5 | :Date: 21st Jan 2011
6 |
7 | Overview
8 | --------
9 |
10 | SimSearch is a dictionary search-by-similarity interface for Japanese kanji,
11 | providing a nice front-end for Kanjidic. It lets you find a kanji you don't
12 | know, using kanji that are visually similar.
13 |
14 | .. image:: http://files.gakusha.info/simsearch/homepage/ss-example-med.png
15 |
16 | If you're viewing this source code, you should be a developer, or someone at
17 | least a little comfortable with Python.
18 |
19 | Developing/running
20 | ------------------
21 |
22 | This is a quick guide to getting SimSearch up and running locally.
23 |
24 | Dependencies
25 | ~~~~~~~~~~~~
26 |
27 | SimSearch uses MongoDB as its database backend. If you don't already have it,
28 | install MongoDB first. By default, it will create and use a database called
29 | ``simsearch`` in MongoDB.
30 |
31 | Next, you need Python (2.6/2.7), pip and virtualenv. Then you can install the
32 | necessary packages in an environment for simsearch::
33 |
34 | $ pip -E ss-env install ./simsearch
35 |
36 | Occasionally a dependency will fail to install cleanly (e.g. NLTK). In that
37 | case, you will need to download a package for it, enter the virtual
38 | environment and install the package from there::
39 |
40 | $ tar xfz nltk-v2.08b.tgz
41 | $ cd nltk-v2.08b
42 | $ source /path/to/simsearch/ss-env/bin/activate
43 | (ss-env) $ python setup.py install
44 |
45 | Building and running
46 | ~~~~~~~~~~~~~~~~~~~~
47 |
48 | Once installed, build the database with::
49 |
50 | $ python -m simsearch.models
51 | Building similarity matrix
52 | Building neighbourhood graph
53 |
54 | You can then run the debug server with the command ``simsearch.py``. The
55 | server will be available at http://localhost:5000/.
56 |
57 | Deploying
58 | ---------
59 |
60 | Please see Flask documentation around deployment. Feel free to email me as
61 | well, if you have any issues.
62 |
63 |
--------------------------------------------------------------------------------
/simsearch/experiments/evaluate_paths.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # -*- coding: utf-8 -*-
3 | #
4 | # evaluate_paths.py
5 | # simsearch
6 | #
7 | # Created by Lars Yencken on 03-09-2010.
8 | # Copyright 2010 Lars Yencken. All rights reserved.
9 | #
10 |
11 | """
12 | A script to generate statistics on a set of query traces (i.e. walks through
13 | the kanji graph generated by simulated search).
14 | """
15 |
16 | import os, sys, optparse
17 |
18 | from simplestats import basic_stats
19 |
20 | from simulate_search import TraceFile
21 |
22 | def evaluate_paths(input_file, limit=5):
23 | print 'Evaluating paths from "%s"' % os.path.basename(input_file)
24 | traces = TraceFile.load(input_file)
25 |
26 | path_lengths = []
27 | successes = []
28 | for (query, target, path) in traces:
29 | if path and path[-1] == target:
30 | successes.append(path)
31 | path_lengths.append(len(path) - 1)
32 | else:
33 | path_lengths.append(limit)
34 |
35 | print u'Success rate: %d/%d (%.02f%%)' % (len(successes),
36 | len(traces), 100.0 * len(successes) / len(traces))
37 |
38 | print u'Mean path length: %.02f (σ = %.02f)' % basic_stats(path_lengths)
39 |
40 | #----------------------------------------------------------------------------#
41 |
42 | def _create_option_parser():
43 | usage = \
44 | """%prog [options] input_file
45 |
46 | Generates evaluation statistics on a collection of traces."""
47 |
48 | parser = optparse.OptionParser(usage)
49 |
50 | return parser
51 |
52 | def main(argv):
53 | parser = _create_option_parser()
54 | (options, args) = parser.parse_args(argv)
55 |
56 | if len(args) != 1:
57 | parser.print_help()
58 | sys.exit(1)
59 |
60 | input_file, = args
61 | evaluate_paths(input_file)
62 |
63 | #----------------------------------------------------------------------------#
64 |
65 | if __name__ == '__main__':
66 | main(sys.argv[1:])
67 |
68 | # vim: ts=4 sw=4 sts=4 et tw=78:
69 |
--------------------------------------------------------------------------------
/simsearch/experiments/check_connectivity.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # -*- coding: utf-8 -*-
3 | #
4 | # check_connectivity.py
5 | # simsearch
6 | #
7 | # Created by Lars Yencken on 03-09-2010.
8 | # Copyright 2010 Lars Yencken. All rights reserved.
9 | #
10 |
11 | """
12 | Performs a connectivity check on the search graph, determining how many kanji
13 | occur within the top-k neighbour list of at least one other kanji.
14 | """
15 |
16 | import sys
17 | import optparse
18 |
19 | from simsearch import settings
20 | from simsearch.search import models
21 |
22 | def check_connectivity(k=settings.N_NEIGHBOURS_RECALLED):
23 | kanji_set = models.Node.get_coverage()
24 | covered_set = set()
25 | for node in models.Node.objects:
26 | covered_set.update(n.kanji for n in node.neighbours[:k])
27 |
28 | print '%d/%d (%.02f%%) covered' % (len(covered_set), len(kanji_set),
29 | 100.0 * len(covered_set) / len(kanji_set))
30 |
31 | #----------------------------------------------------------------------------#
32 |
33 | def _create_option_parser():
34 | usage = \
35 | """%prog [options]
36 |
37 | Performs a connectivity check on the neighbour database to determine how
38 | many kanji are realistically accessible."""
39 |
40 | parser = optparse.OptionParser(usage)
41 |
42 | parser.add_option('-k', action='store', dest='k', type='int',
43 | default=settings.N_NEIGHBOURS_RECALLED,
44 | help='Connected kanji must occur within top-k neighbours [%d]' \
45 | % settings.N_NEIGHBOURS_RECALLED)
46 |
47 | return parser
48 |
49 | def main(argv):
50 | parser = _create_option_parser()
51 | (options, args) = parser.parse_args(argv)
52 |
53 | if args:
54 | parser.print_help()
55 | sys.exit(1)
56 |
57 | check_connectivity(k=options.k)
58 |
59 | #----------------------------------------------------------------------------#
60 |
61 | if __name__ == '__main__':
62 | main(sys.argv[1:])
63 |
64 | # vim: ts=4 sw=4 sts=4 et tw=78:
65 |
--------------------------------------------------------------------------------
/simsearch/settings.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | #
3 | # settings.py
4 | # simsearch
5 | #
6 | # Created by Lars Yencken on 24-08-2010.
7 | # Copyright 2010 Lars Yencken. All rights reserved.
8 | #
9 |
10 | """
11 | Settings for the simsearch project.
12 | """
13 |
14 | import os
15 |
16 | import mongoengine
17 |
18 | # custom MongoDB connection settings
19 | MONGODB_NAME = 'simsearch'
20 | MONGODB_USERNAME = None
21 | MONGODB_PASSWORD = None
22 | MONGODB_HOST = 'localhost'
23 | MONGODB_PORT = 27017
24 |
25 | UTF8_BYTES_PER_CHAR = 3 # for CJK chars
26 |
27 | N_NEIGHBOURS_STORED = 100
28 |
29 | N_NEIGHBOURS_RECALLED = 15
30 |
31 | GOOGLE_ANALYTICS_CODE = None
32 |
33 | # Tradeoff in Pr(a|s) and likelihood of reaching a further target from s'
34 | UPDATE_GAMMA = 0.7
35 |
36 | PROJECT_ROOT = os.path.dirname(__file__)
37 |
38 | # Absolute path to the directory that holds media.
39 | # Example: "/home/media/media.lawrence.com/"
40 | MEDIA_ROOT = os.path.join(PROJECT_ROOT, 'media')
41 |
42 | # URL that handles the media served from MEDIA_ROOT. Make sure to use a
43 | # trailing slash if there is a path component (optional in other cases).
44 | # Examples: "http://media.lawrence.com", "http://example.com/media/"
45 | MEDIA_URL = '/static/'
46 |
47 | # The source of stroke data for each character
48 | STROKE_SOURCE = None
49 |
50 | # The source of frequency counts for each character
51 | FREQ_SOURCE = None
52 |
53 | try:
54 | from local_settings import *
55 | except ImportError:
56 | pass
57 |
58 | # connect to our database
59 | import mongoengine
60 | mongoengine.connect(MONGODB_NAME, username=MONGODB_USERNAME,
61 | password=MONGODB_PASSWORD, host=MONGODB_HOST, port=MONGODB_PORT)
62 |
63 | # static data files needed for building
64 | DATA_DIR = os.path.join(PROJECT_ROOT, 'data')
65 |
66 | # default stroke source
67 | if STROKE_SOURCE is None:
68 | STROKE_SOURCE = os.path.join(DATA_DIR, 'strokes_ulrich')
69 |
70 | # default frequency source
71 | if FREQ_SOURCE is None:
72 | FREQ_SOURCE = os.path.join(DATA_DIR, 'jp_char_corpus_counts.gz')
73 |
74 | # vim: ts=4 sw=4 sts=4 et tw=78:
75 |
--------------------------------------------------------------------------------
/simsearch/heap_cache.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | #
3 | # heap_cache.py
4 | # simsearch
5 | #
6 | # Created by Lars Yencken on 30-08-2010.
7 | # Copyright 2010 Lars Yencken. All rights reserved.
8 | #
9 |
10 | """
11 | Caches to aid similarity caculation, to efficiently maintain only the highest
12 | similarity neighbours.
13 | """
14 |
15 | import heapq
16 |
17 | class TopNHeap(object):
18 | "A heap which only keeps the top-n items and their weights."
19 | __slots__ = '_n', '_backing_list'
20 | def __init__(self, n):
21 | self._n = n
22 | self._backing_list = []
23 |
24 | def add(self, item, weight):
25 | heapq.heappush(self._backing_list, (weight, item))
26 | if len(self._backing_list) > self._n:
27 | heapq.heappop(self._backing_list)
28 |
29 | def get_contents(self):
30 | return self._backing_list
31 |
32 | class FixedSimilarityCache(object):
33 | """
34 | A kanji similarity cache which only keeps the top-n most similar
35 | neighbours.
36 | """
37 | def __init__(self, n):
38 | self._n = n
39 | self._heaps = {}
40 | self._sum = 0.0
41 | self._n_seen = 0.0
42 | self._sum_squared = 0.0
43 |
44 | def add(self, kanji_a, kanji_b, similarity):
45 | """
46 | Attempt to add this similarity score to the cache. If there are
47 | already n closer neighbours for either kanji it will be discarded.
48 | """
49 | self.get_heap(kanji_a).add(kanji_b, similarity)
50 | self.get_heap(kanji_b).add(kanji_a, similarity)
51 | self._n_seen += 1
52 | self._sum += similarity
53 | self._sum_squared += similarity * similarity
54 |
55 |
56 | def __getitem__(self, kanji):
57 | return self.get_heap(kanji)
58 |
59 | def get_heap(self, kanji):
60 | heap = self._heaps.get(kanji)
61 | if heap is None:
62 | heap = self._heaps.setdefault(kanji, TopNHeap(self._n))
63 | return heap
64 |
65 | def get_mean(self):
66 | return self._sum / self._n_seen
67 |
68 | # vim: ts=4 sw=4 sts=4 et tw=78:
69 |
70 |
--------------------------------------------------------------------------------
/simsearch/templates/base.html:
--------------------------------------------------------------------------------
1 |
3 |
4 |
5 |
6 |
7 | Visual kanji search
9 |
27 | Search /
28 | Help /
29 | Feedback /
30 | About
31 |
32 | {% endblock %}
33 |
34 | {% block content %}
35 | {% endblock %}
36 |
37 | {% block copy %}
38 |
39 | Visual kanji search (r{{revision.number}}:{{revision.short}} )
41 |
42 | {% endblock %}
43 |
44 | {% if settings.GOOGLE_ANALYTICS_CODE %}
45 |
56 | {% endif %}
57 |
58 |
59 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/src/print.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | print.css
4 | * Gives you some sensible styles for printing pages.
5 | * See Readme file in this directory for further instructions.
6 |
7 | Some additions you'll want to make, customized to your markup:
8 | #header, #footer, #navigation { display:none; }
9 |
10 | -------------------------------------------------------------- */
11 |
12 | body {
13 | line-height: 1.5;
14 | font-family: "Helvetica Neue", Arial, Helvetica, sans-serif;
15 | color:#000;
16 | background: none;
17 | font-size: 10pt;
18 | }
19 |
20 |
21 | /* Layout
22 | -------------------------------------------------------------- */
23 |
24 | .container {
25 | background: none;
26 | }
27 |
28 | hr {
29 | background:#ccc;
30 | color:#ccc;
31 | width:100%;
32 | height:2px;
33 | margin:2em 0;
34 | padding:0;
35 | border:none;
36 | }
37 | hr.space {
38 | background: #fff;
39 | color: #fff;
40 | visibility: hidden;
41 | }
42 |
43 |
44 | /* Text
45 | -------------------------------------------------------------- */
46 |
47 | h1,h2,h3,h4,h5,h6 { font-family: "Helvetica Neue", Arial, "Lucida Grande", sans-serif; }
48 | code { font:.9em "Courier New", Monaco, Courier, monospace; }
49 |
50 | a img { border:none; }
51 | p img.top { margin-top: 0; }
52 |
53 | blockquote {
54 | margin:1.5em;
55 | padding:1em;
56 | font-style:italic;
57 | font-size:.9em;
58 | }
59 |
60 | .small { font-size: .9em; }
61 | .large { font-size: 1.1em; }
62 | .quiet { color: #999; }
63 | .hide { display:none; }
64 |
65 |
66 | /* Links
67 | -------------------------------------------------------------- */
68 |
69 | a:link, a:visited {
70 | background: transparent;
71 | font-weight:700;
72 | text-decoration: underline;
73 | }
74 |
75 | a:link:after, a:visited:after {
76 | content: " (" attr(href) ")";
77 | font-size: 90%;
78 | }
79 |
80 | /* If you're having trouble printing relative links, uncomment and customize this:
81 | (note: This is valid CSS3, but it still won't go through the W3C CSS Validator) */
82 |
83 | /* a[href^="/"]:after {
84 | content: " (http://www.yourdomain.com" attr(href) ") ";
85 | } */
86 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/src/forms.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | forms.css
4 | * Sets up some default styling for forms
5 | * Gives you classes to enhance your forms
6 |
7 | Usage:
8 | * For text fields, use class .title or .text
9 | * For inline forms, use .inline (even when using columns)
10 |
11 | -------------------------------------------------------------- */
12 |
13 | label { font-weight: bold; }
14 | fieldset { padding:1.4em; margin: 0 0 1.5em 0; border: 1px solid #ccc; }
15 | legend { font-weight: bold; font-size:1.2em; }
16 |
17 |
18 | /* Form fields
19 | -------------------------------------------------------------- */
20 |
21 | input[type=text], input[type=password],
22 | input.text, input.title,
23 | textarea, select {
24 | background-color:#fff;
25 | border:1px solid #bbb;
26 | }
27 | input[type=text]:focus, input[type=password]:focus,
28 | input.text:focus, input.title:focus,
29 | textarea:focus, select:focus {
30 | border-color:#666;
31 | }
32 |
33 | input[type=text], input[type=password],
34 | input.text, input.title,
35 | textarea, select {
36 | margin:0.5em 0;
37 | }
38 |
39 | input.text,
40 | input.title { width: 300px; padding:5px; }
41 | input.title { font-size:1.5em; }
42 | textarea { width: 390px; height: 250px; padding:5px; }
43 |
44 | input[type=checkbox], input[type=radio],
45 | input.checkbox, input.radio {
46 | position:relative; top:.25em;
47 | }
48 |
49 | form.inline { line-height:3; }
50 | form.inline p { margin-bottom:0; }
51 |
52 |
53 | /* Success, notice and error boxes
54 | -------------------------------------------------------------- */
55 |
56 | .error,
57 | .notice,
58 | .success,
59 | .info { padding: 0.8em; margin-bottom: 1em; border: 2px solid #ddd; }
60 |
61 | .error { background: #fbe3e4; color: #8a1f11; border-color: #fbc2c4; }
62 | .notice { background: #fff6bf; color: #514721; border-color: #ffd324; }
63 | .success { background: #e6efc2; color: #264409; border-color: #c6d880; }
64 | .info { background: #d5edf8; color: #205791; border-color: #92cae4; }
65 | .error a { color: #8a1f11; }
66 | .notice a { color: #514721; }
67 | .success a { color: #264409; }
68 | .info a { color: #205791; }
69 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/plugins/buttons/screen.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | buttons.css
4 | * Gives you some great CSS-only buttons.
5 |
6 | Created by Kevin Hale [particletree.com]
7 | * particletree.com/features/rediscovering-the-button-element
8 |
9 | See Readme.txt in this folder for instructions.
10 |
11 | -------------------------------------------------------------- */
12 |
13 | a.button, button {
14 | display:block;
15 | float:left;
16 | margin: 0.7em 0.5em 0.7em 0;
17 | padding:5px 10px 5px 7px; /* Links */
18 |
19 | border:1px solid #dedede;
20 | border-top:1px solid #eee;
21 | border-left:1px solid #eee;
22 |
23 | background-color:#f5f5f5;
24 | font-family:"Lucida Grande", Tahoma, Arial, Verdana, sans-serif;
25 | font-size:100%;
26 | line-height:130%;
27 | text-decoration:none;
28 | font-weight:bold;
29 | color:#565656;
30 | cursor:pointer;
31 | }
32 | button {
33 | width:auto;
34 | overflow:visible;
35 | padding:4px 10px 3px 7px; /* IE6 */
36 | }
37 | button[type] {
38 | padding:4px 10px 4px 7px; /* Firefox */
39 | line-height:17px; /* Safari */
40 | }
41 | *:first-child+html button[type] {
42 | padding:4px 10px 3px 7px; /* IE7 */
43 | }
44 | button img, a.button img{
45 | margin:0 3px -3px 0 !important;
46 | padding:0;
47 | border:none;
48 | width:16px;
49 | height:16px;
50 | float:none;
51 | }
52 |
53 |
54 | /* Button colors
55 | -------------------------------------------------------------- */
56 |
57 | /* Standard */
58 | button:hover, a.button:hover{
59 | background-color:#dff4ff;
60 | border:1px solid #c2e1ef;
61 | color:#336699;
62 | }
63 | a.button:active{
64 | background-color:#6299c5;
65 | border:1px solid #6299c5;
66 | color:#fff;
67 | }
68 |
69 | /* Positive */
70 | body .positive {
71 | color:#529214;
72 | }
73 | a.positive:hover, button.positive:hover {
74 | background-color:#E6EFC2;
75 | border:1px solid #C6D880;
76 | color:#529214;
77 | }
78 | a.positive:active {
79 | background-color:#529214;
80 | border:1px solid #529214;
81 | color:#fff;
82 | }
83 |
84 | /* Negative */
85 | body .negative {
86 | color:#d12f19;
87 | }
88 | a.negative:hover, button.negative:hover {
89 | background-color:#fbe3e4;
90 | border:1px solid #fbc2c4;
91 | color:#d12f19;
92 | }
93 | a.negative:active {
94 | background-color:#d12f19;
95 | border:1px solid #d12f19;
96 | color:#fff;
97 | }
98 |
--------------------------------------------------------------------------------
/simsearch/templates/translate/kanji.html:
--------------------------------------------------------------------------------
1 | {% extends "base.html" %}
2 |
3 | {% block headers %}
4 |
33 |
34 |
35 |
36 |
Readings
37 |
38 |
39 |
40 | On
41 |
42 | {% for r in translation.on_readings %}
43 | {{r}}{% if not loop.last %},{% endif %}
44 | {% endfor %}
45 |
46 |
47 |
48 |
49 | Kun
50 |
51 | {% for r in translation.kun_readings %}
52 | {{r}}{% if not loop.last %},{% endif %}
53 | {% endfor %}
54 |
55 |
56 |
57 |
58 |
59 |
Meaning
60 |
61 |
62 | {% for v in translation.glosses %}
63 | {{v}}
64 | {% endfor %}
65 |
66 |
67 |
Find words
68 |
73 |
74 |
asdf
17 | Best used on prepositions and ampersands. */
18 |
19 | .alt {
20 | color: #666;
21 | font-family: "Warnock Pro", "Goudy Old Style","Palatino","Book Antiqua", Georgia, serif;
22 | font-style: italic;
23 | font-weight: normal;
24 | }
25 |
26 |
27 | /* For great looking quote marks in titles, replace "asdf" with:
28 | “ asdf”
29 | (That is, when the title starts with a quote mark).
30 | (You may have to change this value depending on your font size). */
31 |
32 | .dquo { margin-left: -.5em; }
33 |
34 |
35 | /* Reduced size type with incremental leading
36 | (http://www.markboulton.co.uk/journal/comments/incremental_leading/)
37 |
38 | This could be used for side notes. For smaller type, you don't necessarily want to
39 | follow the 1.5x vertical rhythm -- the line-height is too much.
40 |
41 | Using this class, it reduces your font size and line-height so that for
42 | every four lines of normal sized type, there is five lines of the sidenote. eg:
43 |
44 | New type size in em's:
45 | 10px (wanted side note size) / 12px (existing base size) = 0.8333 (new type size in ems)
46 |
47 | New line-height value:
48 | 12px x 1.5 = 18px (old line-height)
49 | 18px x 4 = 72px
50 | 72px / 5 = 14.4px (new line height)
51 | 14.4px / 10px = 1.44 (new line height in em's) */
52 |
53 | p.incr, .incr p {
54 | font-size: 10px;
55 | line-height: 1.44em;
56 | margin-bottom: 1.5em;
57 | }
58 |
59 |
60 | /* Surround uppercase words and abbreviations with this class.
61 | Based on work by Jørgen Arnor Gårdsø Lom [http://twistedintellect.com/] */
62 |
63 | .caps {
64 | font-variant: small-caps;
65 | letter-spacing: 1px;
66 | text-transform: lowercase;
67 | font-size:1.2em;
68 | line-height:1%;
69 | font-weight:bold;
70 | padding:0 2px;
71 | }
72 |
--------------------------------------------------------------------------------
/simsearch/templates/static/about.html:
--------------------------------------------------------------------------------
1 | {% extends "static/base.html" %}
2 |
3 | {% block inner_content %}
4 | About SimSearch
5 |
6 | What is it?
7 |
8 | SimSearch is a visual search-by-similarity for Japanese kanji. Suppose you
9 | encounter a kanji you don't know, but it looks very similar to one you do know.
10 | In this case, you can enter in the one you know as a query, and navigate
11 | through the similarity space until you find the kanji you are looking for.
12 |
13 | How does it work?
14 |
15 | In order to determine how similar two kanji are to one another, we use the
16 | stroke edit distance between the two kanji [1]. In other words, we
17 | look at the sequence of strokes used to write each character, and determine how
18 | many changes you'd need to make to turn one kanji's series of strokes into that
19 | of the other kanji. Research so far has that this measure best matches human
20 | judgements of similarity [2].
21 |
22 | Secondly, search is also adaptive. That is, it will adapt to whatever people
23 | actually find similar when they do searches using the system. For this, we use
24 | Q-learning [3], a well known algorithm for learning the best action to take in
25 | a state space. Q-learning has been used in a wide variety of search
26 | applications, including game players for board games which learn from
27 | experience.
28 |
29 | Fortunately, systems such as this do not need their own custom dictionaries
30 | of Japanese. For kanji translations and pronunciation, we use the excellent
31 | and free Kanjidic dictionary [4], and for layout we use RaphaelJS [5].
32 |
33 | How can I help?
34 |
35 | If you're a learner of Japanese, or a native speaker, please give the system
36 | a try, and send me some feedback . If you're a
37 | programmer, note that SimSearch is open source, so feel free to suggest new
38 | improvements, or even try running your own site.
39 |
40 | References
43 |
44 | Wikipedia: Levenshtein
46 | distance
47 | Yencken, Lars and Baldwin, Timothy: “Measuring and predicting orthographic
48 | associations: modelling the similarity of Japanese kanji”, in Proceedings of
49 | COLING 2008 , Manchester, UK (2008)
50 | Wikipedia: Q-learning
52 |
53 | The KANJIDIC Home Page (license )
54 |
55 |
56 | Raphaël -- JavaScript Library
57 |
58 |
59 | {% endblock %}
60 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/src/ie.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | ie.css
4 |
5 | Contains every hack for Internet Explorer,
6 | so that our core files stay sweet and nimble.
7 |
8 | -------------------------------------------------------------- */
9 |
10 | /* Make sure the layout is centered in IE5 */
11 | body { text-align: center; }
12 | .container { text-align: left; }
13 |
14 | /* Fixes IE margin bugs */
15 | * html .column, * html .span-1, * html .span-2,
16 | * html .span-3, * html .span-4, * html .span-5,
17 | * html .span-6, * html .span-7, * html .span-8,
18 | * html .span-9, * html .span-10, * html .span-11,
19 | * html .span-12, * html .span-13, * html .span-14,
20 | * html .span-15, * html .span-16, * html .span-17,
21 | * html .span-18, * html .span-19, * html .span-20,
22 | * html .span-21, * html .span-22, * html .span-23,
23 | * html .span-24 { display:inline; overflow-x: hidden; }
24 |
25 |
26 | /* Elements
27 | -------------------------------------------------------------- */
28 |
29 | /* Fixes incorrect styling of legend in IE6. */
30 | * html legend { margin:0px -8px 16px 0; padding:0; }
31 |
32 | /* Fixes wrong line-height on sup/sub in IE. */
33 | sup { vertical-align:text-top; }
34 | sub { vertical-align:text-bottom; }
35 |
36 | /* Fixes IE7 missing wrapping of code elements. */
37 | html>body p code { *white-space: normal; }
38 |
39 | /* IE 6&7 has problems with setting proper Help
16 |
17 | Using similarity search
18 |
19 | First of all, this site can be used as a basic kanji dictionary. Enter the
20 | kanji into the search box and press return, then click on it again, and you'll
21 | be taken to its translation. More interestingly though, it can also be used to
22 | find a kanji which you don't know how to type in . Here's how it
23 | works.
24 |
25 | Suppose you ran into the kanji 動 , which you hadn't seen before. It looks a lot like one
27 | you do know, 働 from 働く
28 | hataraku "to work". If you enter 働 as query,
29 | the search display will show you a number of kanji which look similar.
30 | Searching this way, you can quickly find the unknown kanji you want.
31 |
32 | If you enter a query, and can't immediately see the kanji you're looking
33 | for in the results, you should click on the next closest match. By doing this
34 | a few times, you should hopefully find your target kanji.
35 |
36 | Whilst this form of search is designed for single-kanji queries, you can
37 | also use it for words by querying any of the kanji in the word. When you find
38 | your target, its translation page provides single-click queries to word-level
39 | dictionaries.
40 |
41 | Other good dictionaries
42 |
43 | Whilst searching this way can be fun, it's not appropriate all the time.
44 | You might not know any similar kanji to the one you're trying to find, or you
45 | might just have a difficult one. The best strategy in general is to have
46 | a number of dictionaries you can use, and to pick the best one for the job. We
47 | recommend:
48 |
49 |
50 |
51 | Forgiving Online Kanji Search
54 |
55 | WWWJDIC
58 |
59 | Kansuke
60 | A kanji-level dictionary based on a simplified method of counting strokes.
62 |
63 |
64 | Handwritten kanji lookup
65 | Ben Bullock's handwritten kanji interface, which lets you find a kanji by drawing it.
67 |
68 |
69 | {% endblock %}
70 |
--------------------------------------------------------------------------------
/simsearch/stroke.pyx:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | #
3 | # stroke.pyx
4 | # simsearch
5 | #
6 | # Created by Lars Yencken on 03-09-2010.
7 | # Copyright 2010 Lars Yencken. All rights reserved.
8 | #
9 |
10 | """
11 | Optimised Levenstein distance calculation between stroke signatures for two
12 | kanji.
13 | """
14 |
15 | import os
16 |
17 | from cjktools.common import sopen
18 |
19 | from simsearch import settings
20 |
21 | cdef class StrokeEditDistance:
22 | """The edit distance between stroke sequences for both kanji."""
23 | cdef readonly signatures
24 | cdef readonly object stroke_types
25 | cdef readonly int n_stroke_types
26 |
27 | def __init__(self, input_file=None):
28 | self.stroke_types = {}
29 | self.n_stroke_types = 0
30 |
31 | input_file = input_file or settings.STROKE_SOURCE
32 | self.signatures = {}
33 | i_stream = sopen(input_file)
34 | for i, line in enumerate(i_stream):
35 | kanji, raw_strokes = line.rstrip().split()
36 | raw_strokes = raw_strokes.split(',')
37 | strokes = map(self.get_stroke_type, raw_strokes)
38 | self.signatures[kanji] = strokes
39 | i_stream.close()
40 |
41 | def get_stroke_type(self, stroke):
42 | try:
43 | return self.stroke_types[stroke]
44 | except KeyError:
45 | pass
46 |
47 | self.stroke_types[stroke] = self.n_stroke_types
48 | self.n_stroke_types = self.n_stroke_types + 1
49 |
50 | return self.n_stroke_types - 1
51 |
52 | def raw_distance(self, kanji_a, kanji_b):
53 | s_py = self.signatures[kanji_a]
54 | t_py = self.signatures[kanji_b]
55 |
56 | return edit_distance(s_py, t_py)
57 |
58 | def __call__(self, kanji_a, kanji_b):
59 | s_py = self.signatures[kanji_a]
60 | t_py = self.signatures[kanji_b]
61 |
62 | result = edit_distance(s_py, t_py)
63 | return float(result) / max(len(s_py), len(t_py))
64 |
65 | def __contains__(self, kanji):
66 | return kanji in self.signatures
67 |
68 | #----------------------------------------------------------------------------#
69 |
70 | cdef edit_distance(s_py, t_py):
71 | cdef int m, n, i, j
72 | cdef int table[100][100]
73 | cdef int s[100]
74 | cdef int t[100]
75 | cdef int up, left, diag, cost
76 |
77 | s_len = len(s_py)
78 | t_len = len(t_py)
79 | if s_len > 99 or t_len > 99:
80 | raise ValueError, "stroke sequences too long"
81 |
82 | for 0 <= i < s_len:
83 | table[i][0] = i
84 | s[i] = s_py[i]
85 | table[s_len][0] = s_len
86 |
87 | for 0 <= j < t_len:
88 | table[0][j] = j
89 | t[j] = t_py[j]
90 | table[0][t_len] = t_len
91 |
92 | for 1 <= i <= s_len:
93 | for 1 <= j <= t_len:
94 | if s[i-1] == t[j-1]:
95 | cost = 0
96 | else:
97 | cost = 1
98 |
99 | up = table[i-1][j] + 1
100 | left = table[i][j-1] + 1
101 | diag = table[i-1][j-1] + cost
102 | if up <= left:
103 | if up <= diag:
104 | table[i][j] = up
105 | else:
106 | table[i][j] = diag
107 | else:
108 | if left <= diag:
109 | table[i][j] = left
110 | else:
111 | table[i][j] = diag
112 |
113 | return table[s_len][t_len]
114 |
115 | # vim: ts=4 sw=4 sts=4 et tw=78:
116 |
--------------------------------------------------------------------------------
/SConstruct:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | #
3 | # SConstruct
4 | # simsearch
5 | #
6 | # Created by Lars Yencken on 27-08-2010.
7 | # Copyright 2010 Lars Yencken. All rights reserved.
8 | #
9 |
10 | """
11 | Build instructions for the stroke similarity extension."
12 | """
13 |
14 | #----------------------------------------------------------------------------#
15 |
16 | import os
17 | from distutils import sysconfig
18 |
19 | #----------------------------------------------------------------------------#
20 |
21 | # Default include path for python, version inspecific.
22 | #scons_python_version = sysconfig.get_config_var('VERSION')
23 | scons_python_version = '2.7'
24 | python_version = ARGUMENTS.get('python') or scons_python_version
25 |
26 | print 'Using Python %s' % python_version
27 |
28 | #----------------------------------------------------------------------------#
29 |
30 | def check_libraries(env):
31 | """ Check whether the correct libraries exist, and thus whether building
32 | is possible.
33 | """
34 | # Detect OS X python installation, and attempt to correct for it.
35 | if os.uname()[0] == 'Darwin':
36 | env.Replace(SHLINKFLAGS='$LINKFLAGS -bundle -flat_namespace -undefined suppress')
37 | env.Replace(SHLIBSUFFIX='.so')
38 | if os.path.isdir('/opt/local'):
39 | env.Append(
40 | LIBPATH=['/opt/local/lib'],
41 | CPPPATH=['/opt/local/include']
42 | )
43 |
44 | # Detect the presence of necessary dependencies.
45 | conf = Configure(env)
46 |
47 | if not conf.CheckLibWithHeader('m', 'math.h', 'c'):
48 | print "Can't find standard math libraries."
49 | Exit(1)
50 |
51 | if not conf.CheckLibWithHeader('python%s' % python_version,
52 | 'Python.h', 'c'):
53 | print "Can't find python %s." % python_version
54 | Exit(1)
55 |
56 | env = conf.Finish()
57 |
58 | return env
59 |
60 | #----------------------------------------------------------------------------#
61 | # CONFIGURATION
62 | #----------------------------------------------------------------------------#
63 |
64 | # Set up the compilation environment.
65 | env = Environment(
66 | CPPPATH=sysconfig.get_python_inc().replace(scons_python_version,
67 | python_version),
68 | LIBPATH=[sysconfig.get_config_var('LIBPL').replace(
69 | scons_python_version, python_version)],
70 | SHLIBPREFIX='',
71 | LIBS=['python%s' % python_version],
72 | )
73 |
74 | environmentVars = (
75 | 'CPATH',
76 | 'LD_LIBRARY_PATH',
77 | 'LIBRARY_PATH',
78 | 'PATH',
79 | 'PYTHONPATH',
80 | )
81 |
82 | envDict = env['ENV']
83 | for var in environmentVars:
84 | if var in os.environ:
85 | envDict[var] = os.environ[var]
86 |
87 | # Choose between debugging or optimized mode.
88 | if ARGUMENTS.get('debug'):
89 | print 'Using debug targets'
90 | env.Replace(DEBUG=True, CXXFLAGS='-O0 -g -Wall ', CFLAGS='-O0 -g -Wall ')
91 | else:
92 | print 'Using optimised targets'
93 | env.Replace(DEBUG=False, CXXFLAGS='-O3 -DNDEBUG -Wall ',
94 | CFLAGS='-O3 -DNDEBUG -Wall ')
95 |
96 | # Configure the environment.
97 | env = check_libraries(env)
98 |
99 | pyxbuild = Builder(action='cython -o $TARGET $SOURCE')
100 | env.Append(BUILDERS={'Cython': pyxbuild})
101 |
102 | #----------------------------------------------------------------------------#
103 |
104 | SConscript('simsearch/SConscript', exports='env')
105 |
106 | #----------------------------------------------------------------------------#
107 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/src/typography.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | typography.css
4 | * Sets up some sensible default typography.
5 |
6 | -------------------------------------------------------------- */
7 |
8 | /* Default font settings.
9 | The font-size percentage is of 16px. (0.75 * 16px = 12px) */
10 | html { font-size:100.01%; }
11 | body {
12 | font-size: 75%;
13 | color: #222;
14 | background: #fff;
15 | font-family: "Helvetica Neue", Arial, Helvetica, sans-serif;
16 | }
17 |
18 |
19 | /* Headings
20 | -------------------------------------------------------------- */
21 |
22 | h1,h2,h3,h4,h5,h6 { font-weight: normal; color: #111; }
23 |
24 | h1 { font-size: 3em; line-height: 1; margin-bottom: 0.5em; }
25 | h2 { font-size: 2em; margin-bottom: 0.75em; }
26 | h3 { font-size: 1.5em; line-height: 1; margin-bottom: 1em; }
27 | h4 { font-size: 1.2em; line-height: 1.25; margin-bottom: 1.25em; }
28 | h5 { font-size: 1em; font-weight: bold; margin-bottom: 1.5em; }
29 | h6 { font-size: 1em; font-weight: bold; }
30 |
31 | h1 img, h2 img, h3 img,
32 | h4 img, h5 img, h6 img {
33 | margin: 0;
34 | }
35 |
36 |
37 | /* Text elements
38 | -------------------------------------------------------------- */
39 |
40 | p { margin: 0 0 1.5em; }
41 | p img.left { float: left; margin: 1.5em 1.5em 1.5em 0; padding: 0; }
42 | p img.right { float: right; margin: 1.5em 0 1.5em 1.5em; }
43 |
44 | a:focus,
45 | a:hover { color: #08d; }
46 | a { color: #009; text-decoration: underline; }
47 |
48 | blockquote { margin: 1.5em; color: #666; font-style: italic; }
49 | strong { font-weight: bold; }
50 | em,dfn { font-style: italic; }
51 | dfn { font-weight: bold; }
52 | sup, sub { line-height: 0; }
53 |
54 | abbr,
55 | acronym { border-bottom: 1px dotted #666; }
56 | address { margin: 0 0 1.5em; font-style: italic; }
57 | del { color:#666; }
58 |
59 | pre { margin: 1.5em 0; white-space: pre; }
60 | pre,code,tt { font: 1em 'andale mono', 'lucida console', monospace; line-height: 1.5; }
61 |
62 |
63 | /* Lists
64 | -------------------------------------------------------------- */
65 |
66 | li ul,
67 | li ol { margin: 0; }
68 | ul, ol { margin: 0 1.5em 1.5em 0; padding-left: 3.333em; }
69 |
70 | ul { list-style-type: disc; }
71 | ol { list-style-type: decimal; }
72 |
73 | dl { margin: 0 0 1.5em 0; }
74 | dl dt { font-weight: bold; }
75 | dd { margin-left: 1.5em;}
76 |
77 |
78 | /* Tables
79 | -------------------------------------------------------------- */
80 |
81 | table { margin-bottom: 1.4em; width:100%; }
82 | th { font-weight: bold; }
83 | thead th { background: #c3d9ff; }
84 | th,td,caption { padding: 4px 10px 4px 5px; }
85 | tr.even td { background: #e5ecf9; }
86 | tfoot { font-style: italic; }
87 | caption { background: #eee; }
88 |
89 |
90 | /* Misc classes
91 | -------------------------------------------------------------- */
92 |
93 | .small { font-size: .8em; margin-bottom: 1.875em; line-height: 1.875em; }
94 | .large { font-size: 1.2em; line-height: 2.5em; margin-bottom: 1.25em; }
95 | .hide { display: none; }
96 |
97 | .quiet { color: #666; }
98 | .loud { color: #000; }
99 | .highlight { background:#ff0; }
100 | .added { background:#060; color: #fff; }
101 | .removed { background:#900; color: #fff; }
102 |
103 | .first { margin-left:0; padding-left:0; }
104 | .last { margin-right:0; padding-right:0; }
105 | .top { margin-top:0; padding-top:0; }
106 | .bottom { margin-bottom:0; padding-bottom:0; }
107 |
--------------------------------------------------------------------------------
/simsearch/experiments/simulate_accessibility.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # -*- coding: utf-8 -*-
3 | #
4 | # simulate_accessibility.py
5 | # simsearch
6 | #
7 | # Created by Lars Yencken on 05-09-2010.
8 | # Copyright 2010 Lars Yencken. All rights reserved.
9 | #
10 |
11 | """
12 | A basic simulation of accessibility improvements estimated from use of visual
13 | similarity search.
14 | """
15 |
16 | import os
17 | import sys
18 | import optparse
19 | import codecs
20 | import random
21 |
22 | from simplestats import FreqDist, basic_stats
23 |
24 | from simsearch import settings
25 | from simsearch.search import models
26 |
27 | DEFAULT_THRESHOLD = 0.95
28 |
29 | def simulate_accessibility(output_file, threshold=DEFAULT_THRESHOLD):
30 | print 'Loading frequency distribution'
31 | dist = FreqDist.from_file(settings.FREQ_SOURCE)
32 |
33 | print 'Loading kanji'
34 | kanji_set = list(models._get_kanji())
35 | random.seed(123456789)
36 | random.shuffle(kanji_set)
37 |
38 | kanji_in_order = sorted(kanji_set, key=lambda k: dist.prob(k))
39 |
40 | print 'Loading graph'
41 | graph = RestrictedGraph()
42 |
43 | print 'Dumping frequencies to %s' % os.path.basename(output_file)
44 | n_neighbours = []
45 | with codecs.open(output_file, 'w', 'utf8') as ostream:
46 | print >> ostream, u'#n_known,n_accessible'
47 | print >> ostream, u'%d,%d' % (0, 0)
48 | known_set = set()
49 | accessible_set = set()
50 | for i, kanji in enumerate(kanji_in_order):
51 | known_set.add(kanji)
52 | accessible_set.add(kanji)
53 |
54 | neighbours = graph[kanji]
55 | accessible_set.update(neighbours)
56 | n_neighbours.append(len(neighbours))
57 |
58 | if (i + 1) % 50 == 0:
59 | print >> ostream, u'%d,%d' % (len(known_set),
60 | len(accessible_set))
61 | print >> ostream, u'%d,%d' % (len(known_set), len(accessible_set))
62 |
63 | print 'Average neighbourhood size: %.02f (σ = %.02f)' % \
64 | basic_stats(n_neighbours)
65 |
66 | class RestrictedGraph(object):
67 | def __init__(self, threshold=DEFAULT_THRESHOLD):
68 | self._graph = models.Similarity.load()
69 | self._threshold = threshold
70 |
71 | def __getitem__(self, kanji):
72 | neighbour_heap = self._graph[kanji]
73 | ordered_neighbourhood = sorted(neighbour_heap.get_contents(),
74 | reverse=True)
75 |
76 | first_sim, first_neighbour = ordered_neighbourhood[0]
77 | cutoff_neighbours = set(n for (s, n) in ordered_neighbourhood
78 | if s >= self._threshold * first_sim)
79 |
80 | return cutoff_neighbours
81 |
82 | #----------------------------------------------------------------------------#
83 |
84 | def _create_option_parser():
85 | usage = \
86 | """%prog [options] output_file.csv
87 |
88 | Simulates how many kanji are accessible as kanji are learned, assuming they
89 | are studied in frequency order."""
90 |
91 | parser = optparse.OptionParser(usage)
92 |
93 | parser.add_option('-t', action='store', dest='threshold',
94 | default=DEFAULT_THRESHOLD, type='float',
95 | help='The neighbourhood cutoff threshold [%.02f]' % \
96 | DEFAULT_THRESHOLD)
97 |
98 | return parser
99 |
100 | def main(argv):
101 | parser = _create_option_parser()
102 | (options, args) = parser.parse_args(argv)
103 |
104 | if len(args) != 1:
105 | parser.print_help()
106 | sys.exit(1)
107 |
108 | simulate_accessibility(args[0], threshold=options.threshold)
109 |
110 | #----------------------------------------------------------------------------#
111 |
112 | if __name__ == '__main__':
113 | main(sys.argv[1:])
114 |
115 | # vim: ts=4 sw=4 sts=4 et tw=78:
116 |
--------------------------------------------------------------------------------
/simsearch/templates/search/display.html:
--------------------------------------------------------------------------------
1 | {% extends "base.html" %}
2 |
3 | {% block headers %}
4 | /')
89 | def translate(kanji):
90 | "Updates the query model before redirecting to the real translation."
91 | kanji = kanji or flask.request.args.get('kanji')
92 | if not _is_kanji(kanji):
93 | flask.abort(404)
94 |
95 | path = flask.request.args.get('path')
96 | if path and len(path) > 1 and all(map(_is_kanji, path)) \
97 | and path.endswith(kanji):
98 | models.Node.update(path)
99 | models.Trace.log(flask.request, path)
100 |
101 | t = models.Translation.objects.get(kanji=kanji)
102 | if t is None:
103 | flask.abort(404)
104 |
105 | c = base_context()
106 | c['translation'] = t
107 | return flask.render_template('translate/kanji.html', **c)
108 |
109 | @app.route('/search//')
110 | def search_json(pivot):
111 | "Returns the search display data as JSON."
112 | pivot = pivot or flask.request.args.get('pivot')
113 | node = models.Node.objects.get(pivot=pivot)
114 | neighbours = [n.kanji for n in sorted(node.neighbours, reverse=True)]
115 | neighbours = neighbours[:app.conf['N_NEIGHBOURS_RECALLED']]
116 |
117 | return flask.jsonify(
118 | pivot_kanji=pivot,
119 | tier1=neighbours[:4],
120 | tier2=neighbours[4:9],
121 | tier3=neighbours[9:],
122 | )
123 |
124 | def _is_kanji(kanji):
125 | return isinstance(kanji, unicode) and len(kanji) == 1 \
126 | and scripts.script_type(kanji) == scripts.Script.Kanji
127 |
128 | def base_context():
129 | c = {}
130 | c.update(mercurial_revision())
131 | c.update(site_settings())
132 | return c
133 |
134 | def mercurial_revision():
135 | project_base = os.path.join(app.config['PROJECT_ROOT'], '..')
136 | repo = mercurial.hg.repository(mercurial.ui.ui(), project_base)
137 | fctx = repo.filectx(project_base, 'tip')
138 |
139 | return {'revision': {
140 | 'short': mercurial.node.short(fctx.node()),
141 | 'number': fctx.rev(),
142 | }}
143 |
144 | def site_settings():
145 | return {'settings': app.config, 'MEDIA_URL': app.config['MEDIA_URL']}
146 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/plugins/rtl/screen.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | rtl.css
4 | * Mirrors Blueprint for left-to-right languages
5 |
6 | By Ran Yaniv Hartstein [ranh.co.il]
7 |
8 | -------------------------------------------------------------- */
9 |
10 | body .container { direction: rtl; }
11 | body .column, body .span-1, body .span-2, body .span-3, body .span-4, body .span-5, body .span-6, body .span-7, body .span-8, body .span-9, body .span-10, body .span-11, body .span-12, body .span-13, body .span-14, body .span-15, body .span-16, body .span-17, body .span-18, body .span-19, body .span-20, body .span-21, body .span-22, body .span-23, body .span-24 {
12 | float: right;
13 | margin-right: 0;
14 | margin-left: 10px;
15 | text-align:right;
16 | }
17 |
18 | body div.last { margin-left: 0; }
19 | body table .last { padding-left: 0; }
20 |
21 | body .append-1 { padding-right: 0; padding-left: 40px; }
22 | body .append-2 { padding-right: 0; padding-left: 80px; }
23 | body .append-3 { padding-right: 0; padding-left: 120px; }
24 | body .append-4 { padding-right: 0; padding-left: 160px; }
25 | body .append-5 { padding-right: 0; padding-left: 200px; }
26 | body .append-6 { padding-right: 0; padding-left: 240px; }
27 | body .append-7 { padding-right: 0; padding-left: 280px; }
28 | body .append-8 { padding-right: 0; padding-left: 320px; }
29 | body .append-9 { padding-right: 0; padding-left: 360px; }
30 | body .append-10 { padding-right: 0; padding-left: 400px; }
31 | body .append-11 { padding-right: 0; padding-left: 440px; }
32 | body .append-12 { padding-right: 0; padding-left: 480px; }
33 | body .append-13 { padding-right: 0; padding-left: 520px; }
34 | body .append-14 { padding-right: 0; padding-left: 560px; }
35 | body .append-15 { padding-right: 0; padding-left: 600px; }
36 | body .append-16 { padding-right: 0; padding-left: 640px; }
37 | body .append-17 { padding-right: 0; padding-left: 680px; }
38 | body .append-18 { padding-right: 0; padding-left: 720px; }
39 | body .append-19 { padding-right: 0; padding-left: 760px; }
40 | body .append-20 { padding-right: 0; padding-left: 800px; }
41 | body .append-21 { padding-right: 0; padding-left: 840px; }
42 | body .append-22 { padding-right: 0; padding-left: 880px; }
43 | body .append-23 { padding-right: 0; padding-left: 920px; }
44 |
45 | body .prepend-1 { padding-left: 0; padding-right: 40px; }
46 | body .prepend-2 { padding-left: 0; padding-right: 80px; }
47 | body .prepend-3 { padding-left: 0; padding-right: 120px; }
48 | body .prepend-4 { padding-left: 0; padding-right: 160px; }
49 | body .prepend-5 { padding-left: 0; padding-right: 200px; }
50 | body .prepend-6 { padding-left: 0; padding-right: 240px; }
51 | body .prepend-7 { padding-left: 0; padding-right: 280px; }
52 | body .prepend-8 { padding-left: 0; padding-right: 320px; }
53 | body .prepend-9 { padding-left: 0; padding-right: 360px; }
54 | body .prepend-10 { padding-left: 0; padding-right: 400px; }
55 | body .prepend-11 { padding-left: 0; padding-right: 440px; }
56 | body .prepend-12 { padding-left: 0; padding-right: 480px; }
57 | body .prepend-13 { padding-left: 0; padding-right: 520px; }
58 | body .prepend-14 { padding-left: 0; padding-right: 560px; }
59 | body .prepend-15 { padding-left: 0; padding-right: 600px; }
60 | body .prepend-16 { padding-left: 0; padding-right: 640px; }
61 | body .prepend-17 { padding-left: 0; padding-right: 680px; }
62 | body .prepend-18 { padding-left: 0; padding-right: 720px; }
63 | body .prepend-19 { padding-left: 0; padding-right: 760px; }
64 | body .prepend-20 { padding-left: 0; padding-right: 800px; }
65 | body .prepend-21 { padding-left: 0; padding-right: 840px; }
66 | body .prepend-22 { padding-left: 0; padding-right: 880px; }
67 | body .prepend-23 { padding-left: 0; padding-right: 920px; }
68 |
69 | body .border {
70 | padding-right: 0;
71 | padding-left: 4px;
72 | margin-right: 0;
73 | margin-left: 5px;
74 | border-right: none;
75 | border-left: 1px solid #eee;
76 | }
77 |
78 | body .colborder {
79 | padding-right: 0;
80 | padding-left: 24px;
81 | margin-right: 0;
82 | margin-left: 25px;
83 | border-right: none;
84 | border-left: 1px solid #eee;
85 | }
86 |
87 | body .pull-1 { margin-left: 0; margin-right: -40px; }
88 | body .pull-2 { margin-left: 0; margin-right: -80px; }
89 | body .pull-3 { margin-left: 0; margin-right: -120px; }
90 | body .pull-4 { margin-left: 0; margin-right: -160px; }
91 |
92 | body .push-0 { margin: 0 18px 0 0; }
93 | body .push-1 { margin: 0 18px 0 -40px; }
94 | body .push-2 { margin: 0 18px 0 -80px; }
95 | body .push-3 { margin: 0 18px 0 -120px; }
96 | body .push-4 { margin: 0 18px 0 -160px; }
97 | body .push-0, body .push-1, body .push-2,
98 | body .push-3, body .push-4 { float: left; }
99 |
100 |
101 | /* Typography with RTL support */
102 | body h1,body h2,body h3,
103 | body h4,body h5,body h6 { font-family: Arial, sans-serif; }
104 | html body { font-family: Arial, sans-serif; }
105 | body pre,body code,body tt { font-family: monospace; }
106 |
107 | /* Mirror floats and margins on typographic elements */
108 | body p img { float: right; margin: 1.5em 0 1.5em 1.5em; }
109 | body dd, body ul, body ol { margin-left: 0; margin-right: 1.5em;}
110 | body td, body th { text-align:right; }
111 |
--------------------------------------------------------------------------------
/simsearch/experiments/simulate_search.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # -*- coding: utf-8 -*-
3 | #
4 | # simulate_search.py
5 | # simsearch
6 | #
7 | # Created by Lars Yencken on 03-09-2010.
8 | # Copyright 2010 Lars Yencken. All rights reserved.
9 | #
10 |
11 | """
12 | A script to simulate how users might search with the system. Query paths
13 | generated by this script can be analysed as a form of intrinsic evaluation.
14 | """
15 |
16 | import os
17 | import sys
18 | import optparse
19 | import codecs
20 | import random
21 |
22 | from consoleLog import withProgress
23 |
24 | from simsearch import settings
25 | from simsearch.search import stroke, models
26 |
27 | def simulate_search(output_file, strategy='greedy',
28 | k=settings.N_NEIGHBOURS_RECALLED, error_rate=0.0):
29 | """
30 | Simulate user searches on every query/target pair from the flashcard
31 | dataset, using one of the available strategies. The resulting query paths
32 | are dumped to the specified file.
33 | """
34 | if strategy == 'greedy':
35 | search_fn = _greedy_search
36 | elif strategy == 'shortest':
37 | search_fn = _breadth_first_search
38 | elif strategy == 'random':
39 | random.seed(123456789)
40 | search_fn = _random_stumble
41 | else:
42 | raise ValueError(strategy)
43 |
44 | traces = []
45 | for query, target in withProgress(_load_search_examples()):
46 | path = search_fn(query, target, k=k, error_rate=error_rate)
47 | traces.append((query, target, path))
48 |
49 | TraceFile.save(traces, output_file)
50 | print 'Paths dumped to %s' % output_file
51 |
52 | class TraceFile(object):
53 | "A basic human-readable query path file format."
54 | @staticmethod
55 | def save(traces, filename):
56 | with codecs.open(filename, 'w', 'utf8') as ostream:
57 | print >> ostream, u"#query\ttarget\tvia"
58 | for query, target, path in traces:
59 | if path:
60 | assert path[0] == query
61 | # have at least a partial search
62 | if path[-1] == target:
63 | # success
64 | print >> ostream, u'%s\t%s\t[%s]' % (query, target,
65 | ''.join(path[1:-1]))
66 | else:
67 | # failure with partial path
68 | print >> ostream, u'%s\t(%s)\t[%s]' % (query, target,
69 | ''.join(path[1:]))
70 |
71 | else:
72 | # failure without partial path
73 | print >> ostream, u'%s\t(%s)\tNone' % (query, target)
74 |
75 | @staticmethod
76 | def load(filename):
77 | traces = []
78 | with codecs.open(filename, 'r', 'utf8') as istream:
79 | header = istream.next()
80 | assert header.startswith('#')
81 | for line in istream:
82 | query, target, path = line.rstrip().split('\t')
83 | if len(target) != 1:
84 | target = target.strip('()')
85 | was_success = False
86 | assert len(target) == 1
87 | else:
88 | was_success = True
89 |
90 | if path == 'None':
91 | path = None
92 | else:
93 | path = [query] + list(path.strip('[]'))
94 | if was_success:
95 | path.append(target)
96 |
97 | traces.append((query, target, path))
98 |
99 | return traces
100 |
101 | def _load_search_examples():
102 | flashcard_file = os.path.join(settings.DATA_DIR, 'similarity', 'flashcard')
103 | results = []
104 | with codecs.open(flashcard_file, 'r', 'utf8') as istream:
105 | for line in istream:
106 | _id, query, targets = line.split()
107 | for target in targets:
108 | results.append((query, target))
109 |
110 | return results
111 |
112 | def _greedy_search(query, target, limit=5, k=settings.N_NEIGHBOURS_RECALLED,
113 | error_rate=0.0):
114 | """
115 | Simulate a search between the query and target where the user always
116 | chooses the next kanji which looks closest to the target.
117 | """
118 | assert query != target
119 | if query not in sed or target not in sed:
120 | # we can't simulate this search type without using a distance
121 | # heuristic
122 | return
123 |
124 | path = [query]
125 | while path[-1] != target and len(path) <= limit:
126 | assert path[0] == query
127 |
128 | new_query = path[-1]
129 | neighbours = _get_neighbours(new_query, k=k)
130 |
131 | if target in neighbours:
132 | if error_rate == 0.0 or random.random() < (1 - error_rate)**k:
133 | # Success!
134 | path.append(target)
135 | return path
136 |
137 | # Recognition error =(
138 | neighbours.remove(target)
139 |
140 | # Our options are neighbours we haven't tried yet
141 | options = neighbours.difference(path)
142 |
143 | if not options:
144 | # Search exhausted =(
145 | break
146 |
147 | # Choose the one visually most similar to the target
148 | _d, neighbour = min((sed(n, target), n) for n in options)
149 | path.append(neighbour)
150 |
151 | assert path[0] == query and path[-1] != target
152 |
153 | return path
154 |
155 | def _breadth_first_search(query, target, limit=5,
156 | k=settings.N_NEIGHBOURS_RECALLED, error_rate=0.0):
157 | """
158 | Perform breadth first search to a fixed depth limit, returning the
159 | shortest path from the query to the target (within the limit).
