| Index: branches/lucene-search-2.1/src/org/wikimedia/lsearch/search/SearchEngine.java |
| — | — | @@ -96,7 +96,7 @@ |
| 97 | 97 | if (what.equals("search") || what.equals("explain")) { |
| 98 | 98 | int offset = 0, limit = 20; boolean exactCase = false; |
| 99 | 99 | int iwlimit = 10; int iwoffset = 0; |
| 100 | | - boolean searchOnly = false; |
| | 100 | + boolean searchOnly = false, searchAll = false; |
| 101 | 101 | if (query.containsKey("offset")) |
| 102 | 102 | offset = Math.max(Integer.parseInt((String)query.get("offset")), 0); |
| 103 | 103 | if (query.containsKey("limit")) |
| — | — | @@ -109,10 +109,17 @@ |
| 110 | 110 | exactCase = true; |
| 111 | 111 | if(query.containsKey("searchonly")) |
| 112 | 112 | searchOnly = Boolean.parseBoolean((String)query.get("searchonly")); |
| | 113 | + if(query.containsKey("searchall")) |
| | 114 | + searchAll = Boolean.parseBoolean((String)query.get("searchall")) || |
| | 115 | + ((String)query.get("searchall")).equals("1"); |
| 113 | 116 | if(version <= 2) |
| 114 | 117 | searchOnly = true; |
| 115 | | - NamespaceFilter namespaces = new NamespaceFilter((String)query.get("namespaces")); |
| 116 | | - SearchResults res = search(iid, searchterm, offset, limit, iwoffset, iwlimit, namespaces, what.equals("explain"), exactCase, false, searchOnly); |
| | 118 | + NamespaceFilter namespaces = null; |
| | 119 | + if(searchAll) |
| | 120 | + namespaces = new NamespaceFilter(""); |
| | 121 | + else |
| | 122 | + namespaces = new NamespaceFilter((String)query.get("namespaces")); |
| | 123 | + SearchResults res = search(iid, searchterm, offset, limit, iwoffset, iwlimit, namespaces, what.equals("explain"), exactCase, false, searchOnly, searchAll); |
| 117 | 124 | if(!res.isSuccess()){ |
| 118 | 125 | // note failed search |
| 119 | 126 | if(SearchServer.stats != null) |
| — | — | @@ -132,6 +139,7 @@ |
| 133 | 140 | } else if (what.equals("raw") || what.equals("rawexplain")) { |
| 134 | 141 | int offset = 0, limit = 20; boolean exactCase = false; |
| 135 | 142 | int iwlimit = 10; int iwoffset = 0; |
| | 143 | + boolean searchAll = false; |
| 136 | 144 | if (query.containsKey("offset")) |
| 137 | 145 | offset = Math.max(Integer.parseInt((String)query.get("offset")), 0); |
| 138 | 146 | if (query.containsKey("limit")) |
| — | — | @@ -142,8 +150,10 @@ |
| 143 | 151 | iwlimit = Math.min(Integer.parseInt((String)query.get("iwlimit")), MAXLINES); |
| 144 | 152 | if (query.containsKey("case") && global.exactCaseIndex(iid.getDBname()) && ((String)query.get("case")).equalsIgnoreCase("exact")) |
| 145 | 153 | exactCase = true; |
| | 154 | + if(query.containsKey("searchall")) |
| | 155 | + searchAll = Boolean.parseBoolean((String)query.get("searchall")); |
| 146 | 156 | NamespaceFilter namespaces = new NamespaceFilter((String)query.get("namespaces")); |
| 147 | | - return search(iid, searchterm, offset, limit, iwoffset, iwlimit, namespaces, what.equals("rawexplain"), exactCase, true, true); |
| | 157 | + return search(iid, searchterm, offset, limit, iwoffset, iwlimit, namespaces, what.equals("rawexplain"), exactCase, true, true, searchAll); |
| 148 | 158 | } else if (what.equals("prefix")){ |
| 149 | 159 | int limit = MAXPREFIX; |
| 150 | 160 | if (query.containsKey("limit")) |
| — | — | @@ -577,7 +587,8 @@ |
| 578 | 588 | * the default namespaces filter |
| 579 | 589 | */ |
| 580 | 590 | public SearchResults search(IndexId iid, String searchterm, int offset, int limit, int iwoffset, int iwlimit, |
