001    /*
002     * Copyright (c) 2009 The openGion Project.
003     *
004     * Licensed under the Apache License, Version 2.0 (the "License");
005     * you may not use this file except in compliance with the License.
006     * You may obtain a copy of the License at
007     *
008     *     http://www.apache.org/licenses/LICENSE-2.0
009     *
010     * Unless required by applicable law or agreed to in writing, software
011     * distributed under the License is distributed on an "AS IS" BASIS,
012     * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND,
013     * either express or implied. See the License for the specific language
014     * governing permissions and limitations under the License.
015     */
016    package org.opengion.hayabusa.report2;
017    
018    import java.util.ArrayList;
019    import java.util.List;
020    
021    import org.opengion.hayabusa.common.HybsSystemException;
022    
023    /**
024     * Calc蟶ｳ逾ｨ繧ｷ繧ｹ繝?Β縺ｧ繧ｿ繧ｰ縺ｮ繝代?繧ｹ繧定｡後≧縺溘ａ縺ｮ繧ｯ繝ｩ繧ｹ縺ｧ縺吶?
025     *
026     * 荳ｻ縺ｫ髢句ｧ九ち繧ｰ縲∫ｵゆｺ?ち繧ｰ繧呈欠螳壹＠縺溘ヱ繝ｼ繧ｹ縺ｮ繝ｫ繝ｼ繝怜?逅?ｒ陦後≧縺溘ａ縺ｮ讖溯?繧呈署萓帙＠縺ｾ縺吶?
027     * 蜈ｷ菴鍋噪縺ｫ縺ｯ縲＋@link #doParse(String, String, String)}縺ｫ繧医ｊ縲√ヱ繝ｼ繧ｹ譁?ｭ怜?縲?幕蟋九ち繧ｰ縲∫ｵゆｺ?ち繧ｰ繧?
028     * 謖?ｮ壹＠縲√ヱ繝ｼ繧ｹ繧定｡後＞縺ｾ縺吶?
029     * 繝代?繧ｹ蠕後?譁?ｭ怜?縺ｯ縲＋@link #doParse(String, String, String)}縺ｮ謌ｻ繧雁?縺ｫ縺ｪ繧翫∪縺吶?
030     *
031     * 繝代?繧ｹ螳溯｡御ｸｭ縺ｫ縲∫匱隕九＆繧後◆髢句ｧ九ち繧ｰ縺九ｉ邨ゆｺ?ち繧ｰ縺ｾ縺ｧ縺ｮ髢薙?譁?ｭ怜?縺ｮ蜃ｦ逅??縲＋@link #exec(String, StringBuilder, int)}繧?
032     * 繧ｪ繝ｼ繝舌?繝ｩ繧､繝峨☆繧九％縺ｨ縺ｫ繧医ｊ螳夂ｾｩ縺励∪縺吶?
033     *
034     * 縺ｾ縺溘?縺薙?繧ｯ繝ｩ繧ｹ縺ｧ縺ｯ縲√ヱ繝ｼ繧ｹ縺ｫ蠢?ｦ√↑蜷?ｨｮ繝ｦ繝ｼ繝?ぅ繝ｪ繝?ぅ繝｡繧ｽ繝?ラ縺ｫ縺､縺?※繧ょ酔讒倥↓螳夂ｾｩ縺輔ｌ縺ｦ縺?∪縺吶?
035     *
036     * @og.group 蟶ｳ逾ｨ繧ｷ繧ｹ繝?Β
037     *
038     * @version  4.0
039     * @author   Hiroki.Nakamura
040     * @since    JDK1.6
041     */
042    class TagParser {
043    
044            private int preOffset = 0;
045            private int curOffset = 0;
046    
047            /**
048             * 繝代?繧ｹ蜃ｦ逅?ｒ陦後＞縺ｾ縺吶?
049             *
050             * 繝代?繧ｹ荳ｭ縺ｫ蜿悶ｊ蜃ｺ縺輔ｌ縺滄幕蟋九ち繧ｰ縺九ｉ邨ゆｺ?ち繧ｰ縺ｾ縺ｧ縺ｮ譁?ｭ怜?縺ｮ蜃ｦ逅??縲?
051             * {@link #exec(String, StringBuilder, int)}縺ｧ螳夂ｾｩ縺励∪縺吶?
052             *
053             * 縺ｾ縺溘?isAddTag繧稚rue縺ｫ縺励◆蝣ｴ蜷医?{@link #exec(String, StringBuilder, int)}縺ｫ貂｡縺輔ｌ繧?
