r10825 MediaWiki - Code Review archive

Repository:MediaWiki
Revision:< r10824‎ | r10825 | r10826 >
Date:11:59, 29 August 2005
Author:vibber
Status:old
Tags:
Comment:
Another half-done hacky tool, for taking XML dumps and reformatting them to
SQL dumps for either 1.4 or 1.5 schema. (Not ready to run yet.)
Modified paths:
  • /trunk/mwdumper/mwimport (added) (history)
  • /trunk/mwdumper/mwimport/AssemblyInfo.cs (added) (history)
  • /trunk/mwdumper/mwimport/Contributor.cs (added) (history)
  • /trunk/mwdumper/mwimport/IDumpWriter.cs (added) (history)
  • /trunk/mwdumper/mwimport/Main.cs (added) (history)
  • /trunk/mwdumper/mwimport/Page.cs (added) (history)
  • /trunk/mwdumper/mwimport/Revision.cs (added) (history)
  • /trunk/mwdumper/mwimport/Siteinfo.cs (added) (history)
  • /trunk/mwdumper/mwimport/SqlWriter.cs (added) (history)
  • /trunk/mwdumper/mwimport/SqlWriter14.cs (added) (history)
  • /trunk/mwdumper/mwimport/SqlWriter15.cs (added) (history)
  • /trunk/mwdumper/mwimport/Title.cs (added) (history)
  • /trunk/mwdumper/mwimport/mwimport.mdp (added) (history)
  • /trunk/mwdumper/mwimport/mwimport.mds (added) (history)

Diff [purge]

Index: trunk/mwdumper/mwimport/AssemblyInfo.cs
——@@ -0,0 +1,32 @@
 2+using System.Reflection;
 3+using System.Runtime.CompilerServices;
 4+
 5+// Information about this assembly is defined by the following
 6+// attributes.
 7+//
 8+// change them to the information which is associated with the assembly
 9+// you compile.
 10+
 11+[assembly: AssemblyTitle("")]
 12+[assembly: AssemblyDescription("")]
 13+[assembly: AssemblyConfiguration("")]
 14+[assembly: AssemblyCompany("")]
 15+[assembly: AssemblyProduct("")]
 16+[assembly: AssemblyCopyright("")]
 17+[assembly: AssemblyTrademark("")]
 18+[assembly: AssemblyCulture("")]
 19+
 20+// The assembly version has following format :
 21+//
 22+// Major.Minor.Build.Revision
 23+//
 24+// You can specify all values by your own or you can build default build and revision
 25+// numbers with the '*' character (the default):
 26+
 27+[assembly: AssemblyVersion("1.0.*")]
 28+
 29+// The following attributes specify the key for the sign of your assembly. See the
 30+// .NET Framework documentation for more information about signing.
 31+// This is not required, if you don't want signing let these attributes like they're.
 32+[assembly: AssemblyDelaySign(false)]
 33+[assembly: AssemblyKeyFile("")]
Property changes on: trunk/mwdumper/mwimport/AssemblyInfo.cs
___________________________________________________________________
Added: svn:keywords
134 + Author Date Id Revision
Added: svn:eol-style
235 + native
Index: trunk/mwdumper/mwimport/IDumpWriter.cs
——@@ -0,0 +1,14 @@
 2+// created on 8/29/2005 at 12:06 AM
 3+
 4+interface IDumpWriter {
 5+ void WriteStartWiki();
 6+ void WriteEndWiki();
 7+
 8+ void WriteSiteinfo(Siteinfo info);
 9+
 10+ void WriteStartPage(Page page);
 11+ void WriteEndPage();
 12+
 13+ void WriteRevision(Revision revision);
 14+ //void WriteUpload(Upload upload); // for the future
 15+}
Property changes on: trunk/mwdumper/mwimport/IDumpWriter.cs
___________________________________________________________________
Added: svn:keywords
116 + Author Date Id Revision
Added: svn:eol-style
217 + native
Index: trunk/mwdumper/mwimport/Page.cs
——@@ -0,0 +1,11 @@
 2+// created on 8/28/2005 at 11:50 PM
 3+class Page {
 4+ public Title Title;
 5+ public int Id;
 6+ public string Restrictions;
 7+
 8+ public Page() {
 9+ // <restrictions> is optional...
