path: root/transform/special-forms.scm

(define-module (transform special-forms)
  #:use-module (srfi srfi-1)
  #:use-module (ice-9 regex)
  #:use-module (ice-9 match)
  #:use-module (srfi srfi-26)
  #:use-module (transform sql)
  #:use-module (transform table)
  #:use-module (transform triples)
  #:use-module (transform strings)
  #:export (translate-forms
            collect-forms
            collect-keys
            field->key
            field->assoc-ref
            collect-fields
            find-clause
            remove-namespace
            column-id
            id
            syntax-let
            map-alist
	    with-documentation
            emit-short-turtle
            define-transformer))

(define (emittable-object? o)
  (cond
    ((null? o) #f)
    ((not o)   #f)
    ((and (string? o) (string-blank? o)) #f)
    (else #t)))

(define (emit-short-turtle subject po-alist)
  ;; Hot path: avoid rescanning remaining pairs on each step (O(n^2)).
  ;; Build filtered pairs and count in one pass.
  (let collect ((pairs po-alist) (kept '()) (count 0))
    (if (null? pairs)
        (let loop ((pairs (reverse kept)) (index 0))
          (match pairs
            (((p . o) rest ...)
             (if (zero? index)
                 (format #t "~a " subject)
                 (display "\t"))
             (match o
               ((? symbol?)
                (format #t "~a ~a" p (symbol->string o)))
               ((? string?)
                (format #t "~a \"~a\"" p o))
               (_
                (format #t "~a ~s" p o)))
             (if (< (+ index 1) count)
                 (display " ;\n")
                 (display " .\n"))
             (loop rest (+ index 1)))
            (() #f)))
        (match (car pairs)
          ((p . o)
           (if (emittable-object? o)
               (collect (cdr pairs) (cons (cons p o) kept) (+ count 1))
               (collect (cdr pairs) kept count)))))))

(define (make-short-turtle-emitter subject)
  "Return two procedures: EMIT! and FINISH! for streaming short-turtle output."
  (let ((first? #t)
        (pending #f))
    (define (emit-one pair tail)
      (let ((p (car pair))
            (o (cdr pair)))
        (if first?
            (format #t "~a " subject)
            (display "\t"))
        (match o
          ((? symbol?)
           (format #t "~a ~a" p (symbol->string o)))
          ((? string?)
           (format #t "~a \"~a\"" p o))
          (_
           (format #t "~a ~s" p o)))
        (display tail)
        (set! first? #f)))
    (define (emit! predicate object)
      (when (emittable-object? object)
        (let ((current (cons predicate object)))
          (when pending
            (emit-one pending " ;\n"))
          (set! pending current))))
    (define (finish!)
      (when pending
        (emit-one pending " .\n")
        (set! pending #f)))
    (values emit! finish!)))



(define (key->assoc-ref alist x)
  "Recursively translate (key k) forms in source X to (assoc-ref ALIST
k) forms."
  (translate-forms 'key
                   ;; (syntax-rules (key)
                   ;;   ((key k) (assoc-ref alist k)))
                   (lambda (x)
                     (syntax-case x (key)
                       ((key k) #`(assoc-ref #,alist 'k))))
                   x))

(define (alist-delete* alist keys)
  "Delete entries from ALIST whose key is equal (in the sense of
equal?) to any in KEYS, a list."
  (remove (match-lambda
            ((key . value)
             (member key keys))
            (x (error "malformed alist element" x)))
          alist))

(define (translate-forms from translator x)
  "Recursively pass (FROM ...) forms in source X to TRANSLATOR, and
replace them with the return value."
  (syntax-case x ()
    ;; Handle translation base case.
    ((head tail ...) (eq? (syntax->datum #'head)
                          from)
     (translator x))
    ;; Recurse.
    ((head tail ...)
     (cons (translate-forms from translator #'head)
           (map (cut translate-forms from translator <>)
                #'(tail ...))))
    ;; Handle leaf base case.
    (leaf #'leaf)))

(define (collect-forms target x)
  "Recursively collect (TARGET ...) forms in source X and return them
as a list."
  (syntax-case x ()
    ;; Handle collection base case.
    ((head tail ...) (eq? (syntax->datum #'head)
                          target)
     (list x))
    ;; Recurse.
    ((head tail ...)
     (append (collect-forms target #'head)
             (append-map (cut collect-forms target <>) #'(tail ...))))
    ;; Handle leaf base case.
    (leaf (list))))

(define (collect-keys x)
  "Recursively collect (key k) forms from source X and return as a
list of all K."
  (map (syntax-rules (key)
         ((key k) 'k))
       (collect-forms 'key x)))

(define-syntax map-alist
  (lambda (x)
    "Transform (aka map) ALIST, an association list, into another
association list. The returned association list may contain multiple
associations for the same key. equal? is used in all association list
key comparisons.

