Backport 'Fix for PDF texts not being selectable'

NEWS.rst

@@ -3,9 +3,10 @@
 ======
 
 
-Version 2.0, not released yet
-=============================
+Version 1.0.23, not released yet
+================================
 
+* Make text selectable on generated PDF files
 * Fix markers
 * Fix URL/id handling
 * Test CairoSVG with Travis

cairosvg/__init__.py

@@ -70,6 +70,9 @@ def main():
     option_parser.add_option(
         "-o", "--output",
         default="", help="output filename")
+    option_parser.add_option(
+        "--text-as-text", action="store_true",
+        help="saves text in PDF as text instead of paths")
     options, args = option_parser.parse_args()
 
     # Print help if no argument is given

cairosvg/surface/__init__.py

@@ -75,9 +75,10 @@ class Surface(object):
 
     # Subclasses must either define this or override _create_surface()
     surface_class = None
+    draw_text_as_text = False
 
     @classmethod
-    def convert(cls, bytestring=None, **kwargs):
+    def convert(cls, bytestring=None, draw_text_as_text=None, **kwargs):
         """Convert a SVG document to the format for this class.
 
         Specify the input by passing one of these:
@@ -90,6 +91,9 @@ def convert(cls, bytestring=None, **kwargs):
 
         :param write_to: The filename of file-like object where to write the
                          output. If None or not provided, return a byte string.
+        :param draw_text_as_text: Draw text as text, instead of paths, reducing
+                         the file size in PDFs and allowing text selection. May
+                         not support some path clipping operations.
 
         Only ``source`` can be passed as a positional argument, other
         parameters are keyword-only.
@@ -105,12 +109,14 @@ def convert(cls, bytestring=None, **kwargs):
             output = io.BytesIO()
         else:
             output = write_to
-        cls(tree, output, dpi, None, parent_width, parent_height).finish()
+        cls(tree, output, dpi, None, parent_width, parent_height,
+            draw_text_as_text).finish()
         if write_to is None:
             return output.getvalue()
 
     def __init__(self, tree, output, dpi, parent_surface=None,
-                 parent_width=None, parent_height=None):
+                 parent_width=None, parent_height=None,
+                 draw_text_as_text=None):
         """Create the surface from a filename or a file-like object.
 
         The rendered content is written to ``output`` which can be a filename,
@@ -144,6 +150,8 @@ def __init__(self, tree, output, dpi, parent_surface=None,
         self._old_parent_node = self.parent_node = None
         self.output = output
         self.dpi = dpi
+        if draw_text_as_text is not None:
+            self.draw_text_as_text = draw_text_as_text
         self.font_size = size(self, "12pt")
         self.stroke_and_fill = True
         width, height, viewbox = node_format(self, tree)
@@ -323,6 +331,9 @@ def draw(self, node):
                 self.context.clip()
                 self.context.set_fill_rule(cairo.FILL_RULE_WINDING)
 
+        save_cursor = (self.cursor_position, self.cursor_d_position,
+                       self.text_path_width)
+
         if node.tag in TAGS:
             try:
                 TAGS[node.tag](self, node)
@@ -362,7 +373,12 @@ def draw(self, node):
                 if node.get("fill-rule") == "evenodd":
                     self.context.set_fill_rule(cairo.FILL_RULE_EVEN_ODD)
                 self.context.set_source_rgba(*color(paint_color, fill_opacity))
-            self.context.fill_preserve()
+            if self.draw_text_as_text and TAGS[node.tag] == text:
+                self.cursor_position, self.cursor_d_position, \
+                    self.text_path_width = save_cursor
+                text(self, node)
+            else:
+                self.context.fill_preserve()
             self.context.restore()
 
             # Stroke

cairosvg/surface/text.py

@@ -205,7 +205,12 @@ def text(surface, node):
                 surface.context.rel_move_to(-x_align, y_align)
                 surface.context.rotate(last_r if r is None else r)
 
-            surface.context.text_path(letter)
+            # Only draw characters with 'content' (workaround for bug in cairo)
+            if not letter.isspace():
+                if surface.draw_text_as_text:
+                    surface.context.show_text(letter)
+                else:
+                    surface.context.text_path(letter)
             surface.context.restore()
             if not text_path:
                 surface.cursor_position = cursor_position