From f3080a42b0bfa4ba51e9afbd7d2d2cec2c17625b Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Fri, 16 Nov 2007 18:17:18 +0000
Subject: [PATCH] Make memory usage minimization optional, thereby achieving an
 upto 10x speedup.

---
 src/libprs500/ebooks/lrf/__init__.py          |  2 ++
 src/libprs500/ebooks/lrf/html/convert_from.py | 11 ++++++-----
 2 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/src/libprs500/ebooks/lrf/__init__.py b/src/libprs500/ebooks/lrf/__init__.py
index 64be956f9e..fb00f87606 100644
--- a/src/libprs500/ebooks/lrf/__init__.py
+++ b/src/libprs500/ebooks/lrf/__init__.py
@@ -218,6 +218,8 @@ def option_parser(usage):
                       help='''Be verbose while processing''')
     debug.add_option('--lrs', action='store_true', dest='lrs', \
                       help='Convert to LRS', default=False)
+    parser.add_option('--minimize-memory-usage', action='store_true', default=False,
+                      help=_('Minimize memory usage at the cost of longer processing times. Use this option if you are on a memory constrained machine.'))
     return parser
 
 def find_custom_fonts(options, logger):
diff --git a/src/libprs500/ebooks/lrf/html/convert_from.py b/src/libprs500/ebooks/lrf/html/convert_from.py
index 78f478b3d7..382fac464b 100644
--- a/src/libprs500/ebooks/lrf/html/convert_from.py
+++ b/src/libprs500/ebooks/lrf/html/convert_from.py
@@ -648,8 +648,7 @@ class HTMLConverter(object):
         # Need to make a copy of contents as when
         # extract is called on a child, it will
         # mess up the iteration.
-        contents = [i for i in ptag.contents]
-        for c in contents:
+        for c in copy.copy(ptag.contents):
             if isinstance(c, HTMLConverter.IGNORED_TAGS):
                 continue
             elif isinstance(c, Tag):
@@ -658,7 +657,8 @@ class HTMLConverter(object):
                 self.add_text(c, pcss, ppcss)
         if not self.in_table:
             try:
-                ptag.extract()
+                if self.minimize_memory_usage:
+                    ptag.extract()
             except AttributeError:
                 print ptag, type(ptag)
                     
@@ -1532,8 +1532,9 @@ class HTMLConverter(object):
                 self.logger.debug('Bad table:\n%s', str(tag)[:300])
                 self.in_table = False
                 self.process_children(tag, tag_css, tag_pseudo_css)
-            finally:                
-                tag.extract()
+            finally:
+                if self.minimize_memory_usage:                
+                    tag.extract()
         else:
             self.process_children(tag, tag_css, tag_pseudo_css)        
         if end_page: