| import string | |
| import math | |
| CHARS = string.digits + string.lowercase | |
| assert len(CHARS) == 36 | |
| CHARS += string.uppercase | |
| assert len(CHARS) == 62 | |
| BASE64_CHARS = string.uppercase + string.lowercase + string.digits + "+/" |
| # -*- coding: UTF-8 -*- | |
| import urllib2 | |
| from lxml import etree | |
| from lxml.cssselect import CSSSelector as css | |
| from rdflib import Graph | |
| def check_examples(url, verbose=False): | |
| for title, prologue, text, fmt in find_examples(url): | |
| header = 'Example "%s" as %s: ' % (title, fmt) |
| import re | |
| regexps = [ | |
| r'\d\d', | |
| r'^\d\d$', | |
| r'^\d{2}$', | |
| r'^\d{2,}$', | |
| r'^\d{2,3}$', | |
| r'^\d{2,4}$', | |
| ] |
| @Grab(group='org.ccil.cowan.tagsoup', module='tagsoup', version='0.9.7') | |
| import org.ccil.cowan.tagsoup.Parser | |
| /** | |
| * Parses html files and prints out a tree representing the structure | |
| * (css-selector style). | |
| * | |
| * Shows the frequency for how many pages a specific subtree appears in. Can | |
| * "anonymize" generated ID:s. |
| @property | |
| def identifierare(self): | |
| return "%s %s:%s" % (self.forfattningssamling.kortnamn, | |
| self.arsutgava, self.lopnummer) |
| fs = require 'fs' | |
| {print} = require 'util' | |
| jsdom = require 'jsdom' | |
| jquery = require 'jquery' | |
| convert = (base=null) -> | |
| $ = @$ |
| Canvas = require 'canvas' | |
| fs = require 'fs' | |
| path = require 'path' | |
| render = (width, height, draw, filePath) -> | |
| canvas = new Canvas(width, height) | |
| draw.call canvas.getContext '2d' | |
| out = fs.createWriteStream filePath |
| fpath = process.argv[2] | |
| file = require('fs').readFileSync fpath, 'utf-8' | |
| data = JSON.parse file | |
| window = require('jsdom').jsdom(null, null).createWindow() | |
| $ = require('jquery').create window | |
| for key, items of data | |
| dl = $("<dl id='#{key}'></dl>").appendTo $('body') | |
| for num, value of items | |
| dl.append "<dt>#{num}</dt><dd>#{value}</dd>" |
For the basic case, given data like:
<html lang="en">
<head><base href="http://example.org/"/></head>
<body vocab="http://example.org/vocab#" typeof="Site" lang="sv">
<h1 property="title">The Website</h1>
By:
<ul rel="author">
<li about="/persons/doe" typeof="Person">
| { | |
| "iri": "@subject", | |
| "type": "@type", | |
| "@vocab": "http://rinfo.lagrummet.se/ns/2008/11/rinfo/publ#", | |
| "@language": "sv" | |
| "label": "http://www.w3.org/2000/01/rdf-schema#label", | |
| "comment": "http://www.w3.org/2000/01/rdf-schema#comment", | |
| "title": "http://purl.org/dc/terms/title", | |
| "description": "http://purl.org/dc/terms/description", |