160 | """
161 | paths = [[query]]
162 | shortest = set([query]) # has a shortest path been checked
163 | while paths:
164 | current = paths.pop(0)
165 | current_query = current[-1]
166 | neighbours = _get_neighbours(current_query, k=k)
167 |
168 | if target in neighbours:
169 | current.append(target)
170 | assert current[0] == query
171 | return current
172 |
173 | if len(current) < limit:
174 | # visit in similarity order if possible
175 | try:
176 | neighbours = sorted(neighbours, key=lambda n: sed(n, target))
177 | except KeyError:
178 | pass
179 | neighbours = [n for n in neighbours if n not in shortest]
180 | shortest.update(neighbours)
181 | paths.extend((current + [n]) for n in neighbours)
182 |
183 | def _random_stumble(query, target, limit=5, k=settings.N_NEIGHBOURS_RECALLED,
184 | error_rate=0.0):
185 | """
186 | A worst-case simulation of user search, completely unguided by the
187 | target kanji (except for the initial query).
188 | """
189 | path = [query]
190 | while len(path) <= limit:
191 | neighbours = _get_neighbours(path[-1], k=k)
192 | if target in neighbours:
193 | if error_rate == 0.0 or random.random() < (1 - error_rate)**k:
194 | return path + [target]
195 |
196 | neighbours.remove(target)
197 |
198 | path.append(random.choice(list(neighbours)))
199 |
200 | return path
201 |
202 | class cache(object):
203 | """
204 | A simple cache wrapper whose contents never expire. Useful for reducing
205 | expensive calls on small datasets.
206 | """
207 | def __init__(self, f):
208 | self.f = f
209 | self._cached = {}
210 |
211 | def __call__(self, *args, **kwargs):
212 | key = args + tuple(kwargs.items())
213 | if key not in self._cached:
214 | self._cached[key] = self.f(*args, **kwargs)
215 |
216 | return self._cached[key]
217 |
218 | def __contains__(self, key):
219 | # workaround for StrokeEditDistance also acting like a container
220 | return self.f.__contains__(key)
221 |
222 | @cache
223 | def _get_neighbours(query, k=settings.N_NEIGHBOURS_RECALLED):
224 | neighbours = set(n.kanji for n in models.Node.objects.get(
225 | pivot=query).neighbours[:k])
226 | return neighbours
227 |
228 | sed = cache(stroke.StrokeEditDistance())
229 |
230 | #----------------------------------------------------------------------------#
231 |
232 | def _create_option_parser():
233 | usage = \
234 | """%prog [options] output_file
235 |
236 | Simulate queries through the search graph, dumping the traces to the given
237 | file."""
238 |
239 | parser = optparse.OptionParser(usage)
240 |
241 | parser.add_option('--strategy', action='store', type='choice',
242 | choices=['greedy', 'shortest', 'random'], dest='strategy',
243 | default='greedy',
244 | help='The search strategy to use ([greedy]/shortest/random)')
245 |
246 | parser.add_option('-k', action='store', type='int',
247 | default=settings.N_NEIGHBOURS_RECALLED, dest='k',
248 | help='The number of neighbours displayed each query [%d]' % \
249 | settings.N_NEIGHBOURS_RECALLED)
250 |
251 | parser.add_option('-e', action='store', type='float',
252 | default=0.0, dest='error_rate',
253 | help='Factor in an estimated recognition error rate [0.0]')
254 |
255 | return parser
256 |
257 | def main(argv):
258 | parser = _create_option_parser()
259 | (options, args) = parser.parse_args(argv)
260 |
261 | if len(args) != 1:
262 | parser.print_help()
263 | sys.exit(1)
264 |
265 | simulate_search(args[0], strategy=options.strategy, k=options.k,
266 | error_rate=options.error_rate)
267 |
268 | #----------------------------------------------------------------------------#
269 |
270 | if __name__ == '__main__':
271 | main(sys.argv[1:])
272 |
273 | # vim: ts=4 sw=4 sts=4 et tw=78:
274 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/src/grid.css:
--------------------------------------------------------------------------------
1 | /* --------------------------------------------------------------
2 |
3 | grid.css
4 | * Sets up an easy-to-use grid of 24 columns.
5 |
6 | By default, the grid is 950px wide, with 24 columns
7 | spanning 30px, and a 10px margin between columns.
8 |
9 | If you need fewer or more columns, namespaces or semantic
10 | element names, use the compressor script (lib/compress.rb)
11 |
12 | -------------------------------------------------------------- */
13 |
14 | /* A container should group all your columns. */
15 | .container {
16 | width: 950px;
17 | margin: 0 auto;
18 | }
19 |
20 | /* Use this class on any .span / container to see the grid. */
21 | .showgrid {
22 | background: url(src/grid.png);
23 | }
24 |
25 |
26 | /* Columns
27 | -------------------------------------------------------------- */
28 |
29 | /* Sets up basic grid floating and margin. */
30 | .column, .span-1, .span-2, .span-3, .span-4, .span-5, .span-6, .span-7, .span-8, .span-9, .span-10, .span-11, .span-12, .span-13, .span-14, .span-15, .span-16, .span-17, .span-18, .span-19, .span-20, .span-21, .span-22, .span-23, .span-24 {
31 | float: left;
32 | margin-right: 10px;
33 | }
34 |
35 | /* The last column in a row needs this class. */
36 | .last { margin-right: 0; }
37 |
38 | /* Use these classes to set the width of a column. */
39 | .span-1 {width: 30px;}
40 |
41 | .span-2 {width: 70px;}
42 | .span-3 {width: 110px;}
43 | .span-4 {width: 150px;}
44 | .span-5 {width: 190px;}
45 | .span-6 {width: 230px;}
46 | .span-7 {width: 270px;}
47 | .span-8 {width: 310px;}
48 | .span-9 {width: 350px;}
49 | .span-10 {width: 390px;}
50 | .span-11 {width: 430px;}
51 | .span-12 {width: 470px;}
52 | .span-13 {width: 510px;}
53 | .span-14 {width: 550px;}
54 | .span-15 {width: 590px;}
55 | .span-16 {width: 630px;}
56 | .span-17 {width: 670px;}
57 | .span-18 {width: 710px;}
58 | .span-19 {width: 750px;}
59 | .span-20 {width: 790px;}
60 | .span-21 {width: 830px;}
61 | .span-22 {width: 870px;}
62 | .span-23 {width: 910px;}
63 | .span-24 {width:950px; margin-right:0;}
64 |
65 | /* Use these classes to set the width of an input. */
66 | input.span-1, textarea.span-1, input.span-2, textarea.span-2, input.span-3, textarea.span-3, input.span-4, textarea.span-4, input.span-5, textarea.span-5, input.span-6, textarea.span-6, input.span-7, textarea.span-7, input.span-8, textarea.span-8, input.span-9, textarea.span-9, input.span-10, textarea.span-10, input.span-11, textarea.span-11, input.span-12, textarea.span-12, input.span-13, textarea.span-13, input.span-14, textarea.span-14, input.span-15, textarea.span-15, input.span-16, textarea.span-16, input.span-17, textarea.span-17, input.span-18, textarea.span-18, input.span-19, textarea.span-19, input.span-20, textarea.span-20, input.span-21, textarea.span-21, input.span-22, textarea.span-22, input.span-23, textarea.span-23, input.span-24, textarea.span-24 {
67 | border-left-width: 1px;
68 | border-right-width: 1px;
69 | padding-left: 5px;
70 | padding-right: 5px;
71 | }
72 |
73 | input.span-1, textarea.span-1 { width: 18px; }
74 | input.span-2, textarea.span-2 { width: 58px; }
75 | input.span-3, textarea.span-3 { width: 98px; }
76 | input.span-4, textarea.span-4 { width: 138px; }
77 | input.span-5, textarea.span-5 { width: 178px; }
78 | input.span-6, textarea.span-6 { width: 218px; }
79 | input.span-7, textarea.span-7 { width: 258px; }
80 | input.span-8, textarea.span-8 { width: 298px; }
81 | input.span-9, textarea.span-9 { width: 338px; }
82 | input.span-10, textarea.span-10 { width: 378px; }
83 | input.span-11, textarea.span-11 { width: 418px; }
84 | input.span-12, textarea.span-12 { width: 458px; }
85 | input.span-13, textarea.span-13 { width: 498px; }
86 | input.span-14, textarea.span-14 { width: 538px; }
87 | input.span-15, textarea.span-15 { width: 578px; }
88 | input.span-16, textarea.span-16 { width: 618px; }
89 | input.span-17, textarea.span-17 { width: 658px; }
90 | input.span-18, textarea.span-18 { width: 698px; }
91 | input.span-19, textarea.span-19 { width: 738px; }
92 | input.span-20, textarea.span-20 { width: 778px; }
93 | input.span-21, textarea.span-21 { width: 818px; }
94 | input.span-22, textarea.span-22 { width: 858px; }
95 | input.span-23, textarea.span-23 { width: 898px; }
96 | input.span-24, textarea.span-24 { width: 938px; }
97 |
98 | /* Add these to a column to append empty cols. */
99 |
100 | .append-1 { padding-right: 40px;}
101 | .append-2 { padding-right: 80px;}
102 | .append-3 { padding-right: 120px;}
103 | .append-4 { padding-right: 160px;}
104 | .append-5 { padding-right: 200px;}
105 | .append-6 { padding-right: 240px;}
106 | .append-7 { padding-right: 280px;}
107 | .append-8 { padding-right: 320px;}
108 | .append-9 { padding-right: 360px;}
109 | .append-10 { padding-right: 400px;}
110 | .append-11 { padding-right: 440px;}
111 | .append-12 { padding-right: 480px;}
112 | .append-13 { padding-right: 520px;}
113 | .append-14 { padding-right: 560px;}
114 | .append-15 { padding-right: 600px;}
115 | .append-16 { padding-right: 640px;}
116 | .append-17 { padding-right: 680px;}
117 | .append-18 { padding-right: 720px;}
118 | .append-19 { padding-right: 760px;}
119 | .append-20 { padding-right: 800px;}
120 | .append-21 { padding-right: 840px;}
121 | .append-22 { padding-right: 880px;}
122 | .append-23 { padding-right: 920px;}
123 |
124 | /* Add these to a column to prepend empty cols. */
125 |
126 | .prepend-1 { padding-left: 40px;}
127 | .prepend-2 { padding-left: 80px;}
128 | .prepend-3 { padding-left: 120px;}
129 | .prepend-4 { padding-left: 160px;}
130 | .prepend-5 { padding-left: 200px;}
131 | .prepend-6 { padding-left: 240px;}
132 | .prepend-7 { padding-left: 280px;}
133 | .prepend-8 { padding-left: 320px;}
134 | .prepend-9 { padding-left: 360px;}
135 | .prepend-10 { padding-left: 400px;}
136 | .prepend-11 { padding-left: 440px;}
137 | .prepend-12 { padding-left: 480px;}
138 | .prepend-13 { padding-left: 520px;}
139 | .prepend-14 { padding-left: 560px;}
140 | .prepend-15 { padding-left: 600px;}
141 | .prepend-16 { padding-left: 640px;}
142 | .prepend-17 { padding-left: 680px;}
143 | .prepend-18 { padding-left: 720px;}
144 | .prepend-19 { padding-left: 760px;}
145 | .prepend-20 { padding-left: 800px;}
146 | .prepend-21 { padding-left: 840px;}
147 | .prepend-22 { padding-left: 880px;}
148 | .prepend-23 { padding-left: 920px;}
149 |
150 |
151 | /* Border on right hand side of a column. */
152 | .border {
153 | padding-right: 4px;
154 | margin-right: 5px;
155 | border-right: 1px solid #ddd;
156 | }
157 |
158 | /* Border with more whitespace, spans one column. */
159 | .colborder {
160 | padding-right: 24px;
161 | margin-right: 25px;
162 | border-right: 1px solid #ddd;
163 | }
164 |
165 |
166 | /* Use these classes on an element to push it into the
167 | next column, or to pull it into the previous column. */
168 |
169 |
170 | .pull-1 { margin-left: -40px; }
171 | .pull-2 { margin-left: -80px; }
172 | .pull-3 { margin-left: -120px; }
173 | .pull-4 { margin-left: -160px; }
174 | .pull-5 { margin-left: -200px; }
175 | .pull-6 { margin-left: -240px; }
176 | .pull-7 { margin-left: -280px; }
177 | .pull-8 { margin-left: -320px; }
178 | .pull-9 { margin-left: -360px; }
179 | .pull-10 { margin-left: -400px; }
180 | .pull-11 { margin-left: -440px; }
181 | .pull-12 { margin-left: -480px; }
182 | .pull-13 { margin-left: -520px; }
183 | .pull-14 { margin-left: -560px; }
184 | .pull-15 { margin-left: -600px; }
185 | .pull-16 { margin-left: -640px; }
186 | .pull-17 { margin-left: -680px; }
187 | .pull-18 { margin-left: -720px; }
188 | .pull-19 { margin-left: -760px; }
189 | .pull-20 { margin-left: -800px; }
190 | .pull-21 { margin-left: -840px; }
191 | .pull-22 { margin-left: -880px; }
192 | .pull-23 { margin-left: -920px; }
193 | .pull-24 { margin-left: -960px; }
194 |
195 | .pull-1, .pull-2, .pull-3, .pull-4, .pull-5, .pull-6, .pull-7, .pull-8, .pull-9, .pull-10, .pull-11, .pull-12, .pull-13, .pull-14, .pull-15, .pull-16, .pull-17, .pull-18, .pull-19, .pull-20, .pull-21, .pull-22, .pull-23, .pull-24 {float: left; position:relative;}
196 |
197 |
198 | .push-1 { margin: 0 -40px 1.5em 40px; }
199 | .push-2 { margin: 0 -80px 1.5em 80px; }
200 | .push-3 { margin: 0 -120px 1.5em 120px; }
201 | .push-4 { margin: 0 -160px 1.5em 160px; }
202 | .push-5 { margin: 0 -200px 1.5em 200px; }
203 | .push-6 { margin: 0 -240px 1.5em 240px; }
204 | .push-7 { margin: 0 -280px 1.5em 280px; }
205 | .push-8 { margin: 0 -320px 1.5em 320px; }
206 | .push-9 { margin: 0 -360px 1.5em 360px; }
207 | .push-10 { margin: 0 -400px 1.5em 400px; }
208 | .push-11 { margin: 0 -440px 1.5em 440px; }
209 | .push-12 { margin: 0 -480px 1.5em 480px; }
210 | .push-13 { margin: 0 -520px 1.5em 520px; }
211 | .push-14 { margin: 0 -560px 1.5em 560px; }
212 | .push-15 { margin: 0 -600px 1.5em 600px; }
213 | .push-16 { margin: 0 -640px 1.5em 640px; }
214 | .push-17 { margin: 0 -680px 1.5em 680px; }
215 | .push-18 { margin: 0 -720px 1.5em 720px; }
216 | .push-19 { margin: 0 -760px 1.5em 760px; }
217 | .push-20 { margin: 0 -800px 1.5em 800px; }
218 | .push-21 { margin: 0 -840px 1.5em 840px; }
219 | .push-22 { margin: 0 -880px 1.5em 880px; }
220 | .push-23 { margin: 0 -920px 1.5em 920px; }
221 | .push-24 { margin: 0 -960px 1.5em 960px; }
222 |
223 | .push-1, .push-2, .push-3, .push-4, .push-5, .push-6, .push-7, .push-8, .push-9, .push-10, .push-11, .push-12, .push-13, .push-14, .push-15, .push-16, .push-17, .push-18, .push-19, .push-20, .push-21, .push-22, .push-23, .push-24 {float: right; position:relative;}
224 |
225 |
226 | /* Misc classes and elements
227 | -------------------------------------------------------------- */
228 |
229 | /* In case you need to add a gutter above/below an element */
230 | div.prepend-top, .prepend-top {
231 | margin-top:1.5em;
232 | }
233 | div.append-bottom, .append-bottom {
234 | margin-bottom:1.5em;
235 | }
236 |
237 | /* Use a .box to create a padded box inside a column. */
238 | .box {
239 | padding: 1.5em;
240 | margin-bottom: 1.5em;
241 | background: #E5ECF9;
242 | }
243 |
244 | /* Use this to create a horizontal ruler across a column. */
245 | hr {
246 | background: #ddd;
247 | color: #ddd;
248 | clear: both;
249 | float: none;
250 | width: 100%;
251 | height: 1px;
252 | margin: 0 0 1.45em;
253 | border: none;
254 | }
255 |
256 | hr.space {
257 | background: #fff;
258 | color: #fff;
259 | visibility: hidden;
260 | }
261 |
262 |
263 | /* Clearing floats without extra markup
264 | Based on How To Clear Floats Without Structural Markup by PiE
265 | [http://www.positioniseverything.net/easyclearing.html] */
266 |
267 | .clearfix:after, .container:after {
268 | content: "\0020";
269 | display: block;
270 | height: 0;
271 | clear: both;
272 | visibility: hidden;
273 | overflow:hidden;
274 | }
275 | .clearfix, .container {display: block;}
276 |
277 | /* Regular clearing
278 | apply to column that should drop below previous ones. */
279 |
280 | .clear { clear:both; }
281 |
--------------------------------------------------------------------------------
/simsearch/models.py:
--------------------------------------------------------------------------------
1 | # -*- coding: utf-8 -*-
2 | #
3 | # models.py
4 | # simsearch
5 | #
6 | # Created by Lars Yencken on 28-08-2010.
7 | # Copyright 2010 Lars Yencken. All rights reserved.
8 | #
9 |
10 | """
11 | Database models for similarity search.
12 | """
13 |
14 | import os
15 | import codecs
16 | import gzip
17 | import itertools
18 |
19 | import mongoengine
20 | from cjktools import scripts
21 | from cjktools.resources import kanjidic
22 | from nltk.probability import FreqDist, LaplaceProbDist
23 |
24 | import stroke
25 | import heap_cache
26 |
27 | from simsearch import settings
28 |
29 | class Similarity(mongoengine.Document):
30 | "Raw similarity scores for kanji pairs."
31 | kanji_pair = mongoengine.StringField(max_length=2, primary_key=True)
32 | similarity = mongoengine.FloatField(min_value=0.0, max_value=1.0,
33 | required=True)
34 |
35 | def partner_to(self, kanji):
36 | "Returns the partnering kanji in this pair."
