Merge pull request #169 from jolt-lang/spike/chez-bootstrap

Library class-shim host hooks + FFI byte I/O; remove built-in http-client

host/chez/cli.ss

@@ -20,7 +20,6 @@
 (load "host/chez/seed/image.ss")
 (load "host/chez/compile-eval.ss")
 (load "host/chez/png.ss")          ; jolt.png — a baked namespace before the snapshot
-(load "host/chez/http-client.ss")  ; jolt.http-client (libcurl)
 (load "host/chez/loader.ss")
 ;; jolt.ffi host primitives (memory / library loading) load AFTER the loader's
 ;; baked-ns snapshot, so a library's (require '[jolt.ffi]) still loads jolt.ffi's

host/chez/ffi.ss

@@ -80,6 +80,23 @@
 (def-var! "jolt.ffi" "read-bytes" ffi-read-bytes)
 (def-var! "jolt.ffi" "write-bytes" ffi-write-bytes)
 
+;; --- byte-array buffer I/O (binary-faithful) --------------------------------
+;; Move raw bytes between a jolt byte-array (jolt-array kind 'byte) and foreign
+;; memory, byte-exact (no UTF-8 / latin1 decode) — for socket recv/send and the
+;; zlib / OpenSSL buffers an HTTP client passes through. read-array returns a
+;; fresh byte-array of n bytes; write-array copies a byte-array's bytes into ptr
+;; and returns the count.
+(define (ffi-read-array ptr n)
+  (let* ((n (jnum->exact n)) (p (jnum->exact ptr)) (v (make-vector n 0)))
+    (do ((i 0 (+ i 1))) ((= i n)) (vector-set! v i (foreign-ref 'unsigned-8 p i)))
+    (make-jolt-array v 'byte)))
+(define (ffi-write-array ptr arr)
+  (let* ((v (jolt-array-vec arr)) (n (vector-length v)) (p (jnum->exact ptr)))
+    (do ((i 0 (+ i 1))) ((= i n)) (foreign-set! 'unsigned-8 p i (bitwise-and (exact (vector-ref v i)) #xff)))
+    n))
+(def-var! "jolt.ffi" "read-array" ffi-read-array)
+(def-var! "jolt.ffi" "write-array" ffi-write-array)
+
 ;; --- string / bytevector marshaling ------------------------------------------
 ;; A C string result already comes back as a jolt string (the `string` foreign
 ;; type). For a `void*` that points at a NUL-terminated C string, read it here.

host/chez/host-class.ss

@@ -49,7 +49,16 @@
     ("Charset" . "java.nio.charset.Charset") ("Base64" . "java.util.Base64")
     ("Exception" . "java.lang.Exception")
     ("IllegalArgumentException" . "java.lang.IllegalArgumentException")
+    ("IllegalStateException" . "java.lang.IllegalStateException")
+    ("RuntimeException" . "java.lang.RuntimeException")
+    ("UnsupportedOperationException" . "java.lang.UnsupportedOperationException")
     ("InterruptedException" . "java.lang.InterruptedException")
+    ("IOException" . "java.io.IOException")
+    ("UnknownHostException" . "java.net.UnknownHostException")
+    ("ConnectException" . "java.net.ConnectException")
+    ("SocketTimeoutException" . "java.net.SocketTimeoutException")
+    ("MalformedURLException" . "java.net.MalformedURLException")
+    ("SSLException" . "javax.net.ssl.SSLException")
     ("Throwable" . "java.lang.Throwable")))
 (for-each
   (lambda (pair) (def-var! "clojure.core" (car pair) (cdr pair)))
@@ -74,4 +83,11 @@
   '("java.lang.Long" "java.lang.Integer" "java.lang.Double" "java.lang.Float"
     "java.lang.Number" "java.lang.String" "java.lang.Boolean" "java.lang.Character"
     "java.lang.Object"
+    ;; exception classes compared against (class e): (= java.net.SocketTimeoutException (class e))
+    "java.lang.Exception" "java.lang.Throwable" "java.lang.RuntimeException"
+    "java.lang.IllegalArgumentException" "java.lang.IllegalStateException"
+    "java.lang.UnsupportedOperationException" "java.io.IOException"
+    "java.net.UnknownHostException" "java.net.ConnectException"
+    "java.net.SocketTimeoutException" "java.net.MalformedURLException"
+    "javax.net.ssl.SSLException"
     "clojure.lang.Keyword" "clojure.lang.Symbol" "clojure.lang.Ratio" "clojure.lang.Atom"))

host/chez/host-static.ss

@@ -514,9 +514,31 @@
                                   (error #f "NoSuchElementException")))))))
 
