/[meta-cvs]/meta-cvs/F-26D2A2DCE1CF180F943606C59DC63A51.lisp
ViewVC logotype

Contents of /meta-cvs/F-26D2A2DCE1CF180F943606C59DC63A51.lisp

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.10 - (show annotations)
Mon Jul 8 04:34:49 2002 UTC (11 years, 9 months ago) by kaz
Branch: MAIN
Changes since 1.9: +4 -37 lines
* grab.lisp (added-removed): Function removed.
(mcvs-grab): Use new intersection-difference function
instead of added-removed.
1 ;;; This source file is part of the Meta-CVS program,
2 ;;; which is distributed under the GNU license.
3 ;;; Copyright 2002 Kaz Kylheku
4
5 (require "dirwalk")
6 (require "system")
7 (require "split")
8 (require "seqfuncs")
9 (require "mapping")
10 (require "types")
11 (require "chatter")
12 (require "options")
13 (require "checkout")
14 (require "remove")
15 (require "add")
16 (provide "grab")
17
18 (defun read-word-hash (&optional (byte-stream t))
19 (let ((word-hash (make-hash-table :test #'equalp))
20 token
21 (state :junk))
22 (labels ((new-token ()
23 (if token
24 (setf (fill-pointer token) 0)
25 (setf token (make-array '(8)
26 :element-type 'character
27 :adjustable t
28 :fill-pointer 0))))
29 (save-token ()
30 (unless (gethash token word-hash)
31 (setf (gethash token word-hash) token)
32 (setf token nil))))
33 (do ((byte (read-byte byte-stream nil)
34 (read-byte byte-stream nil)))
35 ((null byte) (if (eq state :word) (save-token)) word-hash)
36 (let ((ch (int-char byte)))
37 (ecase state
38 ((:junk)
39 (when (or (alpha-char-p ch) (digit-char-p ch)
40 (char= ch #\_))
41 (new-token)
42 (vector-push-extend ch token)
43 (setf state :word)))
44 ((:word)
45 (cond
46 ((or (alpha-char-p ch) (digit-char-p ch)
47 (char= ch #\_) (char= ch #\-))
48 (vector-push-extend ch token))
49 (t (setf state :junk)
50 (save-token))))))))))
51
52 (defun word-hash-file (name)
53 (with-open-file (s (parse-posix-namestring name)
54 :direction :input
55 :element-type 'unsigned-byte)
56 (read-word-hash s)))
57
58 (defun correlate-word-hashes (hash-1 hash-2)
59 (let ((hc-1 (hash-table-count hash-1))
60 (hc-2 (hash-table-count hash-2)))
61 (when (> hc-1 hc-2)
62 (psetf hash-1 hash-2 hash-2 hash-1
63 hc-1 hc-2 hc-2 hc-1))
64 (let ((common-count 0))
65 (maphash #'(lambda (key element)
66 (declare (ignore key))
67 (when (gethash element hash-2)
68 (incf common-count)))
69 hash-1)
70 (let ((total-count (- (+ hc-1 hc-2) common-count)))
71 (if (zerop total-count)
72 0
73 (/ common-count total-count))))))
74
75
76 (defun correlate-paths (path-1 path-2)
77 (let* ((split-1 (split-fields path-1 *path-sep*))
78 (split-2 (split-fields path-2 *path-sep*))
79 (longer (max (length split-1) (length split-2)))
80 (lcs-len (length (longest-common-subsequence split-1 split-2
81 :test #'string=))))
82 (case (- longer lcs-len)
83 ((0 1) 1)
84 ((2) 95/100)
85 ((3) 90/100)
86 ((4) 85/100)
87 (otherwise 80/100))))
88
89 (defun determine-common-words (common-hash added-or-removed-files)
90 (dolist (file added-or-removed-files common-hash)
91 (maphash #'(lambda (key value)
92 (declare (ignore value))
93 (let ((existing (gethash key common-hash)))
94 (if existing
95 (setf (gethash key common-hash)
96 (1+ existing))
97 (setf (gethash key common-hash)
98 1))))
99 (second file))))
100
101 (defun eliminate-common-words (common-hash files threshold)
102 (dolist (file files)
103 (maphash #'(lambda (key value)
104 (declare (ignore value))
105 (let ((count (gethash key common-hash)))
106 (if (and count (>= count threshold))
107 (remhash key (second file)))))
108 (second file))))
109
110 (defun move-candidates (added-files removed-files)
111 (let (pairs candidates taken-added taken-removed)
112 (dolist (added added-files)
113 (dolist (removed removed-files)
114 (let ((correlation (correlate-word-hashes (second added)
115 (second removed))))
116 (when (>= correlation 30/100)
117 (push (list added removed
118 (* correlation (correlate-paths (first added)
119 (first removed))))
120 pairs)))))
121 (setf pairs (sort pairs #'> :key #'third))
122 (dolist (pair pairs)
123 (unless (or (member (first pair) taken-added :test #'eq)
