;; -*- coding: utf-8-mcs-er -*- (require 'char-db-util) (setq file-name-coding-system 'utf-8-mcs-er) (concord-assign-genre 'creator@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'journal-volume@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'article@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'book@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'classification@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'region@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'era@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'period@ruimoku "/usr/local/var/ruimoku/db") (concord-assign-genre 'journal@ruimoku "/usr/local/var/ruimoku/db") (defvar chise-wiki-view-url "view.cgi") (defvar chise-wiki-edit-url "edit.cgi") (defvar chise-wiki-bitmap-glyphs-url "http://chise.zinbun.kyoto-u.ac.jp/glyphs") (defvar chise-wiki-glyph-cgi-url "http://chise.zinbun.kyoto-u.ac.jp/chisewiki/glyph.cgi") (defvar chise-wiki-displayed-features nil) (defun decode-uri-string (string &optional coding-system) (if (> (length string) 0) (let ((i 0) dest) (setq string (mapconcat (lambda (char) (if (eq char ?+) " " (char-to-string char))) string "")) (while (string-match "%\\([0-9A-F][0-9A-F]\\)" string i) (setq dest (concat dest (substring string i (match-beginning 0)) (char-to-string (int-char (string-to-int (match-string 1 string) 16)))) i (match-end 0))) (decode-coding-string (concat dest (substring string i)) coding-system)))) (defun www-feature-type (feature-name) (or (char-feature-property feature-name 'type) (let ((str (symbol-name feature-name))) (cond ((string-match "\\*note\\(@[^*]+\\)?$" str) 'stext) ((string-match "\\*sources\\(@[^*]+\\)?$" str) 'domain-list) ((string-match "\\*" str) nil) ((string-match "^\\(->\\|<-\\)" str) 'relation) ((string-match "^ideographic-structure\\(@\\|$\\)" str) 'structure) )))) (defun www-feature-format (feature-name) (or (char-feature-property feature-name 'format) (let (fn parent ret) (setq fn feature-name) (while (and (setq parent (char-feature-name-parent fn)) (null (setq ret (char-feature-property parent 'format)))) (setq fn parent)) ret) '((name) " : " (value)))) (defun www-feature-value-format (feature-name) (or (char-feature-property feature-name 'value-format) (let (fn parent ret) (setq fn feature-name) (while (and (setq parent (char-feature-name-parent fn)) (null (setq ret (char-feature-property parent 'value-format)))) (setq fn parent)) ret) (let ((type (www-feature-type feature-name))) (cond ((eq type 'relation) 'space-separated-char-list) ((eq type 'structure) 'space-separated-ids) ((eq type 'stext) 'wiki-text) )) (if (find-charset feature-name) (if (and (= (charset-dimension feature-name) 2) (= (charset-chars feature-name) 94)) '("0x" (HEX) " (" (decimal) ") <" (ku-ten) "> " (prev-char) (next-char)) '("0x" (HEX) " (" (decimal) ") " (prev-char) (next-char)))))) (defun char-feature-name-at-domain (feature-name domain) (if domain (let ((name (symbol-name feature-name))) (cond ((string-match "@[^*]+$" name) (intern (format "%s/%s" name domain)) ) (t (intern (format "%s@%s" name domain)) ))) feature-name)) (defun char-feature-name-parent (feature-name) (let ((name (symbol-name feature-name))) (if (string-match "@[^@/*]+\\(/[^@/*]+\\)*$" name) (intern (substring name 0 (car (last (match-data) 2))))))) (defun char-feature-name-domain (feature-name) (let ((name (symbol-name feature-name))) (if (string-match "@[^@/*]+\\(/[^@/*]+\\)*$" name) (intern (substring name (1+ (match-beginning 0))))))) (defun char-feature-name-sans-versions (feature) (let ((feature-name (symbol-name feature))) (if (string-match "[@/]\\$rev=latest$" feature-name) (intern (substring feature-name 0 (match-beginning 0))) feature))) (defun www-get-feature-value (object feature) (let ((latest-feature (char-feature-name-at-domain feature '$rev=latest))) (cond ((characterp object) (mount-char-attribute-table latest-feature) (or (char-feature object latest-feature) (char-feature object feature)) ) (t (or (concord-object-get object latest-feature) (concord-object-get object feature)) )))) (defun