ -392,7 +367,7 @@ impl Thread {
 
                 Err(e) => return format!("Error parsing email: {}", e),
 
             };
 
 
-            let html = self.parse_email(&parsed_mail).unwrap_or_else(|_| msg.body.clone());
 
+            let html = parse_email(&parsed_mail).unwrap_or_else(|_| msg.body.clone());
 
             let styles = extract_styles(&html);
 
             let mut body_content = remove_style_tags(&extract_body(&html));
 
             body_content = purple_numbers(&*body_content, "#");

+ 28 - 0
src/templates/util.rs
Xem Tập Tin 

@@ -1,3 +1,6 @@
 
+use std::error::Error;
 
+use mailparse::ParsedMail;
 
+
 
 // less efficient, easier api
 
 pub fn xml_safe(text: &str) -> String {
 
     // note we escape more than we need to
 
@@ -18,3 +21,28 @@ pub fn xml_safe(text: &str) -> String {
 
         })
 
         .collect::<String>()
 
 }
 
+
 
+pub fn extract_html_content(part: &ParsedMail) -> Option<String> {
 
+    if part.ctype.mimetype == "text/html" {
 
+        return part.get_body().ok();
 
+    }
 
+    for subpart in &part.subparts {
 
+        if let Some(html) = extract_html_content(subpart) {
 
+            return Some(html);
 
+        }
 
+    }
 
+    None
 
+}
 
+
 
+pub fn parse_email(parsed_mail: &ParsedMail) -> Result<String, Box<dyn Error>> {
 
+    if let Some(html_content) = extract_html_content(&parsed_mail) {
 
+        return Ok(html_content.replace("\r\n", ""));
 
+    } else if parsed_mail.ctype.mimetype.starts_with("text/plain") {
 
+        // Convert plain text to simple HTML
 
+        let plain_text = parsed_mail.get_body()?.replace("\r\n", "<br>");
 
+        let html_content = format!("<html><body>{}</body></html>", plain_text);
 
+        return Ok(html_content);
 
+    }
 
+
 
+    Err(From::from("No HTML or text content found"))
 
+}

+ 36 - 29
src/templates/xml.rs
Xem Tập Tin 

@@ -1,8 +1,12 @@
 
-use super::util::xml_safe as x;
 
+use std::fs;
 
+use mail_parser::PartType::Html;
 
+use mailparse::parse_mail;
 
+use super::util::{parse_email, xml_safe as x};
 
 use crate::models::*;
 
 use crate::util::unformat_flowed;
 
 // use crate::templates::util::xml_safe;
 
 // use anyhow::{Context, Result};
 
+use scraper::{Html as scraper_html, Selector};
 
 
 fn feed(
 
     feed_title: &str,
 
@@ -58,10 +62,37 @@ fn message(
 
 impl StrMessage {
 
     pub fn to_xml(&self) -> String {
 
         let msg = self;
 
-        let body = match self.flowed {
 
-            true => unformat_flowed(&self.body),
 
-            false => self.body.clone(),
 
+
 
+        let data = match fs::read_to_string(self.original_path.clone()) {
 
+            Ok(content) => content,
 
+            Err(e) => {
 
+                eprintln!("Error reading file: {}", e);
 
+                return "Error loading content".to_string();
 
+            }
 
+        };
 
+        
+        let parsed_mail = match parse_mail(&data.as_bytes()) {
 
+            Ok(mail) => mail,
 
+            Err(e) => return format!("Error parsing email: {}", e),
 
         };
 
+        
+        let html = parse_email(&parsed_mail).unwrap_or_else(|_| msg.body.clone());
 
+        
+        let document = scraper_html::parse_document(&*html);
 
+        let body = Selector::parse("body").unwrap();
 
+        
+        // Traverse the parsed HTML and extract all text from the body
 
+        let text = document.select(&body)
 
+            .next().unwrap()
 
+            .text()
 
+            .collect::<Vec<_>>()
 
+            .join(" ");
 
+        
+        
+        // let body = match self.flowed {
 
+        //     true => unformat_flowed(&self.body),
 
+        //     false => self.body.clone(),
 
+        // };
 
         message(
 
             &x(&msg.subject),
 
             &x(&self.url),
 
@@ -69,31 +100,7 @@ impl StrMessage {
 
             &msg.received.to_rfc3339(),
 
             &x(&msg.from.clone().name.unwrap_or(msg.from.clone().address)),
 
             &x(&msg.from.address),
 
-            &x(&body),
 
-        )
 
-    }
 
-}
 
-
 
-// TODO dedup
 
-impl List {
 
-    pub fn to_xml(&self) -> String {
 
-        let mut entry_list = String::new();
 
-        for msg in &self.recent_messages {
 
-            entry_list.push_str(&msg.to_xml());
 
-        }
 
-        // Sometimes its unclear whether to do stuff like this in models.rs or here. could refactor
 
-        let last_updated = self
 
-            .recent_messages
 
-            .get(0)
 
-            .map(|x| x.received.clone())
 
-            .unwrap_or(crate::util::EPOCH);
 
-        feed(
 
-            &self.config.name,
 
-            &self.url,
 
-            &last_updated.to_rfc3339(),
 
-            &self.config.email,
 
-            &self.config.email,
 
-            &entry_list,
 
+            &x(&text),
 
         )
 
     }
 
 }