From c8bef3906290f6fc31a7b9bba57e5a33fd5658ff Mon Sep 17 00:00:00 2001 From: ronger <1207296860@qq.com> Date: Tue, 31 Jan 2017 12:44:37 +0800 Subject: [PATCH] add check openhub prj to MergerProjectsNew2.java --- .../com/ossean/util/MergeProjectNew2.java | 27 +++++++++++++++---- .../java/com/ossean/util/URLAvailability.java | 4 +-- 2 files changed, 24 insertions(+), 7 deletions(-) diff --git a/project_match/src/main/java/com/ossean/util/MergeProjectNew2.java b/project_match/src/main/java/com/ossean/util/MergeProjectNew2.java index 0f11fa33b..6257f8b01 100644 --- a/project_match/src/main/java/com/ossean/util/MergeProjectNew2.java +++ b/project_match/src/main/java/com/ossean/util/MergeProjectNew2.java @@ -88,10 +88,12 @@ public class MergeProjectNew2 { if(prjHomepage == null || prjHomepage.equals("")) if(model.getSource().equals("apache")) prjHomepage = model.getUrl(); - // if(model.getSource().equals("openhub")) - // if(!isRightUrlByRegex(model.getHomepage())){ - // prjHomepage=null; - // } + //暂时性对openhub项目进行处理 + if(model.getSource().equals("openhub")){ + if(!isRightUrlByRegex(model.getHomepage())||includeSubStringNum(model.getHomepage(), "http")>1){ + prjHomepage=null; + } + } //homepage相同认为一定是相同的项目 if(prjHomepage != null && !prjHomepage.equals("")){ prjHomepage = deleteHttpPre(prjHomepage);//去掉“http&https”前缀及“/”后缀 @@ -422,5 +424,20 @@ public class MergeProjectNew2 { return true; } return false; - } + } + + public int includeSubStringNum(String str,String subStr){ + int count = 0; + if (str.indexOf(subStr)==-1) + { + return 0; + } + else if(str.indexOf(subStr) != -1) + { + count++; + count += includeSubStringNum(str.substring(str.indexOf(subStr)+subStr.length()),subStr); + return count; + } + return 0; + } } diff --git a/project_match/src/main/java/com/ossean/util/URLAvailability.java b/project_match/src/main/java/com/ossean/util/URLAvailability.java index 7ac814c01..fbbfa0507 100644 --- a/project_match/src/main/java/com/ossean/util/URLAvailability.java +++ b/project_match/src/main/java/com/ossean/util/URLAvailability.java @@ -62,8 +62,8 @@ public class URLAvailability { public static void main(String[] args) { //URLAvailability u = new URLAvailability(); - System.out.println(isConnect2("http://www.oschina.nethttps://market.aliyun.com/software?spm=5176.1846502.0.0.CpQnwc")); - //System.out.println(isRightUrlByRegex("https://www.openhub.net/p/tesseract")); + //System.out.println(isConnect2("http://www.oschina.nethttps://market.aliyun.com/software?spm=5176.1846502.0.0.CpQnwc")); + System.out.println(isRightUrlByRegex("https://www.openhub.net/p/tesseracthttp://abc.com")); //System.out.println(isConnect("http://www.freecode.com/projects/pdfsam")); }