File deduplication

Cargo.toml

@@ -110,6 +110,7 @@ toml = "0.8"
 url = "2"
 uuid = { version = "1", features = ["serde", "v4"] }
 webpki-roots = "0.26.7"
+blake3 = "1.5.5"
 
 [dev-dependencies]
 anyhow = { workspace = true, features = ["backtrace"] } # Enable `backtrace` feature in tests.

deltachat-ffi/deltachat.h

@@ -4756,6 +4756,31 @@ void            dc_msg_set_override_sender_name(dc_msg_t* msg, const char* name)
 void            dc_msg_set_file               (dc_msg_t* msg, const char* file, const char* filemime);
 
 
+/**
+ * Sets the file associated with a message.
+ *
+ * If `name` is non-null, it is used as the file name
+ * and the actual current name of the file is ignored.
+ *
+ * If the source file is already in the blobdir, it will be renamed,
+ * otherwise it will be copied to the blobdir first.
+ *
+ * In order to deduplicate files that contain the same data,
+ * the file will be named as a hash of the file data.
+ *
+ * NOTE:
+ * - This function will rename the file. To get the new file path, call `get_file()`.
+ * - The file must not be modified after this function was called.
+ *
+ * @memberof dc_msg_t
+ * @param msg The message object. Must not be NULL.
+ * @param file The path of the file to attach. Must not be NULL.
+ * @param name The original filename of the attachment. If NULL, the current name of `file` will be used instead.
+ * @param filemime The MIME type of the file. NULL if you don't know or don't care.
+ */
+void            dc_msg_set_file_and_deduplicate(dc_msg_t* msg, const char* file, const char* name, const char* filemime);
+
+
 /**
  * Set the dimensions associated with message object.
  * Typically this is the width and the height of an image or video associated using dc_msg_set_file().

deltachat-ffi/src/lib.rs

@@ -3835,6 +3835,33 @@ pub unsafe extern "C" fn dc_msg_set_file(
     )
 }
 
+#[no_mangle]
+pub unsafe extern "C" fn dc_msg_set_file_and_deduplicate(
+    msg: *mut dc_msg_t,
+    file: *const libc::c_char,
+    name: *const libc::c_char,
+    filemime: *const libc::c_char,
+) {
+    if msg.is_null() || file.is_null() {
+        eprintln!("ignoring careless call to dc_msg_set_file_and_deduplicate()");
+        return;
+    }
+    let ffi_msg = &mut *msg;
+    let ctx = &*ffi_msg.context;
+
+    ffi_msg
+        .message
+        .set_file_and_deduplicate(
+            ctx,
+            as_path(file),
+            to_opt_string_lossy(name).as_deref(),
+            to_opt_string_lossy(filemime).as_deref(),
+        )
+        .context("Failed to set file")
+        .log_err(&*ffi_msg.context)
+        .ok();
+}
+
 #[no_mangle]
 pub unsafe extern "C" fn dc_msg_set_dimension(
     msg: *mut dc_msg_t,

python/src/deltachat/message.py

@@ -108,7 +108,9 @@ def set_html(self, html_text):
 
     @props.with_doc
     def filename(self):
-        """filename if there was an attachment, otherwise empty string."""
+        """file path if there was an attachment, otherwise empty string.
+        If you want to get the file extension or a user-visible string,
+        use `basename` instead."""
         return from_dc_charpointer(lib.dc_msg_get_file(self._dc_msg))
 
     def set_file(self, path, mime_type=None):
@@ -120,7 +122,8 @@ def set_file(self, path, mime_type=None):
 
     @props.with_doc
     def basename(self) -> str:
-        """basename of the attachment if it exists, otherwise empty string."""
+        """The user-visible name of the attachment (incl. extension)
+        if it exists, otherwise empty string."""
         # FIXME, it does not return basename
         return from_dc_charpointer(lib.dc_msg_get_filename(self._dc_msg))
 

python/tests/test_1_online.py

@@ -181,15 +181,16 @@ def send_and_receive_message():
     msg = send_and_receive_message()
     assert msg.text == "withfile"
     assert open(msg.filename).read() == "some data"
-    msg.filename.index(basename)
-    assert msg.filename.endswith(ext)
+    msg.basename.index(basename)
+    assert msg.basename.endswith(ext)
 
     msg2 = send_and_receive_message()
     assert msg2.text == "withfile"
     assert open(msg2.filename).read() == "some data"
-    msg2.filename.index(basename)
-    assert msg2.filename.endswith(ext)
-    assert msg.filename != msg2.filename
+    msg2.basename.index(basename)
+    assert msg2.basename.endswith(ext)
+    assert msg.filename == msg2.filename  # The file is deduplicated
+    assert msg.basename == msg2.basename
 
 
 def test_send_file_html_attachment(tmp_path, acfactory, lp):
@@ -214,8 +215,8 @@ def test_send_file_html_attachment(tmp_path, acfactory, lp):
     msg = ac2.get_message_by_id(ev.data2)
 
     assert open(msg.filename).read() == content
-    msg.filename.index(basename)
-    assert msg.filename.endswith(ext)
+    msg.basename.index(basename)
+    assert msg.basename.endswith(ext)
 
 
 def test_html_message(acfactory, lp):