get-previous-code-point (ccs code) (let ((chars (charset-chars ccs)) (dim (charset-dimension ccs)) (i 0) mask byte-min byte-max bytes dest) (cond ((= chars 94) (setq mask #x7F byte-min 33 byte-max 126) ) ((= chars 96) (setq mask #x7F byte-min 32 byte-max 127) ) ((= chars 128) (setq mask #x7F byte-min 0 byte-max #xFF) ) (t ; (= chars 256) (setq mask #xFF byte-min 0 byte-max #xFF) )) (setq bytes (make-vector dim 0)) (while (< i dim) (aset bytes i (logand (lsh code (* i -8)) mask)) (setq i (1+ i))) (setq i 0) (while (and (< i dim) (progn (aset bytes i (1- (aref bytes i))) (< (aref bytes i) byte-min))) (aset bytes i byte-max) (setq i (1+ i))) (when (< i dim) (setq dest (aref bytes 0) i 1) (while (< i dim) (setq dest (logior dest (lsh (aref bytes i) (* i 8))) i (1+ i))) dest))) (defun get-next-code-point (ccs code) (let ((chars (charset-chars ccs)) (dim (charset-dimension ccs)) (i 0) mask byte-min byte-max bytes dest) (cond ((= chars 94) (setq mask #x7F byte-min 33 byte-max 126) ) ((= chars 96) (setq mask #x7F byte-min 32 byte-max 127) ) ((= chars 128) (setq mask #x7F byte-min 0 byte-max #xFF) ) (t ; (= chars 256) (setq mask #xFF byte-min 0 byte-max #xFF) )) (setq bytes (make-vector dim 0)) (while (< i dim) (aset bytes i (logand (lsh code (* i -8)) mask)) (setq i (1+ i))) (setq i 0) (while (and (< i dim) (progn (aset bytes i (1+ (aref bytes i))) (> (aref bytes i) byte-max))) (aset bytes i byte-min) (setq i (1+ i))) (when (< i dim) (setq dest (aref bytes 0) i 1) (while (< i dim) (setq dest (logior dest (lsh (aref bytes i) (* i 8))) i (1+ i))) dest))) (defun find-previous-defined-code-point (ccs code) (let ((i (get-previous-code-point ccs code)) char) (cond ((eq ccs '=jis-x0208) (setq ccs '=jis-x0208@1990)) ((eq ccs '=jis-x0213-1) (setq ccs '=jis-x0213-1@2004))) (while (and i (>= i 0) (null (setq char (decode-char ccs i (unless (eq ccs '=ucs) 'defined-only))))) (setq i (get-previous-code-point ccs i))) char)) (defun find-next-defined-code-point (ccs code) (let ((i (get-next-code-point ccs code)) max char) (setq max (+ code 1000)) (cond ((eq ccs '=jis-x0208) (setq ccs '=jis-x0208@1990)) ((eq ccs '=jis-x0213-1) (setq ccs '=jis-x0213-1@2004))) (while (and i (<= i max) (null (setq char (decode-char ccs i (unless (eq ccs '=ucs) 'defined-only))))) (setq i (get-next-code-point ccs i))) char)) ;;; @ URI representation ;;; (defun www-uri-decode-feature-name (uri-feature) (let (feature) (cond ((string-match "^from\\." uri-feature) (intern (format "<-%s" (substring uri-feature (match-end 0)))) ) ((string-match "^to\\." uri-feature) (intern (format "->%s" (substring uri-feature (match-end 0)))) ) ((string-match "^rep\\." uri-feature) (intern (format "=%s" (substring uri-feature (match-end 0)))) ) ((string-match "^g\\." uri-feature) (intern (format "=>>%s" (substring uri-feature (match-end 0)))) ) ((string-match "^gi\\." uri-feature) (intern (format "=>>>%s" (substring uri-feature (match-end 0)))) ) ((string-match "^gi\\([0-9]+\\)\\." uri-feature) (intern (format "=>>%s%s" (make-string (string-to-int (match-string 1 uri-feature)) ?>) (substring uri-feature (match-end 0)))) ) ((string-match "^a\\." uri-feature) (intern (format "=>%s" (substring uri-feature (match-end 0)))) ) ((string-match "^a\\([0-9]+\\)\\." uri-feature) (intern (format "%s>%s" (make-string (string-to-int (match-string 1 uri-feature)) ?=) (substring uri-feature (match-end 0)))) ) ((and (setq feature (intern (format "=>%s" uri-feature))) (find-charset feature)) feature) ((and (setq feature (intern (format "=>>%s" uri-feature))) (find-charset feature)) feature) ((and (setq feature (intern (format "=>>>%s" uri-feature))) (find-charset feature)) feature) ((and (setq feature (intern (format "=%s" uri-feature))) (find-charset feature)) feature) (t (intern uri-feature))))) (defun www-uri-encode-feature-name (feature-name) (setq feature-name (symbol-name feature-name)) (cond ((string-match "^=\\([^=>]+\\)" feature-name) (concat "rep." (substring feature-name (match-beginning 1))) ) ((string-match "^=>>\\([^=>]+\\)" feature-name) (concat "g." (substring feature-name (match-beginning 1))) ) ((string-match "^=>>>\\([^=>]+\\)" feature-name) (concat "gi." (substring feature-name (match-beginning 1))) ) ((string-match "^=>>\\(>+\\)" feature-name) (format "gi%d.