download file parse filename add RFC 5987 support

support percent-encoded filename

Signed-off-by: zuisong <[email protected]>

Author: zuisong

Cargo.lock

@@ -60,6 +60,7 @@ log = "0.4.21"
 rustls = { version = "0.23.25", optional = true, default-features = false, features = ["logging"] }
 tracing = { version = "0.1.41", default-features = false, features = ["log"] }
 reqwest_cookie_store = { version = "0.8.0", features = ["serde"] }
+percent-encoding = "2.3.1"
 
 [dependencies.reqwest]
 version = "0.12.3"

Cargo.toml

@@ -0,0 +1,127 @@
+use percent_encoding::percent_decode;
+
+/// Parse filename from Content-Disposition header
+/// Prioritizes filename* parameter if present, otherwise uses filename parameter
+pub fn parse_filename_from_content_disposition(content_disposition: &str) -> Option<String> {
+    let parts: Vec<&str> = content_disposition
+        .split(';')
+        .map(|part| part.trim())
+        .collect();
+
+    // First try to find filename* parameter
+    for part in parts.iter() {
+        if part.starts_with("filename*=") {
+            if let Some(filename) = parse_encoded_filename(part) {
+                return Some(filename);
+            }
+        }
+    }
+
+    // If filename* is not found or parsing failed, try regular filename parameter
+    for part in parts {
+        if part.starts_with("filename=") {
+            return parse_regular_filename(part);
+        }
+    }
+
+    None
+}
+
+/// Parse regular filename parameter
+/// Handles both quoted and unquoted filenames
+fn parse_regular_filename(part: &str) -> Option<String> {
+    let filename = part.trim_start_matches("filename=");
+    // Remove quotes if present
+    //
+    // Content-Disposition: attachment; filename="file with \"quotes\".txt"  // This won't occur
+    // Content-Disposition: attachment; filename*=UTF-8''file%20with%20quotes.txt  // This is the actual practice
+    //
+    // We don't need to handle escaped characters in Content-Disposition header parsing because:
+    //
+    // It's not a standard practice
+    // It rarely occurs in real-world scenarios
+    // When filenames contain special characters, they should use the filename* parameter
+    let filename = if filename.starts_with('"') && filename.ends_with('"') {
+        &filename[1..(filename.len() - 1)]
+    } else {
+        filename
+    };
+
+    if filename.is_empty() {
+        return None;
+    }
+
+    Some(filename.to_string())
+}
+
+/// Parse RFC 5987 encoded filename (filename*)
+/// Format: charset'language'encoded-value
+fn parse_encoded_filename(part: &str) -> Option<String> {
+    // Remove "filename*=" prefix
+    let content = part.trim_start_matches("filename*=");
+
+    // According to RFC 5987, format should be: charset'language'encoded-value
+    let parts: Vec<&str> = content.splitn(3, '\'').collect();
+    if parts.len() != 3 {
+        return None;
+    }
+
+    let encoded_filename = parts[2];
+
+    // Decode using percent-encoding
+    let decoded = percent_decode(encoded_filename.as_bytes())
+        .decode_utf8()
+        .ok()?;
+
+    Some(decoded.into_owned())
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_simple_filename() {
+        let header = r#"attachment; filename="example.pdf""#;
+        assert_eq!(
+            parse_filename_from_content_disposition(header),
+            Some("example.pdf".to_string())
+        );
+    }
+
+    #[test]
+    fn test_filename_without_quotes() {
+        let header = "attachment; filename=example.pdf";
+        assert_eq!(
+            parse_filename_from_content_disposition(header),
+            Some("example.pdf".to_string())
+        );
+    }
+
+    #[test]
+    fn test_encoded_filename() {
+        // UTF-8 encoded Chinese filename "测试.pdf"
+        let header = "attachment; filename*=UTF-8''%E6%B5%8B%E8%AF%95.pdf";
+        assert_eq!(
+            parse_filename_from_content_disposition(header),
+            Some("测试.pdf".to_string())
+        );
+    }
+
+    #[test]
+    fn test_both_filenames() {
+        // When both filename and filename* are present, filename* should be preferred
+        let header =
+            r#"attachment; filename="fallback.pdf"; filename*=UTF-8''%E6%B5%8B%E8%AF%95.pdf"#;
+        assert_eq!(
+            parse_filename_from_content_disposition(header),
+            Some("测试.pdf".to_string())
+        );
+    }
+
+    #[test]
+    fn test_no_filename() {
+        let header = "attachment";
+        assert_eq!(parse_filename_from_content_disposition(header), None);
+    }
+}

src/content_disposition.rs

@@ -3,6 +3,9 @@ use std::io::{self, ErrorKind, IsTerminal};
 use std::path::{Path, PathBuf};
 use std::time::Instant;
 
