r9949: delete-directory-and-files update

[kmrcl.git] / strings.lisp

diff --git a/strings.lisp b/strings.lisp
index fd2f37ca071e8dc6a7f8f7b1a849cbf849d19738..9dbe1babdccd7950172033dc0d5c6d9614ff4921 100644 (file)
--- a/strings.lisp
+++ b/strings.lisp
@@ -153,12 +153,16 @@
 (defun is-string-empty (str)
   (zerop (length str)))
 
-(defvar *whitespace-chars* '(#\space #\tab #\return #\linefeed))
+(defvar *whitespace-chars* '(#\space #\tab #\return #\linefeed
+                            #+allegro #\%space
+                            #+lispworks #\No-Break-Space))
 
 (defun is-char-whitespace (c) 
   (declare (character c) (optimize (speed 3) (safety 0)))
   (or (char= c #\Space) (char= c #\Tab) (char= c #\Return)
-      (char= c #\Linefeed)))
+      (char= c #\Linefeed)
+      #+allegro (char= c #\%space)
+      #+lispworks (char= c #\No-Break-Space)))
 
 (defun is-string-whitespace (str)
   "Return t if string is all whitespace"
@@ -409,9 +413,9 @@ for characters in a string"
   (declare (type (integer 0 15) n))
   (schar +hex-chars+ n))
 
-(defconstant +char-code-lower-a+ (char-code #\a))
-(defconstant +char-code-upper-a+ (char-code #\A))
-(defconstant +char-code-0+ (char-code #\0))
+(defconst +char-code-lower-a+ (char-code #\a))
+(defconst +char-code-upper-a+ (char-code #\A))
+(defconst +char-code-0+ (char-code #\0))
 (declaim (type fixnum +char-code-0+ +char-code-upper-a+
               +char-code-0))
 
@@ -423,7 +427,7 @@ for characters in a string"
        (+ 10 (- code +char-code-upper-a+))
        (- code +char-code-0+))))
 
-(defun uriencode-string (query)
+(defun encode-uri-string (query)
   "Escape non-alphanumeric characters for URI fields"
   (declare (simple-string query)
           (optimize (speed 3) (safety 0) (space 0)))
@@ -446,7 +450,7 @@ for characters in a string"
            (setf (schar str dpos) (hexchar (logand c 15))))
        (setf (schar str dpos) ch)))))
 
-(defun uridecode-string (query)
+(defun decode-uri-string (query)
   "Unescape non-alphanumeric characters for URI fields"
   (declare (simple-string query)
           (optimize (speed 3) (safety 0) (space 0)))
@@ -472,9 +476,9 @@ for characters in a string"
 
 
 (eval-when (:compile-toplevel :load-toplevel :execute)
-  (defvar +unambigous-charset+
+  (defvar +unambiguous-charset+
     "abcdefghjkmnpqrstuvwxyz123456789ABCDEFGHJKLMNPQSRTUVWXYZ")
-  (defconstant +unambigous-length+ (length +unambigous-charset+)))
+  (defconstant +unambiguous-length+ (length +unambiguous-charset+)))
 
 (defun random-char (&optional (set :lower-alpha))
   (ecase set
@@ -487,8 +491,8 @@ for characters in a string"
         (code-char (+ +char-code-lower-a+ n)))))
     (:upper-alpha
      (code-char (+ +char-code-upper-a+ (random 26))))
-    (:unambigous
-     (schar +unambigous-charset+ (random +unambigous-length+)))
+    (:unambiguous
+     (schar +unambiguous-charset+ (random +unambiguous-length+)))
     (:upper-lower-alpha
      (let ((n (random 52)))
        (if (>= n 26)
@@ -548,11 +552,24 @@ for characters in a string"
        
 
 (defun string-maybe-shorten (str maxlen)
-  (let ((len (length str)))
-    (if (<= len maxlen)
-       str
-      (concatenate 'string (subseq str 0 (- maxlen 3)) "..."))))
+  (string-elide str maxlen :end))
 
+(defun string-elide (str maxlen position)
+  (declare (fixnum maxlen))
+  (let ((len (length str)))
+    (declare (fixnum len))
+    (cond
+     ((<= len maxlen)
+      str)
+     ((<= maxlen 3)
+      "...")
+     ((eq position :middle)
+      (multiple-value-bind (mid remain) (truncate maxlen 2)
+       (let ((end1 (- mid 1))
+             (start2 (- len (- mid 2) remain))) 
+         (concatenate 'string (subseq str 0 end1) "..." (subseq str start2)))))
+     ((or (eq position :end) t)
+      (concatenate 'string (subseq str 0 (- maxlen 3)) "...")))))
 
 (defun shrink-vector (str size)
   #+allegro
@@ -587,7 +604,12 @@ for characters in a string"
 
 (defun split-alphanumeric-string (string)
   "Separates a string at any non-alphanumeric chararacter"
-  (flet ((is-sep (char) (non-alphanumericp char)))
+  (declare (simple-string string)
+          (optimize (speed 3) (safety 0)))
+  (flet ((is-sep (char) 
+          (declare (character char))
+          (and (non-alphanumericp char)
+               (not (char= #\_ char)))))
     (let ((tokens nil))
       (do* ((token-start
              (position-if-not #'is-sep string) 
@@ -602,4 +624,55 @@ for characters in a string"
         (push (subseq string token-start token-end) tokens)))))
 
 
-
+(defun trim-non-alphanumeric (word)
+  "Strip non-alphanumeric characters from beginning and end of a word."
+  (declare (simple-string word)
+          (optimize (speed 3) (safety 0) (space 0)))
+  (let* ((start 0)
+        (len (length word))
+        (end len))
+    (declare (fixnum start end len))
+    (do ((done nil))
+       ((or done (= start end)))
+      (if (alphanumericp (schar word start))
+         (setq done t)
+       (incf start)))
+    (when (> end start)
+      (do ((done nil))
+         ((or done (= start end)))
+       (if (alphanumericp (schar word (1- end)))
+           (setq done t)
+         (decf end))))
+    (if (or (plusp start) (/= len end))
+       (subseq word start end)
+      word)))
+
+         
+
+(defun collapse-whitespace (s)
+  "Convert multiple whitespace characters to a single space character."
+  (declare (simple-string s)
+          (optimize (speed 3) (safety 0)))
+  (with-output-to-string (stream)
+    (do ((pos 0 (1+ pos))
+        (in-white nil)
+        (len (length s)))
+       ((= pos len))
+      (declare (fixnum pos len))
+      (let ((c (schar s pos)))
+       (declare (character c))
+       (cond
+        ((kl:is-char-whitespace c)
+         (unless in-white
+           (write-char #\space stream))
+         (setq in-white t))
+        (t
+         (setq in-white nil)
+         (write-char c stream)))))))
+
+(defun string->list (string)
+  (let ((eof (list nil)))
+    (with-input-from-string (stream string)
+      (do ((x (read stream nil eof) (read stream nil eof))
+           (l nil (cons x l)))
+          ((eq x eof) (nreverse l))))))