124 (member (second pair) taken-removed :test #'eq))
125 (push (first pair) taken-added)
126 (push (second pair) taken-removed)
127 (push pair candidates)))
128 (values candidates
129 (set-difference added-files taken-added :test #'eq)
130 (set-difference removed-files taken-removed :test #'eq))))
131
132 (defun mcvs-grab (global-options command-options module subdir branch)
133 (declare (ignore command-options))
134 (mcvs-checkout module subdir global-options
135 `(("d" ,*this-dir*) ("r" ,branch))
136 :no-generate t)
137 (in-sandbox-root-dir
138 (let ((old-paths (mapping-read *mcvs-map*))
139 invisible-old-paths new-paths added-files removed-files stable-files)
140 (chatter-info "Scanning directory structure.~%")
141 (multiple-value-setq (old-paths invisible-old-paths)
142 (separate-if #'real-path-exists old-paths
143 :key #'second))
144 (setf old-paths (mapcar #'(lambda (entry)
145 (destructuring-bind (object name &rest stuff)
146 entry
147 (list* object (abstract-to-real-path name)
148 stuff))) old-paths))
149
150 (for-each-file-info (fi *this-dir*)
151 (let* ((path (canonicalize-path (file-name fi))))
152 (cond
153 ((regular-p fi)
154 (push path new-paths))
155 ((directory-p fi)
156 (when (path-equal path *mcvs-dir*)
157 (skip))))))
158
159 (multiple-value-setq (stable-files removed-files added-files)
160 (intersection-difference old-paths new-paths
161 :key1 #'second
162 :test #'equal))
163
164 (chatter-terse "Analyzing ~a added files.~%" (length added-files))
165 (setf added-files (mapcar #'(lambda (name)
166 (list name (word-hash-file name)))
167 added-files))
168
169 (chatter-terse "Analyzing ~a removed files.~%" (length removed-files))
170 (setf removed-files (mapcar #'(lambda (map-entry)
171 (destructuring-bind (f-file path)
172 map-entry
173 (list path
174 (word-hash-file f-file)
175 f-file)))
176 removed-files))
177 (let ((common-word-hash (make-hash-table :test #'equalp)))
178 (determine-common-words common-word-hash added-files)
179 (determine-common-words common-word-hash removed-files)
180 (let ((threshold (max 5 (* 1/5 (+ (length added-files)
181 (length removed-files))))))
182 (eliminate-common-words common-word-hash added-files threshold)
183 (eliminate-common-words common-word-hash removed-files threshold)))
184 (chatter-terse "Determining move candidates.~%")
185 (multiple-value-bind (moved-files added-files removed-files)
186 (move-candidates added-files removed-files)
187 (let ((mapping (mapping-read *mcvs-map*))
188 (f-hash (make-hash-table :test #'equal)))
189 (when moved-files
190 (dolist (pair moved-files)
191 (destructuring-bind ((target-name hash-2)
192 (source-name hash-1 f-file) confidence)
193 pair
194 (declare (ignore hash-1 hash-2))
195 (chatter-terse "moving ~a -> ~a (confidence ~a%)~%"
196 source-name target-name (round (* confidence 100)))
197 (setf (gethash f-file f-hash) target-name)))
198
199 (setf mapping
200 (mapcar #'(lambda (entry)
201 (let ((replacement (gethash (first entry)
202 f-hash)))
203 (if replacement
204 (list (first entry)
205 (real-to-abstract-path replacement))
206 entry)))
207 mapping))
208 (mapping-write mapping *mcvs-map*))
209 (mapping-write mapping *mcvs-map-local*)
210 (dolist (entry mapping)
211 (when (gethash (first entry) f-hash)
212 (unlink (first entry))
213 (link (abstract-to-real-path (second entry)) (first entry))))
214 (dolist (entry stable-files)
215 (unlink (first entry))
216 (link (second entry) (first entry))))
217 (when removed-files
218 (mcvs-remove nil (mapcar #'first removed-files) :no-sync t))
219 (when added-files
220 (mcvs-add nil global-options nil (mapcar #'first added-files)))))))
221
222 (defun mcvs-grab-wrapper (global-options command-options args)
223 (flet ((error ()
224 (error "mcvs-grab: specify module name, branch and optional subdirectory.")))
225 (when (< (length args) 2)
226 (error))
227 (destructuring-bind (module branch &optional subdir &rest superfluous) args
228 (when superfluous
229 (error))
230 (mcvs-grab global-options command-options module subdir branch))))

  ViewVC Help
Powered by ViewVC 1.1.5