add a simple TTS field filter

This simply wraps the field in extra text that the frontend will
deal with. Also added some helpers for extracting and stripping
audio and TTS tags from the rendered text.

proto/backend.proto

@@ -16,6 +16,8 @@ message BackendInput {
         BrowserRowsIn browser_rows = 20;
         RenderCardIn render_card = 21;
         int64 local_minutes_west = 22;
+        string strip_av_tags = 23;
+        string get_av_tags = 24;
     }
 }
 
@@ -28,6 +30,8 @@ message BackendOutput {
         BrowserRowsOut browser_rows = 20;
         RenderCardOut render_card = 21;
         sint32 local_minutes_west = 22;
+        string strip_av_tags = 23;
+        GetAVTagsOut get_av_tags = 24;
 
         BackendError error = 2047;
     }
@@ -44,14 +48,6 @@ message InvalidInputError {
     string info = 1;
 }
 
-message PlusOneIn {
-    int32 num = 1;
-}
-
-message PlusOneOut {
-    int32 num = 1;
-}
-
 message TemplateParseError {
     string info = 1;
 }
@@ -150,3 +146,19 @@ message RenderedTemplateReplacement {
     string current_text = 2;
     repeated string filters = 3;
 }
+
+message GetAVTagsOut {
+    repeated AVTag av_tags = 1;
+}
+
+message AVTag {
+    oneof value {
+        string sound_or_video = 1;
+        TTSTag tts = 2;
+    }
+}
+
+message TTSTag {
+    repeated string args = 1;
+    string text = 2;
+}

pylib/anki/rsbackend.py

@@ -9,6 +9,7 @@ import ankirspy  # pytype: disable=import-error
 import anki.backend_pb2 as pb
 import anki.buildinfo
 from anki.models import AllTemplateReqs
+from anki.sound import AVTag, SoundOrVideoTag, TTSTag
 
 assert ankirspy.buildhash() == anki.buildinfo.buildhash
 
@@ -45,6 +46,14 @@ def proto_template_reqs_to_legacy(
     return legacy_reqs
 
 
+def av_tag_to_native(tag: pb.AVTag) -> AVTag:
+    val = tag.WhichOneof("value")
+    if val == "sound_or_video":
+        return SoundOrVideoTag(filename=tag.sound_or_video)
+    else:
+        return TTSTag(args=list(tag.tts.args), text=tag.tts.text)
+
+
 @dataclass
 class TemplateReplacement:
     field_name: str
@@ -143,3 +152,16 @@ class RustBackend:
         return self._run_command(
             pb.BackendInput(local_minutes_west=stamp)
         ).local_minutes_west
+
+    def strip_av_tags(self, text: str) -> str:
+        return self._run_command(pb.BackendInput(strip_av_tags=text)).strip_av_tags
+
+    def get_av_tags(self, text: str) -> List[AVTag]:
+        return list(
+            map(
+                av_tag_to_native,
+                self._run_command(
+                    pb.BackendInput(get_av_tags=text)
+                ).get_av_tags.av_tags,
+            )
+        )

pylib/anki/sound.py

@@ -1,18 +1,61 @@
 # Copyright: Ankitects Pty Ltd and contributors
 # License: GNU AGPL, version 3 or later; http://www.gnu.org/licenses/agpl.html
 
-import re
-from typing import List
+"""
+Sound/TTS references extracted from card text.
 
-# Shared utils
+Use collection.backend.strip_av_tags(string) to remove all tags,
+and collection.backend.get_av_tags(string) to get a list of AVTags.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import List, Union
+
+
+@dataclass
+class TTSTag:
+    """Records information about a text to speech tag.
+
+    See tts.py for more information.
+    """
+
+    args: List[str]
+    text: str
+
+
+@dataclass
+class SoundOrVideoTag:
+    """Contains the filename inside a [sound:...] tag.
+
+    Video files also use [sound:...].
+    """
+
+    filename: str
+
+
+# note this does not include image tags, which are handled with HTML.
+AVTag = Union[SoundOrVideoTag, TTSTag]
+
+# Legacy utils
 ##########################################################################
+# these will be removed in the future
 
 _soundReg = r"\[sound:(.*?)\]"
 
 
 def allSounds(text) -> List:
-    return re.findall(_soundReg, text)
+    from aqt import mw
+
+    return [
+        x.filename
+        for x in mw.col.backend.get_av_tags(text)
+        if isinstance(x, SoundOrVideoTag)
+    ]
 
 
 def stripSounds(text) -> str:
-    return re.sub(_soundReg, "", text)
+    from aqt import mw
+
+    return mw.col.backend.strip_av_tags(text)

rslib/Cargo.toml

@@ -15,6 +15,7 @@ lazy_static = "1.4.0"
 regex = "1.3.3"
 hex = "0.4.0"
 blake3 = "0.1.0"
+htmlescape = "0.3.1"
 
