Conversion: Add an option under Structure detection to automatically fill in blank alt attributes for images that have alt text in the image file metadata

src/calibre/ebooks/conversion/cli.py

@@ -239,7 +239,7 @@ def add_pipeline_options(parser, plumber):
                       'chapter', 'chapter_mark',
                       'prefer_metadata_cover', 'remove_first_image',
                       'insert_metadata', 'page_breaks_before',
-                      'remove_fake_margins', 'start_reading_at',
+                      'remove_fake_margins', 'start_reading_at', 'add_alt_text_to_img',
                   ]
                   )),
 

src/calibre/ebooks/conversion/config.py

@@ -260,7 +260,7 @@ OPTIONS = {
         'structure_detection': (
             'chapter', 'chapter_mark', 'start_reading_at',
             'remove_first_image', 'remove_fake_margins', 'insert_metadata',
-            'page_breaks_before'),
+            'page_breaks_before', 'add_alt_text_to_img',),
 
         'toc': (
             'level1_toc', 'level2_toc', 'level3_toc',

src/calibre/ebooks/conversion/plumber.py

@@ -420,6 +420,11 @@ OptionRecommendation(name='remove_fake_margins',
                 'case you can disable the removal.')
         ),
 
+OptionRecommendation(name='add_alt_text_to_img',
+    recommended_value=False, level=OptionRecommendation.LOW,
+    help=_('When an <img> tag has no alt attribute, check the associated image file for metadata that specifies alternate text, and'
+            ' use it to fill in the alt attribute. The alt attribute is used by screen readers for assisting the visually challenged.')
+),
 
 OptionRecommendation(name='margin_top',
         recommended_value=5.0, level=OptionRecommendation.LOW,
@@ -1203,6 +1208,12 @@ OptionRecommendation(name='search_replace',
 
         from calibre.ebooks.oeb.transforms.jacket import Jacket
         Jacket()(self.oeb, self.opts, self.user_metadata)
+        pr(0.37)
+        self.flush()
+
+        if self.opts.add_alt_text_to_img:
+            from calibre.ebooks.oeb.transforms.alt_text import AddAltText
+            AddAltText()(self.oeb, self.opts)
         pr(0.4)
         self.flush()
 

src/calibre/ebooks/oeb/transforms/alt_text.py

@@ -0,0 +1,36 @@
+#!/usr/bin/env python
+# License: GPLv3 Copyright: 2024, Kovid Goyal <kovid at kovidgoyal.net>
+
+
+from io import BytesIO
+
+from PIL import Image
+
+from calibre.ebooks.oeb.base import SVG_MIME, urlnormalize, xpath
+from calibre.utils.img import read_alt_text
+
+
+def process_spine_item(item, hrefs, log):
+    html = item.data
+    for elem in xpath(html, '//h:img[@src]'):
+        src = urlnormalize(elem.attrib['src'])
+        image = hrefs.get(item.abshref(src), None)
+        if image and image.media_type != SVG_MIME and not elem.attrib.get('alt'):
+            data = image.bytes_representation
+            try:
+                with Image.open(BytesIO(data)) as im:
+                    alt = read_alt_text(im)
+            except Exception as err:
+                log.warn(f'Failed to read alt text from image {src} with error: {err}')
+            else:
+                if alt:
+                    elem.set('alt', alt)
+
+
+class AddAltText:
+
+    def __call__(self, oeb, opts):
+        oeb.logger.info('Add alt text to images...')
+        hrefs = oeb.manifest.hrefs
+        for item in oeb.spine:
+            process_spine_item(item, hrefs, oeb.log)