 ;; ---- String / BigInteger / MapEntry constructors ----------------------------
-;; (String. x) / (String. bytes): a bytevector decodes UTF-8; else stringify.
+;; (String. bytes [charset]) decodes bytes (a bytevector OR a jolt byte-array)
+;; with the named charset (UTF-8 default; ISO-8859-1/latin1/ascii = one byte per
+;; char); else stringify. clj-http-lite's body coercion is (String. ^[B body cs).
+(define (string-charset-name rest)
+  (if (pair? rest)
+      (let ((c (car rest)))
+        (cond ((string? c) c)
+              ((and (jhost? c) (string=? (jhost-tag c) "charset"))
+               (let ((p (assq 'name (jhost-state c)))) (if p (jolt-str-render-one (cdr p)) "UTF-8")))
+              (else "UTF-8")))
+      "UTF-8"))
+(define (decode-bytevector bv rest)
+  (let ((cs (ascii-string-down (string-charset-name rest))))
+    (cond
+      ((or (string=? cs "utf-8") (string=? cs "utf8")) (utf8->string bv))
+      ((or (string=? cs "iso-8859-1") (string=? cs "latin1") (string=? cs "iso8859-1")
+           (string=? cs "us-ascii") (string=? cs "ascii"))
+       (list->string (map integer->char (bytevector->u8-list bv))))
+      (else (guard (e (#t (list->string (map integer->char (bytevector->u8-list bv))))) (utf8->string bv))))))
 (register-class-ctor! "String"
-  (lambda (x . _) (cond ((bytevector? x) (utf8->string x)) ((string? x) x) (else (jolt-str-render-one x)))))
+  (lambda (x . rest)
+    (cond ((bytevector? x) (decode-bytevector x rest))
+          ((and (jolt-array? x) (eq? (jolt-array-kind x) 'byte)) (decode-bytevector (na-bytearray->bv x) rest))
+          ((string? x) x)
+          (else (jolt-str-render-one x)))))
 (register-class-ctor! "BigInteger"
   (lambda (v) (parse-int-or-throw v 10 "BigInteger")))
 (register-class-ctor! "MapEntry" (lambda (k v) (make-map-entry k v)))
@@ -561,7 +583,9 @@
     (let loop ((l lst) (i 0)) (if (null? l) bv (begin (bytevector-u8-set! bv i (car l)) (loop (cdr l) (+ i 1)))))))
 (register-class-statics! "URLEncoder" (list (cons "encode" url-encode)))
 (register-class-statics! "URLDecoder" (list (cons "decode" url-decode)))
-(register-class-statics! "Charset" (list (cons "forName" (lambda (nm) (make-jhost "charset" (list (cons 'name nm)))))))
+;; Charset/forName yields the canonical name STRING (not an opaque object) so it
+;; threads straight into (.getBytes s cs) / (String. bytes cs), which take a name.
+(register-class-statics! "Charset" (list (cons "forName" (lambda (nm) (jolt-str-render-one nm)))))
 
