Ticket #128: qp_exact_text.patch

xapian-bindings/xapian.i

@@ -955 +955 @@
     Query parse_query(const string &q, unsigned flags);
 
     void add_prefix(const std::string &field, const std::string &prefix);
+    void add_exact_prefix(const std::string &field, const std::string &prefix);
     void add_boolean_prefix(const std::string & field, const std::string &prefix);
 
     TermIterator stoplist_begin() const;

xapian-core/queryparser/queryparser.lemony

@@ -461 +461 @@
                               new Term(prefix, 0), &state);
                         continue;
                     }
+                    
+                    if (type == PrefixInfo::EXACT_TEXT) {
+                        it = p;
+                        if (prefix_needs_colon(prefix, *it))
+                            prefix += ':';
+                        field += ':';
+                        string term;
 
+                        if (*it == '"') {
+                            // Quoted term: continue until unescaped end quote
+                            field += *it;
+                            ++it;
+                            while (it != end && *it != '"') {
+                                field += *it;
+                                if (*it == '\\') {
+                                    ++it;
+                                    if (it == end) {
+                                        term.append(ubuf, to_utf8('\\' , ubuf));
+                                    } else if (*it == '"' || *it == '\\') {
+                                        term.append(ubuf, to_utf8(*it++, ubuf));
+                                    } else {
+                                        term.append(ubuf, to_utf8('\\' , ubuf));
+                                        term.append(ubuf, to_utf8(*it++, ubuf));
+                                    }
+                                } else {
+                                    term.append(ubuf, to_utf8(*it++, ubuf));
+                                }
+                            }
+                            if (it != end)
+                                ++it;
+                        } else {
+                            // Non-quoted term: continue until space or ')'.
+                            while (it != end && *it > ' ' && *it != ')')
+                                term.append(ubuf, to_utf8(*it++, ubuf));
+                            field += term;
+                        }
+
+                        prefix += term;
+                        unstem.insert(make_pair(prefix, field));
+                        Parse(pParser, TERM,
+                              new Term(prefix, term_pos++), &state);
+                        continue;
+                    }
+
                     Assert(type == PrefixInfo::FREE_TEXT);
 
                     if (ch == '"' && (flags & FLAG_PHRASE)) {
@@ -475 +518 @@
                         continue;
                     }
 
+
                     if (ch == '(' && (flags & FLAG_BOOLEAN)) {
                         // Prefixed subexpression, e.g.: title:(fast NEAR food)
                         Parse(pParser, BRA, NULL, &state);
@@ -938 +982 @@
     if (P->hate.can_match()) {
         if (E->empty()) {
             delete E;
-            // Can't just hate!
-            yy_parse_failed(yypParser);
-            return;
+            if (state->flags & QueryParser::FLAG_PURE_NOT) {
+                E = new QpQuery("", 1, 0);
+            } else {
+                // Can't just hate!
+                yy_parse_failed(yypParser);
+                return;
+            }
         }
         *E = QpQuery(Query::OP_AND_NOT, *E, P->hate);
     }

xapian-core/queryparser/queryparser_internal.h

@@ -37 +37 @@
 struct PrefixInfo {
     typedef enum {
         FREE_TEXT,
+        EXACT_TEXT,
         BOOL_FILTER
     } prefix_type;
     PrefixInfo::prefix_type type;

xapian-core/queryparser/queryparser.cc

@@ -130 +130 @@
 }
 
 void
+QueryParser::add_exact_prefix(const std::string &field,
+                              const std::string &prefix)
+{
+    internal->prefixes.insert(
+        make_pair(field, PrefixInfo(PrefixInfo::EXACT_TEXT, prefix)));
+}
+
+void
 QueryParser::add_boolean_prefix(const std::string &field,
                                 const std::string &prefix)
 {

xapian-core/tests/queryparsertest.cc

@@ -87 +87 @@
     { "-site:xapian.org mail", "(mail:(pos=1) AND_NOT Hxapian.org)" },
     { "site:xapian.org", "Hxapian.org" },
     { "mug +site:xapian.org -site:cvs.xapian.org", "((mug:(pos=1) AND_NOT Hcvs.xapian.org) FILTER Hxapian.org)" },
+    { "exact:xapian.org exact:cvs.xapian.org", "(Exapian.org:(pos=1) OR Ecvs.xapian.org:(pos=2))" },
+    { "+exact:xapian.org -exact:cvs.xapian.org", "(Exapian.org:(pos=1) AND_NOT Ecvs.xapian.org:(pos=2))" },
+    { "-exact:cvs.xapian.org", "Ecvs.xapian.org:(pos=1)" },
+    { "mug +exact:xapian.org", "(Exapian.org:(pos=2) AND_MAYBE mug:(pos=1))" },
     { "mug -site:cvs.xapian.org +site:xapian.org", "((mug:(pos=1) AND_NOT Hcvs.xapian.org) FILTER Hxapian.org)" },
     { "NOT windows", "Syntax: <expression> NOT <expression>" },
     { "a AND (NOT b)", "Syntax: <expression> NOT <expression>" },
@@ -564 +568 @@
     { "AND NOT windows", "Syntax: <expression> AND NOT <expression>" },
     { "gordian NOT", "Syntax: <expression> NOT <expression>" },
     { "gordian AND NOT", "Syntax: <expression> AND NOT <expression>" },
+    { "NOT exact:cvs.xapian.org", "(<alldocuments> AND_NOT Ecvs.xapian.org:(pos=1))" },
+    { "-exact:cvs.xapian.org", "(<alldocuments> AND_NOT Ecvs.xapian.org:(pos=1))" },
+    { "-exact:\"hello world\"", "(<alldocuments> AND_NOT Ehello world:(pos=1))" },
+    { "-exact:help", "(<alldocuments> AND_NOT Ehelp:(pos=1))" },
     { NULL, NULL }
 };
 
@@ -576 +584 @@
     queryparser.add_prefix("title", "XT");
     queryparser.add_prefix("subject", "XT");
     queryparser.add_boolean_prefix("site", "H");
+    queryparser.add_exact_prefix("exact", "E");
     for (test *p = test_or_queries; p->query; ++p) {
         string expect, parsed;
         if (p->expect)
@@ -845 +854 @@
     Xapian::QueryParser qp;
     qp.set_stemmer(Xapian::Stem("english"));
     qp.set_stemming_strategy(QueryParser::STEM_SOME);
+    qp.add_exact_prefix("exact", "E");
     for (test *p = test_pure_not_queries; p->query; ++p) {
         string expect, parsed;
         if (p->expect)
@@ -854 +864 @@
         try {
             Xapian::Query qobj = qp.parse_query(p->query,
                                                 QueryParser::FLAG_BOOLEAN |
-                                                QueryParser::FLAG_PURE_NOT);
+                                                QueryParser::FLAG_PURE_NOT |
+                                                QueryParser::FLAG_LOVEHATE);
             parsed = qobj.get_description();
             expect = string("Xapian::Query(") + expect + ')';
         } catch (const Xapian::Error &e) {

xapian-core/include/xapian/queryparser.h

@@ -228 +228 @@
      */
     void add_prefix(const std::string &field, const std::string &prefix);
 
+    /**
+     */
+    void add_exact_prefix(const std::string &field, const std::string &prefix);
+
     /** Add a boolean term prefix allowing the user to restrict a
      *  search with a boolean filter specified in the free text query.
      *