Syntax:

(map-alist alist
  (verb key expression) ...
  (else=> proc))

VERB must be one of set, filter-set, multiset and remove.

For the set VERB, KEY is set to the result of evaluating
EXPRESSION. Multiple set verbs on the same key will result in multiple
associations for that key.

For the filter-set VERB, KEY is set to the result of evaluating
EXPRESSION only if that result is not #f.

For the multiset VERB, EXPRESSION must return a list and KEY is
associated multiple times once with each element of the returned list.

For the remove VERB, KEY is discarded.

EXPRESSIONs must reference elements of ALIST using (key k) forms where
K is the key to be referenced from ALIST. K must not be quoted. That
is, if K is the symbol 'bar, it must be referenced as (key bar), not
as (key 'bar).

The else=> clause is optional.

If the else=> clause is present, PROC is passed all pairs of ALIST
that are not set by an earlier (verb key expression) action. PROC must
return #f or a pair to replace its input pair. If PROC returns #f,
that pair is discarded.

If the else=> clause is absent, all unset pairs are discarded.

Example:

(map-alist '((\"foo\" . 1)
             (bar . 2)
             (foobar . 5)
             (fubar . 3))
  (set spam (1+ (key \"foo\")))
  (set ham (* 2 (key bar)))
  (set eggs (* 3 (key bar)))
  (set aal (+ (key \"foo\")
              (key bar)))
  (multiset vel (iota (* 2 (key bar))))
  (remove foobar)
  (else=> (match-lambda
            ((key . value)
             (cons key (expt 2 value))))))
=> ((spam . 2) (ham . 4) (eggs . 6) (aal . 3)
    (vel . 0) (vel . 1) (vel . 2) (vel . 3) (fubar . 8))"
    (syntax-case x ()
      ((_ alist actions ...)
       ;; TODO: Check that all actions are valid.
       (let* ((actions #'(actions ...))
              (filter-set-keys
               (filter-map (lambda (action)
                             (syntax-case action (filter-set)
                               ((filter-set key expression) #'key)
                               (_ #f)))
                           actions))
              (set/filter-pairs
               (filter-map (lambda (action)
                             (syntax-case action (set filter-set)
                               ((set key expression)
                                #`(cons 'key #,(key->assoc-ref #'evaluated-alist #'expression)))
                               ((filter-set key expression)
                                #`(cons 'key #,(key->assoc-ref #'evaluated-alist #'expression)))
                               (_ #f)))
                           actions))
              (multiset-exprs
               (filter-map (lambda (action)
                             (syntax-case action (multiset)
                               ((multiset key expression)
                                #`(map (cut cons 'key <>)
                                       #,(key->assoc-ref #'evaluated-alist #'expression)))
                               (_ #f)))
                           actions))
              (else-proc
               (any (lambda (action)
                      (syntax-case action (else=>)
                        ((else=> proc) #'proc)
                        (_ #f)))
                    actions))
              (unspecified-keys
               #`(list
                  ;; Keys that were referenced
                  #,@(append-map (lambda (action)
                                   (syntax-case action ()
                                     ((_ key expression)
                                      (collect-keys #'expression))
                                     (_ '())))
                                 actions)
                  ;; Keys that were deleted
                  #,@(filter-map (lambda (action)
                                   (syntax-case action (remove)
                                     ((remove key) #''key)
                                     (_ #f)))
                                 actions)
                  ;; Keys that were set
                  #,@(filter-map (lambda (action)
                                   (syntax-case action ()
                                     ((_ key expression) #''key)
                                     (_ #f)))
                                 actions))))
         #`(let ((evaluated-alist alist))
             (append
              ;; Do set and filter-set.
              #,(if (null? filter-set-keys)
                    #`(list #,@set/filter-pairs)
                    #`(remove (match-lambda
                                ((key . #f)
                                 (member key '#,filter-set-keys))
                                (_ #f))
                              (list #,@set/filter-pairs)))
              ;; Do multiset.
              #,@multiset-exprs
              ;; Apply else=> only when explicitly present.
              #,@(if else-proc
                     (list
                      #`(filter-map #,else-proc
                                    (alist-delete* evaluated-alist
                                                   #,unspecified-keys)))
                     '()))))))))