%s" (length (match-string 1 feature-name)) (substring feature-name (match-end 1))) ) ((string-match "^=>\\([^=>]+\\)" feature-name) (concat "a." (substring feature-name (match-beginning 1))) ) ((string-match "^\\(=+\\)>" feature-name) (format "a%d.%s" (length (match-string 1 feature-name)) (substring feature-name (match-end 0))) ) ((string-match "^->" feature-name) (concat "to." (substring feature-name (match-end 0))) ) ((string-match "^<-" feature-name) (concat "from." (substring feature-name (match-end 0))) ) (t feature-name))) (defun www-uri-make-feature-name-url (uri-feature-name uri-char) (format "%s?feature=%s&char=%s" chise-wiki-view-url uri-feature-name uri-char)) (defun www-uri-decode-object (genre char-rep) (let (ccs cpos) (cond ((string-match "\\(%3A\\|:\\)" char-rep) (setq ccs (substring char-rep 0 (match-beginning 0)) cpos (substring char-rep (match-end 0))) (setq ccs (www-uri-decode-feature-name ccs)) (cond ((string-match "^0x" cpos) (setq cpos (string-to-number (substring cpos (match-end 0)) 16)) ) (t (setq cpos (car (read-from-string cpos))) )) (if (and (eq genre 'character) (numberp cpos)) (decode-char ccs cpos) (concord-decode-object ccs cpos genre)) ) (t (setq char-rep (decode-uri-string char-rep 'utf-8-mcs-er)) (cond ((eq genre 'character) (when (= (length char-rep) 1) (aref char-rep 0)) ) ((eq genre 'feature) (concord-decode-object '=id (www-uri-decode-feature-name char-rep) 'feature) ) (t (concord-decode-object '=id (car (read-from-string char-rep)) genre) )))))) (defun www-uri-encode-object (object) (if (characterp object) (if (encode-char object '=ucs) (mapconcat (lambda (byte) (format "%%%02X" byte)) (encode-coding-string (char-to-string object) 'utf-8-mcs-er) "") (let ((ccs-list '(; =ucs =cns11643-1 =cns11643-2 =cns11643-3 =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7 =gb2312 =gb12345 =jis-x0208 =jis-x0208@1990 =jis-x0212 =cbeta =jef-china3 =jis-x0213-1@2000 =jis-x0213-1@2004 =jis-x0208@1983 =jis-x0208@1978 =zinbun-oracle =>zinbun-oracle =daikanwa =gt =gt-k =>>jis-x0208 =>>jis-x0213-1 =>jis-x0208 =>jis-x0213-1 =>>gt =ruimoku-v6 =big5 =big5-cdp)) ccs ret) (while (and ccs-list (setq ccs (pop ccs-list)) (not (setq ret (encode-char object ccs 'defined-only))))) (cond (ret (format "%s:0x%X" (www-uri-encode-feature-name ccs) ret)) ((and (setq ccs (car (split-char object))) (setq ret (encode-char object ccs))) (format "%s:0x%X" (www-uri-encode-feature-name ccs) ret)) (t (format "system-char-id:0x%X" (encode-char object 'system-char-id)) )))) (format "rep.id:%s" (concord-object-id object)))) (defun est-format-object (object) (if (characterp object) (char-to-string object) (format "%s" (concord-object-id object)))) ;;; @ Feature name presentation ;;; (defun www-format-feature-name-default (feature-name) (mapconcat #'capitalize (split-string (symbol-name feature-name) "-") " ")) (defun www-format-feature-name-as-metadata (feature-name &optional lang) (let ((str (symbol-name feature-name)) base meta) (cond ((string-match "\\*[^*]+$" str) (setq base (substring str 0 (match-beginning 0)) meta (substring str (match-beginning 0))) (concat (www-format-feature-name* (intern base) lang) meta)) (t (www-format-feature-name-default feature-name) )))) (defun www-format-feature-name-as-rel-to (feature-name) (concat "\u2192" (substring (symbol-name feature-name) 2))) (defun www-format-feature-name-as-rel-from (feature-name) (concat "\u2190" (substring (symbol-name feature-name) 2))) (defun www-format-feature-name-as-CCS (feature-name) (let* ((rest (split-string (symbol-name feature-name) "-")) (dest (upcase (pop rest)))) (when (string-match "^=+>*" dest) (setq dest (concat (substring dest 0 (match-end 0)) " " (substring dest (match-end 0))))) (cond (rest (while (cdr rest) (setq dest (concat dest " " (upcase (pop rest))))) (if (string-match "^[0-9]+$" (car rest)) (concat dest "-" (car rest)) (concat dest " " (upcase (car rest)))) ) (t