DOCX Output: Prevent images from being larger than the page size. See #1688690 (links do not work when converted azw3 to docx; pictures too big)

src/calibre/ebooks/docx/writer/container.py

@@ -36,6 +36,20 @@ def page_size(opts):
     return width, height
 
 
+def page_margin(opts, which):
+    val = getattr(opts, 'docx_page_margin_' + which)
+    if val == 0.0:
+        val = getattr(opts, 'margin_' + which)
+    return val
+
+
+def page_effective_area(opts):
+    width, height = page_size(opts)
+    width -= page_margin(opts, 'left') + page_margin(opts, 'right')
+    height -= page_margin(opts, 'top') + page_margin(opts, 'bottom')
+    return width, height  # in pts
+
+
 def create_skeleton(opts, namespaces=None):
     namespaces = namespaces or DOCXNamespace().namespaces
 
@@ -50,9 +64,7 @@ def create_skeleton(opts, namespaces=None):
     width, height = int(20 * width), int(20 * height)
 
     def margin(which):
-        val = getattr(opts, 'docx_page_margin_' + which)
+        val = page_margin(opts, which)
-        if val == 0.0:
-            val = getattr(opts, 'margin_' + which)
         return w(which), str(int(val * 20))
     body.append(E.sectPr(
         E.pgSz(**{w('w'):str(width), w('h'):str(height)}),

src/calibre/ebooks/docx/writer/from_html.py

@@ -421,7 +421,7 @@ class Convert(object):
 
         self.styles_manager = StylesManager(self.docx.namespace, self.log, self.mi.language)
         self.links_manager = LinksManager(self.docx.namespace, self.docx.document_relationships, self.log)
-        self.images_manager = ImagesManager(self.oeb, self.docx.document_relationships)
+        self.images_manager = ImagesManager(self.oeb, self.docx.document_relationships, self.opts)
         self.lists_manager = ListsManager(self.docx)
         self.fonts_manager = FontsManager(self.docx.namespace, self.oeb, self.opts)
         self.blocks = Blocks(self.docx.namespace, self.styles_manager, self.links_manager)

src/calibre/ebooks/docx/writer/images.py

@@ -14,10 +14,12 @@ from future_builtins import map
 
 from lxml import etree
 
+from calibre import fit_image
 from calibre.ebooks.oeb.base import urlunquote
 from calibre.ebooks.docx.images import pt_to_emu
 from calibre.utils.filenames import ascii_filename
 from calibre.utils.imghdr import identify
+from calibre.ebooks.docx.writer.container import page_effective_area
 
 Image = namedtuple('Image', 'rid fname width height fmt item')
 
@@ -40,8 +42,9 @@ def get_image_margins(style):
 
 class ImagesManager(object):
 
-    def __init__(self, oeb, document_relationships):
+    def __init__(self, oeb, document_relationships, opts):
         self.oeb, self.log = oeb, oeb.log
+        self.page_width, self.page_height = page_effective_area(opts)
         self.images = {}
         self.seen_filenames = set()
         self.document_relationships = document_relationships
@@ -100,7 +103,9 @@ class ImagesManager(object):
         self.count += 1
         img = self.images[href]
         name = urlunquote(posixpath.basename(href))
-        width, height = map(pt_to_emu, style.img_size(img.width, img.height))
+        width, height = style.img_size(img.width, img.height)
+        scaled, width, height = fit_image(width, height, self.page_width, self.page_height)
+        width, height = map(pt_to_emu, (width, height))
 
         makeelement, namespaces = self.document_relationships.namespace.makeelement, self.document_relationships.namespace.namespaces