diff --git a/architecture.md b/architecture.md
index 78269a9ac9a4110ccc5780962e8994234cde2d40..3cb6c9933570e1a5036451b5273c2baea13b9078 100644
--- a/architecture.md
+++ b/architecture.md
@@ -17,7 +17,7 @@ job_uuid:
- audio.mkv
- video_language.txt
- metadata.json
- - statefile (new/done/error)
+ - statefile (new/new_transcribe/done/error)
```
### audio.mkv
@@ -31,6 +31,7 @@ Is used by the cronjob script (step 3).
Used by mailservice (step 4).
```json
{
+ "job_type": "subtitle",
"email": "example@example.com",
"language": "de",
"video_language": "de",
@@ -40,7 +41,8 @@ Used by mailservice (step 4).
### statefile
State is tracked through the following files in the jobs folder:
- - new: Job has been submitted by user
+ - new: Subtitle job has been submitted by user
+ - new_transcribe: Transcript job has been submitted by user
- submitted: Job has been scheduled on gpu cluster
- done: Job has been processed without errors
- error: Job has been processed with errors
\ No newline at end of file
diff --git a/forms.py b/forms.py
index 8b251e7191fd7066ab4e48ac385106c16225982d..2b80c66a1a6bdc7521bfc065a62ef4b561481e39 100644
--- a/forms.py
+++ b/forms.py
@@ -2,8 +2,8 @@ import av
from flask import current_app
from flask_wtf import FlaskForm
from flask_wtf.file import FileField, FileRequired, FileSize
-from wtforms import StringField, SelectField
-from wtforms.validators import InputRequired, Email
+from wtforms import StringField, SelectField, IntegerField
+from wtforms.validators import InputRequired, Email, Optional
from wtforms.validators import ValidationError
@@ -18,9 +18,9 @@ def validate_audio(_, field):
if not has_audio:
raise ValidationError("noaudiotrack")
- except av.AVError as e:
+ except av.FFmpegError as e:
current_app.logger.info(
- "Error while checking audio of file '{}': {}".format(file.filename, str(e))
+ "Error while checking audio of file '{}': {}".format(file.filename, str(e.strerror))
)
raise ValidationError("brokenfile")
@@ -41,4 +41,4 @@ class UploadForm(FlaskForm):
FileSize(current_app.config["MAX_CONTENT_LENGTH"]),
validate_audio,
],
- )
+ )
\ No newline at end of file
diff --git a/localisations.toml b/localisations.toml
index 856ea0ef65b46465c97a7baf0d99d4ce9ea91738..8874bf561a102e34e35ee7060a2b2788eff102fd 100644
--- a/localisations.toml
+++ b/localisations.toml
@@ -109,4 +109,36 @@ en = "Your subtitles"
[ mail.content ]
de = "Die Untertitel für Ihre Datei '{}' wurden erstellt.\nSie finden das Resultat im Anhang dieser E-Mail."
-en = "The subtitles for your file '{}' have been created.\nThe result is attached to this email."
\ No newline at end of file
+en = "The subtitles for your file '{}' have been created.\nThe result is attached to this email."
+
+[ transcribe_title ]
+de = "PTB Transkriptionsdienst"
+en = "PTB Transcription Service"
+
+[ transcribe_leadtext ]
+de = "Dieser Dienst erstellt automatisch Transkripte für Video- oder Audioaufnahmen.<br>Die KI-Verarbeitung läuft im HPC-Cluster der PTB."
+en = "This service automatically creates transcripts for video or audio recordings.<br>The AI processing runs in PTB's HPC cluster."
+
+[ transcribe_successtext]
+de = "Daten erfolgreich übermittelt. Ihr Transkript wird demnächst generiert.<br>Sobald der Prozess abgeschlossen ist, erhalten Sie eine E-Mail."
+en = "Data transmitted successfully. Your transcript will be generated shortly.<br>Once the process has completed an email will be sent to you."
+
+[ transcribe_submit ]
+de = "Transkript erstellen"
+en = "Create transcript"
+
+[ footer_subtitle ]
+de = "Stattdessen Transkript erstellen?"
+en = "Create transcript instead?"
+
+[ footer_transcript ]
+de = "Stattdessen Untertitel erstellen?"
+en = "Create subtitles instead?"