dest)))) (defun www-format-feature-name* (feature-name &optional lang) (let (name fn parent ret) (cond ((or (and lang (char-feature-property feature-name (intern (format "name@%s" lang)))) (char-feature-property feature-name 'name))) ((and (setq name (symbol-name feature-name)) (string-match "\\*" name)) (www-format-feature-name-as-metadata feature-name lang)) (t (setq fn feature-name) (while (and (setq parent (char-feature-name-parent fn)) (null (setq ret (or (and lang (char-feature-property parent (intern (format "name@%s" lang)))) (char-feature-property parent 'name))))) (setq fn parent)) (cond (ret (concat ret (substring (symbol-name feature-name) (length (symbol-name parent))))) ((find-charset feature-name) (www-format-feature-name-as-CCS feature-name)) ((string-match "^\\(->\\)" name) (www-format-feature-name-as-rel-to feature-name)) ((string-match "^\\(<-\\)" name) (www-format-feature-name-as-rel-from feature-name)) (t (www-format-feature-name-default feature-name) )) )))) (defun www-format-feature-name (feature-name &optional lang) (www-format-encode-string (www-format-feature-name* feature-name lang))) ;;; @ Feature value presentation ;;; (defun www-format-value-as-kuten (value) (format "%02d-%02d" (- (lsh value -8) 32) (- (logand value 255) 32))) (defun www-format-value-default (value &optional without-tags) (if (listp value) (mapconcat (lambda (unit) (www-format-encode-string (format "%S" unit) without-tags)) value " ") (www-format-encode-string (format "%S" value) without-tags))) (defun www-format-value-as-char-list (value &optional without-tags) (if (listp value) (mapconcat (if without-tags (lambda (unit) (www-format-encode-string (format (if (characterp unit) "%c" "%s") unit) 'without-tags)) (let (genre-o name-f ret) (lambda (unit) (if (characterp unit) (format "%s" chise-wiki-view-url (www-uri-encode-object unit) (www-format-encode-string (char-to-string unit))) (format "%s" chise-wiki-view-url (concord-object-genre unit) (concord-object-id unit) (cond ((setq ret (www-get-feature-value unit (setq name-f (if (setq genre-o (concord-decode-object '=id (concord-object-genre unit) 'genre)) (www-get-feature-value genre-o 'name) 'name)))) (www-format-eval-feature-value unit name-f nil nil nil ret 'without-tags 'without-edit) ) (t (www-format-encode-string (format "%S" unit)) )) unit))))) value " ") (www-format-encode-string (format "%s" value) without-tags))) (defun www-format-value-as-domain-list (value &optional without-tags) (let (name source0 source num dest rest unit start end ddest) (if (listp value) (if without-tags (mapconcat (lambda (unit) (format "%s" unit)) value " ") (setq rest value) (while rest (setq unit (pop rest)) (if (symbolp unit) (setq name (symbol-name unit))) (setq dest (concat dest (cond ((string-match "^zob1968=" name) (setq source (intern (substring name 0 (match-end 0))) num (substring name (match-end 0))) (if (string-match "^\\([0-9]+\\)-\\([0-9]+\\)$" num) (setq start (string-to-number (match-string 1 num)) end (string-to-number (match-string 2 num))) (setq start (string-to-number num) end start)) (setq ddest (if (eq source source0) (format ", %04d" start start) (setq source0 source) (format " %s=%04d" (www-format-encode-string "\u4EAC大人\u6587研甲\u9AA8") start start))) (setq start (1+ start)) (while (<= start end) (setq ddest (concat ddest (format ", %04d" start start))) (setq start (1+ start))) ddest) (t (setq source unit) (if (eq source source0) "" (setq source0 source) (concat " " name)) ))))) dest) (www-format-encode-string (format "%s" value) without-tags)))) (defun www-format-value-as-ids (value &optional without-tags) (if (listp value) (mapconcat (if without-tags (lambda (unit) (www-format-encode-string (format (if (characterp unit) "%c" "%s") unit) 'without-tags)) (lambda (unit) (if (characterp unit) (format "%s" chise-wiki-view-url (www-uri-encode-object unit) (www-format-encode-string (char-to-string unit))) (www-format-encode-string (format "%s" unit))))) (ideographic-structure-to-ids value) " ") (www-format-encode-string (format "%s" value) without-tags))) (defun www-format-value-as-S-exp (value &optional without-tags) (www-format-encode-string (format "%S" value) without-tags)) (defun www-format-value-as-HEX (value) (if (integerp value) (format "%X" value) (www-format-value-as-S-exp value))) (defun www-format-value-as-CCS-default (value) (if (integerp value) (format "0x%s (%d)" (www-format-value-as-HEX value) value) (www-format-value-as-S-exp value))) (defun www-format-value-as-CCS-94x94 (value) (if (integerp value) (format "0x%s [%s] (%d)" (www-format-value-as-HEX value) (www-format-value-as-kuten value) value) (www-format-value-as-S-exp value))) (defun www-format-value-as-kangxi-radical (value) (if (and (integerp value) (<= 0 value) (<= value 214)) (www-format-encode-string (format "%c" (ideographic-radical value))) (www-format-value-as-S-exp value))) (defun www-format-value (object feature-name &optional value format without-tags without-edit) (unless value (setq value (www-get-feature-value object feature-name))) (www-format-apply-value object feature-name format nil value nil nil without-tags without-edit) ) ;;; @ format evaluator ;;; (defun www-format-encode-string (string &optional without-tags) (with-temp-buffer (insert string) (let (plane code start end char variants ret rret) (goto-char (point-min)) (while (search-forward "<" nil t) (replace-match "<" nil t)) (goto-char (point-min)) (while (search-forward ">" nil t) (replace-match ">" nil t)) (if without-tags (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er) (let ((coded-charset-entity-reference-alist (list* '(=gt "GT-" 5 d) '(=cns11643-1 "C1-" 4 X) '(=cns11643-2 "C2-" 4 X) '(=cns11643-3 "C3-" 4 X) '(=cns11643-4 "C4-" 4 X) '(=cns11643-5 "C5-" 4 X) '(=cns11643-6 "C6-" 4 X) '(=cns11643-7 "C7-" 4 X) '(=gb2312 "G0-" 4 X) '(=gb12345 "G1-" 4 X) '(=jis-x0208@1990 "J90-" 4 X) '(=jis-x0212 "JSP-" 4 X) '(=cbeta "CB" 5 d) '(=jis-x0208@1997 "J97-" 4 X) '(=jis-x0208@1978 "J78-" 4 X) '(=jis-x0208@1983 "J83-" 4 X) '(=ruimoku-v6 "RUI6-" 4 X) '(=zinbun-oracle "ZOB-" 4 d) '(=jef-china3 "JC3-" 4 X) '(=daikanwa "M-" 5 d) coded-charset-entity-reference-alist))) (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er) (goto-char (point-min)) (while (re-search-forward "&CB\\([0-9]+\\);" nil t) (setq code (string-to-int (match-string 1))) (replace-match (format "\"CB%05d\"" code chise-wiki-bitmap-glyphs-url (/ code 1000) code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&J\\(78\\|83\\|90\\|97\\|SP\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) (setq plane (match-string 1) code (string-to-int (match-string 2) 16)) (replace-match (format "\"J%s-%04X\"" plane code chise-wiki-bitmap-glyphs-url plane (- (lsh code -8) 32) (- (logand code 255) 32)) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&G\\([01]\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) (setq plane (string-to-int (match-string 1)) code (string-to-int (match-string 2) 16)) (replace-match (format "\"GB%d-%04X\"" plane code chise-wiki-bitmap-glyphs-url plane (- (lsh code -8) 32) (- (logand code 255) 32)) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&C\\([1-7]\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) (setq plane (string-to-int (match-string 1)) code (string-to-int (match-string 2) 16)) (replace-match (format "\"CNS%d-%04X\"" plane code chise-wiki-bitmap-glyphs-url plane code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&JC3-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) (setq code (string-to-int (match-string 1) 16)) (replace-match (format "\"JC3-%04X\"" code code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&\\(A-\\)?ZOB-\\([0-9]+\\);" nil t) (setq code (string-to-int (match-string 2))) (replace-match (format "\"ZOB-%04d\"" code chise-wiki-bitmap-glyphs-url code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&\\(G-\\)?GT-\\([0-9]+\\);" nil t) (setq code (string-to-int (match-string 2))) (replace-match (format "\"GT-%05d\"" code chise-wiki-glyph-cgi-url code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&\\(G-\\)?GT-K\\([0-9]+\\);" nil t) (setq code (string-to-int (match-string 2))) (replace-match (format "\"GT-K%05d\"" code chise-wiki-glyph-cgi-url code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&B-\\([0-9A-F]+\\);" nil t) (setq code (string-to-int (match-string 1) 16)) (replace-match (format "\"B-%04X\"" code chise-wiki-glyph-cgi-url code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&CDP-\\([0-9A-F]+\\);" nil t) (setq code (string-to-int (match-string 1) 16)) (replace-match (format "\"CDP-%04X\"" code chise-wiki-glyph-cgi-url code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&RUI6-\\([0-9A-F]+\\);" nil t) (setq code (string-to-int (match-string 1) 16)) (replace-match (format "\"RUI6-%04X\"" code chise-wiki-glyph-cgi-url code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&\\(UU\\+\\|U-\\)\\([0-9A-F]+\\);" nil t) (setq code (string-to-int (match-string 2) 16)) (replace-match (format "\"UU+%04X\"" code code) t 'literal)) (goto-char (point-min)) (while (re-search-forward "&MCS-\\([0-9A-F]+\\);" nil t) (setq code (string-to-int (match-string 1) 16)) (setq start (match-beginning 0) end (match-end 0)) (setq char (decode-char 'system-char-id code)) (cond ((and (setq variants (or (www-get-feature-value char '->subsumptive) (www-get-feature-value char '->denotational))) (progn (while (and variants (setq ret (www-format-encode-string (char-to-string (car variants)))) (string-match "&MCS-\\([0-9A-F]+\\);" ret)) (setq variants (cdr variants))) ret)) (unless (string-match "&MCS-\\([0-9A-F]+\\);" ret) (goto-char start) (delete-region start end) (insert ret)) ) ((setq ret (or (www-get-feature-value char 'ideographic-combination) (www-get-feature-value char 'ideographic-structure))) (setq ret (mapconcat (lambda (ch) (if (listp ch) (if (characterp (setq rret (find-char ch))) (setq ch rret))) (if (characterp ch) (www-format-encode-string (char-to-string ch) without-tags) (www-format-encode-string (format "%S" ch) without-tags))) ret "")) (when ret (goto-char start) (delete-region start end) (insert ret)) ))) )) ;; (goto-char (point-min)) ;; (while (search-forward ">-" nil t) ;; (replace-match "&GT-" t 'literal)) (buffer-string)))) (defun www-format-props-to-string (props &optional format) (unless format (setq format (plist-get props :format))) (concat "%" (plist-get props :flag) ;; (if (plist-get props :zero-padding) ;; "0") (if (plist-get props :len) (format "0%d" (let ((ret (plist-get props :len))) (if (stringp ret) (string-to-int ret) ret)))) (cond ((eq format 'decimal) "d") ((eq format 'hex) "x") ((eq format 'HEX) "X") ((eq format 'S-exp) "S") (t "s")))) (defun www-format-apply-value (object feature-name format props value &optional uri-object uri-feature without-tags without-edit) (let (ret) (setq ret (cond ((memq format '(decimal hex HEX)) (if (integerp value) (format (www-format-props-to-string props format) value) (www-format-encode-string (format "%s" value) without-tags)) ) ((eq format 'wiki-text) (if without-tags (www-xml-format-list value) (www-format-eval-list value object feature-name nil uri-object without-tags without-edit)) ) ((eq format 'S-exp) (www-format-encode-string (format (www-format-props-to-string props format) value) without-tags)) ((eq format 'ku-ten) (www-format-value-as-kuten value)) ((eq format 'kangxi-radical) (www-format-value-as-kangxi-radical value)) ((eq format 'space-separated-char-list) (www-format-value-as-char-list value without-tags)) ((eq format 'space-separated-ids) (www-format-value-as-ids value without-tags)) ((eq format 'space-separated-domain-list) (www-format-value-as-domain-list value without-tags)) ((eq format 'string) (www-format-encode-string (format "%s" value) without-tags) ) (t (www-format-value-default value without-tags) )) ) (if (or without-tags without-edit (eq (plist-get props :mode) 'peek)) ret (format "%s " ret chise-wiki-edit-url uri-object uri-feature format)))) (defun www-format-eval-feature-value (object feature-name &optional format lang uri-object value without-tags without-edit) (unless value (setq value (www-get-feature-value object feature-name))) (unless format (setq format (www-feature-value-format feature-name))) (cond ((symbolp