public static header_map step1(doc_reader docs)
{
header_map ret = new header_map();
foreach (doc_reader.doc doc in docs) {
ret.push(doc.title, new doc_reader.doc.location(doc.id));
//TODO 同义词
}
return ret;
}
public static header_map step1(doc_reader docs)
{
header_map ret = new header_map();
foreach (doc_reader.doc doc in docs)
{
ret.push(doc.title, new doc_reader.doc.location(doc.id));
//TODO 同义词
}
return(ret);
}
public static link_map step2(doc_reader docs, header_map links)
{
link_map ret = new link_map();
//双循环可以改成多线程做
foreach(doc_reader.doc doc in docs){
foreach(KeyValuePair<string, doc_reader.doc.location> link in links){
Regex law_regex = new Regex(link.Key);
link_map temp= step2_each(doc, link, law_regex);
ret.union(temp);
}
}
return ret;
}
public static link_map step2_each(doc_reader.doc doc, KeyValuePair<string, doc_reader.doc.location> link, Regex law_regex)
{
link_map ret = new link_map();
MatchCollection mc= law_regex.Matches(doc.text);
foreach(Match iter in mc)
{
doc_reader.doc.location match_at= new doc_reader.doc.location(doc.id, iter.Index, iter.Length);
if (step2_flip(doc, match_at, iter))
ret.push(match_at, link.Value);
//TODO push <XXX>XX章
}
return ret;
}
public static link_map step2(doc_reader docs, header_map links)
{
link_map ret = new link_map();
//双循环可以改成多线程做
foreach (doc_reader.doc doc in docs)
{
foreach (KeyValuePair <string, doc_reader.doc.location> link in links)
{
Regex law_regex = new Regex(link.Key);
link_map temp = step2_each(doc, link, law_regex);
ret.union(temp);
}
}
return(ret);
}
public static bool step2_flip(doc_reader.doc doc, doc_reader.doc.location match_at, Match _match)
{
//TODO 过滤
return true;
}
public void push(doc_reader.doc.location from, doc_reader.doc.location dest)
{
links.Add(from, dest);
}
public void push(string name, doc_reader.doc.location dest)
{
try
{
links.Add(name, dest);
}catch(ArgumentException e){
Console.WriteLine(e.Message+" "+name);
}
}