+
+[ mail.transcribe_subject ]
+de = "Ihr Transkript"
+en = "Your transcript"
+
+[ mail.transcribe_content ]
+de = "Das Transkript für Ihre Datei '{}' wurden erstellt.\nSie finden das Resultat im Anhang dieser E-Mail."
+en = "The transcript for your file '{}' have been created.\nThe result is attached to this email."
\ No newline at end of file
diff --git a/mailservice.py b/mailservice.py
index bfe1e67bdc429959194df5ba11e66206406b6258..7627a780fafa081b7a296fd9036064905cc849c5 100644
--- a/mailservice.py
+++ b/mailservice.py
@@ -1,3 +1,4 @@
+import io
import json
import shutil
import smtplib
@@ -5,6 +6,7 @@ import tomllib
from email.message import EmailMessage
from os import scandir
from pathlib import Path
+import re
import requests
from requests.auth import HTTPBasicAuth
@@ -67,26 +69,66 @@ def main(end):
language = metadata["language"]
msg = EmailMessage()
- msg["Subject"] = localisations["mail"]["subject"][language]
- msg["From"] = config["MAIL"]["FROM"]
- msg["To"] = metadata["email"]
- msg.set_content(
- localisations["mail"]["content"][language].format(metadata["filename"])
- )
+ if metadata["job_type"] == "subtitle":
+ msg["Subject"] = localisations["mail"]["subject"][language]
+ msg["From"] = config["MAIL"]["FROM"]
+ msg["To"] = metadata["email"]
- # filename.language.vtt
- if metadata["video_language"] == "None":
- filename = Path(metadata["filename"]).with_suffix(".vtt").name
- else:
- filename = (
- Path(metadata["filename"])
- .with_suffix(".{}.vtt".format(metadata["video_language"]))
- .name
+ msg.set_content(
+ localisations["mail"]["content"][language].format(metadata["filename"])
)
- with open(Path(job).joinpath("audio.vtt")) as f:
- msg.add_attachment(f.read(), filename=filename)
+ # filename.language.vtt
+ if metadata["video_language"] == "None":
+ filename = Path(metadata["filename"]).with_suffix(".vtt").name
+ else:
+ filename = (
+ Path(metadata["filename"])
+ .with_suffix(".{}.vtt".format(metadata["video_language"]))
+ .name
+ )
+
+ with open(Path(job).joinpath("audio.vtt")) as f:
+ msg.add_attachment(f.read(), filename=filename)
+
+ elif metadata["job_type"] == "transcript":
+ msg["Subject"] = localisations["mail"]["transcribe_subject"][language]
+ msg["From"] = config["MAIL"]["FROM"]
+ msg["To"] = metadata["email"]
+
+ msg.set_content(
+ localisations["mail"]["transcribe_content"][language].format(
+ metadata["filename"]
+ )
+ )
+
+ # filename.txt
+ filename = Path(metadata["filename"]).with_suffix(".txt").name
+
+ with open(Path(job).joinpath("audio.txt"), encoding="utf-8") as f:
+ transcript = f.read()
+
+ # reformat transcript
+ lines = []
+ for line in transcript.split("\n"):
+ match = re.fullmatch(transcript_pattern, line)
+
+ if match:
+ lines.append(match.groups())
+
+ transcript = ""
+ first_line = True
+ previous_speaker = None
+ for speaker, line in lines:
+ if previous_speaker == speaker:
+ transcript += f" {line}"
+ else:
+ previous_speaker = speaker
+ transcript += f"{'\n' if not first_line else ''}{speaker}: {line}"
+ first_line = False
+
+ msg.add_attachment(transcript, filename=filename)
s.send_message(msg)
@@ -146,6 +188,8 @@ if __name__ == "__main__":
end = Event()
+ transcript_pattern = re.compile(r"^\[(\w+)]: (.+)$")
+
def handler(signum, frame):
global end
print(signum)
diff --git a/requirements.txt b/requirements.txt
index 440415088a50fbf522ea853655eb68af3a26dfac..f3b0f22286246c305073b8e9ff514e85beaaadee 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
-av~=14.0.0
+pyav~=14.0.0
Flask~=3.1.0
Flask-WTF~=1.2.2
wtforms[email]~=3.2.1
diff --git a/routes.py b/routes.py
index 472e8804bfd2ecfe7ad7c388067de67bd32ce272..ca38c5a20feb391756eb5afb8f6d08a6607044e8 100644
--- a/routes.py
+++ b/routes.py
@@ -16,87 +16,79 @@ from flask import (
from forms import UploadForm
-@current_app.route("/upload/<string:language>", methods=["GET", "POST"])