 ;; ---- Base64 (RFC 4648) ------------------------------------------------------
 (define b64-alphabet "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/")
@@ -658,3 +682,56 @@
   (lambda (name proc) (register-class-ctor! name proc) jolt-nil))
 (def-var! "clojure.core" "__register-class-statics!"
   (lambda (name members) (register-class-statics! name (jmap->static-alist members)) jolt-nil))
+
+;; ---- tagged-table method dispatch + pluggable instance? --------------------
+;; A jolt library can build stateful host objects with (jolt.host/tagged-table
+;; tag) and dispatch (.method obj ...) to handlers registered here, keyed by the
+;; table's "jolt/type" tag — the htable analogue of the jhost method registry
+;; above. jolt-lang/http-client uses this to emulate java.net URL /
+;; HttpURLConnection / java.io byte streams so clj-http-lite runs unchanged.
+(define tagged-methods-tbl (make-hashtable string-hash string=?))   ; tag-key -> (method-ht)
+(define (tag->method-key tag)
+  (if (keyword-t? tag)
+      (let ((ns (keyword-t-ns tag)))
+        (if (and ns (not (jolt-nil? ns))) (string-append ns "/" (keyword-t-name tag)) (keyword-t-name tag)))
+      (jolt-str-render-one tag)))
+(define (register-tagged-methods! tag members)
+  (let* ((key (tag->method-key tag))
+         (h (or (hashtable-ref tagged-methods-tbl key #f)
+                (let ((nh (make-hashtable string-hash string=?)))
+                  (hashtable-set! tagged-methods-tbl key nh) nh))))
+    (for-each (lambda (p) (hashtable-set! h (car p) (cdr p))) members)))
+
+;; htable arm: dispatch (.method obj a*) through the table's tag method registry;
+;; an unregistered method falls through (sorted colls are htables too).
+(define %hs-rmd-htable record-method-dispatch)
+(set! record-method-dispatch
+  (lambda (obj method-name rest-args)
+    (let ((tag (and (htable? obj) (hashtable-ref (htable-h obj) "jolt/type" #f))))
+      (let* ((mh (and tag (hashtable-ref tagged-methods-tbl (tag->method-key tag) #f)))
+             (f  (and mh (hashtable-ref mh method-name #f))))
+        (if f
+            (apply f obj (if (jolt-nil? rest-args) '() (seq->list rest-args)))
+            (%hs-rmd-htable obj method-name rest-args))))))
+
+(def-var! "clojure.core" "__register-class-methods!"
+  (lambda (tag members) (register-tagged-methods! tag (jmap->static-alist members)) jolt-nil))
+
+;; Pluggable instance? — a library registers (fn [class-name-string val] -> true
+;; | false | nil); nil means "not my class, fall through". First non-nil wins.
+(define user-instance-checks '())
+(define %hs-instance-check instance-check)
+(set! instance-check
+  (lambda (type-sym val)
+    (let ((tname (symbol-t-name type-sym)))
+      (let loop ((fs user-instance-checks))
+        (if (null? fs)
+            (%hs-instance-check type-sym val)
+            (let ((r ((car fs) tname val)))
+              (if (jolt-nil? r) (loop (cdr fs)) (if (jolt-truthy? r) #t #f))))))))
+(def-var! "clojure.core" "instance-check" instance-check)
+(def-var! "clojure.core" "__register-instance-check!"
+  (lambda (f) (set! user-instance-checks (append user-instance-checks (list f))) jolt-nil))
+
+;; (jolt.host/table? x) — is x a host tagged-table (the Janet-table replacement)?
+(def-var! "jolt.host" "table?" (lambda (x) (if (htable? x) #t #f)))

host/chez/host-table.ss

@@ -176,3 +176,12 @@
     ((htable-sorted-set? obj) '("PersistentTreeSet" "Sorted" "IPersistentSet"
                                 "Set" "java.util.Set" "Collection" "IPersistentCollection" "Object"))
     (else (%h-value-host-tags obj)))))
+
+;; (class e) on a throwable tagged-table (a library's ex-info envelope carrying a
+;; JVM :class, e.g. jolt-lang/http-client's UnknownHostException) reads that
+;; class name, so clojure.test's (thrown? Class …) / (= Class (class e)) match.
+(define %h-class jolt-class)
+(set! jolt-class (lambda (x)
+  (let ((c (and (htable? x) (hashtable-ref (htable-h x) "class" #f))))
+    (if (and c (string? c)) c (%h-class x)))))
+(def-var! "clojure.core" "class" jolt-class)

