GH 19 - EPUB Import Support

lute/book/forms.py

@@ -25,9 +25,12 @@ class NewBookForm(FlaskForm):
     text = TextAreaField("Text", description=desc)
     textfile = FileField(
         "Text file",
-        # TODO epub: add epub to the list, change prompt.
-        # "Please upload a valid .txt or .epub file"
-        validators=[FileAllowed(["txt"], "Please upload a valid text document")],
+        validators=[
+            FileAllowed(
+                ["txt", "epub"],
+                "Please upload a valid .txt or .epub file.",
+            )
+        ],
     )
     source_uri = StringField("Text source", validators=[Length(max=255)])
     audiofile = FileField(

lute/book/routes.py

@@ -63,8 +63,7 @@ def _get_file_content(filefielddata):
         return str(content, "utf-8")
 
     if ext == ".epub":
-        content = service.get_epub_content(filefielddata)
-        return str(content, "utf-8")
+        return service.get_epub_content(filefielddata)
 
     raise ValueError(f'Unknown file extension "{ext}"')
 
@@ -80,7 +79,10 @@ def new():
     if form.validate_on_submit():
         form.populate_obj(b)
         if form.textfile.data:
-            b.text = _get_file_content(form.textfile.data)
+            content = _get_file_content(form.textfile.data)
+            if not content:
+                return redirect("/")
+            b.text = content
         f = form.audiofile.data
         if f:
             b.audio_filename = service.save_audio_file(f)

lute/book/service.py

@@ -5,8 +5,10 @@
 import os
 from datetime import datetime
 import requests
+from tempfile import TemporaryFile, SpooledTemporaryFile
 from bs4 import BeautifulSoup
 from flask import current_app, flash
+from openepub import Epub, EpubError
 from werkzeug.utils import secure_filename
 from lute.book.model import Book
 
@@ -35,7 +37,25 @@ def get_epub_content(epub_file_field_data):
     """
     Get the content of the epub as a single string.
     """
-    raise ValueError("TODO epub: to be implemented.")
+    content = ""
+    try:
+        if hasattr(epub_file_field_data.stream, "seekable"):
+            epub = Epub(stream=epub_file_field_data.stream)
+            content = epub.get_text()
+        else:
+            # We get a SpooledTemporaryFile from the form but this doesn't
+            # implement all file-like methods until python 3.11. So we need
+            # to rewrite it into a TemporaryFile
+            with TemporaryFile() as tf:
+                epub_file_field_data.stream.seek(0)
+                tf.write(epub_file_field_data.stream.read())
+                epub = Epub(stream=tf)
+                content = epub.get_text()
+    except EpubError as e:
+        msg = f"Could not parse {epub_file_field_data.filename} (error: {str(e)})"
+        flash(msg, "notice")
+        content = ""
+    return content
 
 
 def book_from_url(url):

pyproject.toml

@@ -20,7 +20,6 @@ authors = [
 ]
 readme = "README_PyPi.md"
 
-# TODO epub: add epub parsing library to dependencies
 dependencies = [
   "Flask-SQLAlchemy>=3.1.1,<4",
   "Flask-WTF>=1.2.1,<2",
@@ -31,7 +30,8 @@ dependencies = [
   "beautifulsoup4>=4.12.2,<5",
   "PyYAML>=6.0.1,<7",
   "toml>=0.10.2,<1",
-  "waitress>=2.1.2,<3"
+  "waitress>=2.1.2,<3",
+  "openepub>=0.0.5<1"
 ]
 
 [project.scripts]

requirements.txt

@@ -82,3 +82,4 @@ wrapt==1.15.0
 wsproto==1.2.0
 WTForms==3.0.1
 zipp==3.17.0
+openepub==0.0.5

tests/acceptance/book.feature

@@ -26,13 +26,12 @@ Feature: Books and stats are available
         And the reading pane shows:
             Hola/. /Tengo/ /un/ /perro/.
 
-    # TODO epub: add an epub file to sample_files, activate this test.
-    ### Scenario: I can import an epub file.
-    ###     Given I visit "/"
-    ###     Given a Spanish book "Hola" from file Hola.epub
-    ###     Then the page title is Reading "Hola"
-    ###     And the reading pane shows:
-    ###         Tengo/ /un/ /amigo/.
+    Scenario: I can import an epub file.
+        Given I visit "/"
+        Given a Spanish book "Hola" from file Hola.epub
+        Then the page title is Reading "Hola"
+        And the reading pane shows:
+            Tengo/ /un/ /amigo/.
 
     Scenario: Books and stats are shown on the first page.
         Given I visit "/"