 10+ Restrictions = null;
 11+ }
 12+}
Property changes on: trunk/mwdumper/mwimport/Page.cs
___________________________________________________________________
Added: svn:keywords
113 + Author Date Id Revision
Added: svn:eol-style
214 + native
Index: trunk/mwdumper/mwimport/SqlWriter.cs
——@@ -0,0 +1,186 @@
 2+// created on 8/29/2005 at 12:41 AM
 3+
 4+using System;
 5+using System.Collections;
 6+using System.IO;
 7+using System.Text;
 8+
 9+abstract class SqlWriter : IDumpWriter {
 10+ protected TextWriter _stream;
 11+
 12+ public SqlWriter(TextWriter output) {
 13+ _stream = output;
 14+ }
 15+
 16+ public void WriteStartWiki() {
 17+ _stream.WriteLine("-- MediaWiki XML dump converted to SQL");
 18+ }
 19+
 20+ public void WriteEndWiki() {
 21+ _stream.WriteLine("-- DONE");
 22+ }
 23+
 24+ public void WriteSiteinfo(Siteinfo info) {
 25+ _stream.WriteLine("");
 26+ _stream.WriteLine("-- Site: " + CommentSafe(info.Sitename));
 27+ _stream.WriteLine("-- URL: " + CommentSafe(info.Base));
 28+ _stream.WriteLine("-- Generator: " + CommentSafe(info.Generator));
 29+ _stream.WriteLine("-- Case: " + CommentSafe(info.Case));
 30+ _stream.WriteLine("--");
 31+ _stream.WriteLine("-- Namespaces:");
 32+ foreach (int key in info.Namespaces.Keys) {
 33+ _stream.WriteLine("-- " + key + ": " + info.Namespaces[key]);
 34+ }
 35+ _stream.WriteLine("");
 36+ }
 37+
 38+ public abstract void WriteStartPage(Page page);
 39+
 40+ public abstract void WriteEndPage();
 41+
 42+ public abstract void WriteRevision(Revision revision);
 43+
 44+
 45+
 46+ protected string CommentSafe(string text) {
 47+ // TODO
 48+ return text;
 49+ }
 50+
 51+ protected object InsertRow(string table, IDictionary row) {
 52+ StringBuilder sql = new StringBuilder();
 53+ bool first;
 54+
 55+ sql.Append("INSERT INTO ");
 56+ //sql.Append(_tablePrefix);
 57+ sql.Append(table);
 58+ sql.Append(" (");
 59+
 60+ first = true;
 61+ foreach (string field in row.Keys) {
 62+ if (!first)
 63+ sql.Append(',');
 64+ first = false;
 65+ sql.Append(field);
 66+ }
 67+ sql.Append(") VALUES (");
 68+
 69+ first = true;
 70+ foreach (object val in row.Values) {
 71+ if (!first)
 72+ sql.Append(',');
 73+ first = false;
 74+ sql.Append(SqlSafe(val));
 75+ }
 76+ sql.Append(");");
 77+
 78+ _stream.WriteLine(sql);
 79+ return null;
 80+ }
 81+
 82+ protected void UpdateRow(string table, IDictionary row, string keyField, object keyValue) {
 83+ StringBuilder sql = new StringBuilder();
 84+ bool first;
 85+
 86+ sql.Append("UPDATE ");
 87+ //sql.Append(_tablePrefix);
 88+ sql.Append(table);
 89+ sql.Append(" SET ");
 90+
 91+ first = true;
 92+ foreach (string field in row.Keys) {
 93+ if (!first)
 94+ sql.Append(',');
 95+ first = false;
 96+ sql.Append(field);
 97+ sql.Append('=');
 98+ sql.Append(SqlSafe(row[field]));
 99+ }
 100+
 101+ sql.Append(" WHERE ");
 102+ sql.Append(keyField);
 103+ sql.Append('=');
 104+ sql.Append(SqlSafe(keyValue));
 105+
 106+ sql.Append(";");
 107+
 108+ _stream.WriteLine(sql);
 109+ }
 110+
 111+ protected string SqlSafe(object val) {
 112+ if (val == null)
 113+ return "NULL";
 114+
 115+ Type type = val.GetType();
 116+ int i = 1;
 117+ double d = 1.0;
 118+
 119+ string str = val.ToString();
 120+ if (type == str.GetType()) {
 121+ return SqlEscape(str);
 122+ } else if (type == i.GetType()) {
 123+ return str;
 124+ } else if (type == d.GetType()) {
 125+ return str;
 126+ } else {
 127+ throw new ArgumentException("Unknown thingy in SQL");
 128+ }
 129+ }
 130+
 131+ protected string SqlEscape(string str) {
 132+ StringBuilder sql = new StringBuilder();
 133+ sql.Append('\'');
 134+ int len = str.Length;
 135+ for (int i = 0; i < len; i++) {
