added robots

tts: added error handling
added tts functionality
2025-10-17 05:11:16 +02:00 · 2024-04-22 06:14:32 +02:00 · 2024-04-22 02:29:14 +02:00 · 2024-04-22 02:06:12 +02:00 · 2024-04-21 04:00:41 +02:00 · 2024-04-21 02:58:33 +02:00
10 changed files with 100 additions and 19 deletions
--- a/18
+++ b/18
@@ -2,11 +2,9 @@ FROM python:3

 MAINTAINER tiyn tiyn@mail-mk.eu

-COPY src /blog
-
-WORKDIR /blog
-
-RUN pip3 install -r requirements.txt
+ENV LANG en_US.UTF-8
+ENV LANGUAGE en_US:en
+ENV LC_ALL en_US.UTF-8

 RUN apt-get update && \
    apt-get install -y locales && \
@@ -14,9 +12,13 @@ RUN apt-get update && \
    sed -i '/en_US.UTF-8/s/^# //g' /etc/locale.gen && \
    dpkg-reconfigure --frontend=noninteractive locales

-ENV LANG en_US.UTF-8
-ENV LANGUAGE en_US:en
-ENV LC_ALL en_US.UTF-8
+RUN apt-get install -y espeak
+
+COPY src /blog
+
+WORKDIR /blog
+
+RUN pip3 install -r requirements.txt

 VOLUME /blog/templates/entry

--- a/README.md
+++ b/README.md
@@ -22,6 +22,7 @@ via plain text files.
  - [x] Links to standalone article
 - [x] Standalone article page
  - [x] Links to scrolling blog page
+  - [x] TTS Functionality
 - [x] RSS feed
 - [x] Navigation
  - [x] Header
--- a/src/app.py
+++ b/src/app.py
@@ -121,6 +121,10 @@ def entry(path):
 def feed_re():
  return redirect(url_for("feed"))

+@app.route("/robots.txt")
+def robots():
+  return render_template("robots.txt")
+

@app.route("/feed")
 def feed():
@@ -137,4 +141,5 @@ def feed():


 if __name__ == "__main__":
+  con_gen.prepare_tts()
  app.run(host="0.0.0.0")
--- a/src/config.py
+++ b/src/config.py
@@ -21,3 +21,6 @@ MAIL = "dummy@mail.com"

 # Directory to store entries in
 ENTRY_DIR = "templates/entry"
+
+# Set the timezone of your blog
+TIMEZONE = "+0000"
--- a/src/content.py
+++ b/src/content.py
@@ -1,3 +1,4 @@
+import glob
 import locale
 import os
 import pathlib
@@ -7,6 +8,7 @@ from os import path

 import markdown
 from bs4 import BeautifulSoup
+from gtts import gTTS, gTTSError

 import config
 import search
@@ -15,6 +17,7 @@ WEBSITE = config.WEBSITE
 ENTRY_DIR = config.ENTRY_DIR
 LANGUAGE = config.LANGUAGE
 LOCAL = "de_DE.UTF-8" if LANGUAGE == "de-de" else "en_US.UTF-8"
+TIMEZONE = config.TIMEZONE

 locale.setlocale(locale.LC_TIME, LOCAL)

@@ -143,6 +146,11 @@ def gen_stand_string(path_ex):
    content_string += "<a href=\"" + "/index.html#" + \
        filename_no_end + "\">" + curr_date + "</a>"
    content_string += "<br><br>\n"
+    if os.path.isfile("static/tmp/" + filename_no_end + ".mp3"):
+      content_string += "<audio controls>\n"
+      content_string += '<source src="/static/tmp/' + filename_no_end + '.mp3" type="audio/mp3">\n'
+      content_string += "</audio>\n"
+    content_string += "<br><br>\n"
    if filename.endswith(".html"):
      for line in text:
        content_string += line
@@ -184,7 +192,6 @@ def get_rss_string():
  Returns:
  string: rss-string of everything that is in the ENTRY_DIR.
  """
-  locale.setlocale(locale.LC_TIME, "en_US.UTF-8")
  path_ex = ENTRY_DIR
  content_string = ""
  if path.exists(path_ex):
@@ -202,9 +209,11 @@ def get_rss_string():
      content_string += "<title>" + title + "</title>\n"
      content_string += "<guid>" + WEBSITE + \
          "/index.html#" + filename + "</guid>\n"
+      locale.setlocale(locale.LC_TIME, "en_US.UTF-8")
      content_string += "<pubDate>" + \
          datetime.fromtimestamp(os.path.getmtime(file)).strftime(
-              "%a, %d %b %Y %H:%M:%S") + " +0100</pubDate>\n"
+              "%a, %d %b %Y %H:%M:%S") + " " + TIMEZONE + "</pubDate>\n"
+      locale.setlocale(locale.LC_TIME, LOCAL)
      content_string += "<description>\n<![CDATA[<html>\n<head>\n</head>\n<body>\n"
      html_string = ""
      for line in text:
@@ -212,7 +221,6 @@ def get_rss_string():
      content_string += absolutize_html(html_string)
      content_string += "\n</body></html>\n]]>\n</description>\n"
      content_string += "</item>\n"
-  locale.setlocale(locale.LC_TIME, LOCAL)
  return content_string