host/chez/http-client.ss

@@ -1,160 +0,0 @@
-;; http-client.ss (jolt-90sp) — jolt.http-client: a synchronous HTTP client.
-;;
-;; Backed by the system `curl` binary rather than a direct libcurl FFI: on Apple
-;; Silicon curl_easy_setopt is variadic and its value arg goes on the stack, where
-;; Chez's fixed-signature foreign-procedure can't place it (it would need a
-;; compiled C shim per platform). Shelling to curl uses the same mature native
-;; library — TLS, redirects, gzip and all — with no build step and identical
-;; behavior across platforms. Returns a Ring-ish {:status :headers :body}.
-;;
-;; def-var!'d into the jolt.http-client namespace; loaded by the CLI before the
-;; loader snapshot, so (require '[jolt.http-client]) resolves with no source file.
-
-;; single-quote an argument for the outer `sh -c` (loader.ss's sh-quote isn't
-;; loaded yet at this point).
-(define (hc-shq s)
-  (string-append "'"
-    (apply string-append
-      (map (lambda (c) (if (char=? c #\') "'\\''" (string c))) (string->list s)))
-    "'"))
-
-(define hc-status-sentinel "\nJOLTHTTPSTATUS:")
-
-(define kw-method   (keyword #f "method"))
-(define kw-url      (keyword #f "url"))
-(define kw-headers  (keyword #f "headers"))
-(define kw-body     (keyword #f "body"))
-(define kw-insecure (keyword #f "insecure?"))
-(define kw-follow   (keyword #f "follow?"))
-(define kw-timeout  (keyword #f "timeout-ms"))
-(define kw-status   (keyword #f "status"))
-(define kw-query    (keyword #f "query-params"))
-(define kw-ctype    (keyword #f "content-type"))
-
-;; :content-type :json -> "application/json"; a string passes through.
-(define (hc-ctype->str ct)
-  (cond ((keyword? ct)
-         (let ((n (keyword-t-name ct)))
-           (cond ((string=? n "json") "application/json")
-                 ((string=? n "xml") "application/xml")
-                 ((string=? n "form") "application/x-www-form-urlencoded")
-                 (else n))))
-        (else (jolt-str-render-one ct))))
-
-;; A per-request header file path, unique across processes (PID) and threads (a
-;; mutex-guarded monotonic counter). The previous unguarded `set!` + `mod 90000`
-;; raced: concurrent callers could compute the same path and clobber each other's
-;; -D header dump. getpid is a fast, non-blocking foreign call.
-(define c-getpid (begin (load-shared-object #f) (foreign-procedure "getpid" () int)))
-(define hc-tmp-mutex (make-mutex))
-(define hc-tmp-counter 0)
-(define (hc-tmp-path)
-  (let ((n (with-mutex hc-tmp-mutex (set! hc-tmp-counter (+ hc-tmp-counter 1)) hc-tmp-counter)))
-    (string-append (or (getenv "TMPDIR") "/tmp") "/jolt-http-"
-                   (number->string (c-getpid)) "-" (number->string n) ".hdr")))
-
-(define (hc-trim s)
-  (let* ((n (string-length s))
-         (a (let lp ((i 0)) (if (and (< i n) (char-whitespace? (string-ref s i))) (lp (+ i 1)) i)))
-         (b (let lp ((i n)) (if (and (> i a) (char-whitespace? (string-ref s (- i 1)))) (lp (- i 1)) i))))
-    (substring s a b)))
-(define (hc-lines s)
-  (let loop ((i 0) (start 0) (acc '()))
-    (cond ((>= i (string-length s)) (reverse (if (> i start) (cons (substring s start i) acc) acc)))
-          ((char=? (string-ref s i) #\newline)
-           (loop (+ i 1) (+ i 1) (if (> i start) (cons (substring s start i) acc) acc)))
-          (else (loop (+ i 1) start acc)))))
-;; raw header dump (curl -D) -> a jolt map {lowercased-name -> value}. With
-;; redirects there are several blocks; the LAST status line resets the map so the
-;; final response's headers win.
-(define (hc-parse-headers text)
-  (let ((m (jolt-hash-map)))
-    (for-each
-      (lambda (raw)
-        (let ((line (hc-trim raw)))
-          (cond
-            ((= 0 (string-length line)) #t)
-            ((and (>= (string-length line) 5) (string=? (substring line 0 5) "HTTP/"))
-             (set! m (jolt-hash-map)))             ; new response block
-            (else
-             (let ((ci (let scan ((i 0)) (cond ((>= i (string-length line)) #f)
-                                              ((char=? (string-ref line i) #\:) i)
-                                              (else (scan (+ i 1)))))))
-               (when (and ci (> ci 0))
-                 (set! m (jolt-assoc m (ascii-string-down (hc-trim (substring line 0 ci)))
-                                     (hc-trim (substring line (+ ci 1) (string-length line)))))))))))
-      (hc-lines text))
-    m))
-
-(define (hc-split-status out)