+use crate::content_disposition;
+use crate::decoder::{decompress, get_compression_type};
+use crate::utils::{copy_largebuf, test_pretend_term, HeaderValueExt};
 use anyhow::{anyhow, Context, Result};
 use indicatif::{HumanBytes, ProgressBar, ProgressStyle};
 use mime2ext::mime2ext;
@@ -13,9 +16,6 @@ use reqwest::{
     StatusCode,
 };
 
-use crate::decoder::{decompress, get_compression_type};
-use crate::utils::{copy_largebuf, test_pretend_term, HeaderValueExt};
-
 fn get_content_length(headers: &HeaderMap) -> Option<u64> {
     headers
         .get(CONTENT_LENGTH)
@@ -27,20 +27,12 @@ fn get_content_length(headers: &HeaderMap) -> Option<u64> {
 // of PathBufs
 fn get_file_name(response: &Response, orig_url: &reqwest::Url) -> String {
     fn from_header(response: &Response) -> Option<String> {
-        let quoted = Regex::new("filename=\"([^\"]*)\"").unwrap();
-        // Alternative form:
-        let unquoted = Regex::new("filename=([^;=\"]*)").unwrap();
-        // TODO: support "filename*" version
-
         let header = response
             .headers()
             .get(CONTENT_DISPOSITION)?
             .to_utf8_str()
             .ok()?;
-        let caps = quoted
-            .captures(header)
-            .or_else(|| unquoted.captures(header))?;
-        Some(caps[1].to_string())
+        content_disposition::parse_filename_from_content_disposition(header)
     }
 
     fn from_url(url: &reqwest::Url) -> Option<String> {

src/download.rs

@@ -2,6 +2,7 @@
 mod auth;
 mod buffer;
 mod cli;
+mod content_disposition;
 mod decoder;
 mod download;
 mod error_reporting;

src/main.rs

@@ -119,6 +119,77 @@ fn download_supplied_unicode_filename() {
     );
 }
 
+#[test]
+fn download_support_filename_rfc_5987() {
+    let dir = tempdir().unwrap();
+    let server = server::http(|_req| async move {
+        hyper::Response::builder()
+            .header(
+                "Content-Disposition",
+                r#"attachment; filename*=UTF-8''abcd1234.txt"#,
+            )
+            .body("file".into())
+            .unwrap()
+    });
+
+    get_command()
+        .args(["--download", &server.base_url()])
+        .current_dir(&dir)
+        .assert()
+        .success();
+    assert_eq!(
+        fs::read_to_string(dir.path().join("abcd1234.txt")).unwrap(),
+        "file"
+    );
+}
+#[test]
+fn download_support_filename_rfc_5987_percent_encoded() {
+    let dir = tempdir().unwrap();
+    let server = server::http(|_req| async move {
+        hyper::Response::builder()
+            .header(
+                "Content-Disposition",
+                r#"attachment; filename*=UTF-8''%E6%B5%8B%E8%AF%95.txt"#,
+            )
+            .body("file".into())
+            .unwrap()
+    });
+
+    get_command()
+        .args(["--download", &server.base_url()])
+        .current_dir(&dir)
+        .assert()
+        .success();
+    assert_eq!(
+        fs::read_to_string(dir.path().join("测试.txt")).unwrap(),
+        "file"
+    );
+}
+
+#[test]
+fn download_filename_star_with_high_priority() {
+    let dir = tempdir().unwrap();
+    let server = server::http(|_req| async move {
+        hyper::Response::builder()
+            .header(
+                "Content-Disposition",
+                r#"attachment; filename="fallback.txt"; filename*=UTF-8''%E6%B5%8B%E8%AF%95.txt"#,
+            )
+            .body("file".into())
+            .unwrap()
+    });
+
+    get_command()
+        .args(["--download", &server.base_url()])
+        .current_dir(&dir)
+        .assert()
+        .success();
+    assert_eq!(
+        fs::read_to_string(dir.path().join("测试.txt")).unwrap(),
+        "file"
+    );
+}
+
 #[test]
 fn download_supplied_unquoted_filename() {
     let dir = tempdir().unwrap();