List of usage examples for org.jsoup.nodes Document getElementById
public Element getElementById(String id)
From source file:com.storm.function.GsxtFunction.java
private Map<String, Object> getHtmlInfoMapOfJilin(String area, String keyword, ChannelLogger LOGGER) throws Exception { Map<String, Object> resultHtmlMap = new LinkedHashMap<String, Object>(); String[] command = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/jilin.js", "--web-security=no", "--keyword=" + keyword }; String casperjsResult = CommandUtil.runCommand(command); Elements divDataItems = Jsoup.parse(casperjsResult).getElementsByClass("list"); Elements divNoDataItems = Jsoup.parse(casperjsResult).getElementsByClass("list-a"); if (divDataItems.isEmpty() && !divNoDataItems.isEmpty()) { // ? resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); } else if (divDataItems.isEmpty() && divDataItems.isEmpty()) { // ?? // ????/*from w w w .j a va 2 s .c om*/ if (casperjsResult.contains("")) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.IMAGECODE_ERROR); } else { resultHtmlMap.put("statusCodeDef", StatusCodeDef.FAILURE); } } else if (!divDataItems.isEmpty() && divNoDataItems.isEmpty()) { // ? // ??????? Element nowCookies = Jsoup.parse(casperjsResult).getElementById("nextParams"); Elements tokenEts = Jsoup.parse(casperjsResult).getElementsByAttributeValue("name", "_csrf"); if (null == nowCookies || null == tokenEts || tokenEts.isEmpty()) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.COOKIE_ERROR); return resultHtmlMap; } String nowCookiesJson = nowCookies.text().trim(); String nowCookiesStr = ((String) new GsonBuilder().create().fromJson(nowCookiesJson, Map.class) .get("Cookie")).trim(); String tokenStr = tokenEts.get(0).attr("content"); String HOST_OF_JILIN = "http://211.141.74.198:8081/aiccips/pub/"; String HOST_OF_XQ = "http://211.141.74.198:8081/"; String htmlAnchorHref = ""; for (Element divDataItem : divDataItems) { Element htmlAnchor = divDataItem.getElementsByTag("a").get(0); String htmlAnchorText = htmlAnchor.text(); if (htmlAnchorText.contains(keyword)) { htmlAnchorHref = HOST_OF_JILIN + htmlAnchor.attr("href"); break; } } if (StringUtils.isEmpty(htmlAnchorHref)) { htmlAnchorHref = "http://211.141.74.198:8081/aiccips/pub/" + divDataItems.get(0).getElementsByTag("a").get(0).attr("href"); } String commonUrl = htmlAnchorHref.split("gsgsdetail")[1]; String commonUrlZ = htmlAnchorHref.substring(htmlAnchorHref.lastIndexOf("/") + 1, htmlAnchorHref.length()); // ?->? String[] command11 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/getSimpleRequestPage.js", "--web-security=no", "--url=" + htmlAnchorHref }; String casperjsResult11 = CommandUtil.runCommand(command11); resultHtmlMap.put("gsgsxx", casperjsResult11); Thread.sleep(1000); // ?->?->?? String baxxZyryxxUrl = HOST_OF_JILIN + "gsryxx/1151?encrpripid=" + commonUrlZ; String[] command121 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + baxxZyryxxUrl }; String casperjsResult121 = CommandUtil.runCommand(command121); resultHtmlMap.put("gsgsxx_baxx_zyryxx", casperjsResult121); // ?->?->? String baxxFzjgxxUrl = HOST_OF_JILIN + "gsfzjg/1151?encrpripid=" + commonUrlZ; String[] command123 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + baxxFzjgxxUrl }; String casperjsResult123 = CommandUtil.runCommand(command123); resultHtmlMap.put("gsgsxx_baxx_fzjgxx", casperjsResult123); // ?->?->? String dcdydjxxDcdydjxxUrl = HOST_OF_JILIN + "gsdcdy?encrpripid=" + commonUrlZ; String[] command131 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + dcdydjxxDcdydjxxUrl }; String casperjsResult131 = CommandUtil.runCommand(command131); resultHtmlMap.put("gsgsxx_dcdydjxx_dcdydjxx", casperjsResult131); // ?->??->?? String gqczdjxxGqczdjxxUrl = HOST_OF_JILIN + "gsgqcz?encrpripid=" + commonUrlZ; String[] command141 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + gqczdjxxGqczdjxxUrl }; String casperjsResult141 = CommandUtil.runCommand(command141); resultHtmlMap.put("gsgsxx_gqczdjxx_gqczdjxx", casperjsResult141); // ?->?->? String xzcfxxXzcfxxUrl = HOST_OF_JILIN + "gsxzcfxx?encrpripid=" + commonUrlZ; String[] command151 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + xzcfxxXzcfxxUrl }; String casperjsResult151 = CommandUtil.runCommand(command151); resultHtmlMap.put("gsgsxx_xzcfxx_xzcfxx", casperjsResult151); // ?->???->??? String jyycxxJyycxxUrl = HOST_OF_JILIN + "jyyc/1151?encrpripid=" + commonUrlZ; String[] command161 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + jyycxxJyycxxUrl }; String casperjsResult161 = CommandUtil.runCommand(command161); resultHtmlMap.put("gsgsxx_jyycxx_jyycxx", casperjsResult161); // ?->???->??? String yzwfxxYzwfxxUrl = HOST_OF_JILIN + "yzwfqy?encrpripid=" + commonUrlZ; String[] command171 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + yzwfxxYzwfxxUrl }; String casperjsResult171 = CommandUtil.runCommand(command171); resultHtmlMap.put("gsgsxx_yzwfxx_yzwfxx", casperjsResult171); // ?->?->? String ccjcxxCcjcxxUrl = HOST_OF_JILIN + "ccjcxx?encrpripid=" + commonUrlZ; String[] command181 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + htmlAnchorHref, "--tokenStr=" + tokenStr, "--url=" + ccjcxxCcjcxxUrl }; String casperjsResult181 = CommandUtil.runCommand(command181); resultHtmlMap.put("gsgsxx_ccjcxx_ccjcxx", casperjsResult181); // ?? String qygsUrl = HOST_OF_JILIN + "qygsdetail" + commonUrl; String[] command2 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/getSimpleRequestPage.js", "--web-security=no", "--url=" + qygsUrl }; String casperjsResult2 = CommandUtil.runCommand(command2); resultHtmlMap.put("qygsxx_list", casperjsResult2); // ? ??->?-> Document qygsxxHtml = Jsoup.parseBodyFragment(casperjsResult2); Element qynbDiv = qygsxxHtml.getElementById("qiyenianbao"); if (null != qynbDiv) { Elements qynb_trs = qynbDiv.select("tbody").get(0).select("tr"); if (null != qynb_trs && qynb_trs.size() > 2) { List<Map<String, Object>> qygsxx_qynb_infos = new ArrayList<Map<String, Object>>(); for (int i = 2; i < qynb_trs.size(); i++) { Map<String, Object> qygsxx_qynb_info_map = new LinkedHashMap<String, Object>(); Element wdd = qynb_trs.get(i).select("td").get(1).select("a").get(0); String qygsxx_qynb_list_a_text = wdd.text(); String qygsxx_qynb_list_pubdate = qynb_trs.get(i).select("td").get(2).text(); qygsxx_qynb_info_map.put("qygsxx_qynb_list_a_text", qygsxx_qynb_list_a_text); qygsxx_qynb_info_map.put("qygsxx_qynb_list_pubdate", qygsxx_qynb_list_pubdate); String qynbxqUrl = HOST_OF_XQ + wdd.attr("href"); String[] command21 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/getSimpleRequestPage.js", "--web-security=no", "--url=" + qynbxqUrl }; String casperjsResult21 = CommandUtil.runCommand(command21); qygsxx_qynb_info_map.put("qygsxx_qynb_info_page", casperjsResult21); qygsxx_qynb_infos.add(qygsxx_qynb_info_map); } resultHtmlMap.put("qygsxx_qynb_infos", qygsxx_qynb_infos); } } Thread.sleep(1000); // ??->??->?? String gdjczxxGdjczxxUrl = HOST_OF_JILIN + "qygsjsxxxzczxx?encrpripid=" + commonUrlZ; String[] command221 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + qygsUrl, "--tokenStr=" + tokenStr, "--url=" + gdjczxxGdjczxxUrl }; String casperjsResult221 = CommandUtil.runCommand(command221); resultHtmlMap.put("qygsxx_gdjczxx_gdjczxx", casperjsResult221); // ??->??