Merge pull request #68 from ecmwf/feature/data_keyword

added method to check if a keyword is used for indexing (category: data)

Author: danovaro

share/metkit/axis.yaml

@@ -49,3 +49,4 @@ axes:
     - resolution
     - year
     - month
+    - obsgroup

share/metkit/language.yaml

@@ -350,6 +350,7 @@ _field: &_field
         val: 'prod'
 
   dataset:
+    category: data
     multiple: false
     type: enum
     values:
@@ -1323,7 +1324,6 @@ disseminate:
     type: any
 
   use:
-    category: data
     flatten: false
     multiple: true
     type: enum

src/metkit/mars/MarsLanguage.cc

@@ -84,11 +84,22 @@ MarsLanguage::MarsLanguage(const std::string& verb) : verb_(verb) {
         types_[keyword]->attach();
         keywords_.push_back(keyword);
 
+        std::optional<eckit::Value> aliases;
         if (settings.contains("aliases")) {
-            eckit::Value aliases = settings["aliases"];
-            for (size_t j = 0; j < aliases.size(); ++j) {
-                aliases_[aliases[j]] = keyword;
-                keywords_.push_back(aliases[j]);
+            aliases = settings["aliases"];
+        }
+        if (settings.contains("category") && settings["category"] == "data") {
+            dataKeywords_.insert(keyword);
+            if (aliases) {
+                for (size_t j = 0; j < aliases->size(); ++j) {
+                    dataKeywords_.insert((*aliases)[j]);
+                }
+            }
+        }
+        if (aliases) {
+            for (size_t j = 0; j < aliases->size(); ++j) {
+                aliases_[(*aliases)[j]] = keyword;
+                keywords_.push_back((*aliases)[j]);
             }
         }
     }
@@ -100,9 +111,7 @@ MarsLanguage::MarsLanguage(const std::string& verb) : verb_(verb) {
         }
     }
 
-    std::set<std::string> keywordsInAxis;
     for (const std::string& a : hypercube::AxisOrder::instance().axes()) {
-        keywordsInAxis.insert(a);
         Type* t=nullptr;
         auto it = types_.find(a);
         if(it != types_.end()) {
@@ -111,10 +120,15 @@ MarsLanguage::MarsLanguage(const std::string& verb) : verb_(verb) {
         typesByAxisOrder_.emplace_back(a, t);
     }
     for (const auto& [k,t] : types_) {
-        if (keywordsInAxis.find(k) == keywordsInAxis.end()) { typesByAxisOrder_.emplace_back(k, t); }
+        if (dataKeywords_.find(k) == dataKeywords_.end()) { typesByAxisOrder_.emplace_back(k, t); }
     }
 }
 
+bool MarsLanguage::isData(const std::string& keyword) const {
+    return (dataKeywords_.find(keyword) != dataKeywords_.end());
+}
+
+
 MarsLanguage::~MarsLanguage() {
     for (std::map<std::string, Type*>::iterator j = types_.begin(); j != types_.end(); ++j) {
         (*j).second->detach();

src/metkit/mars/MarsLanguage.h

@@ -53,6 +53,8 @@ class MarsLanguage : private eckit::NonCopyable {
 
     Type* type(const std::string& name) const;
 
+    bool isData(const std::string& keyword) const;
+
 public: // class methods
 
     static std::string expandVerb(const MarsExpandContext&, const std::string& verb);
@@ -80,6 +82,7 @@ class MarsLanguage : private eckit::NonCopyable {
 
     std::string verb_;
     std::map<std::string, Type* > types_;
+    std::set<std::string> dataKeywords_;
     std::vector<std::pair<std::string, Type*>> typesByAxisOrder_;
     std::vector<std::string> keywords_;