Пример #1
0
        public static IEnumerable <jednani> ParseAgenda(string sdatum)
        {
            string html = "";

            using (var net = new Devmasters.Net.HttpClient.URLContent(string.Format(agendaUrl, sdatum)))
            {
                html = net.GetContent().Text;
            }

            DateTime       datum    = DateTime.ParseExact(sdatum, "yyyy-MM-dd", System.Globalization.CultureInfo.GetCultureInfo("en-US"), System.Globalization.DateTimeStyles.AssumeLocal);
            List <string>  usneseni = new List <string>();
            List <jednani> js       = new List <jednani>();
            object         lockObj  = new object();

            var xp = new XPath(html);

            var rows       = xp.GetNodes("//table[@class='lfr-table djv-agenda-table']//tr");
            int ruzneCount = 0;

            foreach (var r in rows)
            {
                if (XPath.Tools.GetNodeAttributeValue(r, "td[1]", "colspan") != null)
                {
                    continue;
                }
                var bod = XPath.Tools.GetNodeText(r, "td[2]") ?? "";

                var obsah = XPath.Tools.GetNode(r, "td[5]");
                if (bod == null && obsah == null)
                {
                    obsah = XPath.Tools.GetNode(r, "td[1]");
                }
                if (bod.Contains("Usnesení č."))
                {
                    usneseni.Add(bod);
                }
                else if (bod.Contains("Příloha č."))
                {
                }
                else
                {
                    jednani j = new jednani();
                    if (string.IsNullOrEmpty(bod))
                    {
                        ruzneCount++;
                        bod = "Různé č." + ruzneCount;
                    }
                    j.bod   = bod.Trim();
                    j.datum = datum;
                    j.vec   = obsah?.InnerText;
                    var docs = XPath.Tools.GetNodes(r, "td[3]/a");
                    if (docs != null)
                    {
                        j.dokumenty = docs.Select(m => new jednani.dokument()
                        {
                            DocumentUrl = NormalizeUrl(m.GetAttributeValue("href", "")),
                            stahnuto    = DateTime.Now,
                            jmeno       = "Znění",
                        }).ToArray();
                    }

                    if (obsah != null && XPath.Tools.GetNodes(obsah, ".//a") != null)
                    {
                        j.veklep = XPath.Tools.GetNodes(obsah, ".//a")
                                   .Select(m => m.GetAttributeValue("href", null))
                                   .Where(n => n?.StartsWith("/veklep") == true)
                                   .Select(m => m.Replace("/veklep-detail?pid=", ""))
                                   .ToArray();
                        j.dokumenty = XPath.Tools.GetNodes(obsah, ".//a")
                                      .Where(m => m.GetAttributeValue("href", null) != null && m.GetAttributeValue("href", null)?.StartsWith("/veklep") == false)
                                      .Select(m => new jednani.dokument()
                        {
                            DocumentUrl = NormalizeUrl(m.GetAttributeValue("href", null)),
                            jmeno       = m.InnerText
                        }
                                              )
                                      .ToArray();
                    }

                    j.SetId();
                    js.Add(j);
                }
            }

            //parse usneseni

            Devmasters.Batch.Manager.DoActionForAll(usneseni,
                                                    (u) =>
            {
                lock (lockObj)
                {
                    js.Add(ParseUsneseni(datum, System.Text.RegularExpressions.Regex.Replace(u, "\\D", "")));
                }

                return(new Devmasters.Batch.ActionOutputData());
            }
                                                    , null
                                                    , null //new Devmasters.Core.Batch.ActionProgressWriter(0.1f).Write
                                                    , true
                                                    , maxDegreeOfParallelism: 5, prefix: "DAT " + sdatum + ":"
                                                    );

            var dids = js.Select(m => m.Id).Distinct().ToArray();

            if (dids.Count() != js.Count())
            {
                for (int i = 1; i < js.Count(); i++)
                {
                    for (int j = 0; j < i; j++)
                    {
                        if (js[i].Id == js[j].Id)
                        {
                            js[i].Id = js[i].Id + "-" + i;
                        }
                    }
                }
            }