->?? String gdjczxxBgxxUrl = HOST_OF_JILIN + "qygsjsxxczxxbgsx?encrpripid=" + commonUrlZ; String[] command222 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + qygsUrl, "--tokenStr=" + tokenStr, "--url=" + gdjczxxBgxxUrl }; String casperjsResult222 = CommandUtil.runCommand(command222); resultHtmlMap.put("qygsxx_gdjczxx_bgxx", casperjsResult222); // ??->???->??? String gqbgxxGqbgxxUrl = HOST_OF_JILIN + "qygsJsxxgqbg?encrpripid=" + commonUrlZ; String[] command231 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + qygsUrl, "--tokenStr=" + tokenStr, "--url=" + gqbgxxGqbgxxUrl }; String casperjsResult231 = CommandUtil.runCommand(command231); resultHtmlMap.put("qygsxx_gqbgxx_gqbgxx", casperjsResult231); // ??->??->?? String xzxkxxXzxkxxUrl = HOST_OF_JILIN + "qygsjsxxxzxk?encrpripid=" + commonUrlZ; String[] command241 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + qygsUrl, "--tokenStr=" + tokenStr, "--url=" + xzxkxxXzxkxxUrl }; String casperjsResult241 = CommandUtil.runCommand(command241); resultHtmlMap.put("qygsxx_xzxkxx_xzxkxx", casperjsResult241); // ??->??->?? String zscqczZscqczUrl = HOST_OF_JILIN + "/qygsjsxxzscqcz?encrpripid=" + commonUrlZ; String[] command251 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + qygsUrl, "--tokenStr=" + tokenStr, "--url=" + zscqczZscqczUrl }; String casperjsResult251 = CommandUtil.runCommand(command251); resultHtmlMap.put("qygsxx_zscqcz_zscqcz", casperjsResult251); // ??->?->? String qygsxxXzcfxxUrl = HOST_OF_JILIN + "qygsjsxxxzcfxx?encrpripid=" + commonUrlZ; String[] command261 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/postJilinSimpleRequestPage.js", "--web-security=no", "--cookieStr=" + nowCookiesStr, "--refererStr=" + qygsUrl, "--tokenStr=" + tokenStr, "--url=" + qygsxxXzcfxxUrl }; String casperjsResult261 = CommandUtil.runCommand(command261); resultHtmlMap.put("qygsxx_zscqcz_zscqcz", casperjsResult261); // ? String qtbmUrl = HOST_OF_JILIN + "qtgsdetail" + commonUrl; String[] command3 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/getSimpleRequestPage.js", "--web-security=no", "--url=" + qtbmUrl }; String casperjsResult3 = CommandUtil.runCommand(command3); resultHtmlMap.put("qtbmgsxx", casperjsResult3); // ???? String sfxzUrl = HOST_OF_JILIN + "sfgsdetail" + commonUrl; String[] command4 = { "casperjs", "/home/ubuntu/nfs-images/casperjscode/getSimpleRequestPage.js", "--web-security=no", "--url=" + sfxzUrl }; String casperjsResult4 = CommandUtil.runCommand(command4); resultHtmlMap.put("sfxzgsxx_list", casperjsResult4); resultHtmlMap.put("statusCodeDef", StatusCodeDef.SCCCESS); } return resultHtmlMap; }
From source file:com.storm.function.GsxtFunction.java
private Map<String, Object> getHtmlInfoMapOfJiangsu(String area, HtmlPage firstInfoPage, String keyword, ChannelLogger LOGGER) throws Exception { LOGGER.info("=========" + area + "=========" + keyword + "========="); Map<String, Object> resultHtmlMap = new LinkedHashMap<String, Object>(); String alertMsg = WebCrawler.getAlertMsg(); if (null == firstInfoPage) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.FAILURE); } else if (null != alertMsg && alertMsg.contains("IP??3")) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.FREQUENCY_LIMITED); } else if (null != alertMsg && alertMsg.contains("??")) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.IMAGECODE_ERROR); } else if (null != alertMsg && alertMsg.contains("?")) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.ILLEGAL_CHAR); resultHtmlMap.put("searchPageHtml", firstInfoPage.asXml()); } else {/*from ww w . j ava 2s . c o m*/ @SuppressWarnings("unchecked") List<HtmlAnchor> anchors = (List<HtmlAnchor>) firstInfoPage .getByXPath("//div[@id='punishResult']/dl/dt/a"); if (null == anchors || anchors.isEmpty()) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); } else { HtmlAnchor htmlAnchor = null; boolean flag = false; for (HtmlAnchor anchor : anchors) { String anchorTitle = anchor.getTextContent().toString().trim(); if (anchorTitle.contains(keyword)) { // ???? htmlAnchor = anchor; flag = true; break; } } if (!flag) { LOGGER.info("????"); resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); } if (flag) { WebWindow webWindow = firstInfoPage.getWebClient().getCurrentWindow(); // ??? HtmlElement target_item_info = (HtmlElement) htmlAnchor.getParentNode().getParentNode(); resultHtmlMap.put("target_item_info", target_item_info.asXml()); // ?? ??->?->?->?? Page gdwd = htmlAnchor.click(); Thread.sleep(3000); HtmlPage gsgsxx_djxx = (HtmlPage) gdwd; Document commonText = Jsoup.parseBodyFragment(gsgsxx_djxx.asXml()); String corp_id = commonText.getElementById("id").attr("value"); String corp_org = commonText.getElementById("org").attr("value"); resultHtmlMap.put("gsgsxx_djxx", gsgsxx_djxx.asXml()); // ?? ??->?-> @SuppressWarnings("unchecked") List<HtmlAnchor> touziren_anchors = (List<HtmlAnchor>) gsgsxx_djxx .getByXPath("//table[@id='touziren']/tbody/tr/td/a"); List<Map<String, Object>> gsgsxx_djxx_tzrxx_xqs = new ArrayList<Map<String, Object>>(); if (null != touziren_anchors && !touziren_anchors.isEmpty()) { for (HtmlAnchor touziren_anchor : touziren_anchors) { Map<String, Object> gsgsxx_djxx_tzrxx_xq = new LinkedHashMap<String, Object>(); HtmlPage gsgsxx_djxx_tzrxx_xq_page = touziren_anchor.click(); gsgsxx_djxx_tzrxx_xq.put("gsgsxx_djxx_tzrxx_xq", gsgsxx_djxx_tzrxx_xq_page.asXml()); gsgsxx_djxx_tzrxx_xqs.add(gsgsxx_djxx_tzrxx_xq); } } resultHtmlMap.put("gsgsxx_djxx_tzrxx_xqs", gsgsxx_djxx_tzrxx_xqs); // ? ??->??->?->? HtmlElement js_tab2 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='2']"); String gsgsxx_baxx = null; if (null != js_tab2 && !StringUtils.isEmpty(js_tab2)) { HtmlPage gsgsxx_baxx_page = js_tab2.click(); gsgsxx_baxx = gsgsxx_baxx_page.asXml(); } resultHtmlMap.put("gsgsxx_baxx", gsgsxx_baxx); // ? ??->? HtmlElement js_tab4 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='4']"); String gsgsxx_dcdydjxx_dcdydjxx = null; if (null != js_tab4 && !StringUtils.isEmpty(js_tab4)) { HtmlPage gsgsxx_dcdydjxx_dcdydjxx_page = js_tab4.click(); gsgsxx_dcdydjxx_dcdydjxx = gsgsxx_dcdydjxx_dcdydjxx_page.asXml(); @SuppressWarnings("unchecked") List<HtmlAnchor> wcqs = (List<HtmlAnchor>) gsgsxx_dcdydjxx_dcdydjxx_page .getByXPath("//div[@id='dongchandiya']/table[1]/tbody/tr/td/a"); if (null != wcqs && wcqs.size() > 0) { List<Map<String, Object>> gsgsxx_dcdydjxx_xqs = new ArrayList<Map<String, Object>>(); for (HtmlAnchor wcq : wcqs) { Map<String, Object> wd = new HashMap<String, Object>(); HtmlPage gsgsxx_dcdydjxx_xq = wcq.click(); wd.put("gsgsxx_dcdydjxx_xq", gsgsxx_dcdydjxx_xq.asXml()); gsgsxx_dcdydjxx_xqs.add(wd); } resultHtmlMap.put("gsgsxx_dcdydjxx_xqs", gsgsxx_dcdydjxx_xqs); } } resultHtmlMap.put("gsgsxx_dcdydjxx_dcdydjxx", gsgsxx_dcdydjxx_dcdydjxx); // ? ???->?? HtmlElement js_tab3 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='3']"); String gsgsxx_gqczdjxx_gqczdjxx = null; if (null != js_tab3 && !StringUtils.isEmpty(js_tab3)) { HtmlPage gsgsxx_gqczdjxx_gqczdjxx_page = js_tab3.click(); gsgsxx_gqczdjxx_gqczdjxx = gsgsxx_gqczdjxx_gqczdjxx_page.asXml(); } resultHtmlMap.put("gsgsxx_gqczdjxx_gqczdjxx", gsgsxx_gqczdjxx_gqczdjxx); // ? ??->? HtmlElement js_tab7 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='7']"); String gsgsxx_xzcfxx_xzcfxx = null; if (null != js_tab7 && !StringUtils.isEmpty(js_tab7)) { HtmlPage gsgsxx_xzcfxx_xzcfxx_page = js_tab7.click(); gsgsxx_xzcfxx_xzcfxx = gsgsxx_xzcfxx_xzcfxx_page.asXml(); // ? ??