format) (www-format-apply-value object feature-name format nil value uri-object (www-uri-encode-feature-name feature-name) without-tags without-edit) ) ((consp format) (cond ((null (cdr format)) (setq format (car format)) (www-format-apply-value object feature-name (car format) (nth 1 format) value uri-object (www-uri-encode-feature-name feature-name) without-tags without-edit) ) (t (www-format-eval-list format object feature-name lang uri-object without-tags without-edit) ))))) (defun www-format-eval-unit (exp object feature-name &optional lang uri-object value without-tags without-edit) (unless value (setq value (www-get-feature-value object feature-name))) (unless uri-object (setq uri-object (www-uri-encode-object object))) (cond ((stringp exp) (www-format-encode-string exp)) ((null exp) "") ((consp exp) (cond ((memq (car exp) '(value decimal hex HEX ku-ten kangxi-radical S-exp string default)) (let ((fn (plist-get (nth 1 exp) :feature)) domain domain-fn ret) (when fn (when (stringp fn) (setq fn (intern fn))) (setq domain (char-feature-name-domain feature-name)) (setq domain-fn (char-feature-name-at-domain fn domain)) (if (setq ret (www-get-feature-value object domain-fn)) (setq feature-name domain-fn value ret) (setq feature-name fn value (www-get-feature-value object fn))) (push feature-name chise-wiki-displayed-features) )) (if (eq (car exp) 'value) (www-format-eval-feature-value object feature-name (plist-get (nth 1 exp) :format) lang uri-object value without-tags without-edit) (www-format-apply-value object feature-name (car exp) (nth 1 exp) value uri-object (www-uri-encode-feature-name feature-name) without-tags without-edit)) ) ((eq (car exp) 'name) (let ((fn (plist-get (nth 1 exp) :feature)) domain domain-fn) (when fn (setq domain (char-feature-name-domain feature-name)) (when (stringp fn) (setq fn (intern fn))) (setq domain-fn (char-feature-name-at-domain fn domain)) (setq feature-name domain-fn))) (if without-tags (www-format-feature-name feature-name lang) (format "%s" (www-uri-make-feature-name-url (www-uri-encode-feature-name feature-name) uri-object) (www-format-feature-name feature-name lang)) ) ) ((eq (car exp) 'name-url) (let ((fn (plist-get (nth 1 exp) :feature)) domain domain-fn) (when fn (setq domain (char-feature-name-domain feature-name)) (when (stringp fn) (setq fn (intern fn))) (setq domain-fn (char-feature-name-at-domain fn domain)) (setq feature-name domain-fn))) (www-uri-make-feature-name-url (www-uri-encode-feature-name feature-name) uri-object) ) ((eq (car exp) 'domain-name) (let ((domain (char-feature-name-domain feature-name))) (if domain (format "@%s" domain)))) ((eq (car exp) 'prev-char) (if without-tags "" (let ((prev-char (find-previous-defined-code-point feature-name value))) (if prev-char (format "\n%s" chise-wiki-view-url (www-uri-encode-object prev-char) "" ;; (www-format-encode-string ;; (char-to-string prev-char)) ) ""))) ) ((eq (car exp) 'next-char) (if without-tags "" (let ((next-char (find-next-defined-code-point feature-name value))) (if next-char (format "%s" chise-wiki-view-url (www-uri-encode-object next-char) "" ;; (www-format-encode-string ;; (char-to-string next-char)) ) ""))) ) ((eq (car exp) 'link) (if without-tags (www-format-eval-list (nthcdr 2 exp) object feature-name lang uri-object without-tags without-edit) (format "%s" (www-format-eval-list (plist-get (nth 1 exp) :ref) object feature-name lang uri-object 'without-tags 'without-edit) (www-format-eval-list (nthcdr 2 exp) object feature-name lang uri-object without-tags without-edit))) ) (t (format "<%s >%s" (car exp) (www-format-eval-list (nthcdr 2 exp) object feature-name lang uri-object without-tags without-edit) (car exp))))))) (defun www-format-eval-list (format-list object feature-name &optional lang uri-object without-tags without-edit) (if (consp format-list) (mapconcat (lambda (exp) (www-format-eval-unit exp object feature-name lang uri-object nil without-tags without-edit)) format-list "") (www-format-eval-unit format-list object feature-name lang uri-object nil without-tags without-edit))) ;;; @ XML generator ;;; (defun www-xml-format-props (props) (let ((dest "") key val) (while props (setq key (pop props) val (pop props)) (if (symbolp key) (setq key (symbol-name key))) (if (eq (aref key 0) ?:) (setq key (substring key 1))) (setq dest (format "%s %s=\"%s\"" dest key (www-format-encode-string (format "%s" val) 'without-tags)))) dest)) (defun www-xml-format-unit (format-unit) (let (name props children ret) (cond ((stringp format-unit) (mapconcat (lambda (c) (cond ((eq c ?