Conversion pipeline: Remove the attempt to detect and autocorrect if text will go off the left edge of the page, as it was a rather crude heuristic. Also do not remove fake margins if the book uses negative text indents on the margined elements

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

@@ -221,7 +221,7 @@ class CSSFlattener(object):
                         value = 0.0
                     cssdict[property] = "%0.5fem" % (value / fsize)
 
-    def flatten_node(self, node, stylizer, names, styles, psize, item_id, left=0):
+    def flatten_node(self, node, stylizer, names, styles, psize, item_id):
         if not isinstance(node.tag, basestring) \
            or namespace(node.tag) != XHTML_NS:
                return
@@ -316,16 +316,6 @@ class CSSFlattener(object):
         if cssdict:
             if self.lineh and self.fbase and tag != 'body':
                 self.clean_edges(cssdict, style, psize)
-            margin = asfloat(style['margin-left'], 0)
-            indent = asfloat(style['text-indent'], 0)
-            left += margin
-            if (left + indent) < 0:
-                try:
-                    percent = (margin - indent) / style['width']
-                    cssdict['margin-left'] = "%d%%" % (percent * 100)
-                except ZeroDivisionError:
-                    pass
-                left -= indent
             if 'display' in cssdict and cssdict['display'] == 'in-line':
                 cssdict['display'] = 'inline'
             if self.unfloat and 'float' in cssdict \
@@ -378,7 +368,7 @@ class CSSFlattener(object):
         if 'style' in node.attrib:
             del node.attrib['style']
         for child in node:
-            self.flatten_node(child, stylizer, names, styles, psize, item_id, left)
+            self.flatten_node(child, stylizer, names, styles, psize, item_id)
 
     def flatten_head(self, item, href, global_href):
         html = item.data

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

@@ -32,6 +32,8 @@ class RemoveAdobeMargins(object):
                         attr = 'margin-'+margin
                         elem.attrib.pop(attr, None)
 
+class NegativeTextIndent(Exception):
+    pass
 
 class RemoveFakeMargins(object):
 
@@ -66,13 +68,25 @@ class RemoveFakeMargins(object):
         self.find_levels()
 
         for level in self.levels:
-            self.process_level(level)
+            try:
+                self.process_level(level)
+            except NegativeTextIndent:
+                self.log.debug('Negative text indent detected at level '
+                        ' %s, ignoring this level'%level)
 
     def get_margins(self, elem):
         cls = elem.get('class', None)
         if cls:
             style = self.selector_map.get('.'+cls, None)
             if style:
+                try:
+                    ti = style['text-indent']
+                except:
+                    pass
+                else:
+                    if ( (hasattr(ti, 'startswith') and ti.startswith('-')) or
+                            isinstance(ti, (int, float)) and ti < 0):
+                        raise NegativeTextIndent()
                 return style.marginLeft, style.marginRight, style
         return '', '', None