->? @SuppressWarnings("unchecked") List<HtmlAnchor> xingzhengchufa_anchors = (List<HtmlAnchor>) gsgsxx_xzcfxx_xzcfxx_page .getByXPath("//div[@id='xingzhengchufa']/table/tbody/tr/td/a"); List<Map<String, Object>> gsgsxx_xzcfxx_xzcfxx_xqs = new ArrayList<Map<String, Object>>(); if (null != xingzhengchufa_anchors && xingzhengchufa_anchors.size() > 0) { for (HtmlAnchor xingzhengchufa_anchor : xingzhengchufa_anchors) { Map<String, Object> gsgsxx_xzcfxx_xzcfxx_xq = new LinkedHashMap<String, Object>(); HtmlPage gsgsxx_xzcfxx_xzcfxx_xq_page = xingzhengchufa_anchor.click(); gsgsxx_xzcfxx_xzcfxx_xq.put("gsgsxx_xzcfxx_xzcfxx_xq", gsgsxx_xzcfxx_xzcfxx_xq_page.asXml()); gsgsxx_xzcfxx_xzcfxx_xqs.add(gsgsxx_xzcfxx_xzcfxx_xq); } } resultHtmlMap.put("gsgsxx_xzcfxx_xzcfxx_xqs", gsgsxx_xzcfxx_xzcfxx_xqs); } resultHtmlMap.put("gsgsxx_xzcfxx_xzcfxx", gsgsxx_xzcfxx_xzcfxx); // ? ?->???->??? HtmlElement js_tab5 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='5']"); String gsgsxx_jyycxx_jyycxx = null; if (null != js_tab5 && !StringUtils.isEmpty(js_tab5)) { HtmlPage gsgsxx_jyycxx_jyycxx_page = js_tab5.click(); gsgsxx_jyycxx_jyycxx = gsgsxx_jyycxx_jyycxx_page.asXml(); } resultHtmlMap.put("gsgsxx_jyycxx_jyycxx", gsgsxx_jyycxx_jyycxx); // ? ?->???->??? HtmlElement js_tab6 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='6']"); String gsgsxx_yzwfxx_yzwfxx = null; if (null != js_tab6 && !StringUtils.isEmpty(js_tab6)) { HtmlPage gsgsxx_yzwfxx_yzwfxx_page = js_tab6.click(); gsgsxx_yzwfxx_yzwfxx = gsgsxx_yzwfxx_yzwfxx_page.asXml(); } resultHtmlMap.put("gsgsxx_yzwfxx_yzwfxx", gsgsxx_yzwfxx_yzwfxx); // ? ?->?->? HtmlElement js_tab8 = (HtmlElement) gsgsxx_djxx .getFirstByXPath("//div[@id='tabs']/ul/li[@id='8']"); String gsgsxx_ccjcxx_ccjcxx = null; if (null != js_tab8 && !StringUtils.isEmpty(js_tab8)) { HtmlPage gsgsxx_ccjcxx_ccjcxx_page = js_tab8.click(); gsgsxx_ccjcxx_ccjcxx = gsgsxx_ccjcxx_ccjcxx_page.asXml(); } resultHtmlMap.put("gsgsxx_ccjcxx_ccjcxx", gsgsxx_ccjcxx_ccjcxx); // ? ??->?-> HtmlElement js_left2 = (HtmlElement) gsgsxx_djxx.getByXPath("//div[@id='leftTabs']/ul/li") .get(1); HtmlPage qygsxx_qynb_list_page = js_left2.click(); resultHtmlMap.put("qygsxx_qynb_list_page", qygsxx_qynb_list_page.asXml()); // ? ??->?-> @SuppressWarnings("unchecked") List<HtmlElement> qygsxx_qynb_list_as = (List<HtmlElement>) qygsxx_qynb_list_page .getByXPath("//div[@id='qiyenianbao']/table/tbody/tr/td/a"); List<Map<String, Object>> qygsxx_qynb_infos = new ArrayList<Map<String, Object>>(); if (null != qygsxx_qynb_list_as && !qygsxx_qynb_list_as.isEmpty()) { for (HtmlElement qygsxx_qynb_list_a : qygsxx_qynb_list_as) { Map<String, Object> qygsxx_qynb_info_map = new LinkedHashMap<String, Object>(); String qygsxx_qynb_list_a_text = qygsxx_qynb_list_a.getTextContent().toString().trim(); // String qygsxx_qynb_list_pubdate = String // .valueOf(((HtmlElement) qygsxx_qynb_list_a // .getParentNode().getNextSibling()) // .getTextContent()); qygsxx_qynb_info_map.put("qygsxx_qynb_list_a_text", qygsxx_qynb_list_a_text); // qygsxx_qynb_info_map.put( // "qygsxx_qynb_list_pubdate", // qygsxx_qynb_list_pubdate); HtmlPage qygsxx_qynb_info_page = (HtmlPage) qygsxx_qynb_list_a.click(); qygsxx_qynb_info_map.put("qygsxx_qynb_info_page", qygsxx_qynb_info_page.asXml()); qygsxx_qynb_infos.add(qygsxx_qynb_info_map); } } resultHtmlMap.put("qygsxx_qynb_infos", qygsxx_qynb_infos); // ? ??->?? HtmlElement js_lefttab3 = (HtmlElement) qygsxx_qynb_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='3']"); String qygsxx_gdjczxx = null; if (null != js_lefttab3 && !StringUtils.isEmpty(js_lefttab3)) { HtmlPage qygsxx_gdjczxx_page = js_lefttab3.click(); qygsxx_gdjczxx = qygsxx_gdjczxx_page.asXml(); } resultHtmlMap.put("qygsxx_gdjczxx", qygsxx_gdjczxx); // ? ??->??? HtmlElement js_lefttab6 = (HtmlElement) qygsxx_qynb_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='6']"); String qygsxx_gqbgxx = null; if (null != js_lefttab6 && !StringUtils.isEmpty(js_lefttab6)) { HtmlPage qygsxx_gqbgxx_page = js_lefttab6.click(); qygsxx_gqbgxx = qygsxx_gqbgxx_page.asXml(); } resultHtmlMap.put("qygsxx_gqbgxx", qygsxx_gqbgxx); // ? ??->?? HtmlElement js_lefttab1 = (HtmlElement) qygsxx_qynb_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='1']"); String qygsxx_xzxkxx = null; if (null != js_lefttab1 && !StringUtils.isEmpty(js_lefttab1)) { HtmlPage qygsxx_xzxkxx_page = js_lefttab1.click(); qygsxx_xzxkxx = qygsxx_xzxkxx_page.asXml(); } resultHtmlMap.put("qygsxx_xzxkxx", qygsxx_xzxkxx); // ? ??->?? HtmlElement js_lefttab2 = (HtmlElement) qygsxx_qynb_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='2']"); String qygsxx_zscqczdjxx = null; if (null != js_lefttab2 && !StringUtils.isEmpty(js_lefttab2)) { HtmlPage qygsxx_zscqczdjxx_page = js_lefttab2.click(); qygsxx_zscqczdjxx = qygsxx_zscqczdjxx_page.asXml(); } resultHtmlMap.put("qygsxx_zscqczdjxx", qygsxx_zscqczdjxx); // ? ??->? HtmlElement js_lefttab4 = (HtmlElement) qygsxx_qynb_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='4']"); String qygsxx_xzcfxx = null; if (null != js_lefttab4 && !StringUtils.isEmpty(js_lefttab4)) { HtmlPage qygsxx_xzcfxx_page = js_lefttab4.click(); qygsxx_xzcfxx = qygsxx_xzcfxx_page.asXml(); } resultHtmlMap.put("qygsxx_xzcfxx", qygsxx_xzcfxx); // ?? // HtmlElement js_left3 = (HtmlElement) gsgsxx_djxx // .getByXPath("//div[@id='leftTabs']/ul/li").get(2); // HtmlPage qtbmgsxx_xzxkxx_page = js_left3.click(); // ? ?->?? WebClient wc = firstInfoPage.getWebClient(); wc.getOptions().setJavaScriptEnabled(false); HtmlPage firstByXPath = wc.getPage(webWindow, new WebRequest(new URL( "http://www.jsgsj.gov.cn:58888/ecipplatform/manyCommonFnQueryServlet.json?query_xingzhengxuke=true&corp_id=" + corp_id + "&corp_org=" + corp_org + "&pageNo=1&pageSize=100"))); String qtbmgsxx_xzxkxx = null; if (null != firstByXPath && !StringUtils.isEmpty(firstByXPath)) { qtbmgsxx_xzxkxx = firstByXPath.asXml(); } resultHtmlMap.put("qtbmgsxx_xzxkxx", qtbmgsxx_xzxkxx); // HtmlElement firstByXPath = (HtmlElement) qtbmgsxx_xzxkxx_page // .getFirstByXPath("//div[@id='tabs']/ul/li[@id='1']"); // String qtbmgsxx_xzxkxx = null; // if (null != firstByXPath // && !StringUtils.isEmpty(firstByXPath)) { // HtmlPage click = firstByXPath.click(); // qtbmgsxx_xzxkxx = click.asXml(); // } // resultHtmlMap.put("qtbmgsxx_xzxkxx", qtbmgsxx_xzxkxx); // ? ?->? HtmlPage qtbm_xzcfxx = wc.getPage(webWindow, new WebRequest(new URL( "http://www.jsgsj.gov.cn:58888/ecipplatform/manyCommonFnQueryServlet.json?query_xingzhengchufa=true&corp_id=" + corp_id + "&corp_org=" + corp_org + "&pageNo=1&pageSize=100"))); String qtbmgsxx_xzcfxx = null; if (null != qtbm_xzcfxx && !StringUtils.isEmpty(qtbm_xzcfxx)) { qtbmgsxx_xzcfxx = qtbm_xzcfxx.asXml(); } resultHtmlMap.put("qtbmgsxx_xzcfxx", qtbmgsxx_xzcfxx); // HtmlElement js_lefttab32 = (HtmlElement) qtbmgsxx_xzxkxx_page // .getFirstByXPath("//div[@id='tabs']/ul/li[@id='2']"); // String qtbmgsxx_xzcfxx = null; // if (null != js_lefttab32 // && !StringUtils.isEmpty(js_lefttab32)) { // HtmlPage qtbmgsxx_xzcfxx_page = js_lefttab32.click(); // qtbmgsxx_xzcfxx = qtbmgsxx_xzcfxx_page.asXml(); // } // resultHtmlMap.put("qtbmgsxx_xzcfxx", qtbmgsxx_xzcfxx); wc.getOptions().setJavaScriptEnabled(true); // ????? int wd = gsgsxx_djxx.getByXPath("//div[@id='leftTabs']/ul/li").size(); HtmlElement js_left23 = null; if (wd > 3) { js_left23 = (HtmlElement) gsgsxx_djxx.getByXPath("//div[@id='leftTabs']/ul/li").get(3); } if (null != js_left23) { HtmlPage sfxzgsxx_gqdjxx_list_page = js_left23.click(); Thread.sleep(2000); // ? ????