 [build-dependencies]
 prost-build = "0.5.0"

rslib/src/backend.rs

@@ -10,6 +10,7 @@ use crate::template::{
     render_card, without_legacy_template_directives, FieldMap, FieldRequirements, ParsedTemplate,
     RenderedNode,
 };
+use crate::text::{av_tags_in_string, strip_av_tags, AVTag};
 use prost::Message;
 use std::collections::{HashMap, HashSet};
 use std::path::PathBuf;
@@ -98,6 +99,8 @@ impl Backend {
             Value::LocalMinutesWest(stamp) => {
                 OValue::LocalMinutesWest(local_minutes_west_for_stamp(stamp))
             }
+            Value::StripAvTags(text) => OValue::StripAvTags(strip_av_tags(&text).into()),
+            Value::GetAvTags(text) => OValue::GetAvTags(self.get_av_tags(&text)),
         })
     }
 
@@ -178,6 +181,24 @@ impl Backend {
             answer_nodes: rendered_nodes_to_proto(anodes),
         })
     }
+
+    fn get_av_tags(&self, text: &str) -> pt::GetAvTagsOut {
+        let tags = av_tags_in_string(text)
+            .map(|avtag| match avtag {
+                AVTag::SoundOrVideo(file) => pt::AvTag {
+                    value: Some(pt::av_tag::Value::SoundOrVideo(file.to_string())),
+                },
+                AVTag::TextToSpeech { args, field_text } => pt::AvTag {
+                    value: Some(pt::av_tag::Value::Tts(pt::TtsTag {
+                        args: args.iter().map(|&s| s.to_string()).collect(),
+                        text: field_text.to_string(),
+                    })),
+                },
+            })
+            .collect();
+
+        pt::GetAvTagsOut { av_tags: tags }
+    }
 }
 
 fn ords_hash_to_set(ords: HashSet<u16>) -> Vec<u32> {

rslib/src/template.rs

@@ -3,7 +3,7 @@
 
 use crate::err::{Result, TemplateError};
 use crate::template_filters::apply_filters;
-use crate::text::strip_sounds;
+use crate::text::strip_av_tags;
 use lazy_static::lazy_static;
 use nom;
 use nom::branch::alt;
@@ -443,7 +443,7 @@ pub fn render_card(
     // if the question side didn't have any unknown filters, we can pass
     // FrontSide in now
     if let [RenderedNode::Text { ref text }] = *qnodes.as_slice() {
-        context.front_text = Some(strip_sounds(text));
+        context.front_text = Some(strip_av_tags(text));
     }
 
     // answer side

rslib/src/template_filters.rs

@@ -75,8 +75,12 @@ fn apply_filter<'a>(
         // an empty filter name (caused by using two colons) is ignored
         "" => text.into(),
         _ => {
-            // unrecognized filter
-            return (false, None);
+            if filter_name.starts_with("tts ") {
+                tts_filter(filter_name, text)
+            } else {
+                // unrecognized filter
+                return (false, None);
+            }
         }
     };
 
@@ -285,6 +289,11 @@ return false;">
     .into()
 }
 
+fn tts_filter(filter_name: &str, text: &str) -> Cow<'static, str> {
+    let args = filter_name.splitn(2, ' ').nth(1).unwrap_or("");
+
+    format!("[anki:tts][{}]{}[/anki:tts]", args, text).into()
+}
 // Tests
 //----------------------------------------
 
@@ -293,7 +302,7 @@ mod test {
     use crate::template::RenderContext;
     use crate::template_filters::{
         apply_filters, cloze_filter, furigana_filter, hint_filter, kana_filter, kanji_filter,
-        type_cloze_filter, type_filter,
+        tts_filter, type_cloze_filter, type_filter,
     };
     use crate::text::strip_html;
 
@@ -368,4 +377,12 @@ field</a>
         ctx.card_ord = 2;
         assert_eq!(cloze_filter(text, &ctx).as_ref(), "");
     }
+
+    #[test]
+    fn test_tts() {
+        assert_eq!(
+            tts_filter("tts lang=en_US", "foo"),
+            "[anki:tts][lang=en_US]foo[/anki:tts]"
+        );
+    }
 }

rslib/src/text.rs

@@ -1,12 +1,22 @@
 // Copyright: Ankitects Pty Ltd and contributors
 // License: GNU AGPL, version 3 or later; http://www.gnu.org/licenses/agpl.html
 
