/// <summary>
/// Check the specified URL for a birth year. This will occur if one sentence
/// is found that has the word born, and a numeric value less than 3000.
/// </summary>
/// <param name="report">Object to report to.</param>
/// <param name="url">The url.</param>
/// <param name="desiredYear">The desired year.</param>
public static void CheckURL(ScanReportable report, Uri url,
int desiredYear)
{
int ch;
StringBuilder sentence = new StringBuilder();
String ignoreUntil = null;
WebRequest http = HttpWebRequest.Create(url);
http.Timeout = 10000;
HttpWebResponse response = (HttpWebResponse)http.GetResponse();
Stream istream = response.GetResponseStream();
ParseHTML html = new ParseHTML(istream);
do
{
ch = html.Read();
if ((ch != -1) && (ch != 0) && (ignoreUntil == null))
{
if (".?!".IndexOf((char)ch) != -1)
{
String str = sentence.ToString();
int year = Text.ExtractYear(str);
if (desiredYear == -1)
{
// looking for any year
if (year != -1)
{
report.ReceiveGoodSentence(str);
}
}
else
{
// looking for a specific year
if (year == desiredYear)
{
report.ReceiveGoodSentence(str);
}
else if (year != -1)
{
report.ReceiveBadSentence(str);
}
}
sentence.Length = 0;
}
else if (ch == ' ')
{
string str = sentence.ToString();
if ((sentence.Length > 0) &&
(str[str.Length - 1] != ' '))
{
sentence.Append(' ');
}
}
else if ((ch != '\n') && (ch != '\t') && (ch != '\r'))
{
if ((ch) < 128)
{
sentence.Append((char)ch);
}
}
}
else if (ch == 0)
{
// clear anything before a body tag
if (html.Tag.Name.Equals("body", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("br", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("li", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("p", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("h1", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("h2", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("h3", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("td", StringComparison.CurrentCultureIgnoreCase) ||
html.Tag.Name.Equals("th", StringComparison.CurrentCultureIgnoreCase))
{
sentence.Length = 0;
}
// ignore everything between script and style tags
if (ignoreUntil == null)
{
if (html.Tag.Name.Equals("script", StringComparison.CurrentCultureIgnoreCase))
{
ignoreUntil = "/script";
}
else if (html.Tag.Name
.Equals("style", StringComparison.CurrentCultureIgnoreCase))
{
ignoreUntil = "/style";
}
}
else
{
if (html.Tag.Name.Equals(ignoreUntil, StringComparison.CurrentCultureIgnoreCase))
{
ignoreUntil = null;
}
}
// add a space after the tag
if (sentence.Length > 0)
{
string str = sentence.ToString();
if (str[str.Length - 1] != ' ')
{
sentence.Append(' ');
}
}
}
} while (ch != -1);
}
/// <summary>
/// Check the specified URL for a birth year. This will occur if one sentence
/// is found that has the word born, and a numeric value less than 3000.
/// </summary>
/// <param name="report">Object to report to.</param>
/// <param name="url">The url.</param>
/// <param name="desiredYear">The desired year.</param>
public static void CheckURL(ScanReportable report, Uri url,
int desiredYear)
{
int ch;
StringBuilder sentence = new StringBuilder();
String ignoreUntil = null;
WebRequest http = HttpWebRequest.Create(url);
http.Timeout = 10000;
HttpWebResponse response = (HttpWebResponse)http.GetResponse();
Stream istream = response.GetResponseStream();
ParseHTML html = new ParseHTML(istream);
do
{
ch = html.Read();
if ((ch != -1) && (ch != 0) && (ignoreUntil == null))
{
if (".?!".IndexOf((char)ch) != -1)
{
String str = sentence.ToString();
int year = Text.ExtractYear(str);
if (desiredYear == -1)
{
// looking for any year
if (year != -1)
{
report.ReceiveGoodSentence(str);
}
}
else
{
// looking for a specific year
if (year == desiredYear)
{
report.ReceiveGoodSentence(str);
}
else if (year != -1)
{
report.ReceiveBadSentence(str);
}
}
sentence.Length = 0;
}
else if (ch == ' ')
{
string str = sentence.ToString();
if ((sentence.Length > 0)
&& (str[str.Length - 1] != ' '))
{
sentence.Append(' ');
}
}
else if ((ch != '\n') && (ch != '\t') && (ch != '\r'))
{
if ((ch) < 128)
{
sentence.Append((char)ch);
}
}
}
else if (ch == 0)
{
// clear anything before a body tag
if (html.Tag.Name.Equals("body", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("br", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("li", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("p", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("h1", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("h2", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("h3", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("td", StringComparison.CurrentCultureIgnoreCase)
|| html.Tag.Name.Equals("th", StringComparison.CurrentCultureIgnoreCase))
{
sentence.Length = 0;
}
// ignore everything between script and style tags
if (ignoreUntil == null)
{
if (html.Tag.Name.Equals("script", StringComparison.CurrentCultureIgnoreCase))
{
ignoreUntil = "/script";
}
else if (html.Tag.Name
.Equals("style", StringComparison.CurrentCultureIgnoreCase))
{
ignoreUntil = "/style";
}
}
else
{
if (html.Tag.Name.Equals(ignoreUntil, StringComparison.CurrentCultureIgnoreCase))
{
ignoreUntil = null;
}
}
// add a space after the tag
if (sentence.Length > 0)
{
string str = sentence.ToString();
if (str[str.Length - 1] != ' ')
{
sentence.Append(' ');
}
}
}
} while (ch != -1);
}