(eval-when (expand load )
  (define (field->datum x)
    (translate-forms
     'field
     (lambda (x)
       (syntax-case x (field)
         ((field (query alias))
          #`(format #f "~a" (syntax->datum #'alias)))
         ((field table column)
          #`(format #f "~a(~a)"
                    (syntax->datum #'table)
                    (syntax->datum #'column)))
         ((field table column alias)
          #`(format #f "~a(~a)"
                    (syntax->datum #'table)
                    (syntax->datum #'alias)))))
     x))

  (define (field->key x)
    (translate-forms 'field
                     (lambda (x)
                       #`(key #,(symbol->string
                                 (syntax->datum
                                  ((syntax-rules (field)
                                     ((field (query alias)) alias)
                                     ((field table column) column)
                                     ((field table column alias) alias))
                                   x)))))
                     x))

  (define (field->assoc-ref alist x)
    "Recursively translate field references in source X to (assoc-ref
ALIST field-name) forms."
    (translate-forms 'field
                     (lambda (x)
                       #`(assoc-ref #,alist
                                    #,(symbol->string
                                       (syntax->datum
                                        ((syntax-rules (field)
                                           ((field (query alias)) alias)
                                           ((field table column) column)
                                           ((field table column alias) alias))
                                         x)))))
                     x))

  (define (collect-fields x)
    (map (syntax-rules (field)
           ((field reference ...)
            (reference ...)))
         (collect-forms 'field x)))

  (define (find-clause clauses key)
    "Find (KEY ...) clause among CLAUSES, a list of syntax forms."
    (find (lambda (x)
            (syntax-case x ()
              ((clause-key _ ...)
               (eq? (syntax->datum #'clause-key)
                    key))))
          clauses))

  (define (remove-namespace str)
    "Remove RDF namespace prefix from STR."
    (substring str (1+ (string-index str #\:))))

  (define (column-id table-name column-name)
    (string->identifier
     "field" (string-append
              ;; We downcase table names in identifiers. So, we
              ;; distinguish between the user and User tables.
              (if (string=? table-name "User")
                  "user2" table-name)
              "__" column-name)))

  (define (id table predicate)
    (symbol->string
     (string->identifier
      "transform"
      (string-append
       table "_" (remove-namespace (symbol->string predicate)))))))

(define-syntax syntax-let
  (syntax-rules ()
    "Like match-let, but for syntax.

(syntax-let ((pattern literals expression))
  body ...)
≡
(syntax-case expression ()
  (pattern literals
   body ...))

literals is optional. So,

(syntax-let ((pattern expression))
  body ...)
≡
(syntax-case expression ()
  (pattern
   body ...))
"
    ((_ () body ...)
     (begin body ...))
    ((_ ((pattern expression)
         bindings ...)
        body ...)
     (syntax-case expression ()
       (pattern
        (syntax-let (bindings ...)
          body ...))))
    ((_ ((pattern literals expression)
         bindings ...)
        body ...)
     (syntax-case expression literals
       (pattern
        (syntax-let (bindings ...)
          body ...))))))

(define-syntax define-transformer
  (lambda (x)
    "Define FUNCTION-NAME, a function that transforms a view of the database.

define-transformer consists of three order-agnostic clauses---tables,
schema-triples and triples---in the form shown below.

(define-transformer function-name
  (tables (table ...) raw-forms ...)
  (schema-triples
   (subject predicate object) ...)
  (triples subject
    (verb predicate object) ...))

The tables clause specifies the database tables to be joined to
construct the view to be transformed. TABLE must be either of the form
TABLE-NAME or of the form (JOIN-OPERATOR TABLE-NAME
RAW-CONDITION). TABLE-NAME must, obviously, be the name of the
table. JOIN-OPERATOR must be one of join, left-join and
inner-join. RAW-CONDITION should be the join condition as a raw
string. This is usually something like
\"USING (SpeciesId)\". RAW-FORMS are expressions that must evaluate to
strings to be appended to the SQL query.

The schema-triples clause specifies the list of triples to be written
once when the transform starts.

The triples clause specifies the triples to be transformed once for each
row in the view. All triples have a common SUBJECT. The (verb
predicate object) clauses are described below.