->?? HtmlElement firstByXPath2 = (HtmlElement) sfxzgsxx_gqdjxx_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='0']"); String sfxzgsxx_gqdjxx_list = null; if (null != firstByXPath2 && !StringUtils.isEmpty(firstByXPath2)) { HtmlPage click1 = firstByXPath2.click(); sfxzgsxx_gqdjxx_list = click1.asXml(); } resultHtmlMap.put("sfxzgsxx_gqdjxx_list", sfxzgsxx_gqdjxx_list); // ? ????->?? HtmlElement js_lefttab231 = (HtmlElement) sfxzgsxx_gqdjxx_list_page .getFirstByXPath("//div[@id='tabs']/ul/li[@id='1']"); String sfxzgsxx_gqbgxx_list = null; if (null != js_lefttab231 && !StringUtils.isEmpty(js_lefttab231)) { HtmlPage sfxzgsxx_gqbgxx_list_page = js_lefttab231.click(); sfxzgsxx_gqbgxx_list = sfxzgsxx_gqbgxx_list_page.asXml(); } resultHtmlMap.put("sfxzgsxx_gqbgxx_list", sfxzgsxx_gqbgxx_list); } resultHtmlMap.put("statusCodeDef", StatusCodeDef.SCCCESS); } } } LOGGER.returnRedisResource(); return resultHtmlMap; }
From source file:com.storm.function.GsxtFunction.java
private Map<String, Object> getHtmlInfoMapOfLiaoning(String area, HtmlPage firstInfoPage, String keyword, ChannelLogger LOGGER) throws Exception { LOGGER.info("=========" + area + "=========" + keyword + "========="); Map<String, Object> resultHtmlMap = new LinkedHashMap<String, Object>(); if (null == firstInfoPage) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.FAILURE); } else {//w w w .j av a2 s . c om WebWindow webWindow = firstInfoPage.getWebClient().getCurrentWindow(); final String HOST_OF_LIAONING = "http://gsxt.lngs.gov.cn"; @SuppressWarnings("unchecked") List<HtmlAnchor> anchors = (List<HtmlAnchor>) firstInfoPage .getByXPath("//div[@id='listContent']/div/ul/li/a"); HtmlElement div_none = firstInfoPage.getFirstByXPath("//div[@class='list-a']"); if (null == anchors || anchors.isEmpty()) { if (null == div_none) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.IMAGECODE_ERROR); } else { if (div_none.asXml().contains("??")) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); } else { resultHtmlMap.put("statusCodeDef", StatusCodeDef.IMAGECODE_ERROR); } } } HtmlAnchor htmlAnchor = null; boolean flag = false; if (anchors != null && !anchors.isEmpty()) { for (HtmlAnchor anchor : anchors) { String anchorTitle = anchor.getTextContent().toString().trim(); if (anchorTitle.contains(keyword)) { // ???? htmlAnchor = anchor; flag = true; break; } } if (!flag) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); LOGGER.info("????"); } } if (flag) { // ? String pripid = ""; String type = ""; // ??? HtmlElement target_item_info = (HtmlElement) htmlAnchor.getParentNode().getParentNode(); resultHtmlMap.put("target_item_info", target_item_info.asXml()); String liaoning_onclick = htmlAnchor.getAttribute("onclick"); if (!StringUtils.isEmpty(liaoning_onclick)) { pripid = liaoning_onclick.split(",")[2].replace("'", "").replace("'", ""); type = liaoning_onclick.split(",")[1].replace("'", "").replace("'", ""); } // ??? HtmlPage gsgsxx = htmlAnchor.click(); Thread.sleep(3000); Document gsgsxx_dm = Jsoup.parseBodyFragment(gsgsxx.asXml()); Element s_gs_dj_1 = gsgsxx_dm.getElementById("s_gs_dj_1"); // Element s_gs_dj_2 = gsgsxx_dm.getElementById("s_gs_dj_2"); // ???->?->? // String jbxx_url = // "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getJbxxAction.action?pripid=" // + pripid + "&type=" + type; // HtmlPage gsgsxx_djxx_jbxx = firstInfoPage.getWebClient() // .getPage(jbxx_url); String gsgsxx_djxx_jbxx_str = null; if (null != s_gs_dj_1) { gsgsxx_djxx_jbxx_str = s_gs_dj_1.html(); } resultHtmlMap.put("gsgsxx_djxx_jbxx", gsgsxx_djxx_jbxx_str); // ??->?->? //String gsgsxx_djxx_tzrxx_str = null; //if (null != s_gs_dj_2) { // gsgsxx_djxx_tzrxx_str = s_gs_dj_2.html(); //} //resultHtmlMap.put("gsgsxx_djxx_tzrxx", gsgsxx_djxx_tzrxx_str); // ????->?-> @SuppressWarnings("unchecked") List<HtmlAnchor> touziren_anchors = (List<HtmlAnchor>) gsgsxx .getByXPath("//tbody[@id='tzr_itemContainer']/tr/td/a"); if (null != touziren_anchors && !touziren_anchors.isEmpty()) { List<Map<String, Object>> gsgsxx_djxx_tzrxx_xqs = new ArrayList<Map<String, Object>>(); for (HtmlAnchor touziren_anchor : touziren_anchors) { Map<String, Object> gsgsxx_djxx_tzrxx_xq = new LinkedHashMap<String, Object>(); HtmlPage gsgsxx_djxx_tzrxx_xq_page = touziren_anchor.click(); gsgsxx_djxx_tzrxx_xq.put("gsgsxx_djxx_tzrxx_xq", gsgsxx_djxx_tzrxx_xq_page.asXml()); gsgsxx_djxx_tzrxx_xqs.add(gsgsxx_djxx_tzrxx_xq); } resultHtmlMap.put("gsgsxx_djxx_tzrxx_xqs", gsgsxx_djxx_tzrxx_xqs); } WebClient wc = firstInfoPage.getWebClient(); wc.getOptions().setJavaScriptEnabled(false); // ??->?->? String tzr_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getTzrxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage tzrPage = wc.getPage(tzr_url); String gsgsxx_djxx_tzrxx_str = null; if (null != tzrPage) { gsgsxx_djxx_tzrxx_str = tzrPage.asXml(); } resultHtmlMap.put("gsgsxx_djxx_tzrxx", gsgsxx_djxx_tzrxx_str); // ??->?->?? String bgxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getBgxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage bgxxPage = wc.getPage(webWindow, new WebRequest(new URL(bgxx_url))); // ??->?->?? String gsgsxx_djxx_bgxx_str = null; if (null != bgxxPage) { gsgsxx_djxx_bgxx_str = bgxxPage.asXml(); } resultHtmlMap.put("gsgsxx_djxx_bgxx", gsgsxx_djxx_bgxx_str); // String[] command = {"casperjs", // "/home/ubuntu/nfs-images/casperjscode/getSimpleRequestPage.js", // "--web-security=no", "--url=" + bgxx_url}; // String casperjsResult = CommandUtil.runCommand(command); // resultHtmlMap.put("gsgsxx_djxx_bgxx", casperjsResult); // ??->?->?? String gsgsxx_baxx_zyryxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getZyryxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_baxx_zyryxx_page = wc.getPage(webWindow, new WebRequest(new URL(gsgsxx_baxx_zyryxx_url))); String gsgsxx_baxx_zyryxx_str = null; if (null != gsgsxx_baxx_zyryxx_page) { gsgsxx_baxx_zyryxx_str = gsgsxx_baxx_zyryxx_page.asXml(); } resultHtmlMap.put("gsgsxx_baxx_zyryxx", gsgsxx_baxx_zyryxx_str); // ??->?->? String gsgsxx_baxx_zgbmxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getTzrxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_baxx_zgbmxx_page = wc.getPage(webWindow, new WebRequest(new URL(gsgsxx_baxx_zgbmxx_url))); String gsgsxx_baxx_zgbmxx_str = null; if (null != gsgsxx_baxx_zgbmxx_page) { gsgsxx_baxx_zgbmxx_str = gsgsxx_baxx_zgbmxx_page.asXml(); } resultHtmlMap.put("gsgsxx_baxx_zgbmxx", gsgsxx_baxx_zgbmxx_str); wc.getOptions().setJavaScriptEnabled(true); // ??->?->? String gsgsxx_baxx_fzjgxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getFgsxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_baxx_fzjgxx_page = firstInfoPage.getWebClient().getPage(gsgsxx_baxx_fzjgxx_url); String gsgsxx_baxx_fzjgxx_str = null; if (null != gsgsxx_baxx_fzjgxx_page) { gsgsxx_baxx_fzjgxx_str = gsgsxx_baxx_fzjgxx_page.asXml(); } resultHtmlMap.put("gsgsxx_baxx_fzjgxx", gsgsxx_baxx_fzjgxx_str); // ??->?