-  ;; out ends with "\nJOLTHTTPSTATUS:<code>"; return (values body code-int).
-  (let ((idx (let ((sl (string-length hc-status-sentinel)) (ol (string-length out)))
-               (let scan ((i (- ol sl)))
-                 (cond ((< i 0) #f)
-                       ((string=? (substring out i (+ i sl)) hc-status-sentinel) i)
-                       (else (scan (- i 1))))))))
-    (if idx
-        (values (substring out 0 idx)
-                (or (string->number (hc-trim (substring out (+ idx (string-length hc-status-sentinel))
-                                                         (string-length out)))) 0))
-        (values out 0))))
-
-(define (hc-request method url opts)
-  (when (or (not url) (jolt-nil? url)) (error #f "jolt.http-client: missing :url"))
-  (let* ((headers (jolt-get opts kw-headers))
-         (body (let ((x (jolt-get opts kw-body))) (and (not (jolt-nil? x)) x)))
-         (insecure? (jolt-truthy? (jolt-get opts kw-insecure)))
-         (follow? (let ((x (jolt-get opts kw-follow))) (or (jolt-nil? x) (jolt-truthy? x))))
-         (timeout (let ((x (jolt-get opts kw-timeout))) (if (jolt-nil? x) 30 (quotient (jnum->exact x) 1000))))
-         (hdrfile (hc-tmp-path))
-         (parts (list "curl -sS"
-                      (if follow? "-L" "")
-                      (if insecure? "-k" "")
-                      "--max-time" (number->string (max 1 timeout))
-                      "-D" (hc-shq hdrfile)
-                      "-X" (hc-shq (string-upcase method))
-                      "-w" (hc-shq (string-append hc-status-sentinel "%{http_code}"))
-                      "--compressed"))
-         (ctype (let ((x (jolt-get opts kw-ctype))) (and (not (jolt-nil? x)) (hc-ctype->str x))))
-         (parts (if ctype (append parts (list "-H" (hc-shq (string-append "Content-Type: " ctype)))) parts))
-         (parts (if (pmap? headers)
-                    (append parts (pmap-fold headers
-                                    (lambda (k v acc)
-                                      (cons "-H" (cons (hc-shq (string-append (jolt-str-render-one k) ": "
-                                                                              (jolt-str-render-one v))) acc)))
-                                    '()))
-                    parts))
-         ;; query-params: let curl URL-encode them onto the URL (-G appends as the
-         ;; query string for a GET).
-         (qp (let ((x (jolt-get opts kw-query))) (and (pmap? x) x)))
-         (parts (if qp
-                    (append parts
-                            (cons "-G"
-                                  (pmap-fold qp (lambda (k v acc)
-                                                  (cons "--data-urlencode"
-                                                        (cons (hc-shq (string-append (jolt-str-render-one k) "="
-                                                                                     (jolt-str-render-one v))) acc)))
-                                             '())))
-                    parts))
-         (parts (if body (append parts (list "--data-binary" (hc-shq (jolt-str-render-one body)))) parts))
-         (parts (append parts (list (hc-shq (jolt-str-render-one url)))))
-         (cmd (let join ((xs parts) (s "")) (if (null? xs) s
-                 (join (cdr xs) (if (string=? (car xs) "") s (string-append s (if (string=? s "") "" " ") (car xs)))))))
-         (out (jolt-sh-out cmd)))
-    (let-values (((bdy code) (hc-split-status out)))
-      (let ((hdr-text (guard (e (#t "")) (read-file-string hdrfile))))
-        (guard (e (#t #f)) (delete-file hdrfile))
-        (jolt-hash-map kw-status code
-                       kw-headers (hc-parse-headers hdr-text)
-                       kw-body bdy)))))
-
-;; --- public jolt.http-client API --------------------------------------------
-(define (hc-verb method) (lambda (url . opt) (hc-request method url (if (pair? opt) (car opt) (jolt-hash-map)))))
-(def-var! "jolt.http-client" "request"
-  (lambda (opts) (hc-request (let ((m (jolt-get opts kw-method))) (if (jolt-nil? m) "GET" (jolt-str-render-one m)))
-                             (jolt-get opts kw-url) opts)))
-(def-var! "jolt.http-client" "get"    (hc-verb "GET"))
-(def-var! "jolt.http-client" "head"   (hc-verb "HEAD"))
-(def-var! "jolt.http-client" "post"   (hc-verb "POST"))
-(def-var! "jolt.http-client" "put"    (hc-verb "PUT"))
-(def-var! "jolt.http-client" "delete" (hc-verb "DELETE"))