37 | if kanji not in self.kanji_pair:
38 | raise ValueError('kanji not part of this pair')
39 |
40 | return self.kanji_pair.replace(kanji, '')
41 |
42 | @classmethod
43 | def build(cls):
44 | print 'Building similarity matrix'
45 | cls.drop_collection()
46 | sed = stroke.StrokeEditDistance()
47 | kanji_set = _get_kanji()
48 |
49 | cache = heap_cache.FixedSimilarityCache(settings.N_NEIGHBOURS_STORED)
50 | pairs = ((a, b) for (a, b) in itertools.product(kanji_set, kanji_set)
51 | if (a < b))
52 |
53 | for kanji_a, kanji_b in pairs:
54 | distance = sed(kanji_a, kanji_b)
55 | cache.add(kanji_a, kanji_b, 1 - distance)
56 |
57 | for kanji in kanji_set:
58 | heap = cache.get_heap(kanji)
59 | for similarity, neighbour in heap.get_contents():
60 | kanji_pair = ''.join(min(
61 | (kanji, neighbour), (neighbour, kanji)
62 | ))
63 | doc = Similarity(
64 | kanji_pair=kanji_pair,
65 | similarity=similarity,
66 | )
67 | doc.save()
68 |
69 | return cache
70 |
71 | @classmethod
72 | def load(cls, n=None):
73 | if n is None:
74 | n = settings.N_NEIGHBOURS_STORED
75 | cache = heap_cache.FixedSimilarityCache(n)
76 | for record in cls.objects:
77 | kanji_pair = record.kanji_pair
78 | cache.add(kanji_pair[0], kanji_pair[1], record.similarity)
79 |
80 | return cache
81 |
82 | def __unicode__(self):
83 | return u'(%s, %s, %f)' % (self.kanji_pair[0], self.kanji_pair[1],
84 | self.similarity)
85 |
86 | class Neighbour(mongoengine.EmbeddedDocument):
87 | "A weighted graph edge."
88 | kanji = mongoengine.StringField(max_length=1)
89 | weight = mongoengine.FloatField(min_value=0.0)
90 |
91 | def __cmp__(self, rhs):
92 | return cmp(self.weight, rhs.weight)
93 |
94 | def __unicode__(self):
95 | return self.kanji
96 |
97 | class Node(mongoengine.Document):
98 | """
99 | A single node in the state graph for Q-learning. The neighbours attribute
100 | stores Q(n, a) for all actions which can be taken from this node.
101 | """
102 | pivot = mongoengine.StringField(max_length=1, primary_key=True)
103 | neighbours = mongoengine.ListField(mongoengine.EmbeddedDocumentField(
104 | Neighbour))
105 | n_updates = mongoengine.IntField(default=0, min_value=0)
106 |
107 | def at(self, kanji):
108 | "Gets the neighbour described by the given kanji."
109 | for neighbour in self.neighbours:
110 | if neighbour.kanji == kanji:
111 | return neighbour
112 |
113 | raise KeyError(kanji)
114 |
115 | @classmethod
116 | def build(cls, cache=None):
117 | "Builds the initial graph for Q learning."
118 | print 'Building neighbourhood graph'
119 | n = settings.N_NEIGHBOURS_RECALLED
120 |
121 | if cache is None:
122 | cache = Similarity.load(n)
123 |
124 | cls.drop_collection()
125 | dist = cls._load_corpus_counts()
126 | for kanji in _get_kanji():
127 | node = Node(pivot=kanji, neighbours=[])
128 |
129 | weights = {}
130 | best_n = sorted(cache.get_heap(kanji).get_contents(),
131 | reverse=True)[:n]
132 | for weight, partner in best_n:
133 | weights[partner] = weight * dist.prob(partner)
134 | total_weights = sum(weights.itervalues())
135 |
136 | for partner, weight in sorted(weights.iteritems(),
137 | key=lambda p: p[1], reverse=True):
138 | node.neighbours.append(Neighbour(kanji=partner,
139 | weight=weight / total_weights))
140 |
141 | node.save()
142 |
143 | @classmethod
144 | def _load_corpus_counts(cls):
145 | input_file = os.path.join(settings.DATA_DIR,
146 | 'corpus', 'jp_char_corpus_counts.gz')
147 | freq_dist = FreqDist()
148 | with open(input_file, 'r') as istream:
149 | istream = gzip.GzipFile(fileobj=istream)
150 | istream = codecs.getreader('utf8')(istream)
151 | for line in istream:
152 | kanji, count = line.split()
153 | freq_dist.inc(kanji, count=int(count))
154 |
155 | return LaplaceProbDist(freq_dist)
156 |
157 | @classmethod
158 | def get_coverage(cls):
159 | "Returns the set of kanji for which neighbours are stored."
160 | db = cls.objects._collection
161 | return set(r['_id'] for r in db.find({}, fields=['_id']))
162 |
163 | @classmethod
164 | def update(cls, path):
165 | path = cls._remove_cycles(path)
166 | nodes = cls.objects.filter(pivot__in=list(path))
167 | if len(nodes) != len(path):
168 | found_pivots = set(n.pivot for n in nodes)
169 | missing_kanji = [k for k in path if k not in found_pivots][0]
170 | raise ValueError('node not found in database for kanji %s' % \
171 | missing_kanji.encode('utf8'))
172 |
173 | # cache Q(s, a) for the subgraph we're interested in
174 | q = cls._cache_subgraph(nodes)
175 |
176 | # Calculate Q'(s, a) in reverse order along the path
177 | # Q'(s, a) = (1 - A(s))Q(s, a) + A(s)*(r(a) + G * max_a Q(s', a))
178 | gamma = settings.UPDATE_GAMMA
179 | for i in xrange(len(path) - 2, -1, -1):
180 | s = path[i]
181 | q_s = q[s]
182 | alpha = 1.0 / (4.0 + 0.5 * q_s.n_updates)
183 |
184 | # update very action available from state s
185 | for a in sorted(q_s.neighbours, key=lambda n: n.weight,
186 | reverse=True):
187 | q_old = a.weight
188 |
189 | r_a = (1 if a.kanji == path[-1] else 0)
190 | q_opt = r_a + gamma * max(q[a.kanji].neighbours).weight
191 |
192 | a.weight = (1.0 - alpha) * q_old + alpha * q_opt
193 |
194 | q_s.n_updates += 1
195 | q_s.save()
196 |
197 | @classmethod
198 | def _cache_subgraph(cls, nodes):
199 | q = {}
200 | missing_neighbours = set()
201 | for node in nodes:
202 | q[node.pivot] = node
203 | missing_neighbours.update(n.kanji for n in node.neighbours)
204 | missing_neighbours.discard(node.pivot for node in nodes)
205 |
206 | extra_nodes = cls.objects.filter(pivot__in=missing_neighbours)
207 | if len(extra_nodes) != len(missing_neighbours):
208 | raise ValueError('cannot cache subgraph -- neighbours missing')
209 |
210 | for node in extra_nodes:
211 | q[node.pivot] = node
212 |
213 | return q
214 |
215 | @staticmethod
216 | def _remove_cycles(path):
217 | # an O(n^2) method is good enough
218 | i = 0
219 | while i < len(path):
220 | kanji = path[i]
221 | j = path.find(kanji, i + 1)
222 | if j >= 0:
223 | # a cycle! trim the path to skip it
224 | path = path[:i] + path[j:]
225 | else:
226 | i += 1
227 |
228 | return path
229 |
230 | def __unicode__(self):
231 | return self.pivot
232 |
233 | class Trace(mongoengine.Document):
234 | "A search path through the graph, as taken by a user."
235 | ip_address = mongoengine.StringField(max_length=15)
236 | path = mongoengine.ListField(mongoengine.StringField(max_length=1))
237 |
238 | @classmethod
239 | def log(cls, request, path):
240 | ip = request.remote_addr
241 | cls(ip_address=ip, path=list(path)).save()
242 |
243 | class Translation(mongoengine.Document):
244 | "A per-kanji dictionary entry of readings and translations."
245 | kanji = mongoengine.StringField(max_length=1, primary_key=True)
246 | on_readings = mongoengine.ListField(mongoengine.StringField())
247 | kun_readings = mongoengine.ListField(mongoengine.StringField())
248 | glosses = mongoengine.ListField(mongoengine.StringField())
249 |
250 | @classmethod
251 | def build(cls):
252 | cls.drop_collection()
253 | kjd = kanjidic.Kanjidic()
254 | for entry in kjd.itervalues():
255 | translation = cls(
256 | kanji=entry.kanji,
257 | on_readings=entry.on_readings,
258 | kun_readings=entry.kun_readings,
259 | glosses = entry.gloss,
260 | )
261 | translation.save()
262 |
263 | def build():
264 | "Builds the database."
265 | cache = Similarity.build()
266 | Node.build(cache)
267 | Translation.build()
268 |
269 | #----------------------------------------------------------------------------#
270 |
271 | def _get_kanji():
272 | "Fetches our canonical list of kanji to work with."
273 | if not hasattr(_get_kanji, '_cached'):
274 | kanji_set = set()
275 | with codecs.open(settings.STROKE_SOURCE, 'r', 'utf8') as istream:
276 | for line in istream:
277 | kanji, rest = line.split()
278 |
279 | # check for a kanji or hanzi; our Chinese data extends into
280 | # the E000-F8FF private use block, so an "Unknown" script is
281 | # ok too
282 | assert len(kanji) == 1 and scripts.script_type(kanji) in \
283 | (scripts.Script.Kanji, scripts.Script.Unknown)
284 |
285 | kanji_set.add(kanji)
286 |
287 | _get_kanji._cached = kanji_set
288 |
289 | return _get_kanji._cached
290 |
291 | #----------------------------------------------------------------------------#
292 |
293 | if __name__ == '__main__':
294 | build()
295 |
296 | # vim: ts=4 sw=4 sts=4 et tw=78:
297 |
--------------------------------------------------------------------------------
/simsearch/static/css/blueprint/screen.css:
--------------------------------------------------------------------------------
1 | /* -----------------------------------------------------------------------
2 |
3 |
4 | Blueprint CSS Framework 0.9
5 | http://blueprintcss.org
6 |
7 | * Copyright (c) 2007-Present. See LICENSE for more info.
8 | * See README for instructions on how to use Blueprint.
9 | * For credits and origins, see AUTHORS.
10 | * This is a compressed file. See the sources in the 'src' directory.
11 |
12 | ----------------------------------------------------------------------- */
13 |
14 | /* reset.css */
15 | html, body, div, span, object, iframe, h1, h2, h3, h4, h5, h6, p, blockquote, pre, a, abbr, acronym, address, code, del, dfn, em, img, q, dl, dt, dd, ol, ul, li, fieldset, form, label, legend, table, caption, tbody, tfoot, thead, tr, th, td, article, aside, dialog, figure, footer, header, hgroup, nav, section {margin:0;padding:0;border:0;font-weight:inherit;font-style:inherit;font-size:100%;font-family:inherit;vertical-align:baseline;}
16 | article, aside, dialog, figure, footer, header, hgroup, nav, section {display:block;}
17 | body {line-height:1.5;}
18 | table {border-collapse:separate;border-spacing:0;}
19 | caption, th, td {text-align:left;font-weight:normal;}
20 | table, td, th {vertical-align:middle;}
21 | blockquote:before, blockquote:after, q:before, q:after {content:"";}
22 | blockquote, q {quotes:"" "";}
23 | a img {border:none;}
24 |
25 | /* typography.css */
26 | html {font-size:100.01%;}
27 | body {font-size:75%;color:#222;background:#fff;font-family:"Helvetica Neue", Arial, Helvetica, sans-serif;}
28 | h1, h2, h3, h4, h5, h6 {font-weight:normal;color:#111;}
29 | h1 {font-size:3em;line-height:1;margin-bottom:0.5em;}
30 | h2 {font-size:2em;margin-bottom:0.75em;}
31 | h3 {font-size:1.5em;line-height:1;margin-bottom:1em;}
32 | h4 {font-size:1.2em;line-height:1.25;margin-bottom:1.25em;}
33 | h5 {font-size:1em;font-weight:bold;margin-bottom:1.5em;}
34 | h6 {font-size:1em;font-weight:bold;}
35 | h1 img, h2 img, h3 img, h4 img, h5 img, h6 img {margin:0;}
36 | p {margin:0 0 1.5em;}
37 | p img.left {float:left;margin:1.5em 1.5em 1.5em 0;padding:0;}
38 | p img.right {float:right;margin:1.5em 0 1.5em 1.5em;}
39 | a:focus, a:hover {color:#08d;}
40 | a {color:#009;text-decoration:underline;}
41 | blockquote {margin:1.5em;color:#666;font-style:italic;}
42 | strong {font-weight:bold;}
43 | em, dfn {font-style:italic;}
44 | dfn {font-weight:bold;}
45 | sup, sub {line-height:0;}
46 | abbr, acronym {border-bottom:1px dotted #666;}
47 | address {margin:0 0 1.5em;font-style:italic;}
48 | del {color:#666;}
49 | pre {margin:1.5em 0;white-space:pre;}
50 | pre, code, tt {font:1em 'andale mono', 'lucida console', monospace;line-height:1.5;}
51 | li ul, li ol {margin:0;}
52 | ul, ol {margin:0 1.5em 1.5em 0;padding-left:3.333em;}
53 | ul {list-style-type:disc;}
54 | ol {list-style-type:decimal;}
55 | dl {margin:0 0 1.5em 0;}
56 | dl dt {font-weight:bold;}
57 | dd {margin-left:1.5em;}
58 | table {margin-bottom:1.4em;width:100%;}
59 | th {font-weight:bold;}
60 | thead th {background:#c3d9ff;}
61 | th, td, caption {padding:4px 10px 4px 5px;}
62 | tr.even td {background:#e5ecf9;}
63 | tfoot {font-style:italic;}
64 | caption {background:#eee;}
65 | .small {font-size:.8em;margin-bottom:1.875em;line-height:1.875em;}
66 | .large {font-size:1.2em;line-height:2.5em;margin-bottom:1.25em;}
67 | .hide {display:none;}
68 | .quiet {color:#666;}
69 | .loud {color:#000;}
70 | .highlight {background:#ff0;}
71 | .added {background:#060;color:#fff;}
72 | .removed {background:#900;color:#fff;}
73 | .first {margin-left:0;padding-left:0;}
74 | .last {margin-right:0;padding-right:0;}
75 | .top {margin-top:0;padding-top:0;}
76 | .bottom {margin-bottom:0;padding-bottom:0;}
77 |
78 | /* forms.css */
79 | label {font-weight:bold;}
80 | fieldset {padding:1.4em;margin:0 0 1.5em 0;border:1px solid #ccc;}
81 | legend {font-weight:bold;font-size:1.2em;}
82 | input[type=text], input[type=password], input.text, input.title, textarea, select {background-color:#fff;border:1px solid #bbb;}
83 | input[type=text]:focus, input[type=password]:focus, input.text:focus, input.title:focus, textarea:focus, select:focus {border-color:#666;}
84 | input[type=text], input[type=password], input.text, input.title, textarea, select {margin:0.5em 0;}
85 | input.text, input.title {width:300px;padding:5px;}
86 | input.title {font-size:1.5em;}
87 | textarea {width:390px;height:250px;padding:5px;}
88 | input[type=checkbox], input[type=radio], input.checkbox, input.radio {position:relative;top:.25em;}
89 | form.inline {line-height:3;}
90 | form.inline p {margin-bottom:0;}
91 | .error, .notice, .success, .info {padding:0.8em;margin-bottom:1em;border:2px solid #ddd;}
92 | .error {background:#fbe3e4;color:#8a1f11;border-color:#fbc2c4;}
93 | .notice {background:#fff6bf;color:#514721;border-color:#ffd324;}
94 | .success {background:#e6efc2;color:#264409;border-color:#c6d880;}
95 | .info {background:#d5edf8;color:#205791;border-color:#92cae4;}
96 | .error a {color:#8a1f11;}
97 | .notice a {color:#514721;}
98 | .success a {color:#264409;}
99 | .info a {color:#205791;}
100 |
101 | /* grid.css */
102 | .container {width:950px;margin:0 auto;}
103 | .showgrid {background:url(src/grid.png);}
104 | .column, .span-1, .span-2, .span-3, .span-4, .span-5, .span-6, .span-7, .span-8, .span-9, .span-10, .span-11, .span-12, .span-13, .span-14, .span-15, .span-16, .span-17, .span-18, .span-19, .span-20, .span-21, .span-22, .span-23, .span-24 {float:left;margin-right:10px;}
105 | .last {margin-right:0;}
106 | .span-1 {width:30px;}
107 | .span-2 {width:70px;}
108 | .span-3 {width:110px;}
109 | .span-4 {width:150px;}
110 | .span-5 {width:190px;}
111 | .span-6 {width:230px;}
112 | .span-7 {width:270px;}
113 | .span-8 {width:310px;}
114 | .span-9 {width:350px;}
115 | .span-10 {width:390px;}
116 | .span-11 {width:430px;}
117 | .span-12 {width:470px;}
118 | .span-13 {width:510px;}
119 | .span-14 {width:550px;}
120 | .span-15 {width:590px;}
121 | .span-16 {width:630px;}
122 | .span-17 {width:670px;}
123 | .span-18 {width:710px;}
124 | .span-19 {width:750px;}
125 | .span-20 {width:790px;}
126 | .span-21 {width:830px;}
127 | .span-22 {width:870px;}
128 | .span-23 {width:910px;}
129 | .span-24 {width:950px;margin-right:0;}
130 | input.span-1, textarea.span-1, input.span-2, textarea.span-2, input.span-3, textarea.span-3, input.span-4, textarea.span-4, input.span-5, textarea.span-5, input.span-6, textarea.span-6, input.span-7, textarea.span-7, input.span-8, textarea.span-8, input.span-9, textarea.span-9, input.span-10, textarea.span-10, input.span-11, textarea.span-11, input.span-12, textarea.span-12, input.span-13, textarea.span-13, input.span-14, textarea.span-14, input.span-15, textarea.span-15, input.span-16, textarea.span-16, input.span-17, textarea.span-17, input.span-18, textarea.span-18, input.span-19, textarea.span-19, input.span-20, textarea.span-20, input.span-21, textarea.span-21, input.span-22, textarea.span-22, input.span-23, textarea.span-23, input.span-24, textarea.span-24 {border-left-width:1px;border-right-width:1px;padding-left:5px;padding-right:5px;}
131 | input.span-1, textarea.span-1 {width:18px;}
132 | input.span-2, textarea.span-2 {width:58px;}
133 | input.span-3, textarea.span-3 {width:98px;}
134 | input.span-4, textarea.span-4 {width:138px;}
135 | input.span-5, textarea.span-5 {width:178px;}
136 | input.span-6, textarea.span-6 {width:218px;}
137 | input.span-7, textarea.span-7 {width:258px;}
138 | input.span-8, textarea.span-8 {width:298px;}
139 | input.span-9, textarea.span-9 {width:338px;}
140 | input.span-10, textarea.span-10 {width:378px;}
141 | input.span-11, textarea.span-11 {width:418px;}
142 | input.span-12, textarea.span-12 {width:458px;}