VERB must either be set or multiset. For the set VERB, a single triple
(SUBJECT PREDICATE OBJECT-VALUE) is written where OBJECT-VALUE is the
result of evaluating OBJECT. For the multiset VERB, OBJECT must
evaluate to a list, and a triple (SUBJECT PREDICATE
OBJECT-VALUE-ELEMENT) is created for each element OBJECT-VALUE-ELEMENT
of that list.

The SUBJECT and OBJECT expressions in the triples clause must
reference database fields using a (field TABLE COLUMN) clause where
TABLE and COLUMN refer to the table and column of the field being
referenced. Database fields can also be referenced using (field TABLE
COLUMN ALIAS) where ALIAS is an alias for that column in the SQL
query. Specification of ALIAS is indeed a leak in the abstraction, and
must be remedied."
    (syntax-case x (tables schema-triples triples)
      ((_ name clauses ...)
       (syntax-let (((tables (primary-table other-tables ...) tables-raw ...) (tables)
                     (find-clause #'(clauses ...) 'tables))
                    (schema-triples-clause (or (find-clause #'(clauses ...) 'schema-triples)
                                               #'(schema-triples)))
                    ((triples subject predicate-clauses ...) (triples)
                     (find-clause #'(clauses ...) 'triples)))
	         #`(define* (name db #:key
	                          (metadata? #f)
	                          (data? #t)
	                          (documentation? #f)
                                  (keyset? #f)
                                  (chunk-key "Id")
	                          (limit #f)
	                          (offset #f)
                                  (cursor #f))
	             (let* ((base-sql
	                     (select-query #,(collect-fields #'(subject predicate-clauses ...))
	                                   (primary-table other-tables ...)
	                                   tables-raw ...))
                            (chunk-keys
                             (if (list? chunk-key)
                                 chunk-key
                                 (list chunk-key)))
                            (chunk-key-exprs
                             (map (lambda (k)
                                    (if (string-index k #\.)
                                        k
                                        (format #f "~a.~a"
                                                #,(symbol->string (syntax->datum #'primary-table))
                                                k)))
                                  chunk-keys))
                            (sql-for-keyset
                             (prepend-select-expr
                              base-sql
                              (string-join
                               (map (lambda (expr idx)
                                      (format #f "~a AS __chunk_key_~a" expr (+ idx 1)))
                                    chunk-key-exprs
                                    (iota (length chunk-key-exprs)))
                               ", ")))
		                    (sql
                             (cond
                              (keyset?
                               (append-order-limit
                                (if cursor
                                    (append-filter-clause
                                     sql-for-keyset
                                     (build-keyset-where chunk-key-exprs cursor))
                                    sql-for-keyset)
                                (build-keyset-order chunk-key-exprs)
                                (or limit 100000)))
			      ((and limit offset)
			       (format #f "~a LIMIT ~a OFFSET ~a"
				       base-sql limit offset))
			      (else base-sql))))
		               (when metadata?
                 #,@(let ((table (symbol->string (syntax->datum #'primary-table)))
                          (subject-type (any (lambda (predicate)
                                               (syntax-case predicate (rdf:type)
                                                 ((_ rdf:type type) #'type)
                                                 (_ #f)))
                                             #'(predicate-clauses ...))))
                      (map (lambda (predicate-clause)
                             (syntax-case predicate-clause ()
                               ((_ predicate _)
                                ;; Dump metadata about the transform itself.
                                #`(begin
                                    (scm->triples
                                     (map-alist '()
	        		       (set rdf:type 'gn-id:transform)
	        		       (set gn-term:createsPredicate 'predicate)
	        		       (filter-set gn-term:forSubjectType #,subject-type)
	        		       (multiset gn-term:dependsOn
	        			         '#,(map (lambda (field)
	        					   (match (syntax->datum field)
	        					     ((table-name column-name _ ...)
	        					      (datum->syntax
	        					       x (column-id (symbol->string table-name)
	        							    (symbol->string column-name))))
	        					     (((query alias))
	        					      (datum->syntax
	        					       x (column-id query (symbol->string alias))))))
	        				         (collect-fields predicate-clause))))
                                     #,(id table (syntax->datum #'predicate)))
                                    ;; Automatically create domain triples
                                    ;; for predicates.
                                    (when #,subject-type
                                      (triple 'predicate 'rdfs:domain #,subject-type))))
                               (_ (error "Invalid predicate clause:" predicate-clause))))
                           #'(predicate-clauses ...))))
               (when documentation?
                 (format #t "~%## '~a'~%~%" (syntax->datum #'name))
                 #,(syntax-case #'schema-triples-clause (schema-triples)
                     ((schema-triples (triple-subject triple-predicate triple-object) ...)
                      #`(begin
                          (when (not (list 'triple-subject ...))
                            (format #t "## Schema Triples:~%~%```text~%")
                            (for-each (lambda (s p o)
                                        (format #t "~a -> ~a -> ~a~%" s p o))
                                      (list 'triple-subject ...)
                                      (list 'triple-predicate ...)
                                      (list 'triple-object ...))
                            (format #t "```"))))
                     (_ (error "Invalid schema triples clause:" #'schema-triples-clause)))
                 (format #t "## Generated Triples:

