diff --git a/match_program/src/main/java/com/ossean/match/matchprocess/Match.java b/match_program/src/main/java/com/ossean/match/matchprocess/Match.java index e229c989d..576af3f9a 100644 --- a/match_program/src/main/java/com/ossean/match/matchprocess/Match.java +++ b/match_program/src/main/java/com/ossean/match/matchprocess/Match.java @@ -75,10 +75,12 @@ public class Match { // 项目名匹配帖子标题 matchMap = LuceneSearch.prjToMemoMatchByLucene(prjName, LuceneIndex.titleFieldName, titleWeight, matchMap, memoIndexReader); +System.out.println("项目名匹配帖子标题"+matchMap.size()); // 项目名匹配帖子标签 matchMap = LuceneSearch.searchMemoTags( prjName, LuceneIndex.memoTagsFieldName, LuceneIndex.memoIdFieldName, memoTagsWeight, matchMap, memoIndexReader); +System.out.println("项目名匹配帖子标签"+matchMap.size()); } if (project.getSynonyms() != null) { synonyms = project.getSynonyms().toLowerCase(); @@ -89,12 +91,14 @@ public class Match { matchMap = LuceneSearch.prjToMemoMatchByLucene(synonym, LuceneIndex.titleFieldName, titleWeight, matchMap, memoIndexReader); } +System.out.println("项目别名匹配帖子标题 "+matchMap.size()); //项目别名匹配帖子标签 TODO for (String synonym : synonymsList) { matchMap = LuceneSearch.searchMemoTags( synonym, LuceneIndex.memoTagsFieldName, LuceneIndex.memoIdFieldName, memoTagsWeight, matchMap, memoIndexReader); } +System.out.println("项目别名匹配帖子标签 "+matchMap.size()); } } if (prjTags != null && prjTags.length() > 0 && matchMap.size() > 0) { @@ -105,12 +109,14 @@ public class Match { prjTag, LuceneIndex.titleFieldName, LuceneIndex.memoIdFieldName, prjTagsToMemoTitleWeight, matchMap, memoIndexReader); } +System.out.println("用项目标签搜索帖子标题 "+matchMap.size()); // 用项目标签搜索帖子标签 for (String prjTag : prjTagsList) { matchMap = LuceneSearch.searchByPrjTag( prjTag, LuceneIndex.memoTagsFieldName, LuceneIndex.memoIdFieldName, prjTagsToMemoTagsWeight, matchMap, memoIndexReader); } +System.out.println("用项目标签搜索帖子标签 "+matchMap.size()); } if (matchMap.size() > 0) insertPrjToMemoMatchResult(prjId, matchMap); @@ -229,9 +235,9 @@ public class Match { memoTags = memo.getTags().toLowerCase(); } memoTitle = memo.getTitle().toLowerCase(); - if (memoTitle.length() == 0) { - continue; - } +// if (memoTitle.length() == 0) { +// continue; +// } if (Normalizer.isAllNumber(memoTitle)) { continue; } diff --git a/match_program/src/main/java/com/ossean/match/matchprocess/MatchIncrement.java b/match_program/src/main/java/com/ossean/match/matchprocess/MatchIncrement.java index c12e89713..a6449711e 100644 --- a/match_program/src/main/java/com/ossean/match/matchprocess/MatchIncrement.java +++ b/match_program/src/main/java/com/ossean/match/matchprocess/MatchIncrement.java @@ -13,56 +13,16 @@ public class MatchIncrement extends Thread implements Runnable { private Logger logger = LoggerFactory.getLogger(getClass()); private int memoHistory; // the record of memo after matching last time private static File record; // record file "record.txt" - private boolean newPrjComing; //新项目进入标识 - private boolean newMemoComing; //新帖子进入标识 - private boolean currentMatchDone = true; //当前批量匹配是否结束 + private boolean newPrjComing; // 新项目进入标识 + private boolean newMemoComing; // 新帖子进入标识 + private boolean currentMatchDone = true; // 当前批量匹配是否结束 private int sleepTime = 0; - - public boolean isNewPrjComing() { - return newPrjComing; - } - - public void setNewPrjComing(boolean newPrjComing) { - this.newPrjComing = newPrjComing; - } - - public boolean isNewMemoComing() { - return newMemoComing; - } - - public void setNewMemoComing(boolean newMemoComing) { - this.newMemoComing = newMemoComing; - } - - public int getSleepTime() { - return sleepTime; - } - - public void setSleepTime(int sleepTime) { - this.sleepTime = sleepTime; - } public MatchIncrement() { record = new File("record.txt"); getMatchRecord(); } - public int getMemoHistory() { - return memoHistory; - } - - public synchronized void setMemoHistory(int memoHistory) { - this.memoHistory = memoHistory; - } - - public boolean isCurrentMatchDone() { - return currentMatchDone; - } - - public synchronized void setCurrentMatchDone(boolean currentMatchDone) { - this.currentMatchDone = currentMatchDone; - } - /** * read the record after matching last time from record file "record.txt", * the format of the record in the file: "project's record" "memo's record" @@ -91,4 +51,45 @@ public class MatchIncrement extends Thread implements Runnable { logger.error("writeIntoRecord FileNotFoundException： " + e); } } + + public boolean isNewPrjComing() { + return newPrjComing; + } + + public void setNewPrjComing(boolean newPrjComing) { + this.newPrjComing = newPrjComing; + } + + public boolean isNewMemoComing() { + return newMemoComing; + } + + public void setNewMemoComing(boolean newMemoComing) { + this.newMemoComing = newMemoComing; + } + + public int getSleepTime() { + return sleepTime; + } + + public void setSleepTime(int sleepTime) { + this.sleepTime = sleepTime; + } + + public int getMemoHistory() { + return memoHistory; + } + + public synchronized void setMemoHistory(int memoHistory) { + this.memoHistory = memoHistory; + } + + public boolean isCurrentMatchDone() { + return currentMatchDone; + } + + public synchronized void setCurrentMatchDone(boolean currentMatchDone) { + this.currentMatchDone = currentMatchDone; + } + } \ No newline at end of file diff --git a/match_program/src/main/java/com/ossean/match/utils/Normalizer.java b/match_program/src/main/java/com/ossean/match/utils/Normalizer.java index 95a8c6627..76ce6bb8b 100644 --- a/match_program/src/main/java/com/ossean/match/utils/Normalizer.java +++ b/match_program/src/main/java/com/ossean/match/utils/Normalizer.java @@ -64,15 +64,17 @@ public class Normalizer { } public static void main(String []args) { - String content = "大声道Corvette123 *&%9&^&%&^%(*_asd's_12 L'Aur--ore, 1766-1775"; - int a = 1 + 11000/11000; - String targetTableName = "relative_memo_to_open_source_projects_" + a; - //System.out.println(StringFilter(content)); - String result = content.replaceAll("[^0-9a-zA-Z]"," "); - System.out.println(result); - System.out.println(normalize(content)); - System.out.println(content.lastIndexOf(" ")); - //System.out.println(StringFilter(content)); +// String content = "大声道Corvette123 *&%9&^&%&^%(*_asd's_12 L'Aur--ore, 1766-1775"; +// int a = 1 + 11000/11000; +// String targetTableName = "relative_memo_to_open_source_projects_" + a; +// //System.out.println(StringFilter(content)); +// String result = content.replaceAll("[^0-9a-zA-Z]"," "); +// System.out.println(result); +// System.out.println(normalize(content)); +// System.out.println(content.lastIndexOf(" ")); +// //System.out.println(StringFilter(content)); + String a = "'One' PHP cache"; + System.out.println(getList(a)); } /** * if a string only contains digital