grab_tags.cc

#include <iostream>
#include <vector>

using namespace std;

vector<string>
getNames(string in)
{
	vector<string> out;
	char c = in[0];
	char prev = 0;
	string buf = "";
	bool grabtag = false;

	int i=0;
	while(i<in.length())
	{
		if(grabtag)
		{
			if (c != '%')
			{
				buf += c;
			}
			else
			{
				grabtag = false;
				out.push_back(buf);
				buf = "";
			}
		}
		if(c =='%' && prev == '<')
		{
			grabtag = true;
		}

		prev = c;
		c = in[++i];
	}

	return out;
}

wstring
mergeSurfaceForms(wstring in)
{
	bool inword = false;
	bool inanalysis = false;
	wstring out = L"";
	wstring blank = L"";
	wchar_t cur = 0;

	for(int i=0; i<=in.size(); i++)
	{
		cur = in[i];
		if(cur == L'^')
		{
			inword = true;
			inanalysis = false;
			if(blank != L"")
			{
				out += L" ";
				blank = L"";
			}
			cur = in[++i];
		}
		if(inword)
		{
			if(cur == L'/')
			{
				inword = false;
				inanalysis = true;
			}
			else if(cur == L'\\')
			{
				out += L'\\';
				out += in[++i];
			}
			else
			{
				out += cur;
			}
		}
		if(inanalysis)
		{
			if(cur == L'$')
			{
				inanalysis = false;
			}
		}
		if(!inword && !inanalysis)
		{
			blank += cur;
		}
	}

	return out;
}

int main (int argc, char** argv)
{
	vector<string> s = getNames("^testing<adj><%nbr%>$ ^foo<n><%nbr%>");
	int count = 1;
	vector<string>::iterator it;

	int i;
	for(it = s.begin(), i=0; it != s.end(); it++, i++)
	{
		cout << count << ' ' << s[i] << endl;
		count++;
	}

	wcout << mergeSurfaceForms(L"^thing/thing<n>$[<b>]^to/to<foo>$ ^check/check<n>$") << endl;
	wcout << mergeSurfaceForms(L"^th\\/ing/thing<n>$[<b>]^to/to<foo>$ ^check/check<n>$^2/2<num$") << endl;
	return 0;
}