public void Match_TwoWordsTextMatchesPattern()
{
var text = "buch katze";
var pattern = "23";
Assert.IsTrue(WordPattern.TextMatchesPattern(text, pattern));
}
public void Given_string_same_When_match_Then_return_false()
{
var pattern = "abba";
var str = "dog dog dog dog";
Assert.AreEqual(false, WordPattern.FindWordPattern(pattern, str));
}
public Dictionary <string, int> CountWords()
{
return(WordPattern.Matches(_text).Cast <Match>()
.Select(m => m.Value.ToLower())
.GroupBy(word => word)
.ToDictionary(g => g.Key, g => g.Count()));
}
public void Given_pattern_str_When_match_Then_return()
{
var pattern = "abba";
var str = "dog cat cat dog";
Assert.AreEqual(true, WordPattern.FindWordPattern(pattern, str));
}
public void DoesntMatch_PatternLongerThenText()
{
var text = "buch hund";
var pattern = "123";
Assert.IsFalse(WordPattern.TextMatchesPattern(text, pattern));
}
public void DoesntMatch_TextLongerThenPattern()
{
var text = "buch hund Katze";
var pattern = "12";
Assert.IsFalse(WordPattern.TextMatchesPattern(text, pattern));
}
public void DoesntMatch_TwoWordsTextMatchesPattern()
{
var text = "buch hund";
var pattern = "22";
Assert.IsFalse(WordPattern.TextMatchesPattern(text, pattern));
}
private string ElementMatchHandler(Definition definition, Match match)
{
if (definition == null)
{
throw new ArgumentNullException(nameof(definition));
}
if (match == null)
{
throw new ArgumentNullException(nameof(match));
}
var pattern = definition.Patterns.First
(
x => match.Groups[x.Key]
.Success
)
.Value;
return(pattern switch
{
BlockPattern blockPattern => ProcessBlockPatternMatch(definition, blockPattern, match),
MarkupPattern markupPattern => ProcessMarkupPatternMatch(definition, markupPattern, match),
WordPattern wordPattern => ProcessWordPatternMatch(definition, wordPattern, match),
_ => match.Value,
});
public void Match_OneWordTextMatchesPattern()
{
var text = "buch";
var pattern = "5";
Assert.IsTrue(WordPattern.TextMatchesPattern(text, pattern));
}
static void Main(string[] args)
{
WordPattern a = new WordPattern();
Console.Out.WriteLine(a.countWords("ABCDEFGHIJKLMNOPQRSTUVWXYZABCDEFGHIJ"));
Console.In.ReadLine();
//
// TODO: Add code to start application here
//
}
// END CUT HERE
// BEGIN CUT HERE
public static void Main()
{
try {
WordPattern ___test = new WordPattern();
___test.run_test(-1);
} catch(Exception e) {
//Console.WriteLine(e.StackTrace);
Console.WriteLine(e.ToString());
}
}
// END CUT HERE
// BEGIN CUT HERE
public static void Main()
{
try {
WordPattern ___test = new WordPattern();
___test.run_test(-1);
} catch (Exception e) {
//Console.WriteLine(e.StackTrace);
Console.WriteLine(e.ToString());
}
}
// recursive function that progressively increases the depth if we get a too high word count
// but maximum 3 permutations (abc???)