&) "&") ;; ((eq c ?<) "&lt;") ;; ((eq c ?>) "&gt;") (t (char-to-string c)))) (www-format-encode-string format-unit 'without-tags) "") ) ((consp format-unit) (setq name (car format-unit) props (nth 1 format-unit) children (nthcdr 2 format-unit)) (when (eq name 'link) (setq ret (plist-get props :ref)) (unless (stringp ret) (setq props (plist-remprop (copy-list props) :ref)) (setq children (cons (list* 'ref nil ret) children)))) (if children (format "<%s%s>%s" name (if props (www-xml-format-props props) "") (www-xml-format-list children) name) (format "<%s%s/>" name (www-xml-format-props props))) ) (t (format "%s" format-unit))))) (defun www-xml-format-list (format-list) (if (atom format-list) (www-xml-format-unit format-list) (mapconcat #'www-xml-format-unit format-list ""))) ;;; @ HTML generator ;;; (defun www-html-display-text (text) (princ (with-temp-buffer (insert text) (goto-char (point-min)) (while (search-forward "<" nil t) (replace-match "<" nil t)) (goto-char (point-min)) (while (search-forward ">" nil t) (replace-match ">" nil t)) (goto-char (point-min)) (while (re-search-forward "\\[\\[\\([^]|[]+\\)|\\([^][]+\\)\\]\\]" nil t) (replace-match (format "%s" (match-string 2) (match-string 1)) nil t)) (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er) (goto-char (point-min)) (while (search-forward ">-" nil t) (replace-match "&GT-" nil t)) (buffer-string)))) (defun www-html-display-paragraph (text) (princ "

") (www-html-display-text text) (princ "

\n")) ;;; @ for GlyphWiki ;;; (defvar coded-charset-GlyphWiki-id-alist '((=ucs "u" 4 x nil) (=ucs@JP "u" 4 x nil) (=ucs@jis "u" 4 x nil) (=ucs@gb "u" 4 x "-g") (=ucs@cns "u" 4 x "-t") (=ucs@ks "u" 4 x "-k") (=ucs@iso "u" 4 x "-u") (=ucs@unicode "u" 4 x "-us") (=adobe-japan1-6 "aj1-" 5 d nil) (=gt "gt-" 5 d nil) (=big5-cdp "cdp-" 4 x nil) (=cbeta "cb" 5 d nil) (=jis-x0208@1978/1pr "j78-" 4 x nil) (=jis-x0208@1978/-4pr "j78-" 4 x nil) (=jis-x0208@1978 "j78-" 4 x nil) (=jis-x0208@1983 "j83-" 4 x nil) (=jis-x0208@1990 "j90-" 4 x nil) (=jis-x0212 "jsp-" 4 x nil) (=jis-x0213-1@2000 "jx1-2000-" 4 x nil) (=jis-x0213-1@2004 "jx1-2004-" 4 x nil) (=jis-x0213-2 "jx2-" 4 x nil) (=cns11643-1 "c1-" 4 x nil) (=cns11643-2 "c2-" 4 x nil) (=cns11643-3 "c3-" 4 x nil) (=cns11643-4 "c4-" 4 x nil) (=cns11643-5 "c5-" 4 x nil) (=cns11643-6 "c6-" 4 x nil) (=cns11643-7 "c7-" 4 x nil) (=daikanwa "dkw-" 5 d nil) (=gt-k "gt-k" 5 d nil) (=jef-china3 "jc3-" 4 x nil) (=big5 "b-" 4 x nil) (=ks-x1001 "k0-" 4 x nil) )) (defun char-GlyphWiki-id (char) (let ((rest coded-charset-GlyphWiki-id-alist) spec ret code) (while (and rest (setq spec (pop rest)) (null (setq ret (char-feature char (car spec)))))) (when ret (or (and (memq (car spec) '(=ucs@unicode '=ucs@iso)) (cond ((and (or (encode-char char '=jis-x0208@1990) (encode-char char '=jis-x0212) (encode-char char '=jis-x0213-1)) (setq code (encode-char char '=ucs@jis))) (format "u%04x" code) ) ((and (or (encode-char char '=gb2312) (encode-char char '=gb12345)) (setq code (encode-char char '=ucs@gb))) (format "u%04x-g" code) ) ((and (or (encode-char char '=cns11643-1) (encode-char char '=cns11643-2) (encode-char char '=cns11643-3) (encode-char char '=cns11643-4) (encode-char char '=cns11643-5) (encode-char char '=cns11643-6) (encode-char char '=cns11643-7)) (setq code (encode-char char '=ucs@cns))) (format "u%04x-t" code) ) ((and (encode-char char '=ks-x1001) (setq code (encode-char char '=ucs@ks))) (format "u%04x-k" code) ))) (format (format "%s%%0%d%s%s" (nth 1 spec) (nth 2 spec) (nth 3 spec) (or (nth 4 spec) "")) ret))))) ;;; @ End. ;;; (provide 'cwiki-common) ;;; cwiki-common.el ends here