Ticket #245: qp-allstop.patch

queryparser/queryparser.lemony

@@ -174 +174 @@
     QueryParser::Internal * qpi;
 
   public:
+    const Stopper * stopper;
     Query query;
     const char * error;
     unsigned flags;
 
     State(QueryParser::Internal * qpi_, unsigned flags_)
-        : qpi(qpi_), error(NULL), flags(flags_) { }
+        : qpi(qpi_), stopper(qpi->stopper), error(NULL), flags(flags_) { }
 
     string stem_term(const string &term) {
         return qpi->stemmer(term);
@@ -220 +221 @@
     Query::op default_op() const { return qpi->default_op; }
 
     bool is_stopword(const Term *term) const {
-        return qpi->stopper && (*qpi->stopper)(term->name);
+        return stopper && (*stopper)(term->name);
     }
 
     Database get_database() const {
@@ -547 +548 @@
     bool value_ranges;
     value_ranges = !valrangeprocs.empty() && (qs.find("..") != string::npos);
 
+    State state(this, flags);
+
+reparse_without_stopwords:
     termpos term_pos = 1;
     Utf8Iterator it(qs), end;
 
-    State state(this, flags);
-
     // To successfully apply more than one spelling correction to a query
     // string, we must keep track of the offset due to previous corrections.
     int correction_offset = 0;
@@ -1029 +1031 @@
     if (mode == IN_QUOTES || mode == IN_PREFIXED_QUOTES)
         Parse(pParser, QUOTE, NULL, &state);
     Parse(pParser, 0, NULL, &state);
+
+    if (!state.error && state.query.empty() && state.stopper &&
+        !stoplist.empty()) {
+        // All terms were stopwords so reparse with stopwords disabled.
+        stoplist.clear();
+        unstem.clear();
+        state.stopper = NULL;
+        goto reparse_without_stopwords;
+    }
     ParseFree(pParser);
 
     errmsg = state.error;
@@ -1447 +1458 @@
         }
         *E = Query(Query::OP_AND_NOT, *E, *P->hate);
     }
-    // FIXME what if E && E->empty() (all terms are stopwords)?
     delete P;
 }
 

tests/queryparsertest.cc

@@ -1007 +1007 @@
     // parse.
     { "test AND the AND queryparser", "(test:(pos=1) AND the:(pos=2) AND queryparser:(pos=3))" },
     // 0.9.6 and earlier ignored a stopword even if it was the only term.
-    // We don't ignore it in this case, which is probably better.  But
-    // an all-stopword query with multiple terms doesn't work, which
-    // prevents 'to be or not to be' for being searchable unless made
-    // into a phrase query.
+    // More recent versions don't ever treat a single term as a stopword.
     { "the", "the:(pos=1)" },
+    // 1.0.5 and earlier ignored an all-stopword query with multiple terms,
+    // which prevents 'to be or not to be' for being searchable unless made
+    // into a phrase query by the user.
+    { "an the a", "(an:(pos=1) AND the:(pos=2) AND a:(pos=3))" },
+    // FIXME: this still doesn't work...
+    { "the AND a an", "" },
     { NULL, NULL }
 };