private void ReadWordsByWordPermutationsRecursive(IWebDriver driver, WordPattern wordPattern, WordHintDbContext db, User adminUser)
{
// get word count
var(wordCount, node, url, page) = GetWordCountByPattern(driver, wordPattern);
if (wordCount == 0)
{
return;
}
else
{
Log.Information("Found {0} words when searching for '{1}' on page {2}", wordCount, wordPattern.Pattern, page + 1);
writer.WriteLine("Found {0} words when searching for '{1}' on page {2}", wordCount, wordPattern.Pattern, page + 1);
}
// if we get too many words back, try to increase the pattern depth
// but maximum 4 levels
if (wordCount <= 108 || wordPattern.Depth > 3)
{
if (wordPattern.Depth > 3)
{
Log.Error("Warning! Pattern search depth is now {0}. Found {1} words when searching for '{2}' on page {3}", wordPattern.Depth, wordCount, wordPattern.Pattern, page + 1);
}
// process each word found using the specified word pattern
ProcessWordsUntilEmpty(wordPattern, driver, db, adminUser, page, node, url);
}
else
{
// increment pattern
var childPatterns = wordPattern.GetWordPatternChildren();
// recursively process children patterns
foreach (var childPattern in childPatterns)
{
if (hasMissedLastWord)
{
return;
}
// if we have a last word - make sure to skip until the pattern matches
if (!hasFoundLastWord && childPattern.LastWord != null && !childPattern.IsMatchLastWord)
{
// skip pattern
Log.Information("Skipping pattern '{0}'.", childPattern.Pattern);
}
else
{
Log.Information("Processing pattern '{0}'.", childPattern.Pattern);
ReadWordsByWordPermutationsRecursive(driver, childPattern, db, adminUser);
}
}
}
}
protected override string ProcessWordPatternMatch(Definition definition, WordPattern pattern, Match match)
{
if (!UseCss)
{
var patternStyle = HtmlEngineHelper.CreatePatternStyle(pattern.Style);
return(string.Format(STYLE_SPAN_FORMAT, patternStyle, match.Value));
}
var cssClassName = HtmlEngineHelper.CreateCssClassName(definition.Name, pattern.Name);
return(string.Format(CLASS_SPAN_FORMAT, cssClassName, match.Value));
}
public override void Save()
{
if (!string.IsNullOrEmpty(SkipPattern))
{
try { new Regex(SkipPattern, RegexOptions.IgnorePatternWhitespace); }
catch (ArgumentException ex) { throw new ModuleException("Invalid skip pattern: " + ex.Message); }
}
if (WordPattern == null || WordPattern.Trim().Length == 0)
{
throw new ModuleException("Empty word pattern is invalid.");
}
try { new Regex(WordPattern, RegexOptions.IgnorePatternWhitespace); }
catch (ArgumentException ex) { throw new ModuleException("Invalid word pattern: " + ex.Message); }
base.Save();
}
public void Generate(MemoryStream ms, params object[] param)
{
//制作生成模板文件需要的实例
var obj = GetObjInstance(param);
var patternFilePath = GetPatternFilePath();
//配置文件生成参数
var setting = new Setting();
setting.PatternTextStyle.SignWithColor = true;
setting.PatternTextStyle.ForecolorName = System.Drawing.Color.Red.Name;
//文件生成
var wordPattern = new WordPattern(setting);
if (!wordPattern.Generate(patternFilePath, obj, ms))
{
throw new ApplicationException("Generate file failed.");
}
}
/// <summary>
/// Give a pattern for the unknown word with known letters filled in and the UnknownToken used for missing letters
/// </summary>
/// <param name="pattern"></param>
public UnknownWord(string pattern)
{
if (pattern == null)
{
throw new ArgumentNullException(nameof(pattern));
}
if (string.IsNullOrWhiteSpace(pattern))
{
throw new ArgumentException($"{nameof(pattern)} cannot be only whitespace");
}
pattern = pattern.Trim();
if (pattern.Any(char.IsWhiteSpace))
{
throw new ArgumentException($"The word pattern had some unexpected white space in it (ie it is several words): {pattern}");
}
WordPattern = pattern.ToLower();
_newWordTemplate = WordPattern.ToCharArray();
UnknownCharCount = WordPattern.Count(c => c == UnknownToken);
}
protected override Span[] ProcessWordPatternMatch(Definition definition, WordPattern pattern, Match match)
{
if (definition == null)
{
throw new ArgumentNullException(nameof(definition));
}
if (pattern == null)
{
throw new ArgumentNullException(nameof(pattern));