The following SQL query was executed:

```sql
~a
```

The above query results to triples that have the form:

```text
"
                         (select-query #,(collect-fields #'(subject predicate-clauses ...))
                                       (primary-table other-tables ...)
                                       tables-raw ...))
                 (for-each (match-lambda
                             ((predicate . object)
                              (format #t "~a -> ~a -> ~a ~%"
                                      (if (symbol? #,(field->datum #'subject))
                                          (symbol->string #,(field->datum #'subject))
                                          #,(field->datum #'subject))
                                      predicate
                                      (if (symbol? object)
                                          (symbol->string object)
                                          object))))
                           (map-alist
                               '()
                             #,@(field->datum #'(predicate-clauses ...))))
                 (format #t "```~%Here's an example query:~%~%```sparql~%")
                 (documentation?)
                 (newline)
                 (let* ((result
                         (map-alist (sql-find
                                     db
                                     (format #f "~a LIMIT 1"
                                             (select-query #,(collect-fields #'(subject predicate-clauses ...))
                                                           (primary-table other-tables ...)
                                                           tables-raw ...)))
        		   #,@(field->key #'(predicate-clauses ...))))
                        (first-n (list-head result
                                            (let ((n
                                                   (min 4 (truncate
                                                           (+ (exact-integer-sqrt (length result)) 1)))))
                                              (if (< n 3)
                                                  (length result)
                                                  n)))))
                   (format #t "SELECT * WHERE { ~%")
                   (for-each (match-lambda
                               ((predicate . object)
                                (match object
                                  ((or (?  symbol? object)
                                       (?  (lambda (el) (string-match "^\\[ .* \\]$" el)) object))
                                   (format #t "    ?s ~a ~a .~%" predicate object))
                                  ((and (? string? object)
                                        (? (lambda (el) (not (string-null? el))) object))
                                   (format #t "    ?s ~a \"~a\" .~%" predicate object))
                                  (_ ""))))
                             first-n)
                   (format #t "    ?s ?p ?o .~%}~%```~%"))
                 (format #t "~%Expected Result:~%~%```rdf~%")
                 (sql-for-each (lambda (row)
                                 (scm->triples
                                  (map-alist row #,@(field->key #'(predicate-clauses ...)))
                                  #,(field->assoc-ref #'row #'subject)
                                  (lambda (s p o)
                                    (triple s p o))))
                               db
                               (format #f "~a LIMIT 1" base-sql))
                 (format #t "```~%~%"))
               (when data?
               #,(syntax-case #'schema-triples-clause (schema-triples)
                   ((schema-triples (triple-subject triple-predicate triple-object) ...)
                    #`(for-each triple
                                (list 'triple-subject ...)
                                (list 'triple-predicate ...)
                                (list 'triple-object ...)))
                   (_ (error "Invalid schema triples clause:" #'schema-triples-clause)))
                        (let ((rows-written 0)
                             (last-cursor cursor))
		         (sql-for-each
			  (lambda (row)
                            (set! rows-written (+ rows-written 1))
                            (when keyset?
                              (set! last-cursor
                                    (map (lambda (idx)
                                           (assoc-ref row (format #f "__chunk_key_~a" (+ idx 1))))
                                         (iota (if (list? chunk-key)
                                                   (length chunk-key)
                                                   1)))))
	                    (let ((subject-val #,(field->assoc-ref #'row #'subject)))
	                      (call-with-values
	                          (lambda () (make-short-turtle-emitter subject-val))
	                        (lambda (emit! finish!)
	                          #,@(map (lambda (predicate-clause)
	                                    (syntax-case predicate-clause (set multiset filter-set)
	                                      ((set predicate object)
	                                       #`(emit! 'predicate
	                                                #,(field->assoc-ref #'row #'object)))
	                                      ((filter-set predicate object)
	                                       #`(let ((val #,(field->assoc-ref #'row #'object)))