054             * 譁?ｭ怜?縺ｫ縲?幕蟋九ち繧ｰ縲∫ｵゆｺ?ち繧ｰ縺悟性縺ｾ繧後∪縺吶?
055             * 騾?↓false縺ｫ縺励◆蝣ｴ蜷医?縲?幕蟋九ち繧ｰ縲∫ｵゆｺ?ち繧ｰ繧帝勁縺阪?{@link #exec(String, StringBuilder, int)}縺ｫ貂｡縺輔ｌ縺ｾ縺吶?
056             *
057             * @og.rev 5.2.2.0 (2010/11/01) 隱ｭ縺ｿ鬟帙?縺励ｒ縺励◆蝣ｴ蜷医↓縲?幕蟋九ち繧ｰ縺梧嶌縺崎ｾｼ縺ｾ繧後↑縺?ヰ繧ｰ繧剃ｿｮ豁｣
058             *
059             * @param content 繝代?繧ｹ蟇ｾ雎｡譁?ｭ怜?
060             * @param startTag 髢句ｧ九ち繧ｰ
061             * @param endTag 邨ゆｺ?ち繧ｰ
062             * @param isAddTag 髢句ｧ九ち繧ｰ繝ｻ邨ゆｺ?ち繧ｰ繧貞性繧?°
063             *
064             * @return 繝代?繧ｹ蠕後?譁?ｭ怜?
065             * @see #exec(String, StringBuilder, int)
066             */
067            public String doParse( final String content, final String startTag, final String endTag, final boolean isAddTag ) {
068                    StringBuilder buf = new StringBuilder();
069                    String errMsg = null;
070    
071                    while( ( preOffset = content.indexOf( startTag, Math.max( preOffset, curOffset ) ) ) >= 0 ) {
072                            buf.append( content.substring( curOffset, preOffset ) );
073                            curOffset = content.indexOf( endTag, preOffset + startTag.length() );
074    
075                            if( checkIgnore( preOffset, curOffset ) ) {
076                                    if( curOffset < 0 ){
077                                            errMsg = "[ERROR]PARSE:髢句ｧ九ち繧ｰ繧堤ｵゆｺ?ち繧ｰ縺ｮ謨ｴ蜷域?縺御ｸ肴ｭ｣縺ｧ縺吶?[髢句ｧ九ち繧ｰ=" + startTag + ":邨ゆｺ?ち繧ｰ=" + endTag + "]";
078                                            throw new HybsSystemException( errMsg );
079                                    }
080                                    preOffset += startTag.length();
081                                    curOffset += endTag.length();
082    
083                                    String str = null;
084                                    if( isAddTag ) {
085                                            str = content.substring( preOffset - startTag.length(), curOffset );
086                                    }
087                                    else {
088                                            str = content.substring( preOffset, curOffset - endTag.length() );
089                                    }
090    
091                                    exec( str, buf, curOffset );
092                            }
093                            else {
094                                    // 5.2.2.0 (2010/11/01) 髢句ｧ九ち繧ｰ縺梧嶌縺崎ｾｼ縺ｾ繧後↑縺?ヰ繧ｰ繧剃ｿｮ豁｣
095                                    buf.append( startTag );
096                                    preOffset += startTag.length();
097                                    curOffset = preOffset;
098                            }
099                    }
100                    buf.append( content.substring( curOffset, content.length() ) );
101    
102                    return buf.toString();
103            }
104    
105            /**
106             * 繝代?繧ｹ蜃ｦ逅?ｒ陦後＞縺ｾ縺吶?
107             *
108             * 隧ｳ邏ｰ縺ｯ縲＋@link #doParse(String, String, String, boolean)}縺ｮJavadoc繧貞盾辣ｧ縺励※荳九＆縺??
109             *
110             * @param content 繝代?繧ｹ蟇ｾ雎｡譁?ｭ怜?
111             * @param startTag 髢句ｧ九ち繧ｰ
112             * @param endTag 邨ゆｺ?ち繧ｰ
113             *
114             * @return 繝代?繧ｹ蠕後?譁?ｭ怜?
115             * @see #doParse(String, String, String, boolean)
116             */
117            public String doParse( final String content, final String startTag, final String endTag ) {
118                    return doParse( content, startTag, endTag, true );
119            }
120    
121            /**
122             * 髢句ｧ九ち繧ｰ縺九ｉ邨ゆｺ?ち繧ｰ縺ｾ縺ｧ縺ｮ譁?ｭ怜?縺ｮ蜃ｦ逅?ｒ螳夂ｾｩ縺励∪縺吶?
