| 1 |
|
package com.atlassian.mail; |
| 2 |
|
|
| 3 |
|
import org.apache.log4j.Category; |
| 4 |
|
|
| 5 |
|
import javax.swing.text.html.HTML; |
| 6 |
|
import javax.swing.text.html.HTMLEditorKit; |
| 7 |
|
import javax.swing.text.html.parser.ParserDelegator; |
| 8 |
|
import java.io.*; |
| 9 |
|
import java.util.ArrayList; |
| 10 |
|
|
| 11 |
|
|
| 12 |
|
|
| 13 |
|
|
| 14 |
|
|
| 15 |
|
|
|
|
|
| 100% |
Uncovered Elements: 0 (8) |
Complexity: 2 |
Complexity Density: 0.33 |
|
| 16 |
|
public class HtmlToTextConverter |
| 17 |
|
{ |
| 18 |
|
private static final Category log = Category.getInstance(HtmlToTextConverter.class); |
| 19 |
|
|
|
|
|
| 44.9% |
Uncovered Elements: 65 (118) |
Complexity: 69 |
Complexity Density: 1.08 |
|
| 20 |
|
private class HTMLCallbackHandler extends HTMLEditorKit.ParserCallback { |
| 21 |
|
|
| 22 |
|
Writer out; |
| 23 |
|
boolean started = false; |
| 24 |
|
boolean inBody = false; |
| 25 |
|
boolean inList = false; |
| 26 |
|
boolean firstTD = true; |
| 27 |
|
int listCount = -1; |
| 28 |
|
ArrayList links = new ArrayList(); |
| 29 |
|
|
| 30 |
|
private static final String NEWLINE = "\n"; |
| 31 |
|
private static final String TAB = "\t"; |
| 32 |
|
private static final String STAR = "*"; |
| 33 |
|
private static final String SPACE = " "; |
| 34 |
|
private static final String PERIOD = "."; |
| 35 |
|
private static final String OPEN_BRACKET = "["; |
| 36 |
|
private static final String CLOSE_BRACKET = "]"; |
| 37 |
|
private static final String DASH_LINE = "----------------------------------------------------------------------------------------"; |
| 38 |
|
|
| 39 |
|
|
| 40 |
|
|
|
|
|
| 100% |
Uncovered Elements: 0 (1) |
Complexity: 1 |
Complexity Density: 1 |
|
| 41 |
6
|
public HTMLCallbackHandler(Writer writer) {... |
| 42 |
6
|
out = writer; |
| 43 |
|
} |
| 44 |
|
|
|
|
|
| 43.9% |
Uncovered Elements: 37 (66) |
Complexity: 47 |
Complexity Density: 1.31 |
|
| 45 |
23
|
public void handleStartTag(HTML.Tag tag, javax.swing.text.MutableAttributeSet set, int position) {... |
| 46 |
23
|
try |
| 47 |
|
{ |
| 48 |
23
|
if (inBody && started && tag.equals(HTML.Tag.P)) |
| 49 |
|
{ |
| 50 |
1
|
out.write(NEWLINE + NEWLINE); |
| 51 |
|
} |
| 52 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.OL) || tag.equals(HTML.Tag.UL)) |
| 53 |
|
{ |
| 54 |
0
|
inList = true; |
| 55 |
0
|
out.write(NEWLINE + NEWLINE); |
| 56 |
0
|
if(tag.equals(HTML.Tag.OL)) |
| 57 |
0
|
listCount = 1; |
| 58 |
|
} |
| 59 |
22
|
else if (inBody && started && inList && tag.equals(HTML.Tag.LI)) |
| 60 |
|
{ |
| 61 |
0
|
out.write(NEWLINE); |
| 62 |
0
|
if(listCount != -1) |
| 63 |
|
{ |
| 64 |
0
|
out.write(listCount + PERIOD + SPACE); |
| 65 |
0
|
listCount++; |
| 66 |
|
} |
| 67 |
|
else |
| 68 |
0
|
out.write(STAR); |
| 69 |
|
} |
| 70 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.TABLE)) |
| 71 |
|
{ |
| 72 |
0
|
out.write(NEWLINE); |
| 73 |
|
} |
| 74 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.TR)) |
| 75 |
|
{ |
| 76 |
0
|
out.write(NEWLINE); |
| 77 |
0
|
firstTD = true; |
| 78 |
|
} |
| 79 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.TD) || tag.equals(HTML.Tag.TH)) |
| 80 |
|
{ |
| 81 |
0
|
if(!firstTD) |
| 82 |
|
{ |
| 83 |
0
|
out.write(TAB); |
| 84 |
|
} |
| 85 |
|
else |
| 86 |
|
{ |
| 87 |
0
|
firstTD = false; |
| 88 |
|
} |
| 89 |
|
} |
| 90 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.PRE)) |
| 91 |
|
{ |
| 92 |
0
|
out.write(NEWLINE); |
| 93 |
|
} |
| 94 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.IMG)) |
| 95 |
|
{ |
| 96 |
|
|
| 97 |
0
|
handleLink((String)set.getAttribute(HTML.Attribute.SRC)); |
| 98 |
|
} |
| 99 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.A)) |
| 100 |
|
{ |
| 101 |
|
|
| 102 |
0
|
handleLink((String)set.getAttribute(HTML.Attribute.HREF)); |
| 103 |
|
} |