->? String gsgsxx_baxx_qsxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQsxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_baxx_qsxx_page = firstInfoPage.getWebClient().getPage(gsgsxx_baxx_qsxx_url); String gsgsxx_baxx_qsxx_str = null; if (null != gsgsxx_baxx_qsxx_page) { gsgsxx_baxx_qsxx_str = gsgsxx_baxx_qsxx_page.asXml(); } resultHtmlMap.put("gsgsxx_baxx_qsxx", gsgsxx_baxx_qsxx_str); // ??->?->? String gsgsxx_dcdydjxx_dcdydjxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getDcdydjAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_dcdydjxx_dcdydjxx_page = firstInfoPage.getWebClient() .getPage(gsgsxx_dcdydjxx_dcdydjxx_url); String gsgsxx_dcdydjxx_dcdydjxx_str = null; if (null != gsgsxx_dcdydjxx_dcdydjxx_page) { gsgsxx_dcdydjxx_dcdydjxx_str = gsgsxx_dcdydjxx_dcdydjxx_page.asXml(); } resultHtmlMap.put("gsgsxx_dcdydjxx_dcdydjxx", gsgsxx_dcdydjxx_dcdydjxx_str); // ??->??->?? String gsgsxx_gqczdjxx_gqczdjxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getGsgsGqczxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_gqczdjxx_gqczdjxx_page = firstInfoPage.getWebClient() .getPage(gsgsxx_gqczdjxx_gqczdjxx_url); String gsgsxx_gqczdjxx_gqczdjxx_str = null; if (null != gsgsxx_gqczdjxx_gqczdjxx_page) { gsgsxx_gqczdjxx_gqczdjxx_str = gsgsxx_gqczdjxx_gqczdjxx_page.asXml(); } resultHtmlMap.put("gsgsxx_gqczdjxx_gqczdjxx", gsgsxx_gqczdjxx_gqczdjxx_str); // ??->?->? String gsgsxx_xzcfxx_xzcfxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getXzcfxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_xzcfxx_xzcfxx_page = firstInfoPage.getWebClient().getPage(gsgsxx_xzcfxx_xzcfxx_url); String gsgsxx_xzcfxx_xzcfxx_str = null; if (null != gsgsxx_xzcfxx_xzcfxx_page) { gsgsxx_xzcfxx_xzcfxx_str = gsgsxx_xzcfxx_xzcfxx_page.asXml(); } resultHtmlMap.put("gsgsxx_xzcfxx_xzcfxx", gsgsxx_xzcfxx_xzcfxx_str); // ??->???->??? String gsgsxx_jyycxx_jyycxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getJyycxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_jyycxx_jyycxx_page = firstInfoPage.getWebClient().getPage(gsgsxx_jyycxx_jyycxx_url); String gsgsxx_jyycxx_jyycxx_str = null; if (null != gsgsxx_jyycxx_jyycxx_page) { gsgsxx_jyycxx_jyycxx_str = gsgsxx_jyycxx_jyycxx_page.asXml(); } resultHtmlMap.put("gsgsxx_jyycxx_jyycxx", gsgsxx_jyycxx_jyycxx_str); // ??->???->??? String gsgsxx_yzwfxx_yzwfxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getYzwfxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_yzwfxx_yzwfxx_page = firstInfoPage.getWebClient().getPage(gsgsxx_yzwfxx_yzwfxx_url); String gsgsxx_yzwfxx_yzwfxx_str = null; if (null != gsgsxx_yzwfxx_yzwfxx_page) { gsgsxx_yzwfxx_yzwfxx_str = gsgsxx_yzwfxx_yzwfxx_page.asXml(); } resultHtmlMap.put("gsgsxx_yzwfxx_yzwfxx", gsgsxx_yzwfxx_yzwfxx_str); // ??->?->? String gsgsxx_ccjcxx_ccjcxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getCcjcxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gsgsxx_ccjcxx_ccjcxx_page = firstInfoPage.getWebClient().getPage(gsgsxx_ccjcxx_ccjcxx_url); String gsgsxx_ccjcxx_ccjcxx_str = null; if (null != gsgsxx_ccjcxx_ccjcxx_page) { gsgsxx_ccjcxx_ccjcxx_str = gsgsxx_ccjcxx_ccjcxx_page.asXml(); } resultHtmlMap.put("gsgsxx_ccjcxx_ccjcxx", gsgsxx_ccjcxx_ccjcxx_str); // ???->?-> String qygsxx_qynb_list_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsQynbxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_qynb_list_page = firstInfoPage.getWebClient().getPage(qygsxx_qynb_list_url); resultHtmlMap.put("qygsxx_qynb_list_page", qygsxx_qynb_list_page.asXml()); // ???->?-> 1_3? ? ? // ???? ????? @SuppressWarnings("unchecked") List<HtmlElement> qygsxx_qynb_list_as = (List<HtmlElement>) qygsxx_qynb_list_page .getByXPath("//tbody[@id='qynbItemContainer']/tr/td[2]/a"); List<Map<String, Object>> qygsxx_qynb_infos = new ArrayList<Map<String, Object>>(); if (qygsxx_qynb_list_as != null && !qygsxx_qynb_list_as.isEmpty()) { for (HtmlElement qygsxx_qynb_list_a : qygsxx_qynb_list_as) { Map<String, Object> qygsxx_qynb_info_map = new LinkedHashMap<String, Object>(); String qygsxx_qynb_list_a_href = HOST_OF_LIAONING + qygsxx_qynb_list_a.getAttribute("href"); String qygsxx_qynb_list_a_text = qygsxx_qynb_list_a.getTextContent(); // String qygsxx_qynb_list_pubdate = ((HtmlElement) // qygsxx_qynb_list_a // .getParentNode().getNextSibling()).getTextContent(); qygsxx_qynb_info_map.put("qygsxx_qynb_list_a_href", qygsxx_qynb_list_a_href); qygsxx_qynb_info_map.put("qygsxx_qynb_list_a_text", qygsxx_qynb_list_a_text); // qygsxx_qynb_info_map.put("qygsxx_qynb_list_pubdate", // qygsxx_qynb_list_pubdate); // ????&(??|??) // ???? // ?? ?????? // ?? // ?? ? HtmlPage qygsxx_qynb_info_page = firstInfoPage.getWebClient() .getPage(qygsxx_qynb_list_a_href); qygsxx_qynb_info_map.put("qygsxx_qynb_info_page", qygsxx_qynb_info_page.asXml()); qygsxx_qynb_infos.add(qygsxx_qynb_info_map); } } resultHtmlMap.put("qygsxx_qynb_infos", qygsxx_qynb_infos); // ???->?? String qygsxx_gdjczxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsJsGdjczxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_gdjczxx_page = firstInfoPage.getWebClient().getPage(qygsxx_gdjczxx_url); String qygsxx_gdjczxx_str = null; if (null != qygsxx_gdjczxx_page) { qygsxx_gdjczxx_str = qygsxx_gdjczxx_page.asXml(); } resultHtmlMap.put("qygsxx_gdjczxx", qygsxx_gdjczxx_str); // ???->??->?? String qygsxx_gdjczxx_bgxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsJsGdjczbgxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_gdjczxx_bgxx_page = firstInfoPage.getWebClient().getPage(qygsxx_gdjczxx_bgxx_url); String qygsxx_gdjczxx_bgxx_str = null; if (null != qygsxx_gdjczxx_page) { qygsxx_gdjczxx_bgxx_str = qygsxx_gdjczxx_bgxx_page.asXml(); } resultHtmlMap.put("qygsxx_gdjczxx_bgxx", qygsxx_gdjczxx_bgxx_str); // ? ??->??? String qygsxx_gqbgxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsJsGqbgxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_gqbgxx_page = firstInfoPage.getWebClient().getPage(qygsxx_gqbgxx_url); String qygsxx_gqbgxx_str = null; if (null != qygsxx_gqbgxx_page) { qygsxx_gqbgxx_str = qygsxx_gqbgxx_page.asXml(); } resultHtmlMap.put("qygsxx_gqbgxx", qygsxx_gqbgxx_str); // ? ??->?? String qygsxx_xzxkxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsJsXzxkxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_xzxkxx_page = firstInfoPage.getWebClient().getPage(qygsxx_xzxkxx_url); String qygsxx_xzxkxx_str = null; if (null != qygsxx_xzxkxx_page) { qygsxx_xzxkxx_str = qygsxx_xzxkxx_page.asXml(); } resultHtmlMap.put("qygsxx_xzxkxx", qygsxx_xzxkxx_str); // ???->?? String qygsxx_zscqczdjxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsJsZscqczxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_zscqczdjxx_page = firstInfoPage.getWebClient().getPage(qygsxx_zscqczdjxx_url); String qygsxx_zscqczdjxx_str = null; if (null != qygsxx_zscqczdjxx_page) { qygsxx_zscqczdjxx_str = qygsxx_zscqczdjxx_page.asXml(); } resultHtmlMap.put("qygsxx_zscqczdjxx", qygsxx_zscqczdjxx_str); // ???->? String qygsxx_xzcfxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getQygsJsXzcfxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage qygsxx_xzcfxx_page = firstInfoPage.getWebClient().getPage(qygsxx_xzcfxx_url); String qygsxx_xzcfxx_str = null; if (null != qygsxx_xzcfxx_page) { qygsxx_xzcfxx_str = qygsxx_xzcfxx_page.asXml(); } resultHtmlMap.put("qygsxx_xzcfxx", qygsxx_xzcfxx_str); // ?????->?? String gqdjxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getSfgsGqdjxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gqdjxx_page = firstInfoPage.getWebClient().getPage(gqdjxx_url); String sfxzgsxx_gqdjxx_str = null; if (null != gqdjxx_page) { sfxzgsxx_gqdjxx_str = gqdjxx_page.asXml(); } resultHtmlMap.put("sfxzgsxx_gqdjxx", sfxzgsxx_gqdjxx_str); // ?????->?? String gdbgxx_url = "http://gsxt.lngs.gov.cn/saicpub/entPublicitySC/entPublicityDC/getSfgsGdbgxxAction.action?pripid=" + pripid + "&type=" + type; HtmlPage gdbgxx_page = firstInfoPage.getWebClient().getPage(gdbgxx_url); String sfxzgsxx_gdbgxx_str = null; if (null != gdbgxx_page) { sfxzgsxx_gdbgxx_str = gdbgxx_page.asXml(); } resultHtmlMap.put("sfxzgsxx_gdbgxx", sfxzgsxx_gdbgxx_str); resultHtmlMap.put("statusCodeDef", StatusCodeDef.SCCCESS); } } LOGGER.returnRedisResource(); return resultHtmlMap; }