-def upload(language: str):
+@current_app.route("/subtitle/<string:language>", methods=["GET", "POST"])
+def subtitle(language: str):
set_language(language)
+ request.type = "subtitle"
- form = UploadForm()
- form.language.choices = [
- (k, v[language]) for k, v in current_app.config["LOCALISATIONS"]["language"]["options"].items()
- ]
+ form = get_form(language)
if form.is_submitted():
if not form.email.data.endswith(current_app.config["MAIL_DOMAIN"]):
form.email.data += current_app.config["MAIL_DOMAIN"]
if form.validate():
+ file = form.file.data
+ video_language = "None" if form.language.data == "auto" else form.language.data
+ metadata = {
+ "job_type": "subtitle",
+ "email": form.email.data,
+ "language": language,
+ "video_language": video_language,
+ "filename": file.filename,
+ }
+ return stream_with_context(process_file(file, video_language, metadata, request.type))
+ else:
+ if form.email.data:
+ form.email.data = form.email.data[: -len(current_app.config["MAIL_DOMAIN"])]
- @stream_with_context
- def process_file():
- # provide immediate user feedback
- yield render_template("success.html")
-
- # do all processing
- file = form.file.data
-
- folder_name = str(uuid4())
- folder_path = path.join(current_app.config["UPLOAD_FOLDER"], folder_name)
- makedirs(folder_path)
-
- # Only save first audio stream
- with av.open(file.stream) as container:
- audio_stream = [stream for stream in container.streams if stream.type == "audio"][0]
-
- with av.open(path.join(folder_path, "audio.mkv"), "w") as out:
- out_stream = out.add_stream_from_template(audio_stream)
-
- for packet in container.demux(audio_stream):
- # Skip the "flushing" packets that `demux` generates.
- if packet.dts is None:
- continue
-
- # Assign the packet to the new stream.
- packet.stream = out_stream
-
- out.mux(packet)
-
- video_language = "None" if form.language.data == "auto" else form.language.data
+ if language != current_app.config["DEFAULT_LANGUAGE"]:
+ form = language_fix_form(form, language)
- metadata = {
- "email": form.email.data,
- "language": language,
- "video_language": video_language,
- "filename": file.filename,
- }
+ return render_template("upload.html", form=form)
- with open(path.join(folder_path, "metadata.json"), "w") as f:
- json.dump(metadata, f)
- with open(path.join(folder_path, "video_language.txt"), "w") as f:
- f.write(video_language)
+@current_app.route("/transcribe/<string:language>", methods=["GET", "POST"])
+def transcribe(language: str):
+ set_language(language)
+ request.type = "transcript"
- open(path.join(folder_path, "new"), "wb").close()
+ form = get_form(language)
- # yield empty string to close response
- yield ""
+ if form.is_submitted():
+ if not form.email.data.endswith(current_app.config["MAIL_DOMAIN"]):
+ form.email.data += current_app.config["MAIL_DOMAIN"]
- return process_file()
+ if form.validate():
+ file = form.file.data
+ video_language = "None" if form.language.data == "auto" else form.language.data
+ metadata = {
+ "job_type": "transcript",
+ "email": form.email.data,
+ "language": language,
+ "video_language": video_language,
+ "filename": file.filename,
+ }
+ return stream_with_context(process_file(file, video_language, metadata, request.type))
else:
if form.email.data:
form.email.data = form.email.data[: -len(current_app.config["MAIL_DOMAIN"])]
if language != current_app.config["DEFAULT_LANGUAGE"]:
- form.email.label.text = current_app.config["LOCALISATIONS"]["email"]["label"][language]
- form.language.label.text = current_app.config["LOCALISATIONS"]["language"]["label"][language]
- form.file.label.text = current_app.config["LOCALISATIONS"]["file"]["label"][language]
+ form = language_fix_form(form, language)
return render_template("upload.html", form=form)
@current_app.route("/")
-@current_app.route("/upload/")