 136+ char c = str[i];
 137+ switch (c) {
 138+ case '\u0000':
 139+ sql.Append("\\0");
 140+ break;
 141+ case '\n':
 142+ sql.Append("\\n");
 143+ break;
 144+ case '\r':
 145+ sql.Append("\\r");
 146+ break;
 147+ case '\u001a':
 148+ sql.Append("\\Z");
 149+ break;
 150+ case '"':
 151+ case '\'':
 152+ case '\\':
 153+ sql.Append('\\');
 154+ goto default;
 155+ default:
 156+ sql.Append(c);
 157+ break;
 158+ }
 159+ }
 160+ sql.Append('\'');
 161+ return sql.ToString();
 162+ }
 163+
 164+ protected string TitleFormat(string title) {
 165+ return title.Replace(' ', '_');
 166+ }
 167+
 168+ protected string TimestampFormat(DateTime time) {
 169+ return string.Format("{0:0000}{1:00}{2:00}{3:00}{4:00}{5:00}",
 170+ time.Year,
 171+ time.Month,
 172+ time.Day,
 173+ time.Hour,
 174+ time.Minute,
 175+ time.Second);
 176+ }
 177+
 178+ protected string InverseTimestamp(DateTime time) {
 179+ return string.Format("{0:0000}{1:00}{2:00}{3:00}{4:00}{5:00}",
 180+ 9999 - time.Year,
 181+ 99 - time.Month,
 182+ 99 - time.Day,
 183+ 99 - time.Hour,
 184+ 99 - time.Minute,
 185+ 99 - time.Second);
 186+ }
 187+}
Property changes on: trunk/mwdumper/mwimport/SqlWriter.cs
___________________________________________________________________
Added: svn:keywords
1188 + Author Date Id Revision
Added: svn:eol-style
2189 + native
Index: trunk/mwdumper/mwimport/mwimport.mds
——@@ -0,0 +1,16 @@
 2+<Combine name="mwimport" fileversion="2.0">
 3+ <Configurations active="Debug">
 4+ <Configuration name="Debug" ctype="CombineConfiguration">
 5+ <Entry configuration="Debug" build="True" name="mwimport" />
 6+ </Configuration>
 7+ <Configuration name="Release" ctype="CombineConfiguration">
 8+ <Entry configuration="Debug" build="True" name="mwimport" />
 9+ </Configuration>
 10+ </Configurations>
 11+ <StartMode startupentry="mwimport" single="True">
 12+ <Execute type="None" entry="mwimport" />
 13+ </StartMode>
 14+ <Entries>
 15+ <Entry filename="./mwimport.mdp" />
 16+ </Entries>
 17+</Combine>
\ No newline at end of file
Property changes on: trunk/mwdumper/mwimport/mwimport.mds
___________________________________________________________________
Added: svn:keywords
118 + Author Date Id Revision
Added: svn:eol-style
219 + native
Index: trunk/mwdumper/mwimport/Revision.cs
——@@ -0,0 +1,25 @@
 2+// created on 8/28/2005 at 11:52 PM
 3+using System;
 4+using System.Collections;
 5+
 6+class Revision {
 7+ public int Id;
 8+ public DateTime Timestamp;
 9+ public Contributor Contributor;
 10+ public string Comment;
 11+ public string Text;
 12+ public bool Minor;
 13+
 14+ public bool IsRedirect {
 15+ get {
 16+ // todo
 17+ return false;
 18+ }
 19+ }
 20+
 21+ public Revision() {
 22+ Comment = "";
 23+ Text = "";
 24+ Minor = false;
 25+ }
 26+}
Property changes on: trunk/mwdumper/mwimport/Revision.cs
___________________________________________________________________
Added: svn:keywords
127 + Author Date Id Revision
Added: svn:eol-style
228 + native
Index: trunk/mwdumper/mwimport/Siteinfo.cs
——@@ -0,0 +1,11 @@
 2+// created on 8/29/2005 at 12:09 AM
 3+using System;
 4+using System.Collections;
 5+
 6+public class Siteinfo {
 7+ public string Sitename;
 8+ public string Base;
 9+ public string Generator;
 10+ public string Case;
 11+ public IDictionary Namespaces;
 12+}
Property changes on: trunk/mwdumper/mwimport/Siteinfo.cs
___________________________________________________________________
Added: svn:keywords
113 + Author Date Id Revision
Added: svn:eol-style
214 + native
Index: trunk/mwdumper/mwimport/SqlWriter14.cs
——@@ -0,0 +1,69 @@
 2+// created on 8/29/2005 at 12:13 AM
 3+using System;
 4+using System.Collections;
 5+using System.IO;
 6+
