On Thu, May 16, 2013 at 09:29:07AM -0400, Joaquim Rocha wrote: > Producing a new SVG layout for the tablets can be a laborious task and should > be automated. > > With that in mind I wrote a Python script to clean the SVG produced by tools > such as Inkscape and to automate certain things, like naming elements of > interest. > No other libs/modules are needs apart from Python itself. > > The script if far from perfect and needs to be tested but it has been already > helpful to me. > I also updated the README in the data/layouts to show how to use it. > > Hopefully you'll also find it useful and we'll include it in the next release.
yep, very useful. thanks. I noticed that the output format is different to the current layout (well, for the bamboo one I looked at anyway). it would make sense to run this over all the current layouts to get a unified description - any disagreement? > From 903c811af2532452ad8a6aed2ca479f09fd52354 Mon Sep 17 00:00:00 2001 > From: Joaquim Rocha <jro...@redhat.com> > Date: Thu, 16 May 2013 15:21:32 +0200 > Subject: [PATCH] tools: Add clean_svg.py script > > To help cleaning the SVG produced by editors such as Inkscape. > It also automates certain things like naming the elements. please sign-off patches to libwacom (or xf86-input-wacom, for that matter) > --- > data/layouts/README | 23 +++++ > tools/clean_svg.py | 280 > ++++++++++++++++++++++++++++++++++++++++++++++++++++ > 2 files changed, 303 insertions(+) > create mode 100755 tools/clean_svg.py > > diff --git a/data/layouts/README b/data/layouts/README > index 9da2321..ae23343 100644 > --- a/data/layouts/README > +++ b/data/layouts/README > @@ -227,3 +227,26 @@ Second touch-strip: > > id="LeaderStrip2Down" > class="Strip2Down Strip2 Leader" > + > +* Tips For Creating New Layouts > + > +Layouts use very simple SVG rules. WISIWYG editors such as Inkscape are > +very convenient to design new layouts but usually produce a much more > +complex SVG markup so files that are produced with those editors should > +be cleaned. To help with this task, there is a script called clean_svg.py > +in the tools folder. > +Besides cleaning the markup and removing editor specific tags, clean_svg.py > +also automates the naming of the elements. > + > + * Automatic Naming with Inkscape and clean_svg.py > + > + On Inkscape, be sure to group the button, leader and label elements > + and assign the group's ID to the desired logical name. E.g.: Assigning > + "A" to the group's ID and running the clean_svg.py script with that > + SVG, will assign "ButtonA"/"B Button" to the ID and class of the first > + rect/circle element found in the group; it also analogously assigns the ID > and class of the first path and text elements found in that group. I've moved this onto two shorter lines. > + > +clean_svg.py needs two arguments, the SVG file path and the name of > +the tablet, e.g.: > + > + ./clean_svg.py /path/to/svg_file.svg "My Brand New Tablet Name" > diff --git a/tools/clean_svg.py b/tools/clean_svg.py > new file mode 100755 > index 0000000..0ad72c6 > --- /dev/null > +++ b/tools/clean_svg.py > @@ -0,0 +1,280 @@ > +#! /usr/bin/env python > +# > +# Copyright (c) 2011 Red Hat, Inc. is this supposed to be 2013? > +# > +# This program is free software; you can redistribute it and/or modify > +# it under the terms of the GNU General Public License as published by > +# the Free Software Foundation; either version 2 of the License, or > +# (at your option) any later version. > +# > +# This program is distributed in the hope that it will be useful, > +# but WITHOUT ANY WARRANTY; without even the implied warranty of > +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > +# GNU General Public License for more details. > +# > +# You should have received a copy of the GNU General Public License > +# along with this program; if not, write to the Free Software > +# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. > +# Author: Joaquim Rocha <jro...@redhat.com> > +# > + > +import sys > +from argparse import ArgumentParser > +from xml.etree import ElementTree as ET > + > +NAMESPACE = "http://www.w3.org/2000/svg"; > +BRACKETS_NAMESPACE = '{' + NAMESPACE + '}' > + > +# List with the attributes as they should be sorted > +ATTRS_ORDER = ['id', 'class', 'x', 'y', 'cx', 'cy', 'width', 'height'] > + > +def human_round(number): > + ''' > + Round to closest .5 > + ''' > + return round(number * 2) / 2.0 > + > +def traverse_and_clean(node): > + ''' > + Clean the tree recursively > + ''' > + # Remove any non-SVG namespace attributes > + for key in node.attrib.keys(): > + if key.startswith('{'): > + del node.attrib[key] > + if node.tag == BRACKETS_NAMESPACE + 'g' and node.attrib.has_key('id'): node_tag_is? or better even a node_tag(foo) function that returns the non-namespaced tag so you can write if