143 | input.span-13, textarea.span-13 {width:498px;}
144 | input.span-14, textarea.span-14 {width:538px;}
145 | input.span-15, textarea.span-15 {width:578px;}
146 | input.span-16, textarea.span-16 {width:618px;}
147 | input.span-17, textarea.span-17 {width:658px;}
148 | input.span-18, textarea.span-18 {width:698px;}
149 | input.span-19, textarea.span-19 {width:738px;}
150 | input.span-20, textarea.span-20 {width:778px;}
151 | input.span-21, textarea.span-21 {width:818px;}
152 | input.span-22, textarea.span-22 {width:858px;}
153 | input.span-23, textarea.span-23 {width:898px;}
154 | input.span-24, textarea.span-24 {width:938px;}
155 | .append-1 {padding-right:40px;}
156 | .append-2 {padding-right:80px;}
157 | .append-3 {padding-right:120px;}
158 | .append-4 {padding-right:160px;}
159 | .append-5 {padding-right:200px;}
160 | .append-6 {padding-right:240px;}
161 | .append-7 {padding-right:280px;}
162 | .append-8 {padding-right:320px;}
163 | .append-9 {padding-right:360px;}
164 | .append-10 {padding-right:400px;}
165 | .append-11 {padding-right:440px;}
166 | .append-12 {padding-right:480px;}
167 | .append-13 {padding-right:520px;}
168 | .append-14 {padding-right:560px;}
169 | .append-15 {padding-right:600px;}
170 | .append-16 {padding-right:640px;}
171 | .append-17 {padding-right:680px;}
172 | .append-18 {padding-right:720px;}
173 | .append-19 {padding-right:760px;}
174 | .append-20 {padding-right:800px;}
175 | .append-21 {padding-right:840px;}
176 | .append-22 {padding-right:880px;}
177 | .append-23 {padding-right:920px;}
178 | .prepend-1 {padding-left:40px;}
179 | .prepend-2 {padding-left:80px;}
180 | .prepend-3 {padding-left:120px;}
181 | .prepend-4 {padding-left:160px;}
182 | .prepend-5 {padding-left:200px;}
183 | .prepend-6 {padding-left:240px;}
184 | .prepend-7 {padding-left:280px;}
185 | .prepend-8 {padding-left:320px;}
186 | .prepend-9 {padding-left:360px;}
187 | .prepend-10 {padding-left:400px;}
188 | .prepend-11 {padding-left:440px;}
189 | .prepend-12 {padding-left:480px;}
190 | .prepend-13 {padding-left:520px;}
191 | .prepend-14 {padding-left:560px;}
192 | .prepend-15 {padding-left:600px;}
193 | .prepend-16 {padding-left:640px;}
194 | .prepend-17 {padding-left:680px;}
195 | .prepend-18 {padding-left:720px;}
196 | .prepend-19 {padding-left:760px;}
197 | .prepend-20 {padding-left:800px;}
198 | .prepend-21 {padding-left:840px;}
199 | .prepend-22 {padding-left:880px;}
200 | .prepend-23 {padding-left:920px;}
201 | .border {padding-right:4px;margin-right:5px;border-right:1px solid #ddd;}
202 | .colborder {padding-right:24px;margin-right:25px;border-right:1px solid #ddd;}
203 | .pull-1 {margin-left:-40px;}
204 | .pull-2 {margin-left:-80px;}
205 | .pull-3 {margin-left:-120px;}
206 | .pull-4 {margin-left:-160px;}
207 | .pull-5 {margin-left:-200px;}
208 | .pull-6 {margin-left:-240px;}
209 | .pull-7 {margin-left:-280px;}
210 | .pull-8 {margin-left:-320px;}
211 | .pull-9 {margin-left:-360px;}
212 | .pull-10 {margin-left:-400px;}
213 | .pull-11 {margin-left:-440px;}
214 | .pull-12 {margin-left:-480px;}
215 | .pull-13 {margin-left:-520px;}
216 | .pull-14 {margin-left:-560px;}
217 | .pull-15 {margin-left:-600px;}
218 | .pull-16 {margin-left:-640px;}
219 | .pull-17 {margin-left:-680px;}
220 | .pull-18 {margin-left:-720px;}
221 | .pull-19 {margin-left:-760px;}
222 | .pull-20 {margin-left:-800px;}
223 | .pull-21 {margin-left:-840px;}
224 | .pull-22 {margin-left:-880px;}
225 | .pull-23 {margin-left:-920px;}
226 | .pull-24 {margin-left:-960px;}
227 | .pull-1, .pull-2, .pull-3, .pull-4, .pull-5, .pull-6, .pull-7, .pull-8, .pull-9, .pull-10, .pull-11, .pull-12, .pull-13, .pull-14, .pull-15, .pull-16, .pull-17, .pull-18, .pull-19, .pull-20, .pull-21, .pull-22, .pull-23, .pull-24 {float:left;position:relative;}
228 | .push-1 {margin:0 -40px 1.5em 40px;}
229 | .push-2 {margin:0 -80px 1.5em 80px;}
230 | .push-3 {margin:0 -120px 1.5em 120px;}
231 | .push-4 {margin:0 -160px 1.5em 160px;}
232 | .push-5 {margin:0 -200px 1.5em 200px;}
233 | .push-6 {margin:0 -240px 1.5em 240px;}
234 | .push-7 {margin:0 -280px 1.5em 280px;}
235 | .push-8 {margin:0 -320px 1.5em 320px;}
236 | .push-9 {margin:0 -360px 1.5em 360px;}
237 | .push-10 {margin:0 -400px 1.5em 400px;}
238 | .push-11 {margin:0 -440px 1.5em 440px;}
239 | .push-12 {margin:0 -480px 1.5em 480px;}
240 | .push-13 {margin:0 -520px 1.5em 520px;}
241 | .push-14 {margin:0 -560px 1.5em 560px;}
242 | .push-15 {margin:0 -600px 1.5em 600px;}
243 | .push-16 {margin:0 -640px 1.5em 640px;}
244 | .push-17 {margin:0 -680px 1.5em 680px;}
245 | .push-18 {margin:0 -720px 1.5em 720px;}
246 | .push-19 {margin:0 -760px 1.5em 760px;}
247 | .push-20 {margin:0 -800px 1.5em 800px;}
248 | .push-21 {margin:0 -840px 1.5em 840px;}
249 | .push-22 {margin:0 -880px 1.5em 880px;}
250 | .push-23 {margin:0 -920px 1.5em 920px;}
251 | .push-24 {margin:0 -960px 1.5em 960px;}
252 | .push-1, .push-2, .push-3, .push-4, .push-5, .push-6, .push-7, .push-8, .push-9, .push-10, .push-11, .push-12, .push-13, .push-14, .push-15, .push-16, .push-17, .push-18, .push-19, .push-20, .push-21, .push-22, .push-23, .push-24 {float:right;position:relative;}
253 | div.prepend-top, .prepend-top {margin-top:1.5em;}
254 | div.append-bottom, .append-bottom {margin-bottom:1.5em;}
255 | .box {padding:1.5em;margin-bottom:1.5em;background:#E5ECF9;}
256 | hr {background:#ddd;color:#ddd;clear:both;float:none;width:100%;height:1px;margin:0 0 1.45em;border:none;}
257 | hr.space {background:#fff;color:#fff;visibility:hidden;}
258 | .clearfix:after, .container:after {content:"\0020";display:block;height:0;clear:both;visibility:hidden;overflow:hidden;}
259 | .clearfix, .container {display:block;}
260 | .clear {clear:both;}
--------------------------------------------------------------------------------
/simsearch/static/js/search.js:
--------------------------------------------------------------------------------
1 | /*
2 | *
3 | * search.py
4 | * simsearch
5 | *
6 | * Created by Lars Yencken on 31-08-2010.
7 | * Copyright 2010 Lars Yencken. All rights reserved.
8 | *
9 | */
10 |
11 | /*
12 | * GLOBALS
13 | */
14 |
15 | // The fraction of the screen's width or height to use for the lookup pane.
16 | var g_useFraction = 0.8;
17 |
18 | // A history of data, so that we can provide backwards and forwards
19 | // navigation.
20 | var g_historyStore = new Array();
21 | var currentIndex = null;
22 |
23 | // The current "mode" of the interface.
24 | var g_currentState = 'seeding';
25 |
26 | // Extra state about what's currently being displayed. Only updated by objects
27 | // that need it on draw() and clean() operations, primarily to avoid excessive
28 | // redraws.
29 | var g_currentObjects = {}
30 | var g_windowDirty = false;
31 |
32 | /*
33 | * drawError()
34 | * Draws an error message to the screen.
35 | */
36 | function drawError(messageEn, messageJp, timeout) {
37 | // Render it, but still hidden, so we can check its size.
38 | g_currentObjects['errorMessage'] = [messageEn, messageJp, timeout]
39 | setOpacity('errorMessage', 0.01);
40 | showElement('errorMessage');
41 |
42 | // Delete any previous messages.
43 | var parentNode = $('errorMessage')
44 | while (parentNode.childNodes.length > 0) {
45 | c = parentNode.childNodes[0];
46 | parentNode.removeChild(c);
47 | delete c;
48 | }
49 |
50 | // Add our messages.
51 | parentNode.appendChild(document.createTextNode(messageEn))
52 | parentNode.appendChild(document.createElement('br'));
53 | parentNode.appendChild(document.createTextNode(messageJp))
54 |
55 | var errorSize = getElementDimensions('errorMessage');
56 | var seedInputSize = getSeedingInputSize();
57 | var seedInputLoc = getElementPosition(document['seedForm'].seedKanji);
58 | var windowSize = getWindowSize();
59 |
60 | var errorLoc = new Coordinates();
61 | errorLoc.x = Math.round(seedInputLoc.x + seedInputSize.w/2 - errorSize.w/2);
62 | errorLoc.y = seedInputLoc.y + seedInputSize.h + 50;
63 |
64 | setElementPosition('errorMessage', errorLoc);
65 |
66 | appear('errorMessage');
67 | drawSearch(false);
68 | if (timeout > 0) {
69 | callLater(timeout, function(){
70 | fade('errorMessage');
71 | delete g_currentObjects["errorMessage"];
72 | });
73 | }
74 | }
75 |
76 | /*
77 | * clearError()
78 | * Hides any error messages that are shown.
79 | */
80 | function clearError() {
81 | hideElement('errorMessage');
82 | }
83 |
84 | /*
85 | * getSeedingInputSize()
86 | * Returns the size of the combined input element.
87 | */
88 | function getSeedingInputSize() {
89 | setOpacity('seedLookup', 0.01);
90 | showElement('seedLookup');
91 | // Get the elements.
92 | var magnifier = document['magnifier'];
93 | var seedKanjiInput = document['seedForm'].seedKanji
94 |
95 | // Determine their sizes.
96 | var inputSize = getElementDimensions(seedKanjiInput);
97 | var magSize = getElementDimensions(magnifier);
98 |
99 | // Calculate the combined size.
100 | var combinedSize = new Dimensions();
101 | var padding = 5;
102 | combinedSize.w = inputSize.w + padding + magSize.w;
103 | combinedSize.h = magSize.h;
104 |
105 | return combinedSize;
106 | }
107 |
108 | /*
109 | * drawSeedingInput()
110 | * Draw the initial input dialog.
111 | */
112 | function drawSeedingInput(useAppear) {
113 | drawSearch(useAppear);
114 | if ("errorMessage" in g_currentObjects) {
115 | var em = g_currentObjects["errorMessage"];
116 | drawError(em[0], em[1], em[2]);
117 | }
118 | }
119 |
120 | function drawSearch(useAppear) {
121 | if (useAppear == null) {
122 | useAppear = true;
123 | }
124 |
125 | // Calculate the coordinates.
126 | var windowSize = getWindowSize();
127 | var inputLoc = new Coordinates();
128 | var combinedSize = getSeedingInputSize();
129 | inputLoc.x = Math.round(0.5*(windowSize.w - combinedSize.w));
130 | inputLoc.y = Math.round(0.5*(windowSize.h - combinedSize.h));
131 |
132 | // Center and make the input appear.
133 | setElementPosition('seedLookup', inputLoc);
134 | if (useAppear) {
135 | appear('seedLookup');
136 | callLater(1, function() { document['seedForm'].seedKanji.focus();});
137 | } else {
138 | setOpacity('seedLookup', 1.0);
139 | document['seedForm'].seedKanji.focus();
140 | }
141 | }
142 |
143 | /*
144 | * clearSeedingInput()
145 | * Clear the initial seeding input dialog.
146 | */
147 | function clearSeedingInput() {
148 | hideElement('seedLookup');
149 | hideElement('errorMessage');
150 | return;
151 | }
152 |
153 | /*
154 | * switchState()
155 | * Switches from one UI state to another.
156 | */
157 | function switchState(newState, stateArg) {
158 | g_clearState[g_currentState]();
159 | g_currentState = newState;
160 | g_initState[newState](stateArg);
161 | }
162 |
163 | /*
164 | * submitSeed()
165 | */
166 | function submitSeed() {
167 | var value = document['seedForm'].seedKanji.value;
168 |
169 | var messageEn = "Please enter a single kanji only.";
170 | var messageJp = '一つの漢字を入力してください。';
171 |
172 | if (value.length != 1) {
173 | drawError(messageEn, messageJp, 6);
174 | } else {
175 | // Check that the input is a kanji.
176 | logDebug("Ok value: " + value);
177 | var valueOrd = ord(value);
178 | if (valueOrd < 12353 || valueOrd > 40869) {
179 | drawError(messageEn, messageJp, 6);
180 | } else {
181 | // Valid!
182 | switchState("lookup", value);
183 | }
184 | }
185 | return false;
186 | }
187 |
188 | /*
189 | * initLookup()
190 | * Loads the kanji which are similar to the current pivot.
191 | */
192 | function initLookup(pivotKanjiVal) {
193 | if (currentIndex != null) {
194 | var previousKanji = g_historyStore[currentIndex].pivotKanji;
195 | var newDoc = loadJSONDoc(g_pivotPath + pivotKanjiVal + "/", {});
196 | } else {
197 | var newDoc = loadJSONDoc(g_pivotPath + pivotKanjiVal + "/", {});
198 | }
199 |
200 | var success = function(obj) {
201 | // Store current values, and redraw the screen.
202 | // Determine whether truncation is needed.
203 | if (currentIndex == null) {
204 | // First time.
205 | currentIndex = 0;
206 | } else if (g_historyStore.length > currentIndex + 1) {
207 | // Truncation needed!
208 | currentIndex++;
209 | g_historyStore.length = currentIndex;
210 | } else {
211 | currentIndex++;
212 | }
213 | g_historyStore[currentIndex] = obj;
214 |
215 | drawBorder();
216 | drawLookup();
217 | }
218 |
219 | var failure = function(err) {
220 | logDebug("Couldn't load data: " + err);
221 | switchState('seeding', null);
222 | drawError('No data found for the kanji ' + pivotKanjiVal + '.', '',
223 | 6);
224 | }
225 |
226 | newDoc.addCallbacks(success, failure);
227 | };
228 |
229 | /*
230 | * fullRedraw()
231 | * Redraws the whole screen, using the existing kanji.
232 | */
233 | function fullRedraw() {
234 | g_windowDirty = true;
235 | g_drawState[g_currentState]();
236 | g_windowDirty = false;
237 | };
238 |
239 | // Redraw the window if it is resized.
240 | window.onresize = fullRedraw;
241 |
242 | /*
243 | * getLookupPlane()
244 | * Fetches the full plane size of the area.
245 | */
246 | function getLookupPlane() {
247 | var windowSize = getWindowSize();
248 | var lookupPlane = {};
249 | lookupPlane.w = g_useFraction*windowSize.w;
250 | lookupPlane.h = g_useFraction*windowSize.h;
251 |
252 | // Make it square.
253 | if (lookupPlane.w < lookupPlane.h) {
254 | lookupPlane.h = lookupPlane.w;
255 | } else {
256 | lookupPlane.w = lookupPlane.h;
257 | }
258 |
259 | // Make it centered.
260 | lookupPlane.left = Math.round((windowSize.w - lookupPlane.w)/2);
261 | lookupPlane.right = lookupPlane.left + lookupPlane.w;
262 | lookupPlane.top = Math.round((windowSize.h - lookupPlane.h)/2);
263 | lookupPlane.bottom = lookupPlane.top + lookupPlane.h;
264 |
265 | lookupPlane.center = new Coordinates(
266 | lookupPlane.left + lookupPlane.w/2,
267 | lookupPlane.top + lookupPlane.h/2
268 | );
269 |
270 | return lookupPlane;
271 | };
272 |
273 | /*
274 | * drawBorder()
275 | * Draws a border around the kanji area.
276 | */
277 | function drawBorder() {
278 | if (g_currentObjects['border'] && !g_windowDirty) {
279 | return;
280 | }
281 | logDebug('Drawing border');
282 | // Work out the dimensions.
283 | var lookupPlane = getLookupPlane();
284 |
285 | // Generate the new border.
286 | var styleString = "position:absolute; ";
287 | styleString += 'left: ' + lookupPlane.left + 'px; '
288 | styleString += 'right: ' + lookupPlane.right + 'px; '
289 | styleString += 'top: ' + lookupPlane.top + 'px; '
290 | styleString += 'bottom: ' + lookupPlane.bottom + 'px; '
291 | styleString += 'width: ' + lookupPlane.w + 'px; '
292 | styleString += 'height: ' + lookupPlane.h + 'px; '
293 | var newBorder = DIV({style: styleString, id:"lookupBorder"}, "");
294 |
295 | // Swap it for the old border.
296 | swapDOM("lookupBorder", newBorder);
297 |
298 | setOpacity("lookupBorder", 1.0);
299 | g_currentObjects['border'] = true;
300 | };
301 |
302 | /*
303 | * clearBorder()
304 | * Clears the border around the kanji area.
305 | */
306 | function clearBorder() {
307 | logDebug('Clearing border');
308 | setOpacity('lookupBorder', 0.0);
309 | g_currentObjects['border'] = false;
310 | }
311 |
312 | /*
313 | * getControlPlane()
314 | * Determine the placement of the control plane.
315 | */
316 | function getControlPlane() {
317 | var windowSize = getWindowSize();
318 | var lookupPlane = getLookupPlane();
319 |
320 | var controlPlane = {};
321 | controlPlane.top = lookupPlane.bottom;
322 | controlPlane.bottom = windowSize.h;
323 | controlPlane.left = lookupPlane.left;
324 | controlPlane.right = lookupPlane.right;
325 | controlPlane.w = controlPlane.right - controlPlane.left;
326 | controlPlane.h = controlPlane.bottom - controlPlane.top;
327 |
328 | controlPlane.center = new Coordinates(
329 | controlPlane.left + controlPlane.w/2,
330 | controlPlane.top + controlPlane.h/2
331 | );
332 |
333 | return controlPlane;
334 | }
335 |
336 | /*
337 | * clearControls()
338 | * Clears the controls from the display.
339 | */
340 | function clearControls() {
341 | hideElement("backControl");
342 | hideElement("forwardControl");
343 | hideElement("resetControl");
344 | return;
345 | }
346 |
347 | /*
348 | * drawControls()
349 | * Repositions the back and forward links dynamically.
350 | */
351 | function drawControls() {
352 | // Fetch location and size information needed for calculations.