 7+class SqlWriter14 : SqlWriter {
 8+ Random _random;
 9+ Page _currentPage;
 10+ Revision _lastRevision;
 11+
 12+ public SqlWriter14(TextWriter output) : base(output) {
 13+ _random = new Random();
 14+ }
 15+
 16+ public override void WriteStartPage(Page page) {
 17+ _currentPage = page;
 18+ _lastRevision = null;
 19+ }
 20+
 21+ public override void WriteEndPage() {
 22+ if (_lastRevision != null)
 23+ WriteCurRevision(_currentPage, _lastRevision);
 24+ _currentPage = null;
 25+ _lastRevision = null;
 26+ }
 27+
 28+ public override void WriteRevision(Revision revision) {
 29+ if (_lastRevision != null)
 30+ WriteOldRevision(_currentPage, _lastRevision);
 31+ _lastRevision = revision;
 32+ }
 33+
 34+ private void WriteOldRevision(Page page, Revision revision) {
 35+ IDictionary row = new Hashtable();
 36+ row["old_id"] = revision.Id;
 37+ row["old_namespace"] = page.Title.Namespace;
 38+ row["old_title"] = TitleFormat(page.Title.Text);
 39+ row["old_text"] = revision.Text;
 40+ row["old_comment"] = revision.Comment;
 41+ row["old_user"] = revision.Contributor.Id;
 42+ row["old_user_text"] = revision.Contributor.Username;
 43+ row["old_timestamp"] = TimestampFormat(revision.Timestamp);
 44+ row["old_minor_edit"] = revision.Minor ? 1 : 0;
 45+ row["old_flags"] = "utf-8";
 46+ row["inverse_timestamp"] = InverseTimestamp(revision.Timestamp);
 47+ InsertRow("old", row);
 48+ }
 49+
 50+ private void WriteCurRevision(Page page, Revision revision) {
 51+ IDictionary row = new Hashtable();
 52+ row["cur_id"] = revision.Id;
 53+ row["cur_namespace"] = page.Title.Namespace;
 54+ row["cur_title"] = TitleFormat(page.Title.Text);
 55+ row["cur_text"] = revision.Text;
 56+ row["cur_comment"] = revision.Comment;
 57+ row["cur_user"] = revision.Contributor.Id;
 58+ row["cur_user_text"] = revision.Contributor.Username;
 59+ row["cur_timestamp"] = TimestampFormat(revision.Timestamp);
 60+ row["cur_restrictions"] = page.Restrictions;
 61+ row["cur_counter"] = 0;
 62+ row["cur_is_redirect"] = revision.IsRedirect ? 1 : 0;
 63+ row["cur_minor_edit"] = revision.Minor ? 1 : 0;
 64+ row["cur_random"] = _random.NextDouble();
 65+ row["cur_touched"] = TimestampFormat(DateTime.UtcNow);
 66+ row["inverse_timestamp"] = InverseTimestamp(revision.Timestamp);
 67+ InsertRow("cur", row);
 68+ }
 69+
 70+}
Property changes on: trunk/mwdumper/mwimport/SqlWriter14.cs
___________________________________________________________________
Added: svn:keywords
171 + Author Date Id Revision
Added: svn:eol-style
272 + native
Index: trunk/mwdumper/mwimport/Title.cs
——@@ -0,0 +1,30 @@
 2+// created on 8/28/2005 at 11:58 PM
 3+using System;
 4+using System.Collections;
 5+
 6+struct Title {
 7+ public int Namespace;
 8+ public string Text;
 9+
 10+ public Title(string prefixedTitle, IDictionary namespaces) {
 11+ foreach (int key in namespaces.Keys) {
 12+ string prefix = (string)namespaces[key];
 13+ int len = prefix.Length;
 14+ if (len > 0
 15+ && (prefixedTitle.Length - len) > 1
 16+ && prefixedTitle.StartsWith(prefix)
 17+ && prefixedTitle[len] == ':') {
 18+ Namespace = key;
 19+ Text = Title.ValidateTitleChars(prefixedTitle.Substring(len + 1));
 20+ return;
 21+ }
 22+ }
 23+ Namespace = 0;
 24+ Text = prefixedTitle;
 25+ }
 26+
 27+ public static string ValidateTitleChars(string text) {
 28+ // FIXME
 29+ return text;
 30+ }
 31+}
Property changes on: trunk/mwdumper/mwimport/Title.cs
___________________________________________________________________
Added: svn:keywords
132 + Author Date Id Revision
Added: svn:eol-style
233 + native
Index: trunk/mwdumper/mwimport/SqlWriter15.cs
——@@ -0,0 +1,71 @@
 2+// created on 8/29/2005 at 12:49 AM
 3+
 4+// Doesn't actually work yet...