}
if (match == null)
{
throw new ArgumentNullException(nameof(match));
}
return(new[] { new Span()
{
Text = match.Value,
Style = GetSpanStyle(pattern.Style)
} });
}
private void ParseText_DoWork(object sender, DoWorkEventArgs e)
{
Lexicon.Clear();
var expElementPattern = new Regex($"({PunctuationPattern.ToString()})|({WordPattern.ToString()})");
var whiteSpacePattern = new Regex(@"[\s\n\r]+", RegexOptions.Singleline | RegexOptions.Multiline);
var worker = sender as BackgroundWorker;
var text = e.Argument as string;
int progress = 0;
string state = string.Empty;
// TODO: handling of paragraph breaks and section headers, etc
foreach (Match p in ParagraphPattern.Matches(text))
{
string paragraphText = p.Value.Trim();
paragraphText = whiteSpacePattern.Replace(paragraphText, " ");
var paragraph = new Lx.Discourse();
Text.Discourse.AddLast(paragraph);
foreach (Match l in LinePattern.Matches(paragraphText))
{
//store line, section up into words and punctuation
string cleanedLine = l.Value.Trim();
//cleanedLine = whiteSpacePattern.Replace(cleanedLine, " ");
state = cleanedLine;
var expression = new Lx.Expression(cleanedLine);
paragraph.Expressions.AddLast(expression);
foreach (Match m in expElementPattern.Matches(expression.Graph))
{
if (m.Groups.Count > 0)
{
// string m => List<Glyphs>
var glyphs = Script.AddGlyphs(m.Value.ToCharArray());
// List<Glyph> => List<Grapheme>
// Pre-analysis, graphemes are 1:1 with glyphs
var graphemes = Orthography.AddGraphemes(glyphs);
// List<Grapheme> => Morpheme
if (string.IsNullOrEmpty(m.Groups[1].Value))
{
//var morph = Text.Lexicon.Add(m.Groups[2].Value);
//morph.GraphemeChain.Add(Lx.SegmentChain<Lx.Grapheme>.NewSegmentChain(graphemes));
Lx.Morpheme morph = Text.Lexicon.Add(graphemes);
expression.Sequence.AddLast(morph);
}
else
{
expression.Sequence.AddLast(Text.Paralexicon.Add(m.Groups[1].Value));
}
}
}
worker.ReportProgress(++progress, state);
}
}
UpdateLocalLexicon();
}
protected override string ProcessWordPatternMatch(Definition definition, WordPattern pattern, Match match)
{
var style = CreateRtfPatternStyle(pattern.Style.Colors.ForeColor, pattern.Style.Colors.BackColor, pattern.Style.Font);
return("{" + String.Format(RtfFormat, style, match.Value) + "}");
}
protected abstract string ProcessWordPatternMatch(Definition definition, WordPattern pattern, Match match);
public void BeforeEach()
{
WordPattern = new WordPattern();
}
protected override string ProcessWordPatternMatch(Definition definition, WordPattern pattern, Match match)
{
HighlightedSpans.Add(new FormattedString(match.Value, pattern.Style.Colors.ForeColor, pattern.Style.Font.IsBold, pattern.Style.Font.IsItalic));
return(match.Value);
}
private Tuple <int, HtmlNode, string, int> GetWordCountByPattern(IWebDriver driver, WordPattern wordPattern)
{
// go to search result page
var query = "";
int page = 0;
string url = string.Format("{0}?a={1}&b={2}&p={3}", "https://www.kryssord.org/search.php", query, wordPattern.Pattern, page);
var(count, node) = GetWordCountByWordPattern(driver, url);
return(new Tuple <int, HtmlNode, string, int>(count, node, url, page));
}
private void ProcessWordsUntilEmpty(WordPattern wordPattern, IWebDriver driver, WordHintDbContext db, User adminUser, int page, HtmlNode documentNode, string url)
{
while (true)
{
Log.Information("Processing pattern search for '{0}' on page {1}", wordPattern.Pattern, page + 1);
writer.WriteLine("Processing pattern search for '{0}' on page {1}", wordPattern.Pattern, page + 1);
// parse all words
var words = ReadWordsAgilityPack(documentNode, adminUser);
foreach (var word in words)
{
if (wordPattern.IsMatchLastWord)
{
Log.Information("The current pattern matches the last-word: {0} = {1}. Current word: {2}", wordPattern.Pattern, wordPattern.LastWord, word.Value);
hasFoundPattern = true;
var wordRemoveDiacriticsToNorwegian = word.Value.RemoveDiacriticsToNorwegian();
// we might have had to add question marks at the end of the string to fix the length bug at the site
if (wordRemoveDiacriticsToNorwegian == wordPattern.LastWord.TrimEnd('?'))