| 104 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.HR)) |
| 105 |
|
{ |
| 106 |
0
|
out.write(NEWLINE + DASH_LINE); |
| 107 |
|
} |
| 108 |
22
|
else if (inBody && started && tag.equals(HTML.Tag.H1) || tag.equals(HTML.Tag.H2) || tag.equals(HTML.Tag.H3) || tag.equals(HTML.Tag.H4) || tag.equals(HTML.Tag.H5) || tag.equals(HTML.Tag.H6)) |
| 109 |
|
{ |
| 110 |
0
|
out.write(NEWLINE); |
| 111 |
|
} |
| 112 |
22
|
else if (tag.equals(HTML.Tag.BODY)) |
| 113 |
|
{ |
| 114 |
6
|
inBody = true; |
| 115 |
|
} |
| 116 |
|
} |
| 117 |
|
catch (IOException e) |
| 118 |
|
{ |
| 119 |
0
|
log.warn("IO error converting HTML to text", e); |
| 120 |
|
} |
| 121 |
|
|
| 122 |
|
} |
| 123 |
|
|
|
|
|
| 0% |
Uncovered Elements: 5 (5) |
Complexity: 2 |
Complexity Density: 0.67 |
|
| 124 |
0
|
private void handleLink(String src) throws IOException... |
| 125 |
|
{ |
| 126 |
0
|
if(src != null) |
| 127 |
|
{ |
| 128 |
0
|
links.add(src); |
| 129 |
0
|
out.write(OPEN_BRACKET + links.size() + CLOSE_BRACKET); |
| 130 |
|
} |
| 131 |
|
} |
| 132 |
|
|
|
|
|
| 29.6% |
Uncovered Elements: 19 (27) |
Complexity: 11 |
Complexity Density: 0.73 |
|
| 133 |
23
|
public void handleEndTag(HTML.Tag tag, int position) {... |
| 134 |
23
|
if (inBody && started && tag.equals(HTML.Tag.OL) || tag.equals(HTML.Tag.UL)) |
| 135 |
|
{ |
| 136 |
0
|
inList = false; |
| 137 |
0
|
if(tag.equals(HTML.Tag.OL)) |
| 138 |
0
|
listCount = -1; |
| 139 |
|
} |
| 140 |
23
|
else if (tag.equals(HTML.Tag.BODY)) |
| 141 |
|
{ |
| 142 |
6
|
if(links.size() != 0) |
| 143 |
|
{ |
| 144 |
|
|
| 145 |
0
|
try |
| 146 |
|
{ |
| 147 |
0
|
out.write(NEWLINE + DASH_LINE + NEWLINE); |
| 148 |
0
|
for (int i = 0; i < links.size(); i++) |
| 149 |
|
{ |
| 150 |
0
|
String src = (String)links.get(i); |
| 151 |
0
|
out.write(OPEN_BRACKET + (i + 1) + CLOSE_BRACKET + SPACE + src); |
| 152 |
0
|
if((i + 1) < links.size()) |
| 153 |
|
{ |
| 154 |
0
|
out.write(NEWLINE); |
| 155 |
|
} |
| 156 |
|
} |
| 157 |
|
} |
| 158 |
|
catch (IOException e) |
| 159 |
|
{ |
| 160 |
0
|
e.printStackTrace(); |
| 161 |
|
} |
| 162 |
|
} |
| 163 |
6
|
inBody = false; |
| 164 |
|
} |
| 165 |
|
} |
| 166 |
|
|
|
|
|
| 85.7% |
Uncovered Elements: 1 (7) |
Complexity: 3 |
Complexity Density: 0.6 |
|
| 167 |
14
|
public void handleText(char[] aChar, int position) {... |
| 168 |
14
|
try |
| 169 |
|
{ |
| 170 |
14
|
if (inBody) |
| 171 |
|
{ |
| 172 |
13
|
out.write(aChar); |
| 173 |
13
|
started = true; |
| 174 |
|
} |
| 175 |
|
} |
| 176 |
|
catch (IOException e) |
| 177 |
|
{ |
| 178 |
0
|
log.warn("IO error converting HTML to text", e); |
| 179 |
|
} |
| 180 |
|
} |
| 181 |
|
|
|
|
|
| 66.7% |
Uncovered Elements: 2 (6) |
Complexity: 5 |
Complexity Density: 1.25 |
|
| 182 |
6
|
public void handleSimpleTag(HTML.Tag tag, javax.swing.text.MutableAttributeSet a, int pos) {... |
| 183 |
6
|
try |
| 184 |
|
{ |
| 185 |
6
|
if (inBody && started && tag.equals(HTML.Tag.BR)) |
| 186 |
6
|
out.write(NEWLINE); |
| 187 |
|
} |
| 188 |
|
catch (IOException e) |
| 189 |
|
{ |
| 190 |
0
|
log.warn("IO error converting HTML to text", e); |
| 191 |
|
} |
| 192 |
|
|
| 193 |
|
} |
| 194 |
|
} |
| 195 |
|
|
| 196 |
|
|
|
|
|
| 100% |
Uncovered Elements: 0 (4) |
Complexity: 1 |
Complexity Density: 0.25 |
|
| 197 |
6
|
public String convert(String html) throws IOException... |
| 198 |
|
{ |
| 199 |
6
|
StringWriter out = new StringWriter(); |
| 200 |
6
|
convert(new StringReader(html), out); |
| 201 |
6
|
out.close(); |
| 202 |
6
|
return out.toString(); |
| 203 |
|
} |
| 204 |
|
|
|
|
|
| 100% |
Uncovered Elements: 0 (2) |
Complexity: 1 |
Complexity Density: 0.5 |
|
| 205 |
6
|
private void convert(Reader reader, Writer writer) throws IOException... |
| 206 |
|
{ |
| 207 |
6
|
HTMLCallbackHandler handler = new HTMLCallbackHandler(writer); |
| 208 |
6
|
new ParserDelegator().parse(reader, handler, true); |
| 209 |
|
} |
| 210 |
|
} |