From source file:com.storm.function.GsxtFunction.java
private Map<String, Object> getHtmlInfoMapOfTianjin(String area, HtmlPage firstInfoPage, String keyword, ChannelLogger LOGGER) throws Exception { Map<String, Object> resultHtmlMap = new HashMap<String, Object>(); //????//from w ww.jav a 2 s . c o m // System.out.println(firstInfoPage.asXml()); WebWindow window = firstInfoPage.getWebClient().getCurrentWindow(); @SuppressWarnings("unchecked") List<HtmlAnchor> divByXPath = (List<HtmlAnchor>) firstInfoPage.getByXPath("//div[@class='result-item']"); HtmlElement firstByXPath = ((HtmlElement) firstInfoPage .getFirstByXPath("//div[@class='content']/div[@style='font-size:12px']")); if (divByXPath.size() == 0 && firstByXPath == null) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.IMAGECODE_ERROR); } else { if (firstByXPath != null) { String textContent = firstByXPath.getTextContent(); if (textContent.indexOf("? 0 ?") > 0) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); } else { resultHtmlMap.put("statusCodeDef", StatusCodeDef.SCCCESS); } } else { resultHtmlMap.put("statusCodeDef", StatusCodeDef.SCCCESS); } } @SuppressWarnings("unchecked") List<HtmlAnchor> anchors = (List<HtmlAnchor>) firstInfoPage.getByXPath("//div[@class='result-item']/div/a"); LOGGER.info(anchors.toString()); if (anchors != null && !anchors.isEmpty()) { boolean matchFlag = false; for (HtmlAnchor anchor : anchors) { String anchorTitle = anchor.getTextContent().toString().trim(); if (anchorTitle.contains(keyword)) { //???? matchFlag = true; //??? HtmlElement target_item_info = (HtmlElement) anchor.getParentNode().getParentNode(); resultHtmlMap.put("target_item_info", target_item_info.asXml()); //*****************? ***************** //?? ?->?->? //?entId //? String ent_id = anchor.getAttribute("href"); if (!StringUtils.isEmpty(ent_id)) { ent_id = ent_id.split("=")[1]; } if (!StringUtils.isEmpty(ent_id)) { /*HtmlPage gsgsxx_djxx_jbxx = anchor.click(); Thread.sleep(3000); resultHtmlMap.put("gsgsxx_djxx_jbxx", gsgsxx_djxx_jbxx.asXml()); */ String gsgsxx_djxx_jbxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=dj"; Page gsgsxx_djxx_jbxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_djxx_jbxx_url))); resultHtmlMap.put("gsgsxx_djxx_jbxx", gsgsxx_djxx_jbxx.getWebResponse().getContentAsString("utf-8")); //?? ?->?->?->?-> Document qygsxxHtml = Jsoup .parseBodyFragment(gsgsxx_djxx_jbxx.getWebResponse().getContentAsString("utf-8")); if (qygsxxHtml != null) { Element qynbDiv = qygsxxHtml.getElementById("touziren"); if (qynbDiv != null) { Elements qynb_trs = qynbDiv.select("tbody").select("tr").select("td").select("a"); if (qynb_trs.size() != 0) { List<String> list = new ArrayList<String>(); for (int i = 0; i < qynb_trs.size(); i++) { //System.out.println(qynb_trs.get(i).toString()); //System.out.println(qynb_trs.get(i).toString().split("\\(\\'")[1].split("\\'\\)")[0].split("\\'\\,\\'")[0]); if (qynb_trs.get(i).toString() != null && qynb_trs.get(i).toString().split("\\(\\'")[1].split("\\'\\)")[0] .split("\\'\\,\\'")[0] != null) { String gsgsxx_djxx_gdxx_detail_url = "http://tjcredit.gov.cn/saicpf/gsgdcz?gdczid=" + qynb_trs.get(i).toString().split("\\(\\'")[1] .split("\\'\\)")[0].split("\\'\\,\\'")[0] + "&entid=" + ent_id + "&issaic=1&hasInfo=0"; Page gsgsxx_djxx_gdxx_detail = firstInfoPage.getWebClient().getPage( window, new WebRequest(new URL(gsgsxx_djxx_gdxx_detail_url))); list.add(gsgsxx_djxx_gdxx_detail.getWebResponse() .getContentAsString("utf-8")); } } if (list.size() != 0) { resultHtmlMap.put("gsgsxx_djxx_gdxx", list); } } } } //? ?->?->?? String gsgsxx_baxx_zyryxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=ba"; Page gsgsxx_baxx_zyryxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_baxx_zyryxx_url))); resultHtmlMap.put("gsgsxx_baxx_zyryxx", gsgsxx_baxx_zyryxx.getWebResponse().getContentAsString("utf-8")); //? ?->?->? String gsgsxx_dcdydjxx_dcdydjxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=dcdydjxx"; Page gsgsxx_dcdydjxx_dcdydjxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_dcdydjxx_dcdydjxx_url))); resultHtmlMap.put("gsgsxx_dcdydjxx_dcdydjxx", gsgsxx_dcdydjxx_dcdydjxx.getWebResponse().getContentAsString("utf-8")); //? ?->??->?? String gsgsxx_gqczdjxx_gqczdjxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=gqczdjxx"; Page gsgsxx_gqczdjxx_gqczdjxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_gqczdjxx_gqczdjxx_url))); resultHtmlMap.put("gsgsxx_gqczdjxx_gqczdjxx", gsgsxx_gqczdjxx_gqczdjxx.getWebResponse().getContentAsString("utf-8")); //? ?->?->? String gsgsxx_xzcfxx_xzcfxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=xzcf"; Page gsgsxx_xzcfxx_xzcfxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_xzcfxx_xzcfxx_url))); resultHtmlMap.put("gsgsxx_xzcfxx_xzcfxx", gsgsxx_xzcfxx_xzcfxx.getWebResponse().getContentAsString("utf-8")); // Document xzcfxxHtml = Jsoup.parseBodyFragment( gsgsxx_xzcfxx_xzcfxx.getWebResponse().getContentAsString("utf-8")); if (xzcfxxHtml != null) { Elements qynbDiv = xzcfxxHtml.getElementsByClass("result-table"); if (qynbDiv != null && qynbDiv.size() != 0) { Elements qynb_trs = qynbDiv.get(0).select("tbody").select("tr").select("td") .select("a"); if (qynb_trs.size() != 0) { List<String> list = new ArrayList<String>(); for (int i = 0; i < qynb_trs.size(); i++) { if (qynb_trs.get(i).toString() != null && qynb_trs.get(i).toString().split("\\(\\'")[1].split("\\'\\)")[0] .split("\\'\\,\\'")[0] != null) { String gsgsxx_djxx_gdxx_detail_url = "http://tjcredit.gov.cn/saicpf/gsxzcf?id=" + qynb_trs.get(i).toString().split("\\(\\'")[1] .split("\\'\\)")[0].split("\\'\\,\\'")[0] + "&entid=" + ent_id + "&issaic=1&hasInfo=0"; Page gsgsxx_djxx_gdxx_detail = firstInfoPage.getWebClient().getPage( window, new WebRequest(new URL(gsgsxx_djxx_gdxx_detail_url))); list.add(gsgsxx_djxx_gdxx_detail.getWebResponse() .getContentAsString("utf-8")); } } if (list.size() != 0) { resultHtmlMap.put("gsgsxx_xzcfxx_detail", list); } } } } //? ?->???->??? String gsgsxx_jyycxx_jyycxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=qyjyycmlxx"; Page gsgsxx_jyycxx_jyycxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_jyycxx_jyycxx_url))); resultHtmlMap.put("gsgsxx_jyycxx_jyycxx", gsgsxx_jyycxx_jyycxx.getWebResponse().getContentAsString("utf-8")); //? ?->???