node_tag(node) == 'g': which is more intuitive to read (imho) or, if this is actually possible, run through the tree once and change every node.tag to a NS-stripped tag so you don't have to worry about it. this is a once-off script, we can afford it if it takes a bit longer. > + apply_id_and_class_from_group(node) > + del node.attrib['id'] > + if node.attrib.has_key('style'): > + if node.tag == BRACKETS_NAMESPACE + 'text': > + node.attrib['style'] = 'text-anchor:start;' > + elif node.tag != BRACKETS_NAMESPACE + 'svg': > + del node.attrib['style'] > + > + remove_transform_if_exists(node) > + > + round_attrib(node, 'd', 'x', 'y', 'rx', 'ry', 'width', 'height', 'cx', > 'cy', 'r') > + > + nodes = node.getchildren() > + for node in nodes: > + traverse_and_clean(node) this overrides the 'node' arg. no big deal, but could lead to confusing later. for n in node.getchildren(): would be better here > + > +def round_attrib(node, *attrs): > + for attr_name in attrs: > + attr_value = node.attrib.get(attr_name) > + if attr_value is None: > + continue > + if attr_name == 'd': > + d = attr_value.replace(',', ' ') > + values = [round_if_number(value) for value in d.split()] couldn't you just use d.split(",") here? > + node.attrib[attr_name] = ' '.join(values) > + else: > + node.attrib[attr_name] = round_if_number(attr_value) if attr_name == 'd': values = d.split(",") else: values = [] + [attr_value] node.attrib(" ".join([round_if_number(value) for value in values])) (ok, this is really just compressing things but was a good idea in my head :) > + > +def round_if_number(value): > + try: > + value = human_round(float(value.strip())) > + except ValueError: > + pass > + return str(value) > + > +def remove_non_svg_nodes(root): > + for elem in root.getchildren(): > + if not elem.tag.startswith(BRACKETS_NAMESPACE) or \ > + elem.tag == BRACKETS_NAMESPACE + 'metadata': node_tag_is (or node_tag) > + root.remove(elem) > + else: > + remove_non_svg_nodes(elem) > + > +def remove_transform_if_exists(node): > + TRANSLATE = 'translate' > + MATRIX = 'matrix' > + > + transform = node.attrib.get('transform') > + if transform is None: > + return > + transform = transform.strip() > + > + if transform.startswith(TRANSLATE): > + values = transform[len(TRANSLATE) + 1:-1].split(',') > + try: > + x, y = float(values[0]), float(values[1]) > + except: > + return > + > + apply_translation(node, 1, 0, 0, 1, x, y) > + elif transform.startswith(MATRIX): > + values = transform[len(MATRIX) + 1:-1].split(',') > + a, b, c, d, e, f = [float(value.strip()) for value in values] > + apply_translation(node, a, b, c, d, e, f) > + apply_scaling(node, a, d) > + del node.attrib['transform'] you have a try/except for TRANSLATE but not for MATRIX. is this intentional? > + > +def apply_translation(node, a, b, c, d, e, f): > + x_attr, y_attr = 'x', 'y' > + if node_tag_is(node, 'circle'): > + x_attr, y_attr = 'cx', 'cy' > + elif node_tag_is(node, 'path'): > + apply_translation_to_path(node, e, f) > + return > + try: > + x, y = float(node.attrib[x_attr]), float(node.attrib[y_attr]) > + new_x = x * a + y * c + 1 * e > + new_y = x * b + y * d + 1 * f > + node.attrib[x_attr] = str(new_x) > + node.attrib[y_attr] = str(new_y) > + except: > + pass > + > +def apply_translation_to_path(node, x, y): > + d = node.attrib.get('d') > + if d is None: > + return > + d = d.replace(',', ' ').split() split(',') > + m_init_index = 0 > + length = len(d) > + m_end_index = length > + operation = 'M' > + i = 0 > + while i < length: > + value = d[i] > + if value.lower() == 'm': > + operation = value > + m_init_index = i + 1 > + elif len(value) == 1 and value.isalpha(): > + m_end_index = i > + break > + i += 1 > + for i in range(m_init_index, m_end_index, 2): > + d[i] = str(float(d[i]) + x) > + d[i + 1] = str(float(d[i + 1]) + y) > + if operation == 'm': > + break > + node.attrib['d'] = ' '.join(d[:m_init_index]) + ' ' + ' > '.join(d[m_init_index:m_end_index]) + ' '.join(d[m_end_index:]) this could really do with a comment on at least what type of data this handles. looking at the first bamboo svg, data is e.g. M 58 100 l 20 0 and this function addds x/y to 58/100. except that it tries to cope with anything that's not M and 2 values anyway? having said that, it would be more intuitive to split this up into multiple lists so you don't have to handle the init/end index. > + > + > +def apply_scaling(node, x, y): > + w_attr, h_attr = 'width', 'height' > + if node_tag_is(node, 'circle'): > + r = float(node.attrib.get('r', 1.0)) > + node.attrib['r'] = str(r * x) > + try: > + w = float(node.attrib[w_attr]) > + h = float(node.attrib[h_attr]) > + node.attrib[w_attr] = str(w * x) > + node.attrib[h_attr] = str(h * y) > + except: > + pass > + > +def node_tag_is(node, tag): > + return node.tag == BRACKETS_NAMESPACE + tag > + > +def