+@current_app.route("/subtitle/")
def default():
- return redirect(url_for("upload", language=current_app.config["DEFAULT_LANGUAGE"]))
+ return redirect(url_for("subtitle", language=current_app.config["DEFAULT_LANGUAGE"]))
+
+
+@current_app.route("/transcribe/")
+def default_transcribe():
+ return redirect(url_for("transcribe", language=current_app.config["DEFAULT_LANGUAGE"]))
def set_language(language):
@@ -106,6 +98,65 @@ def set_language(language):
request.language = language
+def get_form(language):
+ form = UploadForm()
+ form.language.choices = [
+ (k, v[language]) for k, v in current_app.config["LOCALISATIONS"]["language"]["options"].items()
+ ]
+
+ return form
+
+
+def language_fix_form(form, language):
+ form.email.label.text = current_app.config["LOCALISATIONS"]["email"]["label"][language]
+ form.language.label.text = current_app.config["LOCALISATIONS"]["language"]["label"][language]
+ form.file.label.text = current_app.config["LOCALISATIONS"]["file"]["label"][language]
+
+ return form
+
+
+def process_file(file, video_language, metadata, type):
+ # provide immediate user feedback
+ yield render_template("success.html")
+
+ folder_name = str(uuid4())
+ folder_path = path.join(current_app.config["UPLOAD_FOLDER"], folder_name)
+ makedirs(folder_path)
+
+ # Only save first audio stream
+ with av.open(file.stream) as container:
+ audio_stream = [stream for stream in container.streams if stream.type == "audio"][0]
+
+ with av.open(path.join(folder_path, "audio.mkv"), "w") as out:
+ out_stream = out.add_stream_from_template(audio_stream)
+
+ for packet in container.demux(audio_stream):
+ # Skip the "flushing" packets that `demux` generates.
+ if packet.dts is None:
+ continue
+
+ # Assign the packet to the new stream.
+ packet.stream = out_stream
+
+ out.mux(packet)
+
+ with open(path.join(folder_path, "metadata.json"), "w") as f:
+ json.dump(metadata, f)
+
+ with open(path.join(folder_path, "video_language.txt"), "w") as f:
+ f.write(video_language)
+
+ if type == "subtitle":
+ open(path.join(folder_path, "new"), "wb").close()
+ elif type == "transcript":
+ open(path.join(folder_path, "new_transcribe"), "wb").close()
+ else:
+ raise NotImplementedError
+
+ # yield empty string to close response
+ yield ""
+
+
@current_app.route("/health")
def health():
return "", 204
diff --git a/templates/base.html b/templates/base.html
index 6d648beb33606a424104ea1a0c5d91e9c1eefd4d..55b97132cff7fbb82c595d3336515959bd7b665b 100644
--- a/templates/base.html
+++ b/templates/base.html
@@ -44,8 +44,8 @@
<div class="py-5 text-center">
<img class="d-block mx-auto mb-4" src="{{ url_for('static', filename='img/logo.svg') }}" alt="PTB Logo"
width="96" height="96">
- <h1 class="h2">{{ config["LOCALISATIONS"]["title"][request.language] }}</h1>
- <p class="lead">{% autoescape false %}{{ config["LOCALISATIONS"]["leadtext"][request.language] }}{% endautoescape %}</p>
+ <h1 class="h2">{% block title %}{% endblock %}</h1>
+ <p class="lead">{% block leadtext %}{% endblock %}</p>
</div>
{% block content %}{% endblock %}
</main>
@@ -61,6 +61,16 @@
{% endfor %}
</p>
<br>
+ <p class="mb-1">
+ {% if request.type == "subtitle" -%}
+ <a class="link-secondary" href="{{ url_for("transcribe", language=request.language) }}">{{ config["LOCALISATIONS"]["footer_subtitle"][request.language] }}</a>
+ {% endif %}
+ {% if request.type == "transcript" -%}
+ <a class="link-secondary" href="{{ url_for("subtitle", language=request.language) }}">{{ config["LOCALISATIONS"]["footer_transcript"][request.language] }}</a>
+ {% endif %}
+
+ </p>
+ <br>
<p class="mb-1">{{ config["LOCALISATIONS"]["contact"]["text"][request.language] }}:</p>
<p class="mb-1">{{ config["CONTACT"]["ORG"] }} <a class="link-secondary" href="mailto:{{ config["CONTACT"]["MAIL"] }}">{{ config["CONTACT"]["NAME"] }}</a></p>