| 581 | | - NamespaceFilter nsDefault, boolean explain, boolean exactCase, boolean raw, boolean searchOnly){ |
| | 591 | + NamespaceFilter nsDefault, boolean explain, boolean exactCase, boolean raw, boolean searchOnly, |
| | 592 | + boolean searchAllFromRequest){ |
| 582 | 593 | Analyzer analyzer = Analyzers.getSearcherAnalyzer(iid,exactCase); |
| 583 | 594 | if(nsDefault == null || nsDefault.cardinality() == 0) |
| 584 | 595 | nsDefault = new NamespaceFilter("0"); // default to main namespace |
| — | — | @@ -599,8 +610,12 @@ |
| 600 | 611 | return res; |
| 601 | 612 | } |
| 602 | 613 | |
| | 614 | + // check if request is explicitely on one field |
| | 615 | + if(searchAllFromRequest){ |
| | 616 | + nsfw.setNamespaceFilter(new NamespaceFilter()); |
| | 617 | + searchAll = true; |
| 603 | 618 | // if search is over one field, try to use filters |
| 604 | | - if(fields.size()==1){ |
| | 619 | + } else if(fields.size()==1){ |
| 605 | 620 | if(fields.contains(new NamespaceFilter())){ |
| 606 | 621 | nsfw.setNamespaceFilter(new NamespaceFilter()); // empty filter: "all" keyword |
| 607 | 622 | searchAll = true; |
| Index: branches/lucene-search-2.1/src/org/wikimedia/lsearch/analyzers/WikiQueryParser.java |
| — | — | @@ -1035,33 +1035,38 @@ |
| 1036 | 1036 | return query; |
| 1037 | 1037 | } |
| 1038 | 1038 | |
| 1039 | | - /** return true if buffer is wildcard */ |
| | 1039 | + /** |
| | 1040 | + * return true if buffer is wildcard |
| | 1041 | + * the only allowed patterns are *q and q* and not other combinations like *q* or q*r |
| | 1042 | + * |
| | 1043 | + */ |
| 1040 | 1044 | private boolean bufferIsWildCard(){ |
| 1041 | | - if(length < 1) |
| | 1045 | + if(length < 2) |
| 1042 | 1046 | return false; |
| 1043 | 1047 | boolean wild = false; |
| 1044 | 1048 | int index = -1; |
| 1045 | | - for(int i=0;i<length;i++){ |
| 1046 | | - if(buffer[i] == '*'){ |
| 1047 | | - wild = true; |
| 1048 | | - index = i; |
| 1049 | | - break; |
| 1050 | | - } |
| | 1049 | + // only allow '*' at begin and end |
| | 1050 | + if(buffer[0] == '*'){ |
| | 1051 | + index = 0; |
| | 1052 | + wild = true; |
| | 1053 | + } else if( buffer[length-1] == '*' ){ |
| | 1054 | + index = length-1; |
| | 1055 | + wild = true; |
| 1051 | 1056 | } |
| | 1057 | + |
| 1052 | 1058 | // check if it's a valid wildcard |
| 1053 | 1059 | if(wild){ |
| 1054 | | - if((buffer[0] == '*') && (buffer[length-1]=='*')) |
| 1055 | | - return false; // don't support patterns like *a* |
| 1056 | | - //if(index == length-1 && buffer[index]=='?') |
| 1057 | | - // return false; // probably just an ordinary question mark |
| | 1060 | + // check if this is the only asterix |
| | 1061 | + for(int i=0;i<length;i++){ |
| | 1062 | + if( i!= index && buffer[i] == '*'){ |
| | 1063 | + return false; // more than one '*' |
| | 1064 | + } |
| | 1065 | + } |
| 1058 | 1066 | |
| 1059 | | - // don't let * be in middle for performance reasons |
| 1060 | | - if(Character.isLetterOrDigit(buffer[0]) && Character.isLetterOrDigit(buffer[length-1])) |
| 1061 | | - return false; |
| 1062 | | - |
| | 1067 | + // require at least one letter besides the wildcard sign |
| 1063 | 1068 | for(int i=0;i<length;i++){ |
| 1064 | 1069 | if(Character.isLetterOrDigit(buffer[i])) |
| 1065 | | - return true; // +card :P |
| | 1070 | + return true; // found it! |
| 1066 | 1071 | } |
| 1067 | 1072 | } |
| 1068 | 1073 | return false; |