 5+
 6+using System;
 7+using System.Collections;
 8+using System.IO;
 9+
 10+class SqlWriter15 : SqlWriter {
 11+ Random _random;
 12+ Page _currentPage;
 13+ Revision _lastRevision;
 14+
 15+ public SqlWriter15(TextWriter output) : base(output) {
 16+ _random = new Random();
 17+ }
 18+
 19+ public override void WriteStartPage(Page page) {
 20+ IDictionary row = new Hashtable();
 21+ row["page_id"] = page.Id;
 22+ row["page_namespace"] = page.Title.Namespace;
 23+ row["page_title"] = TitleFormat(page.Title.Text);
 24+ row["page_restrictions"] = page.Restrictions;
 25+ row["page_counter"] = 0;
 26+ row["page_is_redirect"] = 0;
 27+ row["page_is_new"] = 0;
 28+ row["page_random"] = _random.NextDouble();
 29+ row["page_touched"] = TimestampFormat(DateTime.UtcNow);
 30+ row["page_latest"] = 0; // We'll touch this up at the end...
 31+ row["page_len"] = 0; // .....
 32+ InsertRow("page", row);
 33+ }
 34+
 35+ public override void WriteEndPage() {
 36+ if (_lastRevision != null)
 37+ UpdatePage(_currentPage, _lastRevision);
 38+ _currentPage = null;
 39+ _lastRevision = null;
 40+ }
 41+
 42+ public override void WriteRevision(Revision revision) {
 43+ IDictionary row = new Hashtable();
 44+ row["old_id"] = null;
 45+ row["old_text"] = revision.Text;
 46+ row["old_flags"] = "utf-8";
 47+ object textId = InsertRow("text", row);
 48+
 49+ row = new Hashtable();
 50+ row["rev_id"] = revision.Id;
 51+ row["rev_page"] = _currentPage.Id;
 52+ row["rev_text_id"] = textId;
 53+ row["rev_comment"] = revision.Comment;
 54+ row["rev_user"] = revision.Contributor.Id;
 55+ row["rev_user_text"] = revision.Contributor.Username;
 56+ row["rev_timestamp"] = TimestampFormat(revision.Timestamp);
 57+ row["rev_minor_edit"] = revision.Minor ? 1 : 0;
 58+ row["rev_deleted"] = 0;
 59+
 60+ InsertRow("rev", row);
 61+ _lastRevision = revision;
 62+ }
 63+
 64+ private void UpdatePage(Page page, Revision revision) {
 65+ IDictionary row = new Hashtable();
 66+ row["page_len"] = revision.Text.Length; // TODO: UTF-8 byte length
 67+ row["page_latest"] = revision.Id;
 68+ row["page_is_redirect"] = revision.IsRedirect ? 1 : 0;
 69+ UpdateRow("page", row, "page_id", page.Id);
 70+ }
 71+
 72+}
Property changes on: trunk/mwdumper/mwimport/SqlWriter15.cs
___________________________________________________________________
Added: svn:keywords
173 + Author Date Id Revision
Added: svn:eol-style
274 + native
Index: trunk/mwdumper/mwimport/Main.cs
——@@ -0,0 +1,221 @@
 2+// project created on 8/28/2005 at 11:08 PM
 3+
 4+/*
 5+ -> read header info
 6+ site name, url, language, namespace keys
 7+
 8+ -> read pages.....