host/chez/io.ss

@@ -151,10 +151,32 @@
                                 ((reader-jhost? rdr) (drain-reader rdr))
                                 (else (jolt-str-render-one rdr))))))
 
+;; (slurp src :encoding "...") — pull the charset from the trailing kwargs.
+(define (slurp-encoding opts)
+  (let loop ((o opts))
+    (cond ((or (null? o) (null? (cdr o))) '())
+          ((and (keyword-t? (car o)) (string=? (keyword-t-name (car o)) "encoding"))
+           (list (jolt-str-render-one (cadr o))))
+          (else (loop (cddr o))))))
+;; drain a byte input-stream shim (tagged-table) one byte at a time to a bytevector.
+(define (drain-byte-stream src)
+  (let loop ((acc '()))
+    (let ((b (record-method-dispatch src "read" jolt-nil)))
+      (if (or (jolt-nil? b) (and (number? b) (< b 0)))
+          (u8-list->bytevector (reverse acc))
+          (loop (cons (bitwise-and (jnum->exact b) #xff) acc))))))
 (define (jolt-slurp src . opts)
   (cond
     ((jfile? src) (read-file-string (jfile-path src)))
     ((reader-jhost? src) (drain-reader src))
+    ;; bytes (a bytevector or a jolt byte-array): decode with :encoding (UTF-8
+    ;; default). clj-http-lite slurps response-body byte arrays.
+    ((bytevector? src) (decode-bytevector src (slurp-encoding opts)))
+    ((and (jolt-array? src) (eq? (jolt-array-kind src) 'byte))
+     (decode-bytevector (na-bytearray->bv src) (slurp-encoding opts)))
+    ;; a byte input-stream shim (e.g. clj-http-lite's :as :stream body): drain it.
+    ((and (htable? src) (jolt-truthy? (jolt-ref-get src (keyword "jolt" "input-stream"))))
+     (decode-bytevector (drain-byte-stream src) (slurp-encoding opts)))
     ((string? src) (read-file-string src))
     (else (error #f "slurp: unsupported source" src))))
 
@@ -230,6 +252,9 @@
     ((jolt-nil? x) jolt-nil)
     ((and (jhost? x) (member (jhost-tag x) '("string-reader" "pushback-reader" "writer")))
      (record-method-dispatch x "close" jolt-nil) jolt-nil)
+    ;; a library's stream shim (tagged-table) closes via its registered .close
+    ;; method (a no-op for in-memory streams); absent method -> no-op.
+    ((htable? x) (guard (e (#t jolt-nil)) (record-method-dispatch x "close" jolt-nil)) jolt-nil)
     ((jfile? x) jolt-nil)
     (else
      (let ((closef (jolt-get x (keyword #f "close") jolt-nil)))
@@ -284,4 +309,11 @@
             ((file-exists? (string-append (car roots) "/" nm)) (make-jfile (string-append (car roots) "/" nm)))
             (else (loop (cdr roots)))))))
 (def-var! "clojure.java.io" "resource" jolt-io-resource)
-(def-var! "clojure.java.io" "as-url" (lambda (x) (if (and (jhost? x) (string=? (jhost-tag x) "url")) x (make-url (jolt-str-render-one x)))))
+;; as-url honors a library-registered URL class (e.g. jolt-lang/http-client's full
+;; java.net.URL shim) so io/as-url and (URL. spec) agree; else the file-only jhost.
+(def-var! "clojure.java.io" "as-url"
+  (lambda (x)
+    (cond ((and (jhost? x) (string=? (jhost-tag x) "url")) x)
+          ((htable? x) x)
+          (else (let ((ctor (lookup-class class-ctors-tbl "URL")))
+                  (if ctor (ctor (jolt-str-render-one x)) (make-url (jolt-str-render-one x))))))))

