SERVER-19510 Move text query parsing to TextMatchExpression::init()

- Introduces FTSQuery, which is now the base class for FTSQueryImpl.
- Introduces a derived class FTSQueryNoop (which
  TextNoOpMatchExpression now wraps).  libfts_query_noop is now linked
  into db/matcher/expressions.
- TextMatchExpression now parses the text query (which acquires a
  collection lock as part of the parsing process), and TextNode now
  stores a parsed version of the query.  The FTSQuery::parse() call in
  buildStages() is removed.

Behavior change: $text against a non-existent collection now returns
an error, instead of an empty result set.

9 files changed, 121 insertions, 91 deletions

diff --git a/src/mongo/db/matcher/SConscript b/src/mongo/db/matcher/SConscript
index af686b9f96b..d8c4fbd8299 100644
--- a/src/mongo/db/matcher/SConscript
+++ b/src/mongo/db/matcher/SConscript
@@ -47,6 +47,7 @@ env.Library(
         '$BUILD_DIR/mongo/base',
         '$BUILD_DIR/mongo/bson/util/bson_extract',
         '$BUILD_DIR/mongo/db/common',
+        '$BUILD_DIR/mongo/db/fts/fts_query_noop',
         '$BUILD_DIR/third_party/shim_pcrecpp',
         'path',
     ],
@@ -123,26 +124,19 @@ env.CppUnitTest(
 )
 
 env.Library(
-    target='expressions_text',
-    source=[
-        'expression_text.cpp',
-    ],
-    LIBDEPS=[
-        '$BUILD_DIR/mongo/db/fts/base',
-        'expressions',
-    ],
-)
-
-env.Library(
     target='expressions_mongod_only',
     source=[
         'extensions_callback_real.cpp',
+        'expression_text.cpp',
         'expression_where.cpp',
     ],
     LIBDEPS=[
         '$BUILD_DIR/mongo/db/auth/authorization_manager_global',
         '$BUILD_DIR/mongo/scripting/scripting_server',
         'expressions',
-        'expressions_text',
+    ],
+    LIBDEPS_TAGS=[
+        # Depends on symbols from serverOnlyFiles.
+        'incomplete',
     ],
 )
diff --git a/src/mongo/db/matcher/expression_text.cpp b/src/mongo/db/matcher/expression_text.cpp
index 8ee279efc1d..b97c2ce2780 100644
--- a/src/mongo/db/matcher/expression_text.cpp
+++ b/src/mongo/db/matcher/expression_text.cpp
@@ -32,34 +32,76 @@
 
 #include "mongo/db/matcher/expression_text.h"
 
+#include "mongo/db/catalog/collection.h"
+#include "mongo/db/db_raii.h"
 #include "mongo/db/fts/fts_language.h"
+#include "mongo/db/fts/fts_spec.h"
+#include "mongo/db/index/fts_access_method.h"
 #include "mongo/stdx/memory.h"
 
 namespace mongo {
 
-TextMatchExpression::TextMatchExpression(TextParams params)
-    : TextMatchExpressionBase(std::move(params)) {}
+Status TextMatchExpression::init(OperationContext* txn,
+                                 const NamespaceString& nss,
+                                 TextParams params) {
+    _ftsQuery.setQuery(std::move(params.query));
+    _ftsQuery.setLanguage(std::move(params.language));
+    _ftsQuery.setCaseSensitive(params.caseSensitive);
+    _ftsQuery.setDiacriticSensitive(params.diacriticSensitive);
 