->??? String gsgsxx_yzwfxx_yzwfxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=yzwfqyxx"; Page gsgsxx_yzwfxx_yzwfxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_yzwfxx_yzwfxx_url))); resultHtmlMap.put("gsgsxx_yzwfxx_yzwfxx", gsgsxx_yzwfxx_yzwfxx.getWebResponse().getContentAsString("utf-8")); //? ?->?->? String gsgsxx_ccjcxx_ccjcxx_url = "http://tjcredit.gov.cn/platform/saic/baseInfo.json?entId=" + ent_id + "&departmentId=scjgw&infoClassId=ccjcxx"; Page gsgsxx_ccjcxx_ccjcxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(gsgsxx_ccjcxx_ccjcxx_url))); resultHtmlMap.put("gsgsxx_ccjcxx_ccjcxx", gsgsxx_ccjcxx_ccjcxx.getWebResponse().getContentAsString("utf-8")); //*****************? ?***************** //*****************?? ***************** //? ??->?-> String qygsxx_qynb_list_url = "http://tjcredit.gov.cn/report/nblist?entid=" + ent_id; Page qygsxx_qynb_list = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_qynb_list_url))); resultHtmlMap.put("qygsxx_qynb_list", qygsxx_qynb_list.getWebResponse().getContentAsString("utf-8")); //? ??->?->-> //? Document qynbHtml = Jsoup .parseBodyFragment(qygsxx_qynb_list.getWebResponse().getContentAsString("utf-8")); if (qynbHtml != null) { Element qynbDiv = qynbHtml.getElementById("touziren"); if (qynbDiv != null) { Elements qynb_trs = qynbDiv.select("tbody").select("tr").select("td").select("a"); if (qynb_trs.size() != 0) { List<String> list = new ArrayList<String>(); for (int i = 0; i < qynb_trs.size(); i++) { //System.out.println(qynb_trs.get(i).toString()); //System.out.println(qynb_trs.get(i).toString().split("\\(\\'")[1].split("\\'\\)")[0].split("\\'\\,\\'")[0]); if (qynb_trs.get(i).toString() != null && qynb_trs.get(i).toString().split("\\(\\'")[1].split("\\'\\)")[0] .split("\\'\\,\\'")[1] != null) { String gsgsxx_djxx_gdxx_detail_url = "http://tjcredit.gov.cn/report/annals?entid=" + ent_id + "&year=" + qynb_trs.get(i).toString().split("\\(\\'")[1] .split("\\'\\)")[0].split("\\'\\,\\'")[1] + "&hasInfo=0"; Page gsgsxx_djxx_gdxx_detail = firstInfoPage.getWebClient().getPage( window, new WebRequest(new URL(gsgsxx_djxx_gdxx_detail_url))); System.out.println(gsgsxx_djxx_gdxx_detail.getWebResponse() .getContentAsString("utf-8")); list.add(gsgsxx_djxx_gdxx_detail.getWebResponse() .getContentAsString("utf-8")); } } if (list.size() != 0) { resultHtmlMap.put("qygsxx_qynb_detail", list); } } } } //? ??->?? String qygsxx_xzxkxx_url = "http://tjcredit.gov.cn/report/xzxk?entid=" + ent_id; Page qygsxx_xzxkxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_xzxkxx_url))); resultHtmlMap.put("qygsxx_xzxkxx", qygsxx_xzxkxx.getWebResponse().getContentAsString("utf-8")); //? ??->?? String qygsxx_gdjczxx_url = "http://tjcredit.gov.cn/report/gdcz?entid=" + ent_id; Page qygsxx_gdjczxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_gdjczxx_url))); resultHtmlMap.put("qygsxx_gdjczxx", qygsxx_gdjczxx.getWebResponse().getContentAsString("utf-8")); //? ??->??? String qygsxx_gqbgxx_url = "http://tjcredit.gov.cn/report/gqbg?entid=" + ent_id; Page qygsxx_gqbgxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_gqbgxx_url))); resultHtmlMap.put("qygsxx_gqbgxx", qygsxx_gqbgxx.getWebResponse().getContentAsString("utf-8")); //? ??->?? String qygsxx_zscqczdjxx_url = "http://tjcredit.gov.cn/report/zscq?entid=" + ent_id; Page qygsxx_zscqczdjxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_zscqczdjxx_url))); resultHtmlMap.put("qygsxx_zscqczdjxx", qygsxx_zscqczdjxx.getWebResponse().getContentAsString("utf-8")); //? ??->? String qygsxx_xzcfxx_url = "http://tjcredit.gov.cn/report/xzcf?entid=" + ent_id; Page qygsxx_xzcfxx = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_xzcfxx_url))); resultHtmlMap.put("qygsxx_xzcfxx", qygsxx_xzcfxx.getWebResponse().getContentAsString("utf-8")); //*****************?? ?***************** //*****************???? ***************** //? ????->?? String sfxzgsxx_gqdjxx_list_url = "http://tjcredit.gov.cn/report/gddjlist?entid=" + ent_id; Page sfxzgsxx_gqdjxx_list = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(sfxzgsxx_gqdjxx_list_url))); resultHtmlMap.put("sfxzgsxx_gqdjxx_list", sfxzgsxx_gqdjxx_list.getWebResponse().getContentAsString("utf-8")); /* //? ????->??->-> @SuppressWarnings("unchecked") List<HtmlAnchor> anchors4 = (List<HtmlAnchor>) sfxzgsxx_gqdjxx_list_page.getByXPath("//table[@id='touziren']/tbody[@id='table2']/tr/td/a"); if (anchors4!=null && !anchors4.isEmpty()) { List<String> detail=new ArrayList<String>(); for (@SuppressWarnings("unused") HtmlAnchor anchor4 : anchors4) { HtmlPage sfxzgsxx_gqdjxx_detail = anchor4.click(); // System.out.println("gsgsxx_qynb_detail.asXml()="+gsgsxx_qynb_detail.asXml()); detail.add(sfxzgsxx_gqdjxx_detail.asXml()); } resultHtmlMap.put("sfxzgsxx_gqdjxx_details",detail); }*/ //? ??->?? String qygsxx_gdbgxx_list_url = "http://tjcredit.gov.cn/report/gdbglist?entid=" + ent_id; Page qygsxx_gdbgxx_list = firstInfoPage.getWebClient().getPage(window, new WebRequest(new URL(qygsxx_gdbgxx_list_url))); resultHtmlMap.put("qygsxx_gdbgxx_list", qygsxx_gdbgxx_list.getWebResponse().getContentAsString("utf-8")); //? ??->??->-> /* @SuppressWarnings("unchecked") List<HtmlAnchor> anchors5 = (List<HtmlAnchor>) qygsxx_gdbgxx_page.getByXPath("//table[@id='touziren']/tbody[@id='table2']/tr/td/a"); if (anchors5!=null && !anchors5.isEmpty()) { List<String> detail=new ArrayList<String>(); for (@SuppressWarnings("unused") HtmlAnchor anchor5 : anchors5) { HtmlPage qygsxx_gdbgxx_detail = anchor5.click(); // System.out.println("gsgsxx_qynb_detail.asXml()="+gsgsxx_qynb_detail.asXml()); detail.add(qygsxx_gdbgxx_detail.asXml()); } resultHtmlMap.put("qygsxx_gdbgxx_details",detail); }*/ //*****************???? ?***************** } break;// } } if (!matchFlag) { resultHtmlMap.put("statusCodeDef", StatusCodeDef.NO_DATA_FOUND); LOGGER.info("????"); } } return resultHtmlMap; }
From source file:com.semfapp.adamdilger.semf.hazardIdActivity.java
public void createPdf() { Document documentTemplate = Pdf.getTemplate(getApplicationContext(), data.getProjectNumber()); try {/*from ww w . j a v a2 s. c o m*/ Document body = Jsoup.parse(getAssets().open("hazardIdentification.html"), "utf-8", "http://www.example.com"); Elements lists = body.select(".list_box"); //Lists html Elements ArrayList<String[]> arrayList = data.getArray(); //editText string arrays //for each Element in lists, add each bullet from arrayList.string[] as a <p> for (int x = 0; x < lists.size(); x++) { String f = ""; for (String bullet : arrayList.get(x)) { f += "<p>" + bullet + "</p>"; } lists.get(x).html(f); } documentTemplate.getElementById("main").html(body.html()); } catch (Exception e) { } //create filename / subject for email name = Emailer.getSubject(Emailer.HAZARD_ID_CODE, data.getProjectNumber()); String filePath = MainActivity.pdf.createFilePath(this, name); MainActivity.pdf.createPdfToFile(this, documentTemplate.html(), filePath, data.getImageArray()); pdfAttatchment = new File(filePath); }
From source file:com.semfapp.adamdilger.semf.NonConformanceActivity.java