host/chez/natives-array.ss

@@ -46,10 +46,20 @@
     (else (make-jolt-array
            (list->vector (map (lambda (c) (if (char? c) c (integer->char (exact (truncate c)))))
                               (seq->list (jolt-seq a)))) 'char))))
+;; (byte-array n [init]) | (byte-array coll). Also coerces the host's OTHER byte
+;; carrier — a Chez bytevector (what String/.getBytes produce) — and a string's
+;; UTF-8 bytes, so bytevector and byte-array interconvert across interop seams.
 (define (na-byte-array a . rest)
-  (if (number? a)
+  (cond
-      (make-jolt-array (make-vector (exact (na-idx a)) (na-byte-of (if (pair? rest) (car rest) 0))) 'byte)
+    ((number? a) (make-jolt-array (make-vector (exact (na-idx a)) (na-byte-of (if (pair? rest) (car rest) 0))) 'byte))
-      (make-jolt-array (list->vector (map na-byte-of (seq->list (jolt-seq a)))) 'byte)))
+    ((bytevector? a) (make-jolt-array (list->vector (bytevector->u8-list a)) 'byte))
+    ((string? a) (make-jolt-array (list->vector (bytevector->u8-list (string->utf8 a))) 'byte))
+    (else (make-jolt-array (list->vector (map na-byte-of (seq->list (jolt-seq a)))) 'byte))))
+;; jolt byte-array -> Chez bytevector (for String decode / utf8->string).
+(define (na-bytearray->bv arr)
+  (let* ((v (jolt-array-vec arr)) (n (vector-length v)) (bv (make-bytevector n)))
+    (do ((i 0 (+ i 1))) ((= i n)) (bytevector-u8-set! bv i (bitwise-and (exact (vector-ref v i)) #xff)))
+    bv))
 (define (na-make-array a . rest)    ; (make-array len) | (make-array type len ...)
   (make-jolt-array (make-vector (exact (na-idx (if (number? a) a (car rest)))) jolt-nil) 'object))
 (define (na-into-array a . rest)    (na-from-seq (if (pair? rest) (car rest) a) 'object))
@@ -195,3 +205,29 @@
     (cons "chunk" na-chunk) (cons "chunk-cons" na-chunk-cons)
     (cons "chunk-first" na-chunk-first) (cons "chunk-rest" na-chunk-rest)
     (cons "chunk-next" na-chunk-next) (cons "chunked-seq?" na-chunked-seq?)))
+
+;; --- clojure.java.io/copy ---------------------------------------------------
+;; Copy src -> dst, JVM-style. Raw bytes (byte-array / bytevector / string) and a
+;; jhost reader write in one shot; any other source (a stream shim with a .read
+;; method, e.g. jolt-lang/http-client's ByteArrayInputStream) drains via .read
+;; into a byte-array buffer and .write to dst — both reached through method
+;; dispatch, so a library's tagged-table streams work without the host knowing
+;; their layout. Lives here (not io.ss) because io.ss loads before byte-array.
+(define (jolt-io-copy src dst . _opts)
+  (define (write-all! bytes)
+    (record-method-dispatch dst "write" (list->cseq (list bytes 0 (vector-length (jolt-array-vec bytes))))))
+  (cond
+    ((or (bytevector? src) (string? src)
+         (and (jolt-array? src) (eq? (jolt-array-kind src) 'byte)))
+     (write-all! (na-byte-array src)))
+    ((and (jhost? src) (member (jhost-tag src) '("string-reader" "pushback-reader")))
+     (write-all! (na-byte-array (drain-reader src))))
+    (else
+     (let ((buf (na-byte-array 8192)))
+       (let loop ()
+         (let ((n (record-method-dispatch src "read" (list->cseq (list buf 0 8192)))))
+           (when (and (number? n) (> (jnum->exact n) 0))
+             (record-method-dispatch dst "write" (list->cseq (list buf 0 n)))
+             (loop)))))))
+  jolt-nil)
+(def-var! "clojure.java.io" "copy" jolt-io-copy)

host/chez/natives-str.ss

@@ -117,7 +117,17 @@
      (string=? (ascii-string-down s) (ascii-string-down (arg 0))))
     ((string=? method "compareTo")
      (let ((o (arg 0))) (cond ((string<? s o) -1.0) ((string>? s o) 1.0) (else 0.0))))
-    ((string=? method "getBytes") (string->utf8 s))
+    ((string=? method "getBytes")
+     ;; (.getBytes s) / (.getBytes s charset). UTF-8 default; ISO-8859-1/latin1/
+     ;; ascii encode one byte per char (clj-http-lite's body-encoding path).
+     (if (null? rest)
+         (string->utf8 s)
+         (let ((cn (ascii-string-down (if (string? (arg 0)) (arg 0) (jolt-str-render-one (arg 0))))))
+           (if (member cn '("iso-8859-1" "latin1" "iso8859-1" "us-ascii" "ascii"))
+               (let* ((n (string-length s)) (bv (make-bytevector n)))
+                 (do ((i 0 (+ i 1))) ((= i n) bv)
+                   (bytevector-u8-set! bv i (bitwise-and (char->integer (string-ref s i)) #xff))))
+               (string->utf8 s)))))
     ((string=? method "matches") (if (irregex-match (str-irx (arg 0)) s) #t #f))
     ((string=? method "replaceAll") (irregex-replace/all (str-irx (arg 0)) s (arg 1)))
     ((string=? method "replaceFirst") (irregex-replace (str-irx (arg 0)) s (arg 1)))

test/chez/ffi-binding-test.ss

@@ -39,6 +39,19 @@
                      (let [v (jolt.ffi/read p :int)] (jolt.ffi/free p) v))"))))
 (ok "sizeof :pointer is a word" (let ((n (jnum->exact (ev "(jolt.ffi/sizeof :pointer)")))) (or (= n 8) (= n 4))))
 
