1 package com.hughes.android.dictionary.parser.wiktionary;
3 import java.util.ArrayList;
6 import java.util.regex.Pattern;
8 import com.hughes.android.dictionary.engine.EntryTypeName;
9 import com.hughes.android.dictionary.engine.HtmlEntry;
10 import com.hughes.android.dictionary.engine.IndexBuilder;
11 import com.hughes.android.dictionary.engine.IndexedEntry;
12 import com.hughes.android.dictionary.parser.WikiTokenizer;
14 public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
16 public static final String NAME = "WholeSectionToHtmlParser";
17 public static final Pattern skipSections = Pattern.compile(".*Translations.*");
19 final IndexBuilder titleIndexBuilder;
21 public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder) {
22 this.titleIndexBuilder = titleIndexBuilder;
26 void parseSection(String heading, String text) {
27 HtmlEntry htmlEntry = new HtmlEntry(entrySource, title);
28 IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
30 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(this);
31 callback.builder = new StringBuilder();
32 callback.indexedEntry = indexedEntry;
33 callback.dispatch(text, null);
35 htmlEntry.html = callback.builder.toString();
36 indexedEntry.isValid = true;
37 titleIndexBuilder.addEntryWithString(indexedEntry, title, EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
41 void removeUselessArgs(Map<String, String> namedArgs) {
44 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
45 public AppendCallback(WholeSectionToHtmlParser parser) {
50 public void onPlainText(String plainText) {
51 super.onPlainText(plainText);
55 public void onWikiLink(WikiTokenizer wikiTokenizer) {
56 super.onWikiLink(wikiTokenizer);
60 public void onFunction(WikiTokenizer wikiTokenizer, String name,
61 List<String> args, Map<String, String> namedArgs) {
62 super.onFunction(wikiTokenizer, name, args, namedArgs);
66 public void onHtml(WikiTokenizer wikiTokenizer) {
67 super.onHtml(wikiTokenizer);
71 public void onNewline(WikiTokenizer wikiTokenizer) {
75 public void onHeading(WikiTokenizer wikiTokenizer) {
76 final String headingText = wikiTokenizer.headingWikiText();
77 final int depth = wikiTokenizer.headingDepth();
78 if (skipSections.matcher(headingText).matches()) {
79 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
80 if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
81 wikiTokenizer.returnToLineStart();
87 onPlainText(String.format("\n<h%d>", depth));
88 dispatch(headingText, null);
89 onPlainText(String.format("</h%d>\n", depth));
92 final List<Character> listPrefixStack = new ArrayList<Character>();
94 public void onListItem(WikiTokenizer wikiTokenizer) {
95 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
98 final String prefix = wikiTokenizer.listItemPrefix();
99 while (listPrefixStack.size() < prefix.length()) {
100 onPlainText(String.format("<%s>", WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
101 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
104 dispatch(wikiTokenizer.listItemWikiText(), null);
105 onPlainText("</li>\n");
107 WikiTokenizer nextToken = wikiTokenizer.nextToken();
108 boolean returnToLineStart = false;
109 if (nextToken != null && nextToken.isNewline()) {
110 nextToken = nextToken.nextToken();
111 returnToLineStart = true;
113 final String nextListHeader;
114 if (nextToken == null || !nextToken.isListItem()) {
117 nextListHeader = nextToken.listItemPrefix();
119 if (returnToLineStart) {
120 wikiTokenizer.returnToLineStart();
122 while (listPrefixStack.size() > nextListHeader.length()) {
123 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
124 onPlainText(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
128 boolean boldOn = false;
129 boolean italicOn = false;
131 public void onMarkup(WikiTokenizer wikiTokenizer) {
132 if ("'''".equals(wikiTokenizer.token())) {
139 } else if ("''".equals(wikiTokenizer.token())) {
143 onPlainText("</em>");
145 italicOn = !italicOn;