X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=cwiki-common.el;h=b5bf4c03861b44b17ddad60e36da1a6c439601e8;hb=80a872505b801316f6918daca3b403dd42e33f99;hp=2fb5094c174abf98548f51df52630fc68b29c53a;hpb=6ab856b7616f26eb6ba423a0abafeab172f3737d;p=chise%2Fest.git diff --git a/cwiki-common.el b/cwiki-common.el index 2fb5094..b5bf4c0 100644 --- a/cwiki-common.el +++ b/cwiki-common.el @@ -1,14 +1,155 @@ ;; -*- coding: utf-8-mcs-er -*- (require 'char-db-util) +;; (require 'concord-images) +(setq file-name-coding-system 'utf-8-mcs-er) + +(concord-assign-genre 'code-point "/usr/local/var/chise-ipld/db") +(concord-assign-genre 'coded-character "/usr/local/var/chise-ipld/db") +(concord-assign-genre 'glyph "/usr/local/var/chise-ipld/db") + +(concord-assign-genre 'image-resource "/usr/local/var/photo/db") +(concord-assign-genre 'glyph-image "/usr/local/var/photo/db") + +(concord-assign-genre 'creator@ruimoku "/usr/local/var/ruimoku/db") +(concord-assign-genre 'person-name@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") +(concord-assign-genre 'journal-name@ruimoku "/usr/local/var/ruimoku/db") +(concord-assign-genre 'publisher@ruimoku "/usr/local/var/ruimoku/db") +(concord-assign-genre 'publisher-name@ruimoku "/usr/local/var/ruimoku/db") + +(mount-char-attribute-table '*instance@ruimoku/bibliography/title) +;; (mount-char-attribute-table '*instance@ruimoku/bibliography/content*note) + +(concord-assign-genre 'entry@zh-classical "/usr/local/var/kanbun/db") +;; (concord-assign-genre 'morpheme-entry@zh-classical "/usr/local/var/kanbun/db") +(concord-assign-genre 'word-class@zh-classical "/usr/local/var/kanbun/db") +(concord-assign-genre 'morpheme@zh-classical "/usr/local/var/kanbun/db") +(concord-assign-genre 'sentence@zh-classical "/usr/local/var/kanbun/db") +;; (concord-assign-genre 'sentence-entry@zh-classical "/usr/local/var/kanbun/db") + +(mount-char-attribute-table '*instance@morpheme-entry/zh-classical) + +(concord-assign-genre 'ud@zh-classical "/usr/local/var/kanbun/db") + + +(concord-assign-genre 'hng-card "/usr/local/var/hng-card/db") + +(mount-char-attribute-table '->HNG) +(mount-char-attribute-table '<-HNG) +(mount-char-attribute-table '->HNG@CN/manuscript) +(mount-char-attribute-table '<-HNG@CN/manuscript) +(mount-char-attribute-table '->HNG@CN/printed) +(mount-char-attribute-table '<-HNG@CN/printed) +(mount-char-attribute-table '->HNG@JP/manuscript) +(mount-char-attribute-table '<-HNG@JP/manuscript) +(mount-char-attribute-table '->HNG@JP/printed) +(mount-char-attribute-table '<-HNG@JP/printed) +(mount-char-attribute-table '->HNG@KR) +(mount-char-attribute-table '<-HNG@KR) +(mount-char-attribute-table '->HNG@MISC) +(mount-char-attribute-table '<-HNG@MISC) + + +(defvar est-hide-cgi-mode nil) +(defvar est-view-url-prefix "..") (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") + "http://www.chise.org/glyphs") + +(defvar chise-wiki-hng-bitmap-glyphs-url + "http://hng.chise.org/glyphs/HNG") (defvar chise-wiki-glyph-cgi-url - "http://chise.zinbun.kyoto-u.ac.jp/chisewiki/glyph.cgi") + "http://www.chise.org/chisewiki/glyph.cgi") + +(defvar chise-wiki-displayed-features nil) + +(defvar est-coded-charset-priority-list + '(; =ucs + =cns11643-1 =cns11643-2 =cns11643-3 + =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7 + =gb2312 =gb12345 + =jis-x0208 =jis-x0208@1990 + =jis-x0213-2 + =jis-x0212 + =adobe-japan1 + =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 + =>>>adobe-japan1 + =>>>jis-x0208 =>>>jis-x0213-1 =>>>jis-x0213-2 + =>>jis-x0208 =>>jis-x0213-1 =>>jis-x0213-2 + =>>adobe-japan1 + =+>jis-x0208 =+>jis-x0213-1 =+>jis-x0213-2 + =+>jis-x0208@1978 + =+>adobe-japan1 + =>jis-x0208 =>jis-x0213-1 + =>>gt + =>ucs@iso =>ucs@unicode + =>ucs@jis =>ucs@cns =>ucs@ks + =+>ucs@iso =+>ucs@unicode + =>>ucs@iso =>>ucs@unicode + =>>ucs@jis =>>ucs@cns =>>ucs@ks + =>>>ucs@iso =>>>ucs@unicode + =>>>ucs@jis =>>>ucs@cns =>>>ucs@ks + ===ucs@iso + =ruimoku-v6 + =big5 + =big5-cdp + =>cbeta + =mj + ==mj + ===mj + =ucs-itaiji-001 + =ucs-itaiji-002 + =ucs-itaiji-003 + =ucs-itaiji-004 + =ucs-itaiji-005 + =ucs-itaiji-006 + =ucs-itaiji-007 + =ucs-itaiji-008 + =ucs-itaiji-009 + =ucs-itaiji-010 + =ucs-itaiji-011 + =ucs-itaiji-084 + =ucs-var-001 + =ucs-var-002 + =ucs-var-003 + =ucs-var-004 + =ucs@unicode + ==ucs@unicode + ===ucs@unicode + ==>ucs@bucs + ===daikanwa/+p + ===gt + =>ucs@iwds-1 + =>ucs@component + =>ucs-itaiji-001 + =>ucs-itaiji-002 + =>ucs-itaiji-003 + =>ucs-itaiji-004 + =>ucs-itaiji-005 + =>ucs-itaiji-006 + =>ucs-itaiji-007 + ===adobe-japan1 + ===cns11643-1 ===cns11643-2 ===cns11643-3 + ===cns11643-4 ===cns11643-5 ===cns11643-6 ===cns11643-7 + )) (defun decode-uri-string (string &optional coding-system) (if (> (length string) 0) @@ -47,11 +188,35 @@ 'structure) )))) +(defun www-feature-format (feature-name) + (or (char-feature-property feature-name 'presentation-format) + (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) + (or (char-feature-property feature-name 'value-presentation-format) + (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 + (or (char-feature-property + parent 'value-presentation-format) + (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) + 'space-separated) ((eq type 'structure) 'space-separated-ids) ((eq type 'stext) @@ -62,17 +227,30 @@ (= (charset-chars feature-name) 94)) '("0x" (HEX) " (" (decimal) ") <" (ku-ten) "> " (prev-char) (next-char)) - '("0x" (HEX) " (" (decimal) ") " (prev-char) (next-char)))))) + '("0x" (HEX) " (" (decimal) ") " (prev-char) (next-char)))) + 'space-separated)) (defun char-feature-name-at-domain (feature-name domain) + (if domain + (let ((name (symbol-name feature-name))) + (cond + ((string-match "@[^*]+$" name) + (intern (format "%s/%s" name domain)) + ) + (t + (intern (format "%s@%s" name domain)) + ))) + feature-name)) + +(defun char-feature-name-parent (feature-name) (let ((name (symbol-name feature-name))) - (cond - ((string-match "@[^*]+$" name) - (intern (format "%s/%s" name domain)) - ) - (t - (intern (format "%s@%s" name domain)) - )))) + (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))) @@ -80,11 +258,27 @@ (intern (substring feature-name 0 (match-beginning 0))) feature))) -(defun www-char-feature (character feature) +(defun est-object-genre (object) + (if (characterp object) + 'character + (concord-object-genre object))) + +(defun www-get-feature-value (object feature) (let ((latest-feature (char-feature-name-at-domain feature '$rev=latest))) - (mount-char-attribute-table latest-feature) - (or (char-feature character latest-feature) - (char-feature character feature)))) + (cond + ((characterp object) + (mount-char-attribute-table latest-feature) + (or (char-feature object latest-feature) + (char-feature object feature)) + ) + (t + (or (condition-case nil + (concord-object-get object latest-feature) + (error nil)) + (condition-case nil + (concord-object-get object feature) + (error nil))) + )))) (defun get-previous-code-point (ccs code) (let ((chars (charset-chars ccs)) @@ -188,7 +382,9 @@ (setq ccs '=jis-x0213-1@2004))) (while (and i (>= i 0) - (null (setq char (decode-char ccs i 'defined-only)))) + (null (setq char (decode-char ccs i + (unless (eq ccs '=ucs) + 'defined-only))))) (setq i (get-previous-code-point ccs i))) char)) @@ -213,42 +409,105 @@ ;;; @ URI representation ;;; +(defun est-uri-decode-feature-name-body (uri-feature) + (let ((len (length uri-feature)) + (i 0) + ch dest) + (while (< i len) + (setq dest + (concat + dest + (if (eq (aref uri-feature i) ?\.) + (if (and (< (+ i 2) len) + (eq (aref uri-feature (+ i 2)) ?\.)) + (prog1 + (cond + ((eq (setq ch (aref uri-feature (1+ i))) ?\.) + "/") + ((eq ch ?-) + "*") + (t + (substring uri-feature i (+ i 3)) + )) + (setq i (+ i 3))) + (setq i (1+ i)) + ".") + (prog1 + (char-to-string (aref uri-feature i)) + (setq i (1+ i))))))) + dest)) + +(defun est-uri-encode-feature-name-body (feature) + (mapconcat (lambda (c) + (cond ((eq c ?*) + ".-.") + ((eq c ?/) + "...") + (t (char-to-string c)))) + feature "")) + (defun www-uri-decode-feature-name (uri-feature) (let (feature) + (setq uri-feature (decode-uri-string uri-feature 'utf-8-mcs-er)) (cond ((string-match "^from\\." uri-feature) - (intern (format "<-%s" (substring uri-feature (match-end 0)))) + (intern (format "<-%s" + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) ) ((string-match "^to\\." uri-feature) - (intern (format "->%s" (substring uri-feature (match-end 0)))) + (intern (format "->%s" + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) ) ((string-match "^rep\\." uri-feature) - (intern (format "=%s" (substring uri-feature (match-end 0)))) + (intern (format "=%s" + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) + ) + ((string-match "^rep[2i]\\." uri-feature) + (intern (format "===%s" + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) ) ((string-match "^g\\." uri-feature) - (intern (format "=>>%s" (substring uri-feature (match-end 0)))) + (intern (format "=>>%s" + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) ) - ((string-match "^gi\\." uri-feature) - (intern (format "=>>>%s" (substring uri-feature (match-end 0)))) + ((string-match "^g[i2]\\." uri-feature) + (intern (format "==%s" + (est-uri-decode-feature-name-body + (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)))) + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) + ) + ((string-match "^o\\." uri-feature) + (intern (format "=+>%s" + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) ) ((string-match "^a\\." uri-feature) - (intern (format "=>%s" (substring uri-feature (match-end 0)))) + (intern (format "=>%s" + (est-uri-decode-feature-name-body + (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)))) + (est-uri-decode-feature-name-body + (substring uri-feature (match-end 0))))) ) - ((and (setq feature (intern (format "=>%s" uri-feature))) + ((and (setq uri-feature (est-uri-decode-feature-name-body uri-feature)) + (setq feature (intern (format "=>%s" uri-feature))) (find-charset feature)) feature) ((and (setq feature (intern (format "=>>%s" uri-feature))) @@ -265,102 +524,173 @@ (defun www-uri-encode-feature-name (feature-name) (setq feature-name (symbol-name feature-name)) (cond + ((string-match "^=\\+>\\([^=>]+\\)" feature-name) + (concat "o." + (est-uri-encode-feature-name-body + (substring feature-name (match-beginning 1)))) + ) ((string-match "^=\\([^=>]+\\)" feature-name) - (concat "rep." (substring feature-name (match-beginning 1))) + (concat "rep." + (est-uri-encode-feature-name-body + (substring feature-name (match-beginning 1)))) + ) + ((string-match "^==\\([^=>]+\\)" feature-name) + (concat "g2." + (est-uri-encode-feature-name-body + (substring feature-name (match-beginning 1)))) + ) + ((string-match "^===\\([^=>]+\\)" feature-name) + (concat "repi." + (est-uri-encode-feature-name-body + (substring feature-name (match-beginning 1)))) ) ((string-match "^=>>\\([^=>]+\\)" feature-name) - (concat "g." (substring feature-name (match-beginning 1))) + (concat "g." + (est-uri-encode-feature-name-body + (substring feature-name (match-beginning 1)))) ) ((string-match "^=>>>\\([^=>]+\\)" feature-name) - (concat "gi." (substring feature-name (match-beginning 1))) + (concat "gi." + (est-uri-encode-feature-name-body + (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))) + (est-uri-encode-feature-name-body + (substring feature-name (match-end 1)))) ) ((string-match "^=>\\([^=>]+\\)" feature-name) - (concat "a." (substring feature-name (match-beginning 1))) + (concat "a." + (est-uri-encode-feature-name-body + (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))) + (est-uri-encode-feature-name-body + (substring feature-name (match-end 0)))) ) ((string-match "^->" feature-name) - (concat "to." (substring feature-name (match-end 0))) + (concat "to." + (est-uri-encode-feature-name-body + (substring feature-name (match-end 0)))) ) ((string-match "^<-" feature-name) - (concat "from." (substring feature-name (match-end 0))) + (concat "from." + (est-uri-encode-feature-name-body + (substring feature-name (match-end 0)))) ) - (t feature-name))) + (t (est-uri-encode-feature-name-body feature-name)))) -(defun www-uri-decode-char (char-rep) +(defun www-uri-make-feature-name-url (uri-genre uri-feature-name uri-object) + (if est-hide-cgi-mode + (format "../feature/%s&%s/%s" + uri-feature-name uri-genre uri-object) + (format "%s?feature=%s&%s=%s" + chise-wiki-view-url uri-feature-name uri-genre uri-object))) + +(defun www-uri-decode-object (genre char-rep) (let (ccs cpos) (cond - ((string-match "\\(%3A\\|:\\)" char-rep) + ((string-match (if est-hide-cgi-mode + "\\(%3D\\|=\\|%3A\\|:\\)" + "\\(%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)) + (setq cpos (est-uri-decode-feature-name-body cpos)) (cond ((string-match "^0x" cpos) (setq cpos (string-to-number (substring cpos (match-end 0)) 16)) ) (t - (setq cpos (string-to-number cpos)) + (setq cpos (car (read-from-string + (decode-uri-string + cpos file-name-coding-system)))) )) - (if (numberp cpos) - (decode-char ccs 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)) - (when (= (length char-rep) 1) - (aref char-rep 0)) - )))) - -(defun www-uri-encode-char (char) - (if (encode-char char '=ucs) - (mapconcat - (lambda (byte) - (format "%%%02X" byte)) - (encode-coding-string (char-to-string char) '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 char ccs 'defined-only))))) - (cond (ret - (format "%s:0x%X" - (www-uri-encode-feature-name ccs) - ret)) - ((and (setq ccs (car (split-char char))) - (setq ret (encode-char char ccs))) - (format "%s:0x%X" - (www-uri-encode-feature-name ccs) - ret)) - (t - (format "system-char-id:0x%X" - (encode-char char 'system-char-id)) - ))))) + (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 est-coded-charset-priority-list) + ccs ret) + (while (and ccs-list + (setq ccs (pop ccs-list)) + (not (setq ret (encode-char object ccs 'defined-only))))) + (cond (ret + (format (if est-hide-cgi-mode + "%s=0x%X" + "%s:0x%X") + (www-uri-encode-feature-name ccs) + ret)) + ((and (setq ccs (car (split-char object))) + (setq ret (encode-char object ccs))) + (format (if est-hide-cgi-mode + "%s=0x%X" + "%s:0x%X") + (www-uri-encode-feature-name ccs) + ret)) + (t + (format (if est-hide-cgi-mode + "system-char-id=0x%X" + "system-char-id:0x%X") + (encode-char object 'system-char-id)) + )))) + (format (if est-hide-cgi-mode + "rep.id=%s" + "rep.id:%s") + (est-uri-encode-feature-name-body + (format "%s" (concord-object-id object)))))) + +(defun est-format-object (object &optional readable) + (if (characterp object) + (char-to-string object) + (let ((ret (or (if readable + (or (concord-object-get object 'name) + (concord-object-get object '=name) + (concord-object-get object 'title))) + (concord-object-id object)))) + (format "%s" ret)))) + +(defun www-uri-make-object-url (object &optional uri-object) + (if est-hide-cgi-mode + (format "%s/%s/%s" + est-view-url-prefix + (est-object-genre object) + (or uri-object + (www-uri-encode-object object))) + (format "%s?%s=%s" + chise-wiki-view-url + (est-object-genre object) + (or uri-object + (www-uri-encode-object object))))) ;;; @ Feature name presentation @@ -414,7 +744,7 @@ (t dest)))) (defun www-format-feature-name* (feature-name &optional lang) - (let (name) + (let (name fn parent ret) (cond ((or (and lang (char-feature-property @@ -422,189 +752,53 @@ (intern (format "name@%s" lang)))) (char-feature-property feature-name 'name))) - ((find-charset feature-name) - (www-format-feature-name-as-CCS feature-name)) ((and (setq name (symbol-name feature-name)) (string-match "\\*" name)) (www-format-feature-name-as-metadata feature-name lang)) - ((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))))) + (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 +;;; @ HTML generator ;;; -(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)) - (lambda (unit) - (if (characterp unit) - (format "%s" - chise-wiki-view-url - (www-uri-encode-char unit) - (www-format-encode-string (char-to-string unit))) - (www-format-encode-string (format "%s" 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\u6587ç ç²\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-char 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-char-feature object feature-name))) - (www-format-apply-value object feature-name - format nil value nil nil - without-tags without-edit) - ) - - -;;; @ format evaluator -;;; +(defvar www-format-char-img-style "vertical-align:bottom;") -(defun www-format-encode-string (string &optional without-tags) +(defun www-format-encode-string (string &optional without-tags as-body) (with-temp-buffer (insert string) - (let (plane code start end char variants ret rret) + (let (plane code subcode start end char variants ret rret) + (when as-body + (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)) @@ -616,6 +810,41 @@ (let ((coded-charset-entity-reference-alist (list* '(=gt "GT-" 5 d) + '(=mj "MJ" 6 d) + '(=hanyo-denshi/ja "HD-JA-" 4 X) + '(=hanyo-denshi/jb "HD-JB-" 4 X) + '(=hanyo-denshi/jc "HD-JC-" 4 X) + '(=hanyo-denshi/jd "HD-JD-" 4 X) + '(=hanyo-denshi/ft "HD-FT-" 4 X) + '(=hanyo-denshi/ia "HD-IA-" 4 X) + '(=hanyo-denshi/ib "HD-IB-" 4 X) + '(=hanyo-denshi/hg "HD-HG-" 4 X) + '(=hanyo-denshi/ip "HD-IP-" 4 X) + '(=hanyo-denshi/jt "HD-JT-" 4 X) + '(=hanyo-denshi/ks "HD-KS-" 6 d) + '(=>>hanyo-denshi/ja "G-HD-JA-" 4 X) + '(=>>hanyo-denshi/jb "G-HD-JB-" 4 X) + '(=>>hanyo-denshi/jc "G-HD-JC-" 4 X) + '(=>>hanyo-denshi/jd "G-HD-JD-" 4 X) + '(=>>hanyo-denshi/ft "G-HD-FT-" 4 X) + '(=>>hanyo-denshi/ia "G-HD-IA-" 4 X) + '(=>>hanyo-denshi/ib "G-HD-IB-" 4 X) + '(=>>hanyo-denshi/hg "G-HD-HG-" 4 X) + '(=>>hanyo-denshi/ip "G-HD-IP-" 4 X) + '(=>>hanyo-denshi/jt "G-HD-JT-" 4 X) + '(=>>hanyo-denshi/ks "G-HD-KS-" 6 d) + '(==mj "g2-MJ" 6 d) + '(==hanyo-denshi/ja "g2-HD-JA-" 4 X) + '(==hanyo-denshi/jb "g2-HD-JB-" 4 X) + '(==hanyo-denshi/jc "g2-HD-JC-" 4 X) + '(==hanyo-denshi/jd "g2-HD-JD-" 4 X) + '(==hanyo-denshi/ft "g2-HD-FT-" 4 X) + '(==hanyo-denshi/ia "g2-HD-IA-" 4 X) + '(==hanyo-denshi/ib "g2-HD-IB-" 4 X) + '(==hanyo-denshi/hg "g2-HD-HG-" 4 X) + '(==hanyo-denshi/ip "g2-HD-IP-" 4 X) + '(==hanyo-denshi/jt "g2-HD-JT-" 4 X) + '(==hanyo-denshi/ks "g2-HD-KS-" 6 d) '(=cns11643-1 "C1-" 4 X) '(=cns11643-2 "C2-" 4 X) '(=cns11643-3 "C3-" 4 X) @@ -623,6 +852,9 @@ '(=cns11643-5 "C5-" 4 X) '(=cns11643-6 "C6-" 4 X) '(=cns11643-7 "C7-" 4 X) + '(=adobe-japan1-6 "AJ1-" 5 d) + '(=big5-cdp "CDP-" 4 X) + '(=>big5-cdp "A-CDP-" 4 X) '(=gb2312 "G0-" 4 X) '(=gb12345 "G1-" 4 X) '(=jis-x0208@1990 "J90-" 4 X) @@ -634,31 +866,111 @@ '(=ruimoku-v6 "RUI6-" 4 X) '(=zinbun-oracle "ZOB-" 4 d) '(=jef-china3 "JC3-" 4 X) - '(=daikanwa "M-" 5 d) + '(=ucs@unicode "UU+" 4 X) + '(=ucs@JP/hanazono "hanaJU+" 4 X) + '(==cns11643-1 "R-C1-" 4 X) + '(==cns11643-2 "R-C2-" 4 X) + '(==cns11643-3 "R-C3-" 4 X) + '(==cns11643-4 "R-C4-" 4 X) + '(==cns11643-5 "R-C5-" 4 X) + '(==cns11643-6 "R-C6-" 4 X) + '(==cns11643-7 "R-C7-" 4 X) + '(=hanziku-1 "HZK01-" 4 X) + '(=hanziku-2 "HZK02-" 4 X) + '(=hanziku-3 "HZK03-" 4 X) + '(=hanziku-4 "HZK04-" 4 X) + '(=hanziku-5 "HZK05-" 4 X) + '(=hanziku-6 "HZK06-" 4 X) + '(=hanziku-7 "HZK07-" 4 X) + '(=hanziku-8 "HZK08-" 4 X) + '(=hanziku-9 "HZK09-" 4 X) + '(=hanziku-10 "HZK10-" 4 X) + '(=hanziku-11 "HZK11-" 4 X) + '(=hanziku-12 "HZK12-" 4 X) + '(==cbeta "CB" 5 d) + '(=big5 "B-" 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))) + (while (re-search-forward "&\\(A-\\|G-\\|g2-\\|R-\\)?CB\\([0-9]+\\);" nil t) + (setq code (string-to-int (match-string 2))) (replace-match - (format "" + (format "" code chise-wiki-bitmap-glyphs-url - (/ code 1000) code) + (/ code 1000) code + www-format-char-img-style) 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)) + (while (re-search-forward "&\\(o-\\|G-\\|g2-\\|R-\\)?J\\(78\\|83\\|90\\|97\\|SP\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) + (setq plane (match-string 2) + code (string-to-int (match-string 3) 16)) + (replace-match + (format "" + plane code + chise-wiki-bitmap-glyphs-url + plane + (- (lsh code -8) 32) + (- (logand code 255) 32) + www-format-char-img-style) + t 'literal)) + + (goto-char (point-min)) + (while (re-search-forward "&\\(o-\\|G-\\|g2-\\|R-\\)?J0-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) + (setq code (string-to-int (match-string 2) 16)) + (replace-match + (format "" + code + chise-wiki-bitmap-glyphs-url + (- (lsh code -8) 32) + (- (logand code 255) 32) + www-format-char-img-style) + t 'literal)) + + (goto-char (point-min)) + (while (re-search-forward "&\\(o-\\|G-\\|g2-\\|R-\\)?HD-\\(JA\\|JB\\|JC\\|JD\\|FT\\|IA\\|IB\\|HG\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) + (setq plane (match-string 2) + code (string-to-int (match-string 3) 16)) (replace-match - (format "" + (format "" plane code chise-wiki-bitmap-glyphs-url plane (- (lsh code -8) 32) - (- (logand code 255) 32)) + (- (logand code 255) 32) + www-format-char-img-style) + t 'literal)) + + (goto-char (point-min)) + (while (re-search-forward "&\\(o-\\|G-\\|g2-\\|R-\\)?HD-\\(IP\\|JT\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) + (setq plane (match-string 2) + code (string-to-int (match-string 3) 16)) + (replace-match + (format "" + plane code + chise-wiki-bitmap-glyphs-url + plane code + www-format-char-img-style) + t 'literal)) + + (goto-char (point-min)) + (while (re-search-forward "&\\(o-\\|G-\\|g2-\\|R-\\)?HD-KS-\\([0-9]+\\);" nil t) + (setq code (string-to-int (match-string 2))) + (replace-match + (format "" + code + chise-wiki-bitmap-glyphs-url + code + www-format-char-img-style) t 'literal)) (goto-char (point-min)) @@ -666,28 +978,32 @@ (setq plane (string-to-int (match-string 1)) code (string-to-int (match-string 2) 16)) (replace-match - (format "" + (format "" plane code chise-wiki-bitmap-glyphs-url plane (- (lsh code -8) 32) - (- (logand code 255) 32)) + (- (logand code 255) 32) + www-format-char-img-style) 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)) + (while (re-search-forward "&\\(R-\\)?C\\([1-7]\\)-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) + (setq plane (string-to-int (match-string 2)) + code (string-to-int (match-string 3) 16)) (replace-match - (format "" + (format "" plane code chise-wiki-bitmap-glyphs-url - plane code) + plane code + www-format-char-img-style) 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)) + (while (re-search-forward "&\\(R-\\)?JC3-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t) + (setq code (string-to-int (match-string 2) 16)) (replace-match (format "" code code) @@ -697,69 +1013,260 @@ (while (re-search-forward "&\\(A-\\)?ZOB-\\([0-9]+\\);" nil t) (setq code (string-to-int (match-string 2))) (replace-match - (format "" + (format "" + code + chise-wiki-bitmap-glyphs-url + code + www-format-char-img-style) + t 'literal)) + + (goto-char (point-min)) + (while (re-search-forward "&SW-JIGUGE\\([45]?\\)-\\([0-9]+\\);" nil t) + (setq subcode (match-string 1) + code (string-to-int (match-string 2))) + (setq plane + (if (string= subcode "") + "5" + subcode)) + (replace-match + (format "