+;; byte-array buffer I/O: write a byte-array into foreign memory and read it back
+;; byte-exact (high bytes preserved, no UTF-8 mangling).
+(ok "byte-array roundtrip (binary-faithful)"
+    (jolt-truthy?
+      (ev "(let [src (byte-array [0 65 200 255 10])
+                  p (jolt.ffi/alloc 5)]
+              (jolt.ffi/write-array p src)
+              (let [back (jolt.ffi/read-array p 5)]
+                (jolt.ffi/free p)
+                (and (= 5 (alength back))
+                     (= 0 (aget back 0)) (= 65 (aget back 1))
+                     (= 200 (aget back 2)) (= 255 (aget back 3)) (= 10 (aget back 4)))))")))
+
 ;; a :blocking foreign call is collect-safe: a thread parked in it must not pin
 ;; the stop-the-world collector. (collect) here would throw "cannot collect when
 ;; multiple threads are active" if usleep weren't emitted __collect_safe.

test/chez/unit.edn

@@ -459,4 +459,17 @@
   {:suite "deftype-mutable" :expr "(do (defprotocol IB (gv [_]) (sv [_ v])) (deftype B [^:unsynchronized-mutable val] IB (gv [_] val) (sv [_ v] (set! val v))) (let [b (->B 1)] (sv b 42) (gv b)))" :expected "42"}
   {:suite "deftype-mutable" :expr "(do (defprotocol IC (bump [_]) (cur [_])) (deftype C [^:unsynchronized-mutable n] IC (bump [this] (set! n (inc n)) n) (cur [_] n)) (let [c (->C 5)] (bump c) (bump c) (bump c) (cur c)))" :expected "8"}
   {:suite "deftype-mutable" :expr "(do (defprotocol IH (g [_]) (s [_ v])) (deftype H [^:volatile-mutable st] IH (g [_] st) (s [this v] (set! (.-st this) v))) (let [h (->H :old)] (s h :new) (g h)))" :expected ":new"}
+  {:suite "hostshim" :expr "(do (__register-class-ctor! \"Box1\" (fn [v] (let [t (jolt.host/tagged-table :my/box)] (jolt.host/ref-put! t :v v) t))) (__register-class-methods! :my/box {\"getV\" (fn [self] (jolt.host/ref-get self :v))}) (.getV (Box1. 42)))" :expected "42"}
+  {:suite "hostshim" :expr "(do (__register-class-ctor! \"Box2\" (fn [v] (let [t (jolt.host/tagged-table :my/box2)] (jolt.host/ref-put! t :v v) t))) (__register-class-methods! :my/box2 {\"add\" (fn [self n] (+ n (jolt.host/ref-get self :v)))}) (.add (Box2. 10) 5))" :expected "15"}
+  {:suite "hostshim" :expr "(do (__register-class-ctor! \"Box3\" (fn [] (jolt.host/tagged-table :my/box3))) (__register-instance-check! (fn [cn val] (if (= cn \"Box3\") (= :my/box3 (jolt.host/ref-get val :jolt/type)) nil))) (instance? Box3 (Box3.)))" :expected "true"}
+  {:suite "hostshim" :expr "(do (__register-class-ctor! \"Box4\" (fn [] (jolt.host/tagged-table :my/box4))) (__register-instance-check! (fn [cn val] (if (= cn \"Box4\") (= :my/box4 (jolt.host/ref-get val :jolt/type)) nil))) (instance? Box4 \"not-a-box\"))" :expected "false"}
+  {:suite "hostshim" :expr "(jolt.host/table? (jolt.host/tagged-table :t))" :expected "true"}
+  {:suite "hostshim" :expr "(jolt.host/table? \"x\")" :expected "false"}
+  {:suite "bytes" :expr "(seq (byte-array \"hi\"))" :expected "(104 105)"}
+  {:suite "bytes" :expr "(seq (byte-array (.getBytes \"AB\" \"UTF-8\")))" :expected "(65 66)"}
+  {:suite "bytes" :expr "(alength (byte-array (.getBytes \"hello\" \"UTF-8\")))" :expected "5"}
+  {:suite "bytes" :expr "(String. (byte-array [104 105]))" :expected "hi"}
+  {:suite "bytes" :expr "(String. (byte-array [104 105]) \"UTF-8\")" :expected "hi"}
+  {:suite "bytes" :expr "(int (first (String. (byte-array [200]) \"ISO-8859-1\")))" :expected "200"}
+  {:suite "bytes" :expr "(String. (.getBytes \"round\" \"UTF-8\") \"UTF-8\")" :expected "round"}
 ]