 9+ <page>
 10+ -> get title, etc
 11+ <revision>
 12+ -> store each revision
 13+ on next one or end of sequence, write out
 14+ [so for 1.4 schema we can be friendly]
 15+
 16+ progress report:
 17+ if possible, a percentage through file. this might not be possible.
 18+ rates and counts definitely
 19+
 20+ input:
 21+ stdin or file
 22+ allow gzip -> autodetect if possible
 23+
 24+ output:
 25+ SQL on stdout
 26+ SQL on file
 27+ SQL directly to a server
 28+
 29+ output formats:
 30+ 1.4 schema
 31+ 1.5 schema
 32+
 33+*/
 34+
 35+using System;
 36+using System.Collections;
 37+using System.IO;
 38+using System.Text;
 39+using System.Xml;
 40+
 41+using ICSharpCode.SharpZipLib.GZip;
 42+
 43+class MainClass {
 44+ public static void Main(string[] args) {
 45+ XmlTextReader reader = new XmlTextReader(Console.In);
 46+ reader.WhitespaceHandling = WhitespaceHandling.Significant;
 47+ SqlWriter14 writer = new SqlWriter14(Console.Out);
 48+
 49+ reader.ReadStartElement();
 50+ writer.WriteStartWiki();
 51+ Siteinfo siteinfo = null;
 52+
 53+ while (!reader.EOF) {
 54+ reader.Read();
 55+ if (reader.NodeType == XmlNodeType.Element) {
 56+ if (reader.LocalName.Equals("page"))
 57+ ReadPage(reader, writer, siteinfo);
 58+ else if (reader.LocalName.Equals("siteinfo"))
 59+ siteinfo = ReadSiteinfo(reader, writer);
 60+ }
 61+ }
 62+ reader.Close();
 63+ writer.WriteEndWiki();
 64+ }
 65+
 66+ static Siteinfo ReadSiteinfo(XmlReader reader, IDumpWriter writer) {
 67+ Siteinfo info = new Siteinfo();
 68+ while (reader.Read()) {
 69+ string name = reader.LocalName;
 70+ if (reader.NodeType == XmlNodeType.Element) {
 71+ reader.MoveToContent();
 72+ if (name.Equals("sitename"))
 73+ info.Sitename = reader.Value;
 74+ else if (name.Equals("base"))
 75+ info.Sitename = reader.Value;
 76+ else if (name.Equals("generator"))
 77+ info.Sitename = reader.Value;
 78+ else if (name.Equals("case"))
 79+ info.Sitename = reader.Value;
 80+ else if (name.Equals("namespaces"))
 81+ info.Namespaces = ReadNamespaces(reader);
 82+ } else if (reader.NodeType == XmlNodeType.EndElement && name.Equals("siteinfo")) {
 83+ return info;
 84+ }
 85+ }
 86+ throw new ArgumentException("Ran out of XML early; incomplete <siteinfo>");
 87+ }
 88+
 89+ static IDictionary ReadNamespaces(XmlReader reader) {
 90+ Hashtable namespaces = new Hashtable();
 91+ while (reader.Read()) {
 92+ string name = reader.LocalName;
 93+ if (reader.NodeType == XmlNodeType.Element && name.Equals("namespace")) {
 94+ int key = XmlConvert.ToInt32(reader.GetAttribute("key"));
 95+ reader.MoveToContent();
 96+ namespaces[key] = reader.Value;
 97+ } else if (reader.NodeType == XmlNodeType.EndElement && name.Equals("namespaces")) {
 98+ return namespaces;
 99+ }
 100+ }
 101+ throw new ArgumentException("Ran out of XML early; incomplete <namespaces>");
 102+ }
 103+
 104+ static string ReadElementContent(XmlReader reader) {
 105+ StringBuilder val = new StringBuilder();
 106+ while (reader.Read()) {
 107+ //Console.WriteLine("XXX: " + reader.NodeType + ", " + reader.LocalName + ", " + reader.Value);
 108+ switch (reader.NodeType) {
 109+ case XmlNodeType.SignificantWhitespace:
 110+ case XmlNodeType.Text:
 111+ reader.MoveToContent();
 112+ val.Append(reader.Value);
 113+ break;
 114+ case XmlNodeType.EndElement:
 115+ return val.ToString();
 116+ default:
 117+ // ignore
 118+ break;
 119+ }
 120+ }
 121+ return val.ToString();
 122+ }
 123+