            return(js);
        }
Пример #2
0
        public static jednani ParseUsneseni(DateTime den, string cislo)
        {
            int    rok  = den.Year;
            string html = "";

            using (var net = new Devmasters.Net.HttpClient.URLContent(string.Format(usneseniUrl, rok, cislo)))
            {
                html = net.GetContent().Text;
            }
            var     xp = new XPath(html);
            jednani j  = new jednani();

            j.datum        = den;
            j.usneseni     = $"{cislo}/{den.Year}";
            j.bod          = "Usnesení č. " + cislo;
            j.vec          = xp.GetNodeText("(//table[@class='lfr-table material-table'])//tr//th[contains(text(),'Název')]/following-sibling::td");
            j.cisloJednaci = xp.GetNodeText("(//table[@class='lfr-table material-table'])//tr//th[contains(text(),'Čj')]/following-sibling::td");
            j.veklep       = new string[] {
                xp.GetNodeText("(//table[@class='lfr-table material-table'])//tr//th[contains(text(),'PID')]/following-sibling::td")
            };

            List <jednani.dokument> jdocs = new List <jednani.dokument>();
            var xprilohyRows = xp.GetNodes("(//table[@class='lfr-table material-table'])[1]/tr");

            if (xprilohyRows != null)
            {
                foreach (var xpri in xprilohyRows)
                {
                    var jdoc = XPath.Tools.GetNodes(xpri, "td[2]/a");
                    if (jdoc != null)
                    {
                        jdocs.AddRange(jdoc.Select(m => new jednani.dokument()
                        {
                            DocumentUrl = NormalizeUrl(m.GetAttributeValue("href", "")),
                            stahnuto    = DateTime.Now,
                            jmeno       = XPath.Tools.GetNodeText(xpri, "td[1]")
                        })
                                       );
                        j.dokumenty = jdocs.ToArray();
                    }
                }
            }
            if (xp.GetNode("//div[@class='resolutionDerogation']") != null)
            {
                var souvisTypes = new string[] { "repeals", "amends", "repealed", "amended" };

                foreach (var styp in souvisTypes)
                {
                    List <jednani.souvis> jsouvis = new List <jednani.souvis>();

                    if (xp.GetNode($"//div[@class='resolutionDerogation']/div[@class='{styp}']/p") != null)
                    {
                        var txt = xp
                                  .GetNodeText($"//div[@class='resolutionDerogation']/div[@class='{styp}']/p")?
                                  .Trim()?
                                  .Replace(" následující", "");

                        foreach (var item in xp.GetNodes($"//div[@class='resolutionDerogation']/div[@class='{styp}']/dl/dd"))
                        {
                            var jss = new jednani.souvis()
                            {
                                zmena           = txt,
                                usneseni        = item.InnerText?.Trim(),
                                usneseniOrigUrl = NormalizeUrl(item.SelectSingleNode(".//a").GetAttributeValue("href", null)),
                            };
                            var srok   = GetRegexGroupValue(jss.usneseni, regexRokFromTxt, "rok")?.Replace("20", "");
                            var scislo = GetRegexGroupValue(jss.usneseni, regexRokFromTxt, "cislo");
                            if (!string.IsNullOrEmpty(srok) && !string.IsNullOrEmpty(scislo))
                            {
                                jss.usneseniCislo = $"{scislo}/{srok}";
                            }

                            jsouvis.Add(jss);
                        }
                    }
                    if (jsouvis != null && jsouvis.Count > 0)
                    {
                        switch (styp)
                        {
                        case "repeals":
                            j.rusi = jsouvis.ToArray();
                            break;

                        case "amends":
                            j.meni = jsouvis.ToArray();
                            break;

                        case "repealed":
                            j.zruseno = jsouvis.ToArray();
                            break;

                        case "amended":
                            j.zmeneno = jsouvis.ToArray();
                            break;

                        default:
                            break;
                        }
                    }
                }
            }
            j.SetId();
            return(j);
        }