<p class="font-monospace"><a class="link-secondary" target="_blank" referrerpolicy="no-referrer" href="https://github.com/m-bain/whisperX">WhisperX</a><br>
diff --git a/templates/success.html b/templates/success.html
index 9d58ebe877e670e0f84eb70ce3859a7f4c9758af..4c2b4d64568f0197db034ee04f8786fac7eca636 100644
--- a/templates/success.html
+++ b/templates/success.html
@@ -5,8 +5,22 @@
<img class="mx-auto mb-4 spinner processing transition-opacity" src="{{ url_for('static', filename='img/arrow-repeat.svg') }}" alt="Spinning arrow" height="136" aria-hidden="true">
<img class="mx-auto mb-4 success hidden transition-opacity" src="{{ url_for('static', filename='img/cloud-check.svg') }}" alt="Cloud with checkmark" height="136" aria-hidden="true" hidden>
<p class="pt-4 lead processing transition-opacity">{% autoescape false %}{{ config["LOCALISATIONS"]["processingtext"][request.language] }}{% endautoescape %}</p>
- <p class="pt-4 lead success hidden transition-opacity" hidden>{% autoescape false %}{{ config["LOCALISATIONS"]["successtext"][request.language] }}{% endautoescape %}</p>
- <a class="pt-4 fs-1 lead success hidden transition-opacity icon-link icon-link-hover link-offset-2" style="--bs-icon-link-transform: translate3d(-.125rem, 0, 0);" href="{{ url_for('upload', language=request.language) }}">
+ <p class="pt-4 lead success hidden transition-opacity" hidden>
+ {% if request.type == "subtitle" -%}
+ {% autoescape false %}{{ config["LOCALISATIONS"]["successtext"][request.language] }}{% endautoescape %}
+ {% endif %}
+ {% if request.type == "transcript" -%}
+ {% autoescape false %}{{ config["LOCALISATIONS"]["transcribe_successtext"][request.language] }}{% endautoescape %}
+ {% endif %}
+
+ </p>
+ <a class="pt-4 fs-1 lead success hidden transition-opacity icon-link icon-link-hover link-offset-2" style="--bs-icon-link-transform: translate3d(-.125rem, 0, 0);" href="
+ {% if request.type == "subtitle" -%}
+ {{ url_for('subtitle', language=request.language) }}
+ {% endif %}
+ {% if request.type == "transcript" -%}
+ {{ url_for('transcribe', language=request.language) }}
+ {% endif %}">
<img class="bi" aria-label='{{ config["LOCALISATIONS"]["return"][request.language] }}' src="{{ url_for('static', filename='img/arrow-left.svg') }}" alt="Return arrow" height="4em">
</a>
</div>
diff --git a/templates/upload.html b/templates/upload.html
index f123f9de3266dcd9bed99694e887b47b38c07d6e..68bc5a65398d13365df9ea7f9bd356a034cfc345 100644
--- a/templates/upload.html
+++ b/templates/upload.html
@@ -1,5 +1,23 @@
{% extends "base.html" %}
+{% block title %}
+ {% if request.type == "subtitle" -%}
+ {{ config["LOCALISATIONS"]["title"][request.language] }}
+ {% endif %}
+ {% if request.type == "transcript" -%}
+ {{ config["LOCALISATIONS"]["transcribe_title"][request.language] }}
+ {% endif %}
+{% endblock %}
+
+{% block leadtext %}
+ {% if request.type == "subtitle" -%}
+ {% autoescape false %}{{ config["LOCALISATIONS"]["leadtext"][request.language] }}{% endautoescape %}
+ {% endif %}
+ {% if request.type == "transcript" -%}
+ {% autoescape false %}{{ config["LOCALISATIONS"]["transcribe_leadtext"][request.language] }}{% endautoescape %}
+ {% endif %}
+{% endblock %}
+
{% block content %}
<div>
<form id="form" class="transition-opacity" method="post" enctype="multipart/form-data">
@@ -81,7 +99,14 @@
</div>
<div class="row justify-content-center">
<div class="col-sm-9 col-md-7 pt-4">
- <button class="w-100 btn btn-dark btn-lg" type="submit">{{ config["LOCALISATIONS"]["submit"][request.language] }}</button>
+ <button class="w-100 btn btn-dark btn-lg" type="submit">
+ {% if request.type == "subtitle" -%}
+ {{ config["LOCALISATIONS"]["submit"][request.language] }}
+ {% endif %}
+ {% if request.type == "transcript" -%}
+ {{ config["LOCALISATIONS"]["transcribe_submit"][request.language] }}
+ {% endif %}
+ </button>
</div>
</div>
</form>