	                                           (when val
	                                             (emit! 'predicate val))))
	                                      ((multiset predicate object)
	                                       #`(for-each (lambda (val)
	                                                     (emit! 'predicate val))
	                                                   #,(field->assoc-ref #'row #'object)))
	                                      (_ (error "Invalid predicate clause:" predicate-clause))))
	                                  #'(predicate-clauses ...))
	                          (finish!)))))
	                  db
	                  sql)
                         (if keyset?
                             (cons rows-written last-cursor)
                             rows-written)))))))
      (_ (error "Invalid define-transformer syntax:" (syntax->datum x))))))

(define (sql-literal v)
  (cond
   ((number? v) (number->string v))
   ((string? v)
    (format #f "'~a'" (replace-substrings v '(("'" . "''")))))
   ((symbol? v)
    (format #f "'~a'" (symbol->string v)))
   (else
    (format #f "'~a'" v))))

(define (normalize-keyset-cursor cursor key-count)
  (cond
   ((not cursor) #f)
   ((and (= key-count 1) (not (list? cursor)))
    (list cursor))
   ((and (list? cursor) (= (length cursor) key-count))
    cursor)
   (else
    (error "Invalid keyset cursor, expected "
           key-count " values, got: " cursor))))

(define (normalize-key-exprs key-spec)
  "Accept either a list of SQL key expressions or a key count."
  (cond
   ((list? key-spec) key-spec)
   ((number? key-spec)
    (map (lambda (idx)
           (format #f "__chunk_key_~a" (+ idx 1)))
         (iota key-spec)))
   (else
    (error "Invalid key specification: " key-spec))))

(define (build-keyset-order key-spec)
  (string-join
   (normalize-key-exprs key-spec)
   ", "))

(define (build-keyset-where key-spec cursor)
  (let* ((key-exprs (normalize-key-exprs key-spec))
         (key-count (length key-exprs))
         (vals (normalize-keyset-cursor cursor key-count)))
    (unless vals
      (error "build-keyset-where called with empty cursor"))
    (string-join
     (map (lambda (i)
            (let ((eq-prefix
                   (string-join
                    (map (lambda (j)
                           (format #f "~a = ~a"
                                   (list-ref key-exprs j)
                                   (sql-literal (list-ref vals j))))
                         (iota i))
                    " AND "))
                  (gt-part
                   (format #f "~a > ~a"
                           (list-ref key-exprs i)
                           (sql-literal (list-ref vals i)))))
              (if (zero? i)
                  gt-part
                  (string-append "(" eq-prefix " AND " gt-part ")"))))
          (iota key-count))
     " OR ")))

(define (find-sql-tail-start sql)
  (let ((positions
         (filter-map identity
                     (list (string-contains sql " GROUP BY ")
                           (string-contains sql " ORDER BY ")
                           (string-contains sql " LIMIT ")))))
    (if (null? positions) #f (apply min positions))))

(define (append-filter-clause sql clause)
  (let* ((tail-start (find-sql-tail-start sql))
         (head (if tail-start (substring sql 0 tail-start) sql))
         (tail (if tail-start (substring sql tail-start) "")))
    (if (string-contains head " WHERE ")
        (string-append head " AND (" clause ")" tail)
        (string-append head " WHERE (" clause ")" tail))))

(define (append-order-limit sql order-expr limit)
  (let* ((tail-start (find-sql-tail-start sql))
         (head (if tail-start (substring sql 0 tail-start) sql))
         (tail (if tail-start (substring sql tail-start) "")))
    (string-append head tail
                   " ORDER BY " order-expr
                   " LIMIT " (number->string limit))))

(define (prepend-select-expr sql expr)
  "Prepend EXPR to the SELECT list of SQL."
  (cond
   ((and (>= (string-length sql) 16)
         (string=? (substring sql 0 16) "SELECT DISTINCT "))
    (string-append "SELECT DISTINCT " expr ", " (substring sql 16)))
   ((and (>= (string-length sql) 7)
         (string=? (substring sql 0 7) "SELECT "))
    (string-append "SELECT " expr ", " (substring sql 7)))
   (else
    (error "Expected SQL SELECT query" sql))))

(define (get-keyword-value args keyword default)
  (let ((kv (memq keyword args)))
    (if (and kv (>= (length kv) 2))
        (cadr kv)
        default)))

(define-syntax with-documentation
  (syntax-rules ()
    ((_ (key value) ...)
     (let* ((alist `((key . ,value) ...))
            (name (assoc-ref alist 'name))
            (connection (assoc-ref alist 'connection))
            (table-metadata? (assoc-ref alist 'table-metadata?))
            (prefixes (assoc-ref alist 'prefixes))
            (inputs (assoc-ref alist 'inputs))
	            (outputs (assoc-ref alist 'outputs))
	            (total-rows (assoc-ref alist 'total-rows))
	            (rows-per-chunk (assoc-ref alist 'rows-per-chunk))
                    (keyset-pagination? (assoc-ref alist 'keyset-pagination?))
                    (chunk-key-field (assoc-ref alist 'chunk-key-field))
	            (chunking? (and total-rows rows-per-chunk))
            (chunks (if chunking?
                        (ceiling (/ total-rows rows-per-chunk))
                        1))
