tttocea608: implement word wrapping in roll-up modes.

In roll-up modes, we open new lines when the last column is reached.
This commit implements lookahead on a word basis, in order to avoid
splitting words unless absolutely necessary (when a word won't fit
on a full row)

video/closedcaption/src/tttocea608/imp.rs

@@ -540,6 +540,10 @@ impl TtToCea608 {
         ret
     }
 
+    fn peek_word_length(&self, chars: std::iter::Peekable<std::str::Chars>) -> u32 {
+        chars.take_while(|c| !c.is_ascii_whitespace()).count() as u32
+    }
+
     fn generate(
         &self,
         mut state: &mut State,
@@ -548,6 +552,7 @@ impl TtToCea608 {
         duration: gst::ClockTime,
         lines: Lines,
     ) -> Result<gst::FlowSuccess, gst::FlowError> {
+        let origin_column = self.settings.lock().unwrap().origin_column;
         let mut row = 13;
         let mut bufferlist = gst::BufferList::new();
         let mut_list = bufferlist.get_mut().unwrap();
@@ -585,7 +590,7 @@ impl TtToCea608 {
                     Cea608Mode::RollUp2 | Cea608Mode::RollUp3 | Cea608Mode::RollUp4 => {
                         state.send_roll_up_preamble = true;
                     }
-                    _ => col = self.settings.lock().unwrap().origin_column,
+                    _ => col = origin_column,
                 }
             }
         }
@@ -597,7 +602,7 @@ impl TtToCea608 {
                 if state.mode != Cea608Mode::PopOn && state.mode != Cea608Mode::PaintOn {
                     state.send_roll_up_preamble = true;
                 }
-                col = self.settings.lock().unwrap().origin_column;
+                col = origin_column;
             }
         }
 
@@ -637,7 +642,7 @@ impl TtToCea608 {
                 }
                 col = line_column;
             } else if state.mode == Cea608Mode::PopOn || state.mode == Cea608Mode::PaintOn {
-                col = self.settings.lock().unwrap().origin_column;
+                col = origin_column;
             }
 
             for (j, chunk) in line.chunks.iter().enumerate() {
@@ -722,20 +727,23 @@ impl TtToCea608 {
 
                     col += 1;
 
-                    if col > 31 {
+                    if state.mode.is_rollup() {
-                        match state.mode {
+                        /* In roll-up mode, we introduce carriage returns automatically.
-                            Cea608Mode::PaintOn | Cea608Mode::PopOn => {
+                         * Instead of always wrapping once the last column is reached, we
-                                if chars.peek().is_some() {
+                         * want to look ahead and check whether the following word will fit
-                                    gst_warning!(
+                         * on the current row. If it won't, we insert a carriage return,
-                                        CAT,
+                         * unless it won't fit on a full row either, in which case it will need
-                                        obj: element,
+                         * to be broken up.
-                                        "Dropping characters after 32nd column: {}",
+                         */
-                                        c
+                        let next_word_length = if c.is_ascii_whitespace() {
-                                    );
+                            self.peek_word_length(chars.clone())
-                                }
+                        } else {
-                                break;
+                            0
-                            }
+                        };
-                            Cea608Mode::RollUp2 | Cea608Mode::RollUp3 | Cea608Mode::RollUp4 => {
+
+                        if (next_word_length <= 32 - origin_column && col + next_word_length > 31)
+                            || col > 31
+                        {
                             if prev_char != 0 {
                                 state.cc_data(element, mut_list, prev_char);
                                 prev_char = 0;
@@ -751,7 +759,16 @@ impl TtToCea608 {
                                 Some(true),
                             );
                         }
+                    } else if col > 31 {
+                        if chars.peek().is_some() {
+                            gst_warning!(
+                                CAT,
+                                obj: element,
+                                "Dropping characters after 32nd column: {}",
+                                c
+                            );
                         }
+                        break;
                     }
                 }
             }

video/closedcaption/tests/tttocea608.rs

@@ -371,3 +371,62 @@ fn test_one_timed_buffer_and_eos_roll_up2() {
     let event = h.pull_event().unwrap();
     assert_eq!(event.type_(), gst::EventType::Eos);
 }
+
+/* Here we test that tttocea608 introduces carriage returns in
+ * judicious places and avoids to break words without rhyme or
+ * reason.
+ */
+#[test]
+fn test_word_wrap_roll_up() {
+    init();
+
+    let mut h = gst_check::Harness::new_parse("tttocea608 mode=roll-up2 origin-column=24");
+    h.set_src_caps_str("text/x-raw");
+
+    while h.events_in_queue() != 0 {
+        let _event = h.pull_event().unwrap();
+    }
+
+    let inbuf = new_timed_buffer(&"Hello World", gst::SECOND, gst::SECOND);
+    assert_eq!(h.push(inbuf), Ok(gst::FlowSuccess::Ok));
+
+    /* Padding */
+    loop {
+        let outbuf = h.pull().unwrap();
+        if outbuf.pts() + outbuf.duration() >= gst::SECOND {
+            break;
+        }
+
+        let data = outbuf.map_readable().unwrap();
+        assert_eq!(&*data, &[0x80, 0x80]);
+    }
+
+    let expected: [(gst::ClockTime, gst::ClockTime, [u8; 2usize]); 11] = [
+        (1_000_000_000.into(), 33_333_333.into(), [0x94, 0x25]), /* roll_up_2 */
+        (1_033_333_333.into(), 33_333_334.into(), [0x94, 0x7c]), /* preamble */
+        (1_066_666_667.into(), 33_333_333.into(), [0xc8, 0xe5]), /* H e */
+        (1_100_000_000.into(), 33_333_333.into(), [0xec, 0xec]), /* l l */
+        (1_133_333_333.into(), 33_333_334.into(), [0xef, 0x20]), /* o SPACE */
+        (1_166_666_667.into(), 33_333_333.into(), [0x94, 0xad]), /* carriage return */
+        (1_200_000_000.into(), 33_333_333.into(), [0x94, 0x25]), /* roll_up_2 */
+        (1_233_333_333.into(), 33_333_334.into(), [0x94, 0x7c]), /* preamble */
+        (1_266_666_667.into(), 33_333_333.into(), [0x57, 0xef]), /* W o */
+        (1_300_000_000.into(), 33_333_333.into(), [0xf2, 0xec]), /* r l */
+        (1_333_333_333.into(), 33_333_334.into(), [0x64, 0x80]), /* d nil */
+    ];
+
+    for (i, e) in expected.iter().enumerate() {
+        let outbuf = h.try_pull().unwrap();
+
+        assert_eq!(e.0, outbuf.pts(), "Unexpected PTS for {}th buffer", i + 1);
+        assert_eq!(
+            e.1,
+            outbuf.duration(),
+            "Unexpected duration for {}th buffer",
+            i + 1
+        );
+
+        let data = outbuf.map_readable().unwrap();
+        assert_eq!(e.2, &*data);
+    }
+}