@@ -269,3 +277,55 @@ def create_preview(path, is_markdown):
    preview = "\n<p>" + first_p.text + "</p>\n"
  preview += "...<br>"
  return preview
+
+
+def get_text_only(filename):
+  """
+  Convert a file to text only to use in tts
+
+  Parameters:
+  path (string): path to the article
+
+  Returns:
+  string: unformatted string containing the contents of the file
+  """
+  # filename = os.path.join(ENTRY_DIR, path)
+  clean_text = ""
+  if path.exists(filename):
+    title = open(filename).readline().rstrip("\n")
+    text = open(filename).readlines()[1:]
+    filename_no_end = filename.split(".", 1)[0]
+    filename_no_end = filename_no_end.split("/")[-1]
+    content_string = ""
+    if filename.endswith(".html"):
+      for line in text:
+        content_string += line
+    if filename.endswith(".md"):
+      content_string += gen_md_content(filename, 1)
+    content_string = absolutize_html(content_string)
+    soup = BeautifulSoup(content_string, "html.parser")
+    tag_to_remove = soup.find("figure")
+    if tag_to_remove:
+      tag_to_remove.decompose()
+    clean_text = soup.get_text(separator=" ")
+    clean_text = title + "\n\n" + clean_text
+  return clean_text
+
+
+def prepare_tts():
+  files = glob.glob('static/tmp/*')
+  for f in files:
+    os.remove(f)
+  files = glob.glob('templates/entry/*')
+  clean_text = ""
+  for f in files:
+    clean_text = get_text_only(f)
+    _, tail = os.path.split(f)
+    new_filename = "static/tmp/" + os.path.splitext(tail)[0] + ".mp3"
+    try:
+      tts = gTTS(clean_text, lang=LANGUAGE.split("-")[0])
+      tts.save(new_filename)
+    except gTTSError as e:
+      print("Too many request to the google servers. Try it again later.")
+      os.remove(new_filename)
+      return e
--- a/src/requirements.txt
+++ b/src/requirements.txt
@@ -5,3 +5,4 @@ WTForms
 Flask_WTF
 Font-Awesome-Flask
 BeautifulSoup4
+gTTS
--- a/src/static/css/style.css
+++ b/src/static/css/style.css
@@ -31,6 +31,7 @@ html {
 footer {
  height: 100px;
  padding-top: 20px;
+  width: 100%;
 }

 footer .center {
@@ -38,11 +39,12 @@ footer .center {
 }

 .container {
-  min-height: 100%;
  padding-bottom: 50px;
  padding-left: 10%;
  padding-right: 10%;
-  padding-top: 5%;
+  padding-top: 5vh;
+  /* position: relative; */
+  min-height: calc(100vh - 50px - 5vh - 100px - 100px);
 }

 .container .flash {
@@ -163,11 +165,10 @@ form {
 }

 .standalone {
-  max-height: 100%;
+  min-height: 100vh;
 }

 .entry {
-  max-height: 100%;
  /* border-radius: 0 10px 30px 0; */
  margin-bottom: 20px;
  padding-left: 20px;
@@ -252,7 +253,7 @@ form.search input[type=text] {
 form.search button[type=submit] {
  float: left;
  width: 5%;
-  padding: 12px;
+  padding: 10px;
  font-size: 17px;
  border: 1px solid grey;
  border-left: none; /* Prevent double borders */
@@ -268,6 +269,12 @@ form.search::after {
 .standalone img {
  width: 50%;
 }
+
 .entry img {
  width: 50%;
 }
+
+audio {
+    width: 50%;
+    height: 5vh;
+}
--- a/src/static/tmp/.gitkeep
+++ b/src/static/tmp/.gitkeep
--- a/src/templates/robots.txt
+++ b/src/templates/robots.txt
@@ -0,0 +1,2 @@
+User-agent: *
+Disallow: /static/
--- a/src/templates/template.html
+++ b/src/templates/template.html
@@ -35,9 +35,9 @@
  <!-- Content -->
  <footer>
    <div class="center">
-      <a href="{{ url_for('imprint') }}">
-        {% if language=="de-de" %}Impressum und Kontakt{% else %}Imprint and Contact{% endif %}
-      </a><br>
+      <a href="{{ url_for('imprint') }}">{% if language=="de-de" %}Impressum und Kontakt{% else %}Imprint and Contact{%
+        endif %}</a>.<br>
+      <a href="{{ url_for('feed') }}">{% if language=="de-de" %}RSS-Feed{% else %}RSS feed{% endif %}</a>.<br>
      Made with <a href="https://github.com/tiyn/beaker-blog">Beaker Blog</a>.
    </div>
  </footer>
Author	SHA1	Message	Date
tiyn	9fd714eef3	added robots	2024-04-22 06:14:32 +02:00
tiyn	a927a18e39	tts: added error handling	2024-04-22 02:29:14 +02:00
tiyn	6e844a3cb1	added tts functionality	2024-04-22 02:06:12 +02:00
tiyn	9ea9ab3c53	improved css for footer placement	2024-04-21 04:00:41 +02:00
tiyn	73d9faea42	added feed to the footer	2024-04-21 02:58:33 +02:00
tiyn	586549a7f9	added changable timezone for rss feed	2024-04-21 02:14:11 +02:00