src/calibre/gui2/convert/structure_detection.ui

@@ -14,39 +14,9 @@
    <string>Form</string>
   </property>
   <layout class="QGridLayout" name="gridLayout">
-   <item row="2" column="3">
-    <widget class="QCheckBox" name="opt_remove_fake_margins">
-     <property name="text">
-      <string>Remove &amp;fake margins</string>
-     </property>
-    </widget>
-   </item>
-   <item row="4" column="0" colspan="4">
-    <widget class="QLabel" name="label_2">
-     <property name="text">
-      <string>The header and footer removal options have been replaced by the Search &amp; replace options. Click the Search &amp; replace category in the bar to the left to use these options. Leave the replace field blank and enter your header/footer removal regexps into the search field.</string>
-     </property>
-     <property name="wordWrap">
-      <bool>true</bool>
-     </property>
-    </widget>
-   </item>
-   <item row="5" column="0" rowspan="2" colspan="4">
+   <item row="7" column="0" rowspan="2" colspan="4">
     <widget class="XPathEdit" name="opt_page_breaks_before" native="true"/>
    </item>
-   <item row="3" column="0" colspan="4">
-    <widget class="QCheckBox" name="opt_insert_metadata">
-     <property name="text">
-      <string>Insert &amp;metadata as page at start of book</string>
-     </property>
-    </widget>
-   </item>
-   <item row="7" column="0" colspan="4">
-    <widget class="XPathEdit" name="opt_start_reading_at" native="true"/>
-   </item>
-   <item row="0" column="0" colspan="4">
-    <widget class="XPathEdit" name="opt_chapter" native="true"/>
-   </item>
    <item row="1" column="0" colspan="2">
     <widget class="QLabel" name="label">
      <property name="text">
@@ -57,12 +27,8 @@
      </property>
     </widget>
    </item>
-   <item row="1" column="2">
-    <widget class="QComboBox" name="opt_chapter_mark">
-     <property name="minimumContentsLength">
-      <number>20</number>
-     </property>
-    </widget>
+   <item row="0" column="0" colspan="4">
+    <widget class="XPathEdit" name="opt_chapter" native="true"/>
    </item>
    <item row="1" column="3">
     <spacer name="horizontalSpacer">
@@ -77,14 +43,24 @@
      </property>
     </spacer>
    </item>
-   <item row="2" column="0" colspan="3">
-    <widget class="QCheckBox" name="opt_remove_first_image">
-     <property name="text">
-      <string>Remove first &amp;image</string>
+   <item row="1" column="2">
+    <widget class="QComboBox" name="opt_chapter_mark">
+     <property name="minimumContentsLength">
+      <number>20</number>
      </property>
     </widget>
    </item>
-   <item row="8" column="0">
+   <item row="6" column="0" colspan="4">
+    <widget class="QLabel" name="label_2">
+     <property name="text">
+      <string>The header and footer removal options have been replaced by the Search &amp; replace options. Click the Search &amp; replace category in the bar to the left to use these options. Leave the replace field blank and enter your header/footer removal regexps into the search field.</string>
+     </property>
+     <property name="wordWrap">
+      <bool>true</bool>
+     </property>
+    </widget>
+   </item>
+   <item row="10" column="0">
     <spacer name="verticalSpacer">
      <property name="orientation">
       <enum>Qt::Vertical</enum>
@@ -97,6 +73,37 @@
      </property>
     </spacer>
    </item>
+   <item row="4" column="0">
+    <widget class="QCheckBox" name="opt_insert_metadata">
+     <property name="text">
+      <string>Insert &amp;metadata as page at start of book</string>
+     </property>
+    </widget>
+   </item>
+   <item row="9" column="0" colspan="4">
+    <widget class="XPathEdit" name="opt_start_reading_at" native="true"/>
+   </item>
+   <item row="5" column="0">
+    <widget class="QCheckBox" name="opt_remove_fake_margins">
+     <property name="text">
+      <string>Remove &amp;fake margins</string>
+     </property>
+    </widget>
+   </item>
+   <item row="5" column="1">
+    <widget class="QCheckBox" name="opt_remove_first_image">
+     <property name="text">
+      <string>Remove first &amp;image</string>
+     </property>
+    </widget>
+   </item>
+   <item row="4" column="1">
+    <widget class="QCheckBox" name="opt_add_alt_text_to_img">
+     <property name="text">
+      <string>Add &amp;alt text to images</string>
+     </property>
+    </widget>
+   </item>
   </layout>
  </widget>
  <customwidgets>

src/pyj/book_list/conversion_widgets.pyj

@@ -360,6 +360,7 @@ def structure_detection(container):
     g.appendChild(choices('chapter_mark', _('Chap&ter mark:'), ['pagebreak', 'rule', 'both', 'none']))
     g.appendChild(checkbox('remove_first_image', _('Remove first &image')))
     g.appendChild(checkbox('remove_fake_margins', _('Remove &fake margins')))
+    g.appendChild(checkbox('add_alt_text_to_img', _('Add &alt text to images')))
     g.appendChild(checkbox('insert_metadata', _('Insert metadata at start of book')))
     g.appendChild(lineedit('page_breaks_before', _('Insert page breaks before'), 50))
     g.appendChild(lineedit('start_reading_at', _('Start reading at'), 50))