-Status TextMatchExpression::init() {
-    // Validate language, but defer construction of FTSQueryImpl (which requires access to the
-    // target namespace) until stage building time.
-    if (!getLanguage().empty()) {
-        if (!fts::FTSLanguage::make(getLanguage(), fts::TEXT_INDEX_VERSION_2).isOK()) {
-            return {ErrorCodes::BadValue, "$language specifies unsupported language"};
+    fts::TextIndexVersion version;
+    {
+        // Find text index.
+        ScopedTransaction transaction(txn, MODE_IS);
+        AutoGetDb autoDb(txn, nss.db(), MODE_IS);
+        Lock::CollectionLock collLock(txn->lockState(), nss.ns(), MODE_IS);
+        Database* db = autoDb.getDb();
+        if (!db) {
+            return {ErrorCodes::IndexNotFound,
+                    str::stream() << "text index required for $text query (no such collection '"
+                                  << nss.ns() << "')"};
         }
+        Collection* collection = db->getCollection(nss);
+        if (!collection) {
+            return {ErrorCodes::IndexNotFound,
+                    str::stream() << "text index required for $text query (no such collection '"
+                                  << nss.ns() << "')"};
+        }
+        std::vector<IndexDescriptor*> idxMatches;
+        collection->getIndexCatalog()->findIndexByType(txn, IndexNames::TEXT, idxMatches);
+        if (idxMatches.empty()) {
+            return {ErrorCodes::IndexNotFound, "text index required for $text query"};
+        }
+        if (idxMatches.size() > 1) {
+            return {ErrorCodes::IndexNotFound, "more than one text index found for $text query"};
+        }
+        invariant(idxMatches.size() == 1);
+        IndexDescriptor* index = idxMatches[0];
+        const FTSAccessMethod* fam =
+            static_cast<FTSAccessMethod*>(collection->getIndexCatalog()->getIndex(index));
+        invariant(fam);
+
+        // Extract version and default language from text index.
+        version = fam->getSpec().getTextIndexVersion();
+        if (_ftsQuery.getLanguage().empty()) {
+            _ftsQuery.setLanguage(fam->getSpec().defaultLanguage().str());
+        }
+    }
+
+    Status parseStatus = _ftsQuery.parse(version);
+    if (!parseStatus.isOK()) {
+        return parseStatus;
     }
 
     return initPath("_fts");
 }
 
 std::unique_ptr<MatchExpression> TextMatchExpression::shallowClone() const {
-    TextParams params;
-    params.query = getQuery();
-    params.language = getLanguage();
-    params.caseSensitive = getCaseSensitive();
-    params.diacriticSensitive = getDiacriticSensitive();
-    auto expr = stdx::make_unique<TextMatchExpression>(std::move(params));
-    expr->init();
+    auto expr = stdx::make_unique<TextMatchExpression>();
+    // We initialize _ftsQuery here directly rather than calling init(), to avoid needing to examine
+    // the index catalog.
+    expr->_ftsQuery = _ftsQuery;
+    invariantOK(expr->initPath("_fts"));
     if (getTag()) {
         expr->setTag(getTag()->clone());
     }
diff --git a/src/mongo/db/matcher/expression_text.h b/src/mongo/db/matcher/expression_text.h
index dfe4cbd8204..3bd44b29da3 100644
--- a/src/mongo/db/matcher/expression_text.h
+++ b/src/mongo/db/matcher/expression_text.h
@@ -30,17 +30,27 @@
 
 #pragma once
 
+#include "mongo/db/fts/fts_query_impl.h"
 #include "mongo/db/matcher/expression_text_base.h"
+#include "mongo/db/namespace_string.h"
 
 namespace mongo {
 
+class NamespaceString;
+class OperationContext;
+
 class TextMatchExpression : public TextMatchExpressionBase {
 public:
-    TextMatchExpression(TextParams params);
+    Status init(OperationContext* txn, const NamespaceString& nss, TextParams params);
 
-    Status init();
+    const fts::FTSQuery& getFTSQuery() const final {
+        return _ftsQuery;
+    }
 
     std::unique_ptr<MatchExpression> shallowClone() const final;
+
+private:
+    fts::FTSQueryImpl _ftsQuery;
 };
 
 }  // namespace mongo
diff --git a/src/mongo/db/matcher/expression_text_base.cpp b/src/mongo/db/matcher/expression_text_base.cpp
index 04b62ecd6b1..82fa7f9fd3e 100644
--- a/src/mongo/db/matcher/expression_text_base.cpp
+++ b/src/mongo/db/matcher/expression_text_base.cpp
@@ -30,23 +30,21 @@
 
 #include "mongo/db/matcher/expression_text_base.h"
 
+#include "mongo/db/fts/fts_query.h"
+
 namespace mongo {
 
 const bool TextMatchExpressionBase::kCaseSensitiveDefault = false;
 const bool TextMatchExpressionBase::kDiacriticSensitiveDefault = false;
 