123             *
124             * 縺薙?螳溯｣?〒縺ｯ縲∽ｽ輔ｂ蜃ｦ逅?ｒ陦後＞縺ｾ縺帙ｓ縲?蛻?ｊ蜃ｺ縺励◆譁?ｭ怜?縺ｯ繧｢繝壹Φ繝峨＆繧後∪縺帙ｓ)
125             * 繧ｵ繝悶け繝ｩ繧ｹ縺ｧ繧ｪ繝ｼ繝舌?繝ｩ繧､繝峨＠縺ｦ螳滄圀縺ｮ蜃ｦ逅?ｒ螳溯｣?＠縺ｦ荳九＆縺??
126             *
127             * @param str 髢句ｧ九ち繧ｰ縺九ｉ邨ゆｺ?ち繧ｰ縺ｾ縺ｧ縺ｮ譁?ｭ怜?(髢句ｧ九ち繧ｰ繝ｻ邨ゆｺ?ち繧ｰ繧貞性繧?
128             * @param buf 蜃ｺ蜉帙ｒ陦後≧譁?ｭ怜?繝舌ャ繝輔ぃ
129             * @param offset 邨ゆｺ?ち繧ｰ縺ｮ繧ｪ繝輔そ繝?ヨ
130             */
131            protected void exec( final String str, final StringBuilder buf, final int offset ) {
132                    // Document empty method 蟇ｾ遲?
133            }
134    
135            /**
136             * 髢句ｧ九ち繧ｰ縺九ｉ邨ゆｺ?ち繧ｰ縺ｾ縺ｧ縺ｮ譁?ｭ怜?縺ｮ蜃ｦ逅?ｒ螳溯｡後☆繧九°縺ｩ縺?°繧貞ｮ夂ｾｩ縺励∪縺吶?
137             *
138             * false縺瑚ｿ斐＆繧後◆蝣ｴ蜷医?菴輔ｂ蜃ｦ逅?＆繧後★({@link #exec(String, StringBuilder, int)}縺悟ｮ溯｡後＆繧後↑縺?縲?
139             * 蜈??譁?ｭ怜?縺後◎縺ｮ縺ｾ縺ｾ蜃ｺ蜉帙＆繧後∪縺吶?
140             *
141             * @param strOffset 髢句ｧ九ち繧ｰ縺ｮ繧ｪ繝輔そ繝?ヨ
142             * @param endOffset 邨ゆｺ?ち繧ｰ縺ｮ繧ｪ繝輔そ繝?ヨ
143             *
144             * @return 蜃ｦ逅?ｒ陦後≧縺九←縺?°(true:蜃ｦ逅?ｒ陦後≧ false:蜃ｦ逅?ｒ陦後ｏ縺ｪ縺?
145             */
146            protected boolean checkIgnore( final int strOffset, final int endOffset ) {
147                    return true;
148            }
149    
150            /**
151             * 繝代?繧ｹ螳溯｡御ｸｭ縺ｮoffset蛟､繧貞､夜Κ縺九ｉ繧ｻ繝?ヨ縺励∪縺吶?
152             *
153             * 縺薙?繝｡繧ｽ繝?ラ縺ｯ縲＋@link #exec(String, StringBuilder, int)}縺ｧ縲∝?逅?ｵ先棡縺ｫ繧医ｊ縲｛ffset蛟､繧?
154             * 騾ｲ繧√※縺翫￥蠢?ｦ√′縺ゅｋ蝣ｴ蜷医↓蛻ｩ逕ｨ縺輔ｌ縺ｾ縺吶?(縺､縺ｾ繧企?蟶ｸ縺ｯ蛻ｩ逕ｨ縺吶ｋ蠢?ｦ√?縺ゅｊ縺ｾ縺帙ｓ)
155             *
156             * @param offset 繧ｪ繝輔そ繝?ヨ
157             * @see #exec(String, StringBuilder, int)
158             */
159            public void setOffset( final int offset ) {
160                    curOffset = offset;
161            }
162    
163            /**
164             * 蠑墓焚縺ｮ譁?ｭ怜?繧呈欠螳壹＆繧後◆髢句ｧ九ち繧ｰ縲∫ｵゆｺ?ち繧ｰ縺ｧ隗｣譫舌＠驟榊?縺ｨ縺励※霑斐☆縲√Θ繝ｼ繝?ぅ繝ｪ繝?ぅ繝｡繧ｽ繝?ラ縺ｧ縺吶?