353 | var controlPlane = getControlPlane();
354 |
355 | //var backSize = elementDimensions(document["backControl"]);
356 | //var forwardSize = elementDimensions(document["forwardControl"]);
357 | var backSize = new Dimensions(30, 35);
358 | var forwardSize = new Dimensions(30, 35);
359 | var resetSize = new Dimensions(35, 35);
360 |
361 | var top = Math.round(controlPlane.top + controlPlane.h/2 - 35/2);
362 | var backPos = new Coordinates(controlPlane.left, top);
363 |
364 | var forwardPos = new Coordinates(controlPlane.right - backSize.w, top);
365 |
366 | var resetPos = new Coordinates(
367 | Math.round(controlPlane.left + controlPlane.w/2 - resetSize.w/2),
368 | top
369 | );
370 |
371 | setElementPosition("backControl", backPos);
372 | setElementPosition("forwardControl", forwardPos);
373 | setElementPosition("resetControl", resetPos);
374 |
375 | appear("resetControl");
376 |
377 | if (hasHistory()) {
378 | appear("backControl");
379 | } else {
380 | hideElement("backControl");
381 | }
382 |
383 | if (hasFuture()) {
384 | appear("forwardControl");
385 | } else {
386 | hideElement("forwardControl");
387 | }
388 |
389 | g_currentObjects['controls'] = true;
390 |
391 | return;
392 | };
393 |
394 | /*
395 | * clearKanji()
396 | */
397 | function clearKanji() {
398 | // Hide the existing elements.
399 | var existingElements = getElementsByTagAndClassName("div", "similarKanji");
400 | for (var i = 0; i < existingElements.length; i++) {
401 | removeElement(existingElements[i]);
402 | }
403 | setOpacity('pivotKanji', 0.0);
404 |
405 | g_currentObjects['controls'] = false;
406 | }
407 |
408 | /*
409 | * clearLookup()
410 | * Clears any kanji currently displayed.
411 | */
412 | function clearLookup() {
413 | logDebug('Clearing lookup');
414 |
415 | clearKanji();
416 | clearControls();
417 | clearBorder();
418 |
419 | // Clear the history too.
420 | g_historyStore.length = 0;
421 | currentIndex = null;
422 | return;
423 | }
424 |
425 | /*
426 | * drawLookup()
427 | * Draws the existing dataset to the screen.
428 | */
429 | function drawLookup() {
430 | drawBorder();
431 | drawControls();
432 | drawKanji();
433 | }
434 |
435 | /*
436 | * drawKanji()
437 | * Draws the lookup kanji to the window.
438 | */
439 | function drawKanji() {
440 | logDebug("Drawing kanji");
441 | clearKanji();
442 |
443 | // Draw the new pivot.
444 | var lookupPlane = getLookupPlane();
445 | var pivotLoc = toCornerLoc(lookupPlane.center);
446 | var pivotKanji = g_historyStore[currentIndex].pivot_kanji;
447 | var path = ""
448 | for (var i = 0; i < g_historyStore.length; i++) {
449 | path += g_historyStore[i]['pivot_kanji']
450 | }
451 |
452 | newPivot = DIV(
453 | {
454 | id: "pivotKanji",
455 | style: locToStyle(pivotLoc)
456 | },
457 | A(
458 | {href: g_translatePath + pivotKanji + "/?path=" + path},
459 | pivotKanji
460 | )
461 | );
462 | swapDOM("pivotKanji", newPivot);
463 |
464 | var tier1 = g_historyStore[currentIndex].tier1;
465 | var tier2 = g_historyStore[currentIndex].tier2;
466 | var tier3 = g_historyStore[currentIndex].tier3;
467 |
468 | drawTier(tier1, 1);
469 | drawTier(tier2, 2);
470 | drawTier(tier3, 3);
471 |
472 | callLater(0.1, function() {
473 | showClass('tier1Kanji');
474 | callLater(0.1, function() {
475 | showClass('tier2Kanji');
476 | callLater(0.1, function() {
477 | showClass('tier3Kanji');
478 | });
479 | });
480 | });
481 |
482 | return;
483 | }
484 |
485 | /*
486 | * locAdd(locA, locB)
487 | * Addition of coordinates, returns the result.
488 | */
489 | function locAdd(locA, locB) {
490 | return new Coordinates(locA.x + locB.x, locA.y + locB.y);
491 | }
492 |
493 | /*
494 | * drawTier(kanjiArray, tierNumber)
495 | * Draws the given numbered tier (can be 1, 2 or 3), and spaces out the
496 | * given kanji evenly upon that tier.
497 | */
498 | function drawTier(kanjiArray, tierNumber) {
499 | logDebug("Drawing tier " + tierNumber);
500 | var lookupPlane = getLookupPlane();
501 |
502 | var center = lookupPlane.center;
503 | var angleFraction = 2*Math.PI/kanjiArray.length;
504 | var radius = tierNumber*lookupPlane.w/8;
505 | var initialFraction = Math.random()*angleFraction;
506 |
507 | // Generate new elements for the kanji.
508 | var newElements = new Array();
509 | for (var i = 0; i < kanjiArray.length; i++) {
510 | var kanji = kanjiArray[i];
511 | var loadCommand = "initLookup('" + kanji + "')";
512 |
513 | var kanjiAngle = initialFraction + i*angleFraction;
514 | var loc = new Coordinates(
515 | radius*Math.cos(kanjiAngle),
516 | radius*Math.sin(kanjiAngle)
517 | )
518 | loc = locAdd(loc, lookupPlane.center);
519 | loc = toCornerLoc(loc);
520 | newElements[i] = DIV(
521 | {
522 | class: "tier" + tierNumber + "Kanji similarKanji",
523 | style: "opacity:0.0; " + locToStyle(loc),
524 | },
525 | A(
526 | {
527 | href: "javascript:;",
528 | onclick: loadCommand
529 | },
530 | kanji
531 | )
532 | );
533 | }
534 |
535 | var body = document.getElementsByTagName("body")[0];
536 | appendChildNodes(body, newElements);
537 |
538 | return;
539 | };
540 |
541 | /*
542 | * showClass()
543 | * Make all the DIV elements of this class appear.
544 | */
545 | function showClass(className) {
546 | newElements = getElementsByTagAndClassName("div", className);
547 | for (var i = 0; i < newElements.length; i++) {
548 | setOpacity(newElements[i], 1.0);
549 | }
550 | }
551 |
552 | /*
553 | * getWindowSize()
554 | * Returns the size in pixels of the browser window as a (width, height)
555 | * array of 2 elements.
556 | */
557 | function getWindowSize() {
558 | var windowSize = new Dimensions();
559 | if (navigator.appName == "Netscape") {
560 | windowSize.w = window.innerWidth;
561 | windowSize.h = window.innerHeight;
562 |
563 | } else if (navigator.appName == "Microsoft Internet Explorer") {
564 | windowSize.w = document.body.clientWidth;
565 | windowSize.h = document.body.clientHeight;
566 | }
567 |
568 | return windowSize;
569 | };
570 |
571 | /*
572 | * toCornerLoc(centerLoc, object=kanji)
573 | * Given the coordinates for the center of the object, returns the
574 | * coordinates for the top-left corner of the object, which is used
575 | * to lay it out.
576 | */
577 | function toCornerLoc(centerLoc, object) {
578 | var lookupPlane = getLookupPlane();
579 |
580 | if (object == null) {
581 | var objectSize = getElementDimensions("pivotKanji");
582 | } else {
583 | var objectSize = getElemendDimensions(object);
584 | }
585 |
586 | var retVal = new Coordinates();
587 | retVal.x = Math.round(centerLoc.x - objectSize.w/2);
588 | retVal.y = Math.round(centerLoc.y - objectSize.h/2);
589 |
590 | return retVal;
591 | };
592 |
593 | /*
594 | * hasHistory()
595 | * Returns true if there is a previous lookup state recorded.
596 | */
597 | function hasHistory() {
598 | return (currentIndex > 0);
599 | }
600 |
601 | /*
602 | * hasFuture()
603 | * Returns true if the user has browsed back from a lookup state.
604 | */
605 | function hasFuture() {
606 | if (currentIndex == null) {
607 | return false;
608 | } else {
609 | return (currentIndex < g_historyStore.length - 1);
610 | }
611 | }
612 |
613 | /*
614 | * previousPivot()
615 | * Fetches the previous pivot (if one exists) and redraws.
616 | */
617 | function previousPivot() {
618 | if (hasHistory()) {
619 | currentIndex--;
620 | clearKanji();
621 | drawKanji();
622 | drawControls();
623 | }
624 | };
625 |
626 | /*
627 | * nextPivot()
628 | * Fetches the next pivot (if one exists) and redraws.
629 | */
630 | function nextPivot() {
631 | if (hasFuture()) {
632 | currentIndex++;
633 | clearKanji();
634 | drawKanji();
635 | drawControls();
636 | }
637 | }
638 |
639 | /*
640 | * ord()
641 | * Returns the character code of a character.
642 | */
643 | function ord(char) {
644 | return char.charCodeAt(0);
645 | }
646 |
647 | /*
648 | * chr(charCode)
649 | * Returns the character indicated by the given character code.
650 | */
651 | function chr(charCode) {
652 | return String.fromCharCode(charCode);
653 | }
654 |
655 | var g_clearState = {
656 | 'seeding': clearSeedingInput,
657 | 'lookup': clearLookup,
658 | }
659 |
660 | var g_drawState = {
661 | 'seeding': drawSeedingInput,
662 | 'lookup': drawLookup,
663 | }
664 |
665 | var g_initState = {
666 | 'seeding': drawSeedingInput,
667 | 'lookup': initLookup,
668 | }
669 |
670 | function initInterface() {
671 | drawSeedingInput();
672 | callLater(4, function() {
673 | if (emptyInput()) {
674 | drawError('Enter a kanji similar to',
675 | 'to the one you want to find.', 0);
676 | }
677 | });
678 | }
679 |
680 | function emptyInput() {
681 | return g_currentState == 'seeding'
682 | && document['seedForm'].seedKanji.value == '';
683 | }
684 |
685 | /*
686 | * Converts coordinates to a style string.
687 | */
688 | function locToStyle(loc) {
689 | return "position:absolute; left:" + loc.x + "px; top:" + loc.y + "px; ";
690 | }
691 |
692 | // Image rollover function.
693 | function roll(imageName, imageSrc) {
694 | document[imageName].src = imageSrc;
695 | }
696 |
697 |
--------------------------------------------------------------------------------
/simsearch/static/js/raphael-min.js:
--------------------------------------------------------------------------------
1 | /*
2 | * Raphael 1.5.2 - JavaScript Vector Library
3 | *
4 | * Copyright (c) 2010 Dmitry Baranovskiy (http://raphaeljs.com)
5 | * Licensed under the MIT (http://raphaeljs.com/license.html) license.
6 | */
7 | (function(){function a(){if(a.is(arguments[0],G)){var b=arguments[0],d=bV[m](a,b.splice(0,3+a.is(b[0],E))),e=d.set();for(var g=0,h=b[w];g";bg=bf.firstChild;bg.style.behavior="url(#default#VML)";if(!(bg&&typeof bg.adj=="object"))return a.type=null;bf=null}a.svg=!(a.vml=a.type=="VML");j[e]=a[e];k=j[e];a._id=0;a._oid=0;a.fn={};a.is=function(a,b){b=x.call(b);if(b=="finite")return!O[f](+a);return b=="null"&&a===null||b==typeof a||b=="object"&&a===Object(a)||b=="array"&&Array.isArray&&Array.isArray(a)||J.call(a).slice(8,-1).toLowerCase()==b};a.angle=function(b,c,d,e,f,g){{if(f==null){var h=b-d,i=c-e;if(!h&&!i)return 0;return((h<0)*180+y.atan(-i/-h)*180/D+360)%360}return a.angle(b,c,f,g)-a.angle(d,e,f,g)}};a.rad=function(a){return a%360*D/180};a.deg=function(a){return a*180/D%360};a.snapTo=function(b,c,d){d=a.is(d,"finite")?d:10;if(a.is(b,G)){var e=b.length;while(e--)if(B(b[e]-c)<=d)return b[e]}else{b=+b;var f=c%b;if(fb-d)return c-f+b}return c};function bh(){var a=[],b=0;for(;b<32;b++)a[b]=(~(~(y.random()*16)))[H](16);a[12]=4;a[16]=(a[16]&3|8)[H](16);return"r-"+a[v]("")}a.setWindow=function(a){h=a;g=h.document};var bi=function(b){if(a.vml){var c=/^\s+|\s+$/g,d;try{var e=new ActiveXObject("htmlfile");e.write("");e.close();d=e.body}catch(a){d=createPopup().document.body}var f=d.createTextRange();bi=bm(function(a){try{d.style.color=r(a)[Y](c,p);var b=f.queryCommandValue("ForeColor");b=(b&255)<<16|b&65280|(b&16711680)>>>16;return"#"+("000000"+b[H](16)).slice(-6)}catch(a){return"none"}})}else{var h=g.createElement("i");h.title="Raphaël Colour Picker";h.style.display="none";g.body[l](h);bi=bm(function(a){h.style.color=a;return g.defaultView.getComputedStyle(h,p).getPropertyValue("color")})}return bi(b)},bj=function(){return"hsb("+[this.h,this.s,this.b]+")"},bk=function(){return"hsl("+[this.h,this.s,this.l]+")"},bl=function(){return this.hex};a.hsb2rgb=function(b,c,d,e){if(a.is(b,"object")&&"h"in b&&"s"in b&&"b"in b){d=b.b;c=b.s;b=b.h;e=b.o}return a.hsl2rgb(b,c,d/2,e)};a.hsl2rgb=function(b,c,d,e){if(a.is(b,"object")&&"h"in b&&"s"in b&&"l"in b){d=b.l;c=b.s;b=b.h}if(b>1||c>1||d>1){b/=360;c/=100;d/=100}var f={},g=["r","g","b"],h,i,j,k,l,m;if(c){d<0.5?h=d*(1+c):h=d+c-d*c;i=2*d-h;for(var n=0;n<3;n++){j=b+1/3*-(n-1);j<0&&j++;j>1&&j--;j*6<1?f[g[n]]=i+(h-i)*6*j:j*2<1?f[g[n]]=h:j*3<2?f[g[n]]=i+(h-i)*(2/3-j)*6:f[g[n]]=i}}else f={r:d,g:d,b:d};f.r*=255;f.g*=255;f.b*=255;f.hex="#"+(16777216|f.b|f.g<<8|f.r<<16).toString(16).slice(1);a.is(e,"finite")&&(f.opacity=e);f.toString=bl;return f};a.rgb2hsb=function(b,c,d){if(c==null&&a.is(b,"object")&&"r"in b&&"g"in b&&"b"in b){d=b.b;c=b.g;b=b.r}if(c==null&&a.is(b,F)){var e=a.getRGB(b);b=e.r;c=e.g;d=e.b}if(b>1||c>1||d>1){b/=255;c/=255;d/=255}var f=z(b,c,d),g=A(b,c,d),h,i,j=f;{if(g==f)return{h:0,s:0,b:f,toString:bj};var k=f-g;i=k/f;b==f?h=(c-d)/k:c==f?h=2+(d-b)/k:h=4+(b-c)/k;h/=6;h<0&&h++;h>1&&h--}return{h:h,s:i,b:j,toString:bj}};a.rgb2hsl=function(b,c,d){if(c==null&&a.is(b,"object")&&"r"in b&&"g"in b&&"b"in b){d=b.b;c=b.g;b=b.r}if(c==null&&a.is(b,F)){var e=a.getRGB(b);b=e.r;c=e.g;d=e.b}if(b>1||c>1||d>1){b/=255;c/=255;d/=255}var f=z(b,c,d),g=A(b,c,d),h,i,j=(f+g)/2,k;if(g==f)k={h:0,s:0,l:j};else{var l=f-g;i=j<0.5?l/(f+g):l/(2-f-g);b==f?h=(c-d)/l:c==f?h=2+(d-b)/l:h=4+(b-c)/l;h/=6;h<0&&h++;h>1&&h--;k={h:h,s:i,l:j}}k.toString=bk;return k};a._path2string=function(){return this.join(",")[Y](ba,"$1")};function bm(a,b,c){function d(){var g=Array[e].slice.call(arguments,0),h=g[v]("►"),i=d.cache=d.cache||{},j=d.count=d.count||[];if(i[f](h))return c?c(i[h]):i[h];j[w]>=1000&&delete i[j.shift()];j[L](h);i[h]=a[m](b,g);return c?c(i[h]):i[h]}return d}a.getRGB=bm(function(b){if(!b||!(!((b=r(b)).indexOf("-")+1)))return{r:-1,g:-1,b:-1,hex:"none",error:1};if(b=="none")return{r:-1,g:-1,b:-1,hex:"none"};!