to_string_rec(node, level=0): > + i = '\n' + level * ' ' please don't use 'i' here, that's for loops only > + > + tag_name = node.tag[len(BRACKETS_NAMESPACE):] > + > + # Remove 'defs' element. This cannot be done in the traver_and_clean > + # because somehow it is never found > + if tag_name == 'defs': > + return '' > + > + # use a list to put id and class as the first arguments > + attribs = [] > + for attr in ATTRS_ORDER: > + attr_value = node.attrib.get(attr) > + if attr_value is not None: > + attribs.append(i + ' %s="%s"' % (attr, attr_value)) > + del node.attrib[attr] > + attribs += [i + ' %s="%s"' % (key, value) for key, value in > node.attrib.items()] > + > + string = i + '<' + tag_name + ''.join(attribs) > + if len(node) or node.text: > + string += '>' > + if not node.text or not node.text.strip(): > + node.text = i + ' ' > + else: > + string += node.text > + if list(node): > + for child in get_node_children_sorted(node): > + string += to_string_rec(child, level+1) > + string += i > + string += '</%s>' % tag_name > + else: > + string += ' />' > + return string > + > +def get_node_children_sorted(node): > + ''' > + Returns a list with the children rect nodes, > + the path nodes and any other nodes, in this order. > + ''' > + other_nodes_index = 1 > + children = [] > + for child in node.getchildren(): > + if node_tag_is(child, 'rect') or node_tag_is(child, 'circle'): > + children = [child] + children > + other_nodes_index += 1 > + elif node_tag_is(child, 'path'): > + children.insert(other_nodes_index, child) > + else: > + children.append(child) > + return children try this (with adjustment to get the tag into the right namespaced tag) def custom_tag_sort(arg): ''' Use as key functon in sorted(). Pre-fix arg tag name by a number in the sort order we want. Anything unspecified defaults to 9. i.e. circle → 1circle, thus sorts lower than other tags. ''' tag_order = { 'circle' : 1, 'rect' : 2, 'path' : 3 } return str(tag_order.get(arg.tag, 9)) + arg.tag def get_node_children_sorted(node): return sorted(children, key=custom_tag_sort) easier to maintain, I think. plus we're guaranteed that all other children will be sorted too, not sure if node.getchildren() returns a sorted list by default. could do the same for the attrib sort above. Cheers, Peter > + > +def apply_id_and_class_from_group(group_node): > + button_assigned = label_assigned = path_assigned = False > + _id = group_node.attrib.get('id') > + if _id is None: > + return > + for child in group_node.getchildren(): > + if node_tag_is(child, 'rect') or node_tag_is(child, 'circle'): > + if button_assigned: > + continue > + child.attrib['id'] = 'Button%s' % _id > + child.attrib['class'] = '%s Button' % _id > + button_assigned = True > + elif node_tag_is(child, 'path'): > + if path_assigned: > + continue > + child.attrib['id'] = 'Leader%s' % _id > + child.attrib['class'] = '%s Leader' % _id > + path_assigned = True > + elif node_tag_is(child, 'text'): > + if label_assigned: > + continue > + child.attrib['id'] = 'Label%s' % _id > + child.attrib['class'] = '%s Label' % _id > + child.text = _id > + label_assigned = True > + > +def to_string(root): > + header = '''<?xml version="1.0" standalone="no"?> > +<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" > + "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd";>''' > + return header + to_string_rec(root) > + > +def clean_svg(root, tabletname): > + remove_non_svg_nodes(root) > + title = root.find(BRACKETS_NAMESPACE + 'title') > + if title is not None: > + title.text = tabletname > + root.attrib['xmlns'] = 'http://www.w3.org/2000/svg' > + traverse_and_clean(root) > + > +if __name__ == '__main__': > + parser = ArgumentParser(description='Clean SVG files for libwacom') > + parser.add_argument('filename', nargs=1, type=str, > + help='SVG file to clean', metavar='FILE') > + parser.add_argument('tabletname', nargs=1, type=str, > + help='The name of the tablet', metavar='TABLET_NAME') > + args = parser.parse_args() > + > + ET.register_namespace('', NAMESPACE) > + try: > + tree = ET.parse(args.filename[0]) > + except Exception, e: > + sys.stderr.write(str(e) + '\n') > + exit(1) > + root = tree.getroot() > + clean_svg(root, args.tabletname[0]) > + print to_string(root) > -- > 1.8.1.4 > ------------------------------------------------------------------------------ Try New Relic Now & We'll Send You this Cool Shirt New Relic is the only SaaS-based application performance monitoring service that delivers powerful full stack analytics. Optimize and monitor your browser, app, & servers with just a few lines of code. Try New Relic and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may _______________________________________________ Linuxwacom-devel mailing list Linuxwacom-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxwacom-devel