 124+ static void ReadPage(XmlReader reader, IDumpWriter writer, Siteinfo siteinfo) {
 125+ Page page = new Page();
 126+
 127+ while (reader.Read()) {
 128+ string name = reader.LocalName;
 129+ if (reader.NodeType == XmlNodeType.Element) {
 130+ if (name.Equals("revision"))
 131+ break; // Move on to Stage Two
 132+
 133+ string val = ReadElementContent(reader);
 134+
 135+ if (name.Equals("title"))
 136+ page.Title = new Title(val, siteinfo.Namespaces);
 137+ else if (name.Equals("id"))
 138+ page.Id = XmlConvert.ToInt32(val);
 139+ else if (name.Equals("restrictions"))
 140+ page.Restrictions = val;
 141+ }
 142+ }
 143+
 144+ writer.WriteStartPage(page);
 145+ do {
 146+ string name = reader.LocalName;
 147+ if (reader.NodeType == XmlNodeType.Element && name.Equals("revision"))
 148+ ReadRevision(reader, writer, siteinfo);
 149+ else if (reader.NodeType == XmlNodeType.EndElement && name.Equals("page"))
 150+ break;
 151+ } while(reader.Read());
 152+ writer.WriteEndPage();
 153+ }
 154+
 155+ static void ReadRevision(XmlReader reader, IDumpWriter writer, Siteinfo siteinfo) {
 156+ Revision rev = new Revision();
 157+ while (reader.Read()) {
 158+ string name = reader.LocalName;
 159+ if (reader.NodeType == XmlNodeType.Element) {
 160+ string val = ReadElementContent(reader);
 161+ if (name.Equals("id"))
 162+ rev.Id = XmlConvert.ToInt32(val);
 163+ else if (name.Equals("timestamp"))
 164+ rev.Timestamp = XmlConvert.ToDateTime(val);
 165+ else if (name.Equals("contributor"))
 166+ rev.Contributor = new Contributor("test");
 167+ else if (name.Equals("minor"))
 168+ rev.Minor = true;
 169+ else if (name.Equals("comment"))
 170+ rev.Comment = val;
 171+ else if (name.Equals("text"))
 172+ rev.Text = val;
 173+ } else if (reader.NodeType == XmlNodeType.EndElement && name.Equals("revision")) {
 174+ writer.WriteRevision(rev);
 175+ return;
 176+ }
 177+ }
 178+ }
 179+
 180+ public static void Test(string[] args) {
 181+ Siteinfo info = new Siteinfo();
 182+ info.Sitename = "OneFive";
 183+ info.Base = "http://localhost/head/index.php/Main_Page";
 184+ info.Generator = "MediaWiki 1.6alpha";
 185+ info.Case = "first-letter";
 186+ info.Namespaces = new Hashtable();
 187+ info.Namespaces[-2] = "Media";
 188+ info.Namespaces[-1] = "Special";
 189+ info.Namespaces[0] = "";
 190+ info.Namespaces[1] = "Talk";
 191+
 192+ Page page = new Page();
 193+ page.Id = 1;
 194+ page.Title = new Title("Talk:Main Page", info.Namespaces);
 195+ page.Restrictions = "";
 196+
 197+ Revision revision = new Revision();
 198+ revision.Id = 1;
 199+ revision.Text = "This is a bunch of stuff\nyo momma!";
 200+ revision.Minor = true;
 201+ revision.Timestamp = DateTime.UtcNow;
 202+ revision.Contributor = new Contributor("WikiSysop", 1);
 203+ revision.Comment = "wacky edit o doom (it's all good)";
 204+
 205+ Revision revision2 = new Revision();
 206+ revision2.Id = 2;
 207+ revision2.Text = "''''''''\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\"\" VANDALE!!!!!";
 208+ revision2.Minor = false;
 209+ revision2.Timestamp = DateTime.UtcNow;
 210+ revision2.Contributor = new Contributor("127.0.0.1");
 211+ revision2.Comment = "/* fuk uuuu */";
 212+
 213+ SqlWriter14 writer = new SqlWriter14(Console.Out);
 214+ writer.WriteStartWiki();
 215+ writer.WriteSiteinfo(info);
 216+ writer.WriteStartPage(page);
 217+ writer.WriteRevision(revision);
 218+ writer.WriteRevision(revision2);
 219+ writer.WriteEndPage();
 220+ writer.WriteEndWiki();
 221+ }
 222+}