public void createPdf() { Document document = Pdf.getTemplate(getApplicationContext(), data.getJobNumber()); try {// ww w . jav a 2 s . com Document body = Jsoup.parse(getAssets().open("nonConformance.html"), "utf-8", "http://www.example.com"); Element site = body.getElementById("site"); Element siteLocation = body.getElementById("site_location"); Element recipient = body.getElementById("recipient"); Element recipientEmail = body.getElementById("recipient_email"); Element description = body.getElementById("description_list"); Element actions = body.getElementById("actions_list"); String[] descriptionArray, actionsArray; descriptionArray = data.getDescription().split(System.lineSeparator()); actionsArray = data.getActions().split(System.lineSeparator()); site.text(data.getSite()); siteLocation.text(data.getLocation()); recipient.html("<p>" + data.getRecipient() + "</p>"); recipientEmail.html("<p>" + data.getRecipientEmail() + "</p>"); //add each bullet from arrays as a <p> for (int x = 0; x < descriptionArray.length; x++) { String f = ""; for (String bullet : descriptionArray) { f += "<p>" + bullet + "</p>"; } description.html(f); } for (int x = 0; x < actionsArray.length; x++) { String f = ""; for (String bullet : actionsArray) { f += "<p>" + bullet + "</p>"; } actions.html(f); } document.getElementById("main").html(body.html()); } catch (Exception e) { System.out.println("ERROR: " + e.toString()); } String filePath = MainActivity.pdf.createFilePath(this, "Non Conformance"); MainActivity.pdf.createPdfToFile(this, document.html(), filePath, null); pdfAttatchment = new File(filePath); }
From source file:com.semfapp.adamdilger.semf.protectPlanActivity.java
public void createPdf() { Document documentTemplate = null; Element body = null;// www .j a v a 2 s. c o m try { documentTemplate = Pdf.getTemplate(getApplicationContext(), null); body = Jsoup.parse(getAssets().open("protectPlan.html"), "utf-8", "http://www.example.com"); Elements lists = body.select(".list_box"); //Lists html Elements ArrayList<String[]> arrayList = data.getArray(); //editText string arrays //for each Element in lists, add each bullet from arrayList.string[] as a <p> for (int x = 0; x < lists.size(); x++) { String f = ""; for (String bullet : arrayList.get(x)) { f += "<p>" + bullet + "</p>"; } lists.get(x).html(f); } } catch (Exception e) { System.out.println("ERROR: " + e.toString()); } documentTemplate.getElementById("main").html(body.html()); String filePath = MainActivity.pdf.createFilePath(this, "Protect Plan"); MainActivity.pdf.createPdfToFile(this, documentTemplate.html(), filePath, null); pdfAttatchment = new File(filePath); }
From source file:com.semfapp.adamdilger.semf.SiteInstructionActivity.java
public void createPdf() { Document documentTemplate = Pdf.getTemplate(getApplicationContext(), data.getJobNumber()); try {/* w w w . j a v a2 s .c o m*/ Document body = Jsoup.parse(getAssets().open("siteInstruction.html"), "utf-8", "http://www.example.com"); Element site = body.getElementById("site"); Element siteLocation = body.getElementById("site_location"); Element recipient = body.getElementById("recipient"); Element recipientEmail = body.getElementById("recipient_email"); Element description = body.getElementById("description_list"); String[] descriptionArray; descriptionArray = data.getDescription().split(System.lineSeparator()); site.text(data.getSite()); siteLocation.text(data.getLocation()); recipient.html("<p>" + data.getRecipient() + "</p>"); recipientEmail.html("<p>" + data.getRecipientEmail() + "</p>"); //add each bullet from arrays as a <p> for (int x = 0; x < descriptionArray.length; x++) { String f = ""; for (String bullet : descriptionArray) { f += "<p>" + bullet + "</p>"; } description.html(f); } documentTemplate.getElementById("main").html(body.html()); } catch (Exception e) { System.out.println("ERROR: " + e.toString()); } name = Emailer.getSubject(Emailer.SITE_INSTRUCTION_CODE, data.getJobNumber()); String filePath = MainActivity.pdf.createFilePath(this, name); MainActivity.pdf.createPdfToFile(this, documentTemplate.html(), filePath, data.getImageArray()); pdfAttatchment = new File(filePath); }
From source file:im.ene.lab.attiq.ui.activities.ItemDetailActivity.java
@SuppressWarnings("unused") public void onEventMainThread(ItemDetailEvent event) { Article article = event.article;//www. j a v a 2 s .c om String userName = null; if (article != null) { mArticle = article; mRealm.beginTransaction(); mRealm.copyToRealmOrUpdate(mArticle); mRealm.commitTransaction(); User user = article.getUser(); mArticleName.setText(article.getTitle()); mSpannableTitle = new SpannableString(article.getTitle()); userName = user.getId(); mSpannableSubtitle = new SpannableString(userName); updateTitle(); buildArticleComments(article); buildArticleMenu(article); final String html; try { if (PrefUtil.isMathJaxEnabled()) { html = IOUtil.readAssets("html/article_mathjax.html"); } else { html = IOUtil.readAssets("html/article.html"); } Document doc = Jsoup.parse(html); Element elem = doc.getElementById("content"); elem.append(article.getRenderedBody()); String result = doc.outerHtml(); mContentView.loadDataWithBaseURL(article.getUrl(), result, "text/html", "utf-8", null); } catch (IOException e) { e.printStackTrace(); } } final CharSequence subTitle; if (article != null && !UIUtil.isEmpty(userName)) { subTitle = Html.fromHtml(getString(R.string.item_user_info, userName, userName, TimeUtil.beautify(article.getCreatedAt()))); } else { subTitle = getString(R.string.item_detail_subtitle, userName); } mArticleDescription.setText(subTitle); }
From source file:im.ene.lab.attiq.ui.activities.ItemDetailActivity.java
@SuppressWarnings("unused") public void onEventMainThread(ItemCommentsEvent event) { if (!UIUtil.isEmpty(event.comments)) { mCommentsView.setVisibility(View.VISIBLE); List<Comment> comments = event.comments; mCommentCount.setText(comments.size() + ""); String info = comments.size() == 1 ? getString(R.string.comment_singular) : getString(R.string.comment_plural); // FIXME should use plural strings mCommentInfo.setText(getString(R.string.article_comment, comments.size(), info)); final String html; try {/*from ww w . j a v a2 s. co m*/ html = IOUtil.readAssets("html/comments.html"); Document fullBody = Jsoup.parse(html); Element content = fullBody.getElementById("content"); for (Comment comment : comments) { String commentHtml = IOUtil.readAssets("html/comment.html"); commentHtml = commentHtml.replace("{user_icon_url}", comment.getUser().getProfileImageUrl()) .replace("{user_name}", comment.getUser().getId()) .replace("{comment_time}", TimeUtil.commentTime(comment.getCreatedAt())) .replace("{article_uuid}", mItemUuid).replace("{comment_id}", comment.getId()); Document commentDoc = Jsoup.parse(commentHtml); Element eComment = commentDoc.getElementsByClass("comment-box").first(); eComment.getElementsByClass("message").first().append(comment.getRenderedBody()); // remove comment edit block if it is not from current user if (mMyProfile == null || !mMyProfile.getId().equals(comment.getUser().getId())) { String commentId = "comment_{comment_id}_{user_name}" .replace("{comment_id}", comment.getId()) .replace("{user_name}", comment.getUser().getId()); Element commentEditor = commentDoc.getElementById(commentId); commentEditor.remove(); } content.appendChild(eComment); } String result = fullBody.outerHtml(); mCommentsView.loadDataWithBaseURL("http://qiita.com/", result, null, null, null); } catch (IOException e) { e.printStackTrace(); } } else { mCommentCount.setText("0"); mCommentInfo.setText(getString(R.string.article_comment, 0, getString(R.string.comment_plural))); mCommentsView.setVisibility(View.GONE); } }