            (rdf-path (get-keyword-value outputs #:rdf #f))
            (doc-path (get-keyword-value outputs #:documentation #f)))
       (call-with-target-database
        connection
        (lambda (db)
          (when doc-path
            (with-output-to-file        ;
                doc-path
              (lambda ()
                (format #t "# ~a" name)
                (for-each
                 (lambda (proc)
                   (proc db
                         #:metadata? #f
                         #:data? #f
                         #:documentation?
                         (lambda () (for-each
                                     (match-lambda
                                       ((k v)
                                        (begin
                                          (prefix k v #f))))
                                     prefixes))))
                 inputs))
              #:encoding "UTF-8"))

	          ;; Dumping the actual data
	          (when rdf-path
                    (if keyset-pagination?
                        ;; Cursor-based chunking with configurable key column.
                        (let loop ((chunk-index 0)
                                   (states (map (lambda (_) (cons #f #f)) inputs)))
                          (unless (every cdr states)
                            (let* ((out-file
                                    (if (= chunk-index 0)
                                        rdf-path
                                        (string-append (path-without-extension rdf-path)
                                                       "." (number->string (+ chunk-index 1)) ".ttl")))
                                   (next-states
                                    (with-output-to-file
                                        out-file
                                      (lambda ()
                                        (for-each
                                         (match-lambda
                                           ((k v)
                                            (prefix k v)))
                                         prefixes)
                                        (newline)
                                        (let walk ((procs inputs)
                                                   (st states)
                                                   (acc '()))
                                          (if (null? procs)
                                              (reverse acc)
                                              (let* ((state (car st))
                                                     (cursor (car state))
                                                     (done? (cdr state)))
                                                (if done?
                                                    (walk (cdr procs) (cdr st) (cons state acc))
                                                    (let* ((result ((car procs)
                                                                    db
                                                                    #:metadata? table-metadata?
                                                                    #:keyset? #t
                                                                    #:chunk-key (or chunk-key-field "Id")
                                                                    #:limit rows-per-chunk
                                                                    #:cursor cursor))
                                                           (rows (car result))
                                                           (next-cursor (cdr result))
                                                           (proc-done? (or (= rows 0)
                                                                           (< rows rows-per-chunk))))
                                                      (walk (cdr procs)
                                                            (cdr st)
                                                            (cons (cons next-cursor proc-done?) acc))))))))
                                      #:encoding "UTF-8")))
                              (loop (+ chunk-index 1) next-states))))
                        ;; Legacy OFFSET chunking.
                        (do ((i 0 (+ i 1)))
                            ((>= i chunks))
                          (let* ((offset (* i (or rows-per-chunk 0)))
                                 (out-file
                                  (if (= chunks 1)
                                      rdf-path
                                      (string-append (path-without-extension rdf-path)
                                                     "." (number->string (+ i 1)) ".ttl"))))
                            (with-output-to-file
                                out-file
                              (lambda ()
                                (for-each
                                 (match-lambda
                                   ((k v)
                                    (prefix k v)))
                                 prefixes)
                                (newline)
                                (for-each
                                 (lambda (proc)
                                   (proc db #:metadata? table-metadata?
                                         #:limit rows-per-chunk
                                         #:offset offset))
                                 inputs))
                              #:encoding "UTF-8")))))))))))