-TextMatchExpressionBase::TextMatchExpressionBase(TextParams params)
-    : LeafMatchExpression(TEXT),
-      _query(std::move(params.query)),
-      _language(std::move(params.language)),
-      _caseSensitive(params.caseSensitive),
-      _diacriticSensitive(params.diacriticSensitive) {}
+TextMatchExpressionBase::TextMatchExpressionBase() : LeafMatchExpression(TEXT) {}
 
 void TextMatchExpressionBase::debugString(StringBuilder& debug, int level) const {
+    const fts::FTSQuery& ftsQuery = getFTSQuery();
     _debugAddSpace(debug, level);
-    debug << "TEXT : query=" << _query << ", language=" << _language
-          << ", caseSensitive=" << _caseSensitive << ", diacriticSensitive=" << _diacriticSensitive
-          << ", tag=";
+    debug << "TEXT : query=" << ftsQuery.getQuery() << ", language=" << ftsQuery.getLanguage()
+          << ", caseSensitive=" << ftsQuery.getCaseSensitive()
+          << ", diacriticSensitive=" << ftsQuery.getDiacriticSensitive() << ", tag=";
     MatchExpression::TagData* td = getTag();
     if (NULL != td) {
         td->debugString(&debug);
@@ -57,9 +55,11 @@ void TextMatchExpressionBase::debugString(StringBuilder& debug, int level) const
 }
 
 void TextMatchExpressionBase::toBSON(BSONObjBuilder* out) const {
+    const fts::FTSQuery& ftsQuery = getFTSQuery();
     out->append("$text",
-                BSON("$search" << _query << "$language" << _language << "$caseSensitive"
-                               << _caseSensitive << "$diacriticSensitive" << _diacriticSensitive));
+                BSON("$search" << ftsQuery.getQuery() << "$language" << ftsQuery.getLanguage()
+                               << "$caseSensitive" << ftsQuery.getCaseSensitive()
+                               << "$diacriticSensitive" << ftsQuery.getDiacriticSensitive()));
 }
 
 bool TextMatchExpressionBase::equivalent(const MatchExpression* other) const {
@@ -68,19 +68,7 @@ bool TextMatchExpressionBase::equivalent(const MatchExpression* other) const {
     }
     const TextMatchExpressionBase* realOther = static_cast<const TextMatchExpressionBase*>(other);
 
-    if (realOther->getQuery() != _query) {
-        return false;
-    }
-    if (realOther->getLanguage() != _language) {
-        return false;
-    }
-    if (realOther->getCaseSensitive() != _caseSensitive) {
-        return false;
-    }
-    if (realOther->getDiacriticSensitive() != _diacriticSensitive) {
-        return false;
-    }
-    return true;
+    return getFTSQuery().equivalent(realOther->getFTSQuery());
 }
 
 }  // namespace mongo
diff --git a/src/mongo/db/matcher/expression_text_base.h b/src/mongo/db/matcher/expression_text_base.h
index 8ca25cf60eb..187a91bb2d4 100644
--- a/src/mongo/db/matcher/expression_text_base.h
+++ b/src/mongo/db/matcher/expression_text_base.h
@@ -32,6 +32,10 @@
 
 namespace mongo {
 
+namespace fts {
+class FTSQuery;
+}  // namespace fts
+
 /**
  * Common base class for $text match expression implementations.
  */
@@ -47,23 +51,12 @@ public:
     static const bool kCaseSensitiveDefault;
     static const bool kDiacriticSensitiveDefault;
 
-    TextMatchExpressionBase(TextParams params);
+    TextMatchExpressionBase();
 
-    const std::string& getQuery() const {
-        return _query;
-    }
-
-    const std::string& getLanguage() const {
-        return _language;
-    }
-
-    bool getCaseSensitive() const {
-        return _caseSensitive;
-    }
-
-    bool getDiacriticSensitive() const {
-        return _diacriticSensitive;
-    }
+    /**
+     * Returns a reference to the parsed text query that this TextMatchExpressionBase owns.
+     */
+    virtual const fts::FTSQuery& getFTSQuery() const = 0;
 
     //
     // Methods inherited from MatchExpression.
@@ -89,12 +82,6 @@ public:
     void toBSON(BSONObjBuilder* out) const final;
 
     bool equivalent(const MatchExpression* other) const final;
-
-private:
-    const std::string _query;
-    const std::string _language;
-    const bool _caseSensitive;
-    const bool _diacriticSensitive;
 };
 
