(in-package :cl-user)

(defun kegg-generic-formula-match ()
  "Wouldn't be matched by KEGG because not definedBy"
  (set-difference (sparql '(:select (?mol1 ?mol2) (:distinct t)
			    (?mol1 !rdf:type !bug:smallMolecule) 
			    (?mol1 !bug:classifiedByKEGG ?kegg)
			    (?mol2 !bug:classifiedByKEGG ?kegg)
			    (?mol1 !bug:hasChemicalFormula ?formula)
			    (?mol2 !bug:hasChemicalFormula ?formula)
			    (:filter (and (is-canonical ?mol1) (not (equal ?mol1 ?mol2))))) :use-reasoner :jena) 
		  (sparql '(:select (?mol1 ?mol2) ()
			    (?mol1 !rdf:type !bug:smallMolecule)
			    (?mol2 !owl:sameAs ?mol1)
			    (:filter (and (is-canonical ?mol1) (not (equal ?mol1 ?mol2))))) :use-reasoner :jena) :test 'equalp))

(defun sames-sparql ()
  "Which pairs of small molecules are considered to be the same?"
  (sparql '(:select (?mol1 ?mol2) ()
	    (?mol1 !rdf:type !bug:smallMolecule)
	    (?mol2 !owl:sameAs ?mol1)
	    (:filter (and (is-canonical ?mol1) (not (equal ?mol1 ?mol2)))))
	  :use-reasoner :jena))

(defun fuzzy-name-matches ()
  (let ((has-same (make-hash-table :test 'equal)))
    (loop for (this that) in (sames-sparql)
	 do (setf (gethash this  has-same ) that)
	 (setf (gethash that  has-same ) this))
    (loop with name2mol = (make-hash-table :test 'equal) and them
       for (mol name) in 
       (sparql '(:select (?mol1 ?name) (:distinct t)
		 (?mol1 !rdf:type !bug:smallMolecule)
		 (?mol1 !bug:hasName ?name))
	       :use-reasoner :jena)
       for fuzzed = (string-downcase (#"replaceAll" name "[^a-zA-Z0-9]" ""))
       do (unless (gethash mol has-same) (push (list mol name) (gethash fuzzed name2mol)))
       finally
       (maphash (lambda(k v)
		  (when (> (length v) 1)
		     (push (mapcar 'car v) them)))
		name2mol)
	 (return them))))

(defun sameindividual-owl-for-fuzzy-name-matches (&optional (stream t))
  (loop for set in (fuzzy-name-matches)
       do
       (format stream "SameIndividual(~{<~a>~^ ~})~%" (mapcar 'uri-full set))))

(defun generate-sames-by-name ()
  (with-open-file (f "bug:sames-by-name.absowl" :direction :output :if-exists :supersede)
    (sameindividual-owl-for-fuzzy-name-matches f)))