Property changes on: trunk/mwdumper/mwimport/Main.cs
___________________________________________________________________
Added: svn:keywords
1223 + Author Date Id Revision
Added: svn:eol-style
2224 + native
Index: trunk/mwdumper/mwimport/mwimport.mdp
——@@ -0,0 +1,36 @@
 2+<Project name="mwimport" fileversion="2.0" language="C#" ctype="DotNetProject">
 3+ <Configurations active="Debug">
 4+ <Configuration name="Debug" ctype="DotNetProjectConfiguration">
 5+ <Output directory="./bin/Debug" assembly="mwimport" />
 6+ <Build debugmode="True" target="Exe" />
 7+ <Execution runwithwarnings="True" consolepause="True" runtime="MsNet" />
 8+ <CodeGeneration compiler="Csc" warninglevel="4" optimize="True" unsafecodeallowed="False" generateoverflowchecks="True" generatexmldocumentation="False" ctype="CSharpCompilerParameters" />
 9+ </Configuration>
 10+ <Configuration name="Release" ctype="DotNetProjectConfiguration">
 11+ <Output directory="./bin/Release" assembly="mwimport" />
 12+ <Build debugmode="False" target="Exe" />
 13+ <Execution runwithwarnings="True" consolepause="True" runtime="MsNet" />
 14+ <CodeGeneration compiler="Csc" warninglevel="4" optimize="True" unsafecodeallowed="False" generateoverflowchecks="True" generatexmldocumentation="False" ctype="CSharpCompilerParameters" />
 15+ </Configuration>
 16+ </Configurations>
 17+ <DeploymentInformation strategy="File">
 18+ <excludeFiles />
 19+ </DeploymentInformation>
 20+ <Contents>
 21+ <File name="./Main.cs" subtype="Code" buildaction="Compile" />
 22+ <File name="./AssemblyInfo.cs" subtype="Code" buildaction="Compile" />
 23+ <File name="./Page.cs" subtype="Code" buildaction="Compile" />
 24+ <File name="./Revision.cs" subtype="Code" buildaction="Compile" />
 25+ <File name="./Contributor.cs" subtype="Code" buildaction="Compile" />
 26+ <File name="./Title.cs" subtype="Code" buildaction="Compile" />
 27+ <File name="./IDumpWriter.cs" subtype="Code" buildaction="Compile" />
 28+ <File name="./Siteinfo.cs" subtype="Code" buildaction="Compile" />
 29+ <File name="./SqlWriter14.cs" subtype="Code" buildaction="Compile" />
 30+ <File name="./SqlWriter.cs" subtype="Code" buildaction="Compile" />
 31+ <File name="./SqlWriter15.cs" subtype="Code" buildaction="Compile" />
 32+ </Contents>
 33+ <References>
 34+ <ProjectReference type="Assembly" localcopy="True" refto="../libs/ICSharpCode.SharpZipLib.dll" />
 35+ <ProjectReference type="Gac" localcopy="True" refto="System.Xml, Version=1.0.5000.0, Culture=neutral, PublicKeyToken=b77a5c561934e089" />
 36+ </References>
 37+</Project>
\ No newline at end of file
Property changes on: trunk/mwdumper/mwimport/mwimport.mdp
___________________________________________________________________
Added: svn:keywords
138 + Author Date Id Revision
Added: svn:eol-style
239 + native
Index: trunk/mwdumper/mwimport/Contributor.cs
——@@ -0,0 +1,21 @@
 2+// created on 8/28/2005 at 11:53 PM
 3+struct Contributor {
 4+ public string Username;
 5+ public int Id;
 6+
 7+ public string Address {
 8+ get {
 9+ return Username;
 10+ }
 11+ }
 12+
 13+ public Contributor(string username, int id) {
 14+ Username = username;
 15+ Id = id;
 16+ }
 17+
 18+ public Contributor(string ip) {
 19+ Username = ip;
 20+ Id = 0;
 21+ }
 22+}
Property changes on: trunk/mwdumper/mwimport/Contributor.cs
___________________________________________________________________
Added: svn:keywords
123 + Author Date Id Revision
Added: svn:eol-style
224 + native

Status & tagging log