165             *
166             * 髢句ｧ九ち繧ｰ繧医ｊ蜑阪?譁?ｭ怜?縺ｯ0逡ｪ逶ｮ縺ｫ縲∫ｵゆｺ?ち繧ｰ繧医ｊ蠕後?譁?ｭ怜?縺ｯ1逡ｪ逶ｮ縺ｫ譬ｼ邏阪＆繧後∪縺吶?
167             * 2逡ｪ逶ｮ莉･髯阪↓縲?幕蟋九ち繧ｰ縲∫ｵゆｺ?ち繧ｰ縺ｮ驛ｨ蛻?′譬ｼ邏阪＆繧後∪縺吶?
168             *
169             * @param str           隗｣譫舌☆繧区枚蟄怜?
170             * @param startTag      髢句ｧ九ち繧ｰ
171             * @param endTag        邨ゆｺ?ち繧ｰ
172             *
173             * @return 隗｣譫千ｵ先棡縺ｮ驟榊?
174             */
175            public static String[] tag2Array( final String str, final String startTag, final String endTag ) {
176                    String header = null;
177                    String footer = null;
178                    List<String> body = new ArrayList<String>();
179    
180                    int preOffset = -1;
181                    int curOffset = 0;
182    
183                    while( true ) {
184                            curOffset = str.indexOf( startTag, preOffset + 1 );
185                            if( curOffset < 0 ) {
186                                    curOffset = str.lastIndexOf( endTag ) + endTag.length();
187                                    body.add( str.substring( preOffset, curOffset ) );
188    
189                                    footer = str.substring( curOffset );
190                                    break;
191                            }
192                            else if( preOffset == -1 ) {
193                                    header = str.substring( 0, curOffset );
194                            }
195                            else {
196                                    body.add( str.substring( preOffset, curOffset ) );
197                            }
198                            preOffset = curOffset;
199                    }
200    
201                    String[] arr = new String[body.size()+2];
202                    arr[0] = header;
203                    arr[1] = footer;
204                    for( int i=0; i<body.size(); i++ ) {
205                            arr[i+2] = body.get(i);
206                    }
207    
208                    return arr;
209            }
210    
211            /**
212             * 蠑墓焚縺ｮ譁?ｭ怜?縺ｮ髢句ｧ区枚蟄励→邨ゆｺ?枚蟄励?髢薙?譁?ｭ怜?繧貞叙繧雁?縺吶?繝ｦ繝ｼ繝?ぅ繝ｪ繝?ぅ繝｡繧ｽ繝?ラ縺ｧ縺吶?
213             * 窶ｻ霑斐＆繧後ｋ譁?ｭ怜?縺ｫ縲?幕蟋区枚蟄励?邨ゆｺ?枚蟄励?蜷ｫ縺ｾ繧後∪縺帙ｓ縲?
214             *
215             * @param str   隗｣譫舌☆繧区枚蟄怜?
216             * @param start 髢句ｧ区枚蟄?
217             * @param end   邨ゆｺ?枚蟄?
218             *
219             * @return 隗｣譫千ｵ先棡縺ｮ譁?ｭ?
220             */
221            public static String getValueFromTag( final String str, final String start, final String end ) {
222                    int startOffset = str.indexOf( start );
223    //              int startOffset = str.indexOf( start ) + start.length();
224                    // 4.2.4.0 (2008/06/02) 蟄伜惠縺励↑縺??蜷医?null縺ｧ霑斐☆
225                    if( startOffset == -1 ) {
226                            return null;
227                    }
228                    startOffset += start.length();
229    
230                    int endOffset = str.indexOf( end, startOffset );
231                    String value = str.substring( startOffset, endOffset );
232    
233                    return value;
234            }
235    
236            /**
237             * 蠑墓焚縺ｮ繧ｭ繝ｼ縺九ｉ荳崎ｦ√↑繧ｭ繝ｼ繧貞叙繧企勁縺上?繝ｦ繝ｼ繝?ぅ繝ｪ繝?ぅ繝｡繧ｽ繝?ラ縺ｧ縺吶?