 }  // namespace mongo
diff --git a/src/mongo/db/matcher/expression_text_noop.cpp b/src/mongo/db/matcher/expression_text_noop.cpp
index 7d2fb5c84b7..bfd61a54081 100644
--- a/src/mongo/db/matcher/expression_text_noop.cpp
+++ b/src/mongo/db/matcher/expression_text_noop.cpp
@@ -34,21 +34,24 @@
 
 namespace mongo {
 
-TextNoOpMatchExpression::TextNoOpMatchExpression(TextParams params)
-    : TextMatchExpressionBase(std::move(params)) {}
-
-Status TextNoOpMatchExpression::init() {
+Status TextNoOpMatchExpression::init(TextParams params) {
+    _ftsQuery.setQuery(std::move(params.query));
+    _ftsQuery.setLanguage(std::move(params.language));
+    _ftsQuery.setCaseSensitive(params.caseSensitive);
+    _ftsQuery.setDiacriticSensitive(params.diacriticSensitive);
+    invariantOK(_ftsQuery.parse(fts::TEXT_INDEX_VERSION_INVALID));
     return initPath("_fts");
 }
 
 std::unique_ptr<MatchExpression> TextNoOpMatchExpression::shallowClone() const {
     TextParams params;
-    params.query = getQuery();
-    params.language = getLanguage();
-    params.caseSensitive = getCaseSensitive();
-    params.diacriticSensitive = getDiacriticSensitive();
-    auto expr = stdx::make_unique<TextNoOpMatchExpression>(std::move(params));
-    expr->init();
+    params.query = _ftsQuery.getQuery();
+    params.language = _ftsQuery.getLanguage();
+    params.caseSensitive = _ftsQuery.getCaseSensitive();
+    params.diacriticSensitive = _ftsQuery.getDiacriticSensitive();
+
+    auto expr = stdx::make_unique<TextNoOpMatchExpression>();
+    invariantOK(expr->init(std::move(params)));
     if (getTag()) {
         expr->setTag(getTag()->clone());
     }
diff --git a/src/mongo/db/matcher/expression_text_noop.h b/src/mongo/db/matcher/expression_text_noop.h
index de02bcfa3cf..8e01d40e70b 100644
--- a/src/mongo/db/matcher/expression_text_noop.h
+++ b/src/mongo/db/matcher/expression_text_noop.h
@@ -28,17 +28,23 @@
 
 #pragma once
 
+#include "mongo/db/fts/fts_query_noop.h"
 #include "mongo/db/matcher/expression_text_base.h"
 
 namespace mongo {
 
 class TextNoOpMatchExpression : public TextMatchExpressionBase {
 public:
-    TextNoOpMatchExpression(TextParams params);
+    Status init(TextParams params);
 
-    Status init();
+    const fts::FTSQuery& getFTSQuery() const final {
+        return _ftsQuery;
+    }
 
     std::unique_ptr<MatchExpression> shallowClone() const final;
+
+private:
+    fts::FTSQueryNoop _ftsQuery;
 };
 
 }  // namespace mongo
diff --git a/src/mongo/db/matcher/extensions_callback_noop.cpp b/src/mongo/db/matcher/extensions_callback_noop.cpp
index e5941f73562..10026370e64 100644
--- a/src/mongo/db/matcher/extensions_callback_noop.cpp
+++ b/src/mongo/db/matcher/extensions_callback_noop.cpp
@@ -41,8 +41,8 @@ StatusWithMatchExpression ExtensionsCallbackNoop::parseText(BSONElement text) co
         return textParams.getStatus();
     }
 
-    auto expr = stdx::make_unique<TextNoOpMatchExpression>(std::move(textParams.getValue()));
-    Status initStatus = expr->init();
+    auto expr = stdx::make_unique<TextNoOpMatchExpression>();
+    Status initStatus = expr->init(std::move(textParams.getValue()));
     if (!initStatus.isOK()) {
         return initStatus;
     }
diff --git a/src/mongo/db/matcher/extensions_callback_real.cpp b/src/mongo/db/matcher/extensions_callback_real.cpp
index 161e0fb2e79..0b7e815ca58 100644
--- a/src/mongo/db/matcher/extensions_callback_real.cpp
+++ b/src/mongo/db/matcher/extensions_callback_real.cpp
@@ -45,8 +45,8 @@ StatusWithMatchExpression ExtensionsCallbackReal::parseText(BSONElement text) co
         return textParams.getStatus();
     }
 
-    auto exp = stdx::make_unique<TextMatchExpression>(std::move(textParams.getValue()));
-    Status status = exp->init();
+    auto exp = stdx::make_unique<TextMatchExpression>();
+    Status status = exp->init(_txn, *_nss, std::move(textParams.getValue()));
     if (!status.isOK()) {
         return status;
     }