(_[f](b.toLowerCase().substring(0,2))||b.charAt()=="#")&&(b=bi(b));var c,d,e,g,h,i,j,k=b.match(N);if(k){if(k[2]){g=T(k[2].substring(5),16);e=T(k[2].substring(3,5),16);d=T(k[2].substring(1,3),16)}if(k[3]){g=T((i=k[3].charAt(3))+i,16);e=T((i=k[3].charAt(2))+i,16);d=T((i=k[3].charAt(1))+i,16)}if(k[4]){j=k[4][s]($);d=S(j[0]);j[0].slice(-1)=="%"&&(d*=2.55);e=S(j[1]);j[1].slice(-1)=="%"&&(e*=2.55);g=S(j[2]);j[2].slice(-1)=="%"&&(g*=2.55);k[1].toLowerCase().slice(0,4)=="rgba"&&(h=S(j[3]));j[3]&&j[3].slice(-1)=="%"&&(h/=100)}if(k[5]){j=k[5][s]($);d=S(j[0]);j[0].slice(-1)=="%"&&(d*=2.55);e=S(j[1]);j[1].slice(-1)=="%"&&(e*=2.55);g=S(j[2]);j[2].slice(-1)=="%"&&(g*=2.55);(j[0].slice(-3)=="deg"||j[0].slice(-1)=="°")&&(d/=360);k[1].toLowerCase().slice(0,4)=="hsba"&&(h=S(j[3]));j[3]&&j[3].slice(-1)=="%"&&(h/=100);return a.hsb2rgb(d,e,g,h)}if(k[6]){j=k[6][s]($);d=S(j[0]);j[0].slice(-1)=="%"&&(d*=2.55);e=S(j[1]);j[1].slice(-1)=="%"&&(e*=2.55);g=S(j[2]);j[2].slice(-1)=="%"&&(g*=2.55);(j[0].slice(-3)=="deg"||j[0].slice(-1)=="°")&&(d/=360);k[1].toLowerCase().slice(0,4)=="hsla"&&(h=S(j[3]));j[3]&&j[3].slice(-1)=="%"&&(h/=100);return a.hsl2rgb(d,e,g,h)}k={r:d,g:e,b:g};k.hex="#"+(16777216|g|e<<8|d<<16).toString(16).slice(1);a.is(h,"finite")&&(k.opacity=h);return k}return{r:-1,g:-1,b:-1,hex:"none",error:1}},a);a.getColor=function(a){var b=this.getColor.start=this.getColor.start||{h:0,s:1,b:a||0.75},c=this.hsb2rgb(b.h,b.s,b.b);b.h+=0.075;if(b.h>1){b.h=0;b.s-=0.2;b.s<=0&&(this.getColor.start={h:0,s:1,b:b.b})}return c.hex};a.getColor.reset=function(){delete this.start};a.parsePathString=bm(function(b){if(!b)return null;var c={a:7,c:6,h:1,l:2,m:2,q:4,s:4,t:2,v:1,z:0},d=[];a.is(b,G)&&a.is(b[0],G)&&(d=bo(b));d[w]||r(b)[Y](bb,function(a,b,e){var f=[],g=x.call(b);e[Y](bc,function(a,b){b&&f[L](+b)});if(g=="m"&&f[w]>2){d[L]([b][n](f.splice(0,2)));g="l";b=b=="m"?"l":"L"}while(f[w]>=c[g]){d[L]([b][n](f.splice(0,c[g])));if(!c[g])break}});d[H]=a._path2string;return d});a.findDotsAtSegment=function(a,b,c,d,e,f,g,h,i){var j=1-i,k=C(j,3)*a+C(j,2)*3*i*c+j*3*i*i*e+C(i,3)*g,l=C(j,3)*b+C(j,2)*3*i*d+j*3*i*i*f+C(i,3)*h,m=a+2*i*(c-a)+i*i*(e-2*c+a),n=b+2*i*(d-b)+i*i*(f-2*d+b),o=c+2*i*(e-c)+i*i*(g-2*e+c),p=d+2*i*(f-d)+i*i*(h-2*f+d),q=(1-i)*a+i*c,r=(1-i)*b+i*d,s=(1-i)*e+i*g,t=(1-i)*f+i*h,u=90-y.atan((m-o)/(n-p))*180/D;(m>o||n1){x=y.sqrt(x);c=x*c;d=x*d}var z=c*c,A=d*d,C=(f==g?-1:1)*y.sqrt(B((z*A-z*u*u-A*t*t)/(z*u*u+A*t*t))),E=C*c*u/d+(a+h)/2,F=C*-d*t/c+(b+i)/2,G=y.asin(((b-F)/d).toFixed(9)),H=y.asin(((i-F)/d).toFixed(9));G=aH&&(G=G-D*2);!g&&H>G&&(H=H-D*2)}var I=H-G;if(B(I)>k){var J=H,K=h,L=i;H=G+k*(g&&H>G?1:-1);h=E+c*y.cos(H);i=F+d*y.sin(H);m=bt(h,i,c,d,e,0,g,K,L,[H,J,E,F])}I=H-G;var M=y.cos(G),N=y.sin(G),O=y.cos(H),P=y.sin(H),Q=y.tan(I/4),R=4/3*c*Q,S=4/3*d*Q,T=[a,b],U=[a+R*N,b-S*M],V=[h+R*P,i-S*O],W=[h,i];U[0]=2*T[0]-U[0];U[1]=2*T[1]-U[1];{if(j)return[U,V,W][n](m);m=[U,V,W][n](m)[v]()[s](",");var X=[];for(var Y=0,Z=m[w];Y"1e12"&&(l=0.5);B(n)>"1e12"&&(n=0.5);if(l>0&&l<1){q=bu(a,b,c,d,e,f,g,h,l);p[L](q.x);o[L](q.y)}if(n>0&&n<1){q=bu(a,b,c,d,e,f,g,h,n);p[L](q.x);o[L](q.y)}i=f-2*d+b-(h-2*f+d);j=2*(d-b)-2*(f-d);k=b-d;l=(-j+y.sqrt(j*j-4*i*k))/2/i;n=(-j-y.sqrt(j*j-4*i*k))/2/i;B(l)>"1e12"&&(l=0.5);B(n)>"1e12"&&(n=0.5);if(l>0&&l<1){q=bu(a,b,c,d,e,f,g,h,l);p[L](q.x);o[L](q.y)}if(n>0&&n<1){q=bu(a,b,c,d,e,f,g,h,n);p[L](q.x);o[L](q.y)}return{min:{x:A[m](0,p),y:A[m](0,o)},max:{x:z[m](0,p),y:z[m](0,o)}}}),bw=bm(function(a,b){var c=bq(a),d=b&&bq(b),e={x:0,y:0,bx:0,by:0,X:0,Y:0,qx:null,qy:null},f={x:0,y:0,bx:0,by:0,X:0,Y:0,qx:null,qy:null},g=function(a,b){var c,d;if(!a)return["C",b.x,b.y,b.x,b.y,b.x,b.y];!(a[0]in{T:1,Q:1})&&(b.qx=b.qy=null);switch(a[0]){case"M":b.X=a[1];b.Y=a[2];break;case"A":a=["C"][n](bt[m](0,[b.x,b.y][n](a.slice(1))));break;case"S":c=b.x+(b.x-(b.bx||b.x));d=b.y+(b.y-(b.by||b.y));a=["C",c,d][n](a.slice(1));break;case"T":b.qx=b.x+(b.x-(b.qx||b.x));b.qy=b.y+(b.y-(b.qy||b.y));a=["C"][n](bs(b.x,b.y,b.qx,b.qy,a[1],a[2]));break;case"Q":b.qx=a[1];b.qy=a[2];a=["C"][n](bs(b.x,b.y,a[1],a[2],a[3],a[4]));break;case"L":a=["C"][n](br(b.x,b.y,a[1],a[2]));break;case"H":a=["C"][n](br(b.x,b.y,a[1],b.y));break;case"V":a=["C"][n](br(b.x,b.y,b.x,a[1]));break;case"Z":a=["C"][n](br(b.x,b.y,b.X,b.Y));break}return a},h=function(a,b){if(a[b][w]>7){a[b].shift();var e=a[b];while(e[w])a.splice(b++,0,["C"][n](e.splice(0,6)));a.splice(b,1);k=z(c[w],d&&d[w]||0)}},i=function(a,b,e,f,g){if(a&&b&&a[g][0]=="M"&&b[g][0]!="M"){b.splice(g,0,["M",f.x,f.y]);e.bx=0;e.by=0;e.x=a[g][1];e.y=a[g][2];k=z(c[w],d&&d[w]||0)}};for(var j=0,k=z(c[w],d&&d[w]||0);j0.5)*2-1;C(e-0.5,2)+C(f-0.5,2)>0.25&&(f=y.sqrt(0.25-C(e-0.5,2))*g+0.5)&&f!=0.5&&(f=f.toFixed(5)-0.00001*g)}return p});b=b[s](/\s*\-\s*/);if(d=="linear"){var i=b.shift();i=-S(i);if(isNaN(i))return null;var j=[0,0,y.cos(i*D/180),y.sin(i*D/180)],k=1/(z(B(j[2]),B(j[3]))||1);j[2]*=k;j[3]*=k;if(j[2]<0){j[0]=-j[2];j[2]=0}if(j[3]<0){j[1]=-j[3];j[3]=0}}var m=bx(b);if(!m)return null;var n=a.getAttribute(I);n=n.match(/^url\(#(.*)\)$/);n&&c.defs.removeChild(g.getElementById(n[1]));var o=bG(d+"Gradient");o.id=bh();bG(o,d=="radial"?{fx:e,fy:f}:{x1:j[0],y1:j[1],x2:j[2],y2:j[3]});c.defs[l](o);for(var q=0,t=m[w];q1?G.opacity/100:G.opacity});case"stroke":G=a.getRGB(o);h[R](n,G.hex);n=="stroke"&&G[f]("opacity")&&bG(h,{"stroke-opacity":G.opacity>1?G.opacity/100:G.opacity});break;case"gradient":(({circle:1,ellipse:1})[f](c.type)||r(o).charAt()!="r")&&bI(h,o,c.paper);break;case"opacity":i.gradient&&!i[f]("stroke-opacity")&&bG(h,{"stroke-opacity":o>1?o/100:o});case"fill-opacity":if(i.gradient){var H=g.getElementById(h.getAttribute(I)[Y](/^url\(#|\)$/g,p));if(H){var J=H.getElementsByTagName("stop");J[J[w]-1][R]("stop-opacity",o)}break}default:n=="font-size"&&(o=T(o,10)+"px");var K=n[Y](/(\-.)/g,function(a){return V.call(a.substring(1))});h.style[K]=o;h[R](n,o);break}}}bM(c,d);m?c.rotate(m.join(q)):S(j)&&c.rotate(j,true)},bL=1.2,bM=function(b,c){if(b.type!="text"||!(c[f]("text")||c[f]("font")||c[f]("font-size")||c[f]("x")||c[f]("y")))return;var d=b.attrs,e=b.node,h=e.firstChild?T(g.defaultView.getComputedStyle(e.firstChild,p).getPropertyValue("font-size"),10):10;if(c[f]("text")){d.text=c.text;while(e.firstChild)e.removeChild(e.firstChild);var i=r(c.text)[s]("\n");for(var j=0,k=i[w];jb.height&&(b.height=e.y+e.height-b.y);e.x+e.width-b.x>b.width&&(b.width=e.x+e.width-b.x)}}a&&this.hide();return b};bN[e].attr=function(b,c){if(this.removed)return this;if(b==null){var d={};for(var e in this.attrs)this.attrs[f](e)&&(d[e]=this.attrs[e]);this._.rt.deg&&(d.rotation=this.rotate());(this._.sx!=1||this._.sy!=1)&&(d.scale=this.scale());d.gradient&&d.fill=="none"&&(d.fill=d.gradient)&&delete d.gradient;return d}if(c==null&&a.is(b,F)){if(b=="translation")return cz.call(this);if(b=="rotation")return this.rotate();if(b=="scale")return this.scale();if(b==I&&this.attrs.fill=="none"&&this.attrs.gradient)return this.attrs.gradient;return this.attrs[b]}if(c==null&&a.is(b,G)){var g={};for(var h=0,i=b.length;h"));m.W=h.w=m.paper.span.offsetWidth;m.H=h.h=m.paper.span.offsetHeight;m.X=h.x;m.Y=h.y+Q(m.H/2);switch(h["text-anchor"]){case"start":m.node.style["v-text-align"]="left";m.bbx=Q(m.W/2);break;case"end":m.node.style["v-text-align"]="right";m.bbx=-Q(m.W/2);break;default:m.node.style["v-text-align"]="center";break}}};bI=function(a,b){a.attrs=a.attrs||{};var c=a.attrs,d,e="linear",f=".5 .5";a.attrs.gradient=b;b=r(b)[Y](bd,function(a,b,c){e="radial";if(b&&c){b=S(b);c=S(c);C(b-0.5,2)+C(c-0.5,2)>0.25&&(c=y.sqrt(0.25-C(b-0.5,2))*((c>0.5)*2-1)+0.5);f=b+q+c}return p});b=b[s](/\s*\-\s*/);if(e=="linear"){var g=b.shift();g=-S(g);if(isNaN(g))return null}var h=bx(b);if(!h)return null;a=a.shape||a.node;d=a.getElementsByTagName(I)[0]||cd(I);!d.parentNode&&a.appendChild(d);if(h[w]){d.on=true;d.method="none";d.color=h[0].color;d.color2=h[h[w]-1].color;var i=[];for(var j=0,k=h[w];j")}}catch(a){cd=function(a){return g.createElement("<"+a+" xmlns=\"urn:schemas-microsoft.com:vml\" class=\"rvml\">")}}bV=function(){var b=by[m](0,arguments),c=b.container,d=b.height,e,f=b.width,h=b.x,i=b.y;if(!c)throw new Error("VML container not found.");var k=new j,n=k.canvas=g.createElement("div"),o=n.style;h=h||0;i=i||0;f=f||512;d=d||342;f==+f&&(f+="px");d==+d&&(d+="px");k.width=1000;k.height=1000;k.coordsize=b_*1000+q+b_*1000;k.coordorigin="0 0";k.span=g.createElement("span");k.span.style.cssText="position:absolute;left:-9999em;top:-9999em;padding:0;margin:0;line-height:1;display:inline;";n[l](k.span);o.cssText=a.format("top:0;left:0;width:{0};height:{1};display:inline-block;position:relative;clip:rect(0 {0} {1} 0);overflow:hidden",f,d);if(c==1){g.body[l](n);o.left=h+"px";o.top=i+"px";o.position="absolute"}else c.firstChild?c.insertBefore(n,c.firstChild):c[l](n);bz.call(k,k,a.fn);return k};k.clear=function(){this.canvas.innerHTML=p;this.span=g.createElement("span");this.span.style.cssText="position:absolute;left:-9999em;top:-9999em;padding:0;margin:0;line-height:1;display:inline;";this.canvas[l](this.span);this.bottom=this.top=null};k.remove=function(){this.canvas.parentNode.removeChild(this.canvas);for(var a in this)this[a]=bF(a);return true}}var ce=navigator.userAgent.match(/Version\\x2f(.*?)\s/);navigator.vendor=="Apple Computer, Inc."&&(ce&&ce[1]<4||navigator.platform.slice(0,2)=="iP")?k.safari=function(){var a=this.rect(-99,-99,this.width+99,this.height+99).attr({stroke:"none"});h.setTimeout(function(){a.remove()})}:k.safari=function(){};var cf=function(){this.returnValue=false},cg=function(){return this.originalEvent.preventDefault()},ch=function(){this.cancelBubble=true},ci=function(){return this.originalEvent.stopPropagation()},cj=(function(){{if(g.addEventListener)return function(a,b,c,d){var e=o&&u[b]?u[b]:b,g=function(e){if(o&&u[f](b))for(var g=0,h=e.targetTouches&&e.targetTouches.length;g1&&(a=Array[e].splice.call(arguments,0,arguments[w]));return new cC(a)};k.setSize=bU;k.top=k.bottom=null;k.raphael=a;function co(){return this.x+q+this.y}bO.resetScale=function(){if(this.removed)return this;this._.sx=1;this._.sy=1;this.attrs.scale="1 1"};bO.scale=function(a,b,c,d){if(this.removed)return this;if(a==null&&b==null)return{x:this._.sx,y:this._.sy,toString:co};b=b||a;!(+b)&&(b=a);var e,f,g,h,i=this.attrs;if(a!=0){var j=this.getBBox(),k=j.x+j.width/2,l=j.y+j.height/2,m=B(a/this._.sx),o=B(b/this._.sy);c=+c||c==0?c:k;d=+d||d==0?d:l;var r=this._.sx>0,s=this._.sy>0,t=~(~(a/B(a))),u=~(~(b/B(b))),x=m*t,y=o*u,z=this.node.style,A=c+B(k-c)*x*(k>c==r?1:-1),C=d+B(l-d)*y*(l>d==s?1:-1),D=a*t>b*u?o:m;switch(this.type){case"rect":case"image":var E=i.width*m,F=i.height*o;this.attr({height:F,r:i.r*D,width:E,x:A-E/2,y:C-F/2});break;case"circle":case"ellipse":this.attr({rx:i.rx*m,ry:i.ry*o,r:i.r*D,cx:A,cy:C});break;case"text":this.attr({x:A,y:C});break;case"path":var G=bp(i.path),H=true,I=r?x:m,J=s?y:o;for(var K=0,L=G[w];Kr)p=n.data[r*l];else{p=a.findDotsAtSegment(b,c,d,e,f,g,h,i,r/l);n.data[r]=p}r&&(k+=C(C(o.x-p.x,2)+C(o.y-p.y,2),0.5));if(j!=null&&k>=j)return p;o=p}if(j==null)return k},cr=function(b,c){return function(d,e,f){d=bw(d);var g,h,i,j,k="",l={},m,n=0;for(var o=0,p=d.length;oe){if(c&&!l.start){m=cq(g,h,i[1],i[2],i[3],i[4],i[5],i[6],e-n);k+=["C",m.start.x,m.start.y,m.m.x,m.m.y,m.x,m.y];if(f)return k;l.start=k;k=["M",m.x,m.y+"C",m.n.x,m.n.y,m.end.x,m.end.y,i[5],i[6]][v]();n+=j;g=+i[5];h=+i[6];continue}if(!b&&!c){m=cq(g,h,i[1],i[2],i[3],i[4],i[5],i[6],e-n);return{x:m.x,y:m.y,alpha:m.alpha}}}n+=j;g=+i[5];h=+i[6]}k+=i}l.end=k;m=b?n:c?l:a.findDotsAtSegment(g,h,i[1],i[2],i[3],i[4],i[5],i[6],1);m.alpha&&(m={x:m.x,y:m.y,alpha:m.alpha});return m}},cs=cr(1),ct=cr(),cu=cr(0,1);bO.getTotalLength=function(){if(this.type!="path")return;if(this.node.getTotalLength)return this.node.getTotalLength();return cs(this.attrs.path)};bO.getPointAtLength=function(a){if(this.type!="path")return;return ct(this.attrs.path,a)};bO.getSubpath=function(a,b){if(this.type!="path")return;if(B(this.getTotalLength()-b)<"1e-6")return cu(this.attrs.path,a).end;var c=cu(this.attrs.path,b,1);return a?cu(c,a).end:c};a.easing_formulas={linear:function(a){return a},"<":function(a){return C(a,3)},">":function(a){return C(a-1,3)+1},"<>":function(a){a=a*2;if(a<1)return C(a,3)/2;a-=2;return(C(a,3)+2)/2},backIn:function(a){var b=1.70158;return a*a*((b+1)*a-b)},backOut:function(a){a=a-1;var b=1.70158;return a*a*((b+1)*a+b)+1},elastic:function(a){if(a==0||a==1)return a;var b=0.3,c=b/4;return C(2,-10*a)*y.sin((a-c)*(2*D)/b)+1},bounce:function(a){var b=7.5625,c=2.75,d;if(a<1/c)d=b*a*a;else if(a<2/c){a-=1.5/c;d=b*a*a+0.75}else if(a<2.5/c){a-=2.25/c;d=b*a*a+0.9375}else{a-=2.625/c;d=b*a*a+0.984375}return d}};var cv=[],cw=function(){var b=+(new Date);for(var c=0;cd)return d;while(cf?c=e:d=e;e=(d-c)/2+c}return e}return n(a,1/(200*f))}bO.onAnimation=function(a){this._run=a||0;return this};bO.animate=function(c,d,e,g){var h=this;h.timeouts=h.timeouts||[];if(a.is(e,"function")||!e)g=e||null;if(h.removed){g&&g.call(h);return h}var i={},j={},k=false,l={};for(var m in c)if(c[f](m)){if(X[f](m)||h.paper.customAttributes[f](m)){k=true;i[m]=h.attr(m);i[m]==null&&(i[m]=W[m]);j[m]=c[m];switch(X[m]){case"along":var n=cs(c[m]),o=ct(c[m],n*!(!c.back)),p=h.getBBox();l[m]=n/d;l.tx=p.x;l.ty=p.y;l.sx=o.x;l.sy=o.y;j.rot=c.rot;j.back=c.back;j.len=n;c.rot&&(l.r=S(h.rotate())||0);break;case E:l[m]=(j[m]-i[m])/d;break;case"colour":i[m]=a.getRGB(i[m]);var q=a.getRGB(j[m]);l[m]={r:(q.r-i[m].r)/d,g:(q.g-i[m].g)/d,b:(q.b-i[m].b)/d};break;case"path":var t=bw(i[m],j[m]);i[m]=t[0];var u=t[1];l[m]=[];for(var v=0,x=i[m][w];v