238             *
239             * @og.rev 5.1.8.0 (2010/07/01) span繧ｿ繧ｰ繧貞炎髯､
240             *
241             * @param key   繧ｪ繝ｪ繧ｸ繝翫Ν縺ｮ繧ｭ繝ｼ
242             * @param sb    繧ｭ繝ｼ縺ｮ螟悶↓蜷ｫ縺ｾ繧後ｋa繧ｿ繧ｰ繧貞炎髯､縺吶ｋ縺溘ａ縺ｮ縲√ヰ繝?ヵ繧｡
243             *
244             * @return 蜑企勁蠕後?繧ｭ繝ｼ
245             */
246            public static String checkKey( final String key, final StringBuilder sb ) {
247                    if( key.indexOf( '<' ) < 0 && key.indexOf( '>' ) < 0 ) { return key; }
248    
249                    StringBuilder rtn = new StringBuilder( key );
250                    String tagEnd = ">";
251                    int rtnOffset = -1;
252    
253                    // <text:a ...>{@XXX</text:a>縺ｮ荳崎ｦ√ち繧ｰ繧貞炎髯､
254                    String delTagStart1 = "<text:a ";
255                    String delTagEnd1 = "</text:a>";
256                    while( ( rtnOffset = rtn.lastIndexOf( delTagEnd1 ) ) >= 0 ) {
257                            boolean isDel = false;
258                            // 繧ｭ繝ｼ閾ｪ霄ｫ縺ｫ蜷ｫ縺ｾ繧後ｋa繧ｿ繧ｰ繧貞炎髯､
259                            int startOffset = rtn.lastIndexOf( delTagStart1, rtnOffset );
260                            if( startOffset >= 0 ) {
261                                    int endOffset = rtn.indexOf( tagEnd, startOffset );
262                                    if( endOffset >= 0 ) {
263                                            rtn.delete( rtnOffset, rtnOffset + delTagEnd1.length() );
264                                            rtn.delete( startOffset, endOffset + tagEnd.length() );
265                                            isDel = true;
266                                    }
267                            }
268                            else {
269                                    // 繧ｭ繝ｼ縺ｮ螟悶↓蜷ｫ縺ｾ繧後ｋa繧ｿ繧ｰ繧貞炎髯､
270                                    startOffset = sb.lastIndexOf( delTagStart1 );
271                                    if( startOffset >= 0 ) {
272                                            int endOffset = sb.indexOf( tagEnd, startOffset );
273                                            if( endOffset >= 0 ) {
274                                                    rtn.delete( rtnOffset, rtnOffset + delTagEnd1.length() );
275                                                    sb.delete( startOffset, endOffset + tagEnd.length() );
276                                                    isDel = true;
277                                            }
278                                    }
279                            }
280                            if( !isDel ) { break; }
281                    }
282    
283                    // 5.1.8.0 (2010/07/01) span繧ｿ繧ｰ繧貞炎髯､
284                    String delTagStart2 = "<text:span ";
285                    String delTagEnd2 = "</text:span>";
286                    while( ( rtnOffset = rtn.lastIndexOf( delTagEnd2 ) ) >= 0 ) {
287                            boolean isDel = false;
288                            // 繧ｭ繝ｼ閾ｪ霄ｫ縺ｫ蜷ｫ縺ｾ繧後ｋspan繧ｿ繧ｰ繧貞炎髯､
289                            int startOffset = rtn.lastIndexOf( delTagStart2, rtnOffset );
290                            if( startOffset >= 0 ) {
291                                    int endOffset = rtn.indexOf( tagEnd, startOffset );
292                                    if( endOffset >= 0 ) {
293                                            rtn.delete( rtnOffset, rtnOffset + delTagEnd2.length() );
294                                            rtn.delete( startOffset, endOffset + tagEnd.length() );
295                                            isDel = true;
296                                    }
297                            }
298                            else {
299                                    // 繧ｭ繝ｼ縺ｮ螟悶↓蜷ｫ縺ｾ繧後ｋspan繧ｿ繧ｰ繧貞炎髯､
300                                    startOffset = sb.lastIndexOf( delTagStart2 );
301                                    if( startOffset >= 0 ) {
302                                            int endOffset = sb.indexOf( tagEnd, startOffset );
303                                            if( endOffset >= 0 ) {
304                                                    rtn.delete( rtnOffset, rtnOffset + delTagEnd2.length() );
305                                                    sb.delete( startOffset, endOffset + tagEnd.length() );
306                                                    isDel = true;
307                                            }
308                                    }
309                            }
310                            if( !isDel ) { break; }
311                    }
312    
313                    return rtn.toString();
314            }
315    }
316