+use htmlescape;
 use lazy_static::lazy_static;
 use regex::Regex;
 use std::borrow::Cow;
 use std::collections::HashSet;
 use std::ptr;
 
+#[derive(Debug, PartialEq)]
+pub enum AVTag<'a> {
+    SoundOrVideo(Cow<'a, str>),
+    TextToSpeech {
+        args: Vec<&'a str>,
+        field_text: Cow<'a, str>,
+    },
+}
+
 lazy_static! {
     static ref HTML: Regex = Regex::new(concat!(
         "(?si)",
@@ -22,9 +32,16 @@ lazy_static! {
         r#"(?i)<img[^>]+src=["']?([^"'>]+)["']?[^>]*>"#
     ).unwrap();
 
-    static ref SOUND_TAG: Regex = Regex::new(
-        r"\[sound:(.*?)\]"
-    ).unwrap();
+    // videos are also in sound tags
+    static ref AV_TAGS: Regex = Regex::new(
+        r#"(?xs)
+            \[sound:(.*?)\]     # 1 - the filename in a sound tag
+            |
+            \[anki:tts\]
+                \[(.*?)\]       # 2 - arguments to tts call
+                (.*?)           # 3 - field text
+            \[/anki:tts\]
+            "#).unwrap();
 
     static ref CLOZED_TEXT: Regex = Regex::new(
         r"(?s)\{\{c(\d+)::.+?\}\}"
@@ -35,8 +52,43 @@ pub fn strip_html(html: &str) -> Cow<str> {
     HTML.replace_all(html, "")
 }
 
-pub fn strip_sounds(html: &str) -> Cow<str> {
-    SOUND_TAG.replace_all(html, "")
+pub fn decode_entities(html: &str) -> Cow<str> {
+    if html.contains('&') {
+        match htmlescape::decode_html(html) {
+            Ok(text) => text,
+            Err(e) => format!("{:?}", e),
+        }
+        .into()
+    } else {
+        // nothing to do
+        html.into()
+    }
+}
+
+pub fn strip_html_for_tts(html: &str) -> Cow<str> {
+    match HTML.replace_all(html, " ") {
+        Cow::Borrowed(_) => decode_entities(html),
+        Cow::Owned(s) => decode_entities(&s).to_string().into(),
+    }
+}
+
+pub fn strip_av_tags(text: &str) -> Cow<str> {
+    AV_TAGS.replace_all(text, "")
+}
+
+pub fn av_tags_in_string(text: &str) -> impl Iterator<Item = AVTag> {
+    AV_TAGS.captures_iter(text).map(|caps| {
+        if let Some(av_file) = caps.get(1) {
+            AVTag::SoundOrVideo(decode_entities(av_file.as_str()))
+        } else {
+            let args = caps.get(2).unwrap();
+            let field_text = caps.get(3).unwrap();
+            AVTag::TextToSpeech {
+                args: args.as_str().split(' ').collect(),
+                field_text: strip_html_for_tts(field_text.as_str()),
+            }
+        }
+    })
 }
 
 pub fn strip_html_preserving_image_filenames(html: &str) -> Cow<str> {
@@ -64,7 +116,10 @@ pub fn cloze_numbers_in_string(html: &str) -> HashSet<u16> {
 
 #[cfg(test)]
 mod test {
-    use crate::text::{cloze_numbers_in_string, strip_html, strip_html_preserving_image_filenames};
+    use crate::text::{
+        av_tags_in_string, cloze_numbers_in_string, strip_av_tags, strip_html,
+        strip_html_preserving_image_filenames, AVTag,
+    };
     use std::collections::HashSet;
 
     #[test]
@@ -95,4 +150,20 @@ mod test {
             vec![1, 2].into_iter().collect::<HashSet<u16>>()
         );
     }
+
+    #[test]
+    fn test_audio() {
+        let s = "abc[sound:fo&amp;o.mp3]def[anki:tts][lang=en_US voices=Bob,Jane]foo<br>1&gt;2[/anki:tts]gh";
+        assert_eq!(strip_av_tags(s), "abcdefgh");
+        assert_eq!(
+            av_tags_in_string(s).collect::<Vec<_>>(),
+            vec![
+                AVTag::SoundOrVideo("fo&o.mp3".into()),
+                AVTag::TextToSpeech {
+                    args: vec!["lang=en_US", "voices=Bob,Jane"],
+                    field_text: "foo 1>2".into()
+                },
+            ]
+        );
+    }
 }