{
Log.Information("The current word matches the last-word: {0} = {1}", word.Value, wordPattern.LastWord);
hasFoundLastWord = true;
}
}
else
{
if (!hasFoundLastWord && hasFoundPattern)
{
// if the pattern not any longer match, we never found the word - has it been deleted?
Log.Error("Warning! The current pattern does not any longer match the last-word: {0} = {1}. Current word: {2}", wordPattern.Pattern, wordPattern.LastWord, word.Value);
writer.WriteLine("Warning! The current pattern does not any longer match the last-word: {0} = {1}. Current word: {2}", wordPattern.Pattern, wordPattern.LastWord, word.Value);
hasMissedLastWord = true;
return;
}
}
if (hasFoundLastWord)
{
string currentValue = word.Value;
// check if this is one of the buggy words from their site where the words found don't have the same length as the pattern says it should have
if (wordPattern.Length != word.Value.Length)
{
Log.Error("Warning! The current word doesn't match the length of the query pattern: {0} = {1}", word.Value, wordPattern.Pattern);
writer.WriteLine("Warning! The current word doesn't match the length of the query pattern: {0} = {1}", word.Value, wordPattern.Pattern);
if (wordPattern.Length > word.Value.Length)
{
currentValue = currentValue + new string('?', wordPattern.Length - word.Value.Length);
}
else
{
currentValue = currentValue.Substring(0, wordPattern.Length);
}
}
else
{
// everything is OK
}
// update that we are processing this word, ignore length and comment
WordDatabaseService.UpdateState(db, source, new Word()
{
Value = currentValue.ToUpper(), Source = source, CreatedDate = DateTime.Now
}, writer);
GetWordSynonyms(word, driver, db, adminUser);
}
}
// go to next page if exist
var(hasFoundNextPage, pageNumber, pageUrl, pageNode) = NavigateToNextPageIfExist(driver, documentNode);
if (hasFoundNextPage)
{
url = pageUrl;
page = pageNumber;
documentNode = pageNode;
}
else
{
break;
}
}
}
protected override string ProcessWordPatternMatch(Definition definition, WordPattern pattern, Match match)
{
return(ProcessPatternMatch(pattern, match));
}
private List <List <List <string> > > AnalysePatterns(string[] blocks)
{
// Find occurance indeces for each word
Dictionary <string, List <int> > histogram = new Dictionary <string, List <int> >();
for (int blockIndex = 0; blockIndex < blocks.Length; blockIndex++)
{
string blockText = blocks[blockIndex];
string[] blockParts = blockText.Split(' ');
foreach (string blockPart in blockParts)
{
if (!histogram.ContainsKey(blockPart))
{
histogram.Add(blockPart, new List <int>());
}
histogram[blockPart].Add(blockIndex);
}
}
/*
* Count length of lists => list length that occures the most will be the one we're looking for
* Dictionary
* Key => length of word occurence list
* Value
* - times this list length was found
* - All words that have this length in occurences
*/
var metaHistogram = new Dictionary <int, WordPattern>();
foreach (KeyValuePair <string, List <int> > wordOccurence in histogram)
{
if (!metaHistogram.ContainsKey(wordOccurence.Value.Count))
{
metaHistogram.Add(wordOccurence.Value.Count, new WordPattern(0, new List <string>()));
}
metaHistogram[wordOccurence.Value.Count].NumOfOccurences++;
metaHistogram[wordOccurence.Value.Count].Words.Add(wordOccurence.Key);
}
// Remove lists that only occur once
var keys = metaHistogram.Keys.ToArray();
for (var i = keys.Length - 1; i >= 0; i--)
{
if (keys[i] == 1)
{
metaHistogram.Remove(keys[i]);
}
}
// Remove lists that span more than 100 wordblocks
keys = metaHistogram.Keys.ToArray();
for (var i = keys.Length - 1; i >= 0; i--)
{
List <string> words = metaHistogram[keys[i]].Words;
int firstWordIndex = histogram[words[0]][0];
int lastWordIndex = histogram[words[words.Count - 1]][0];
if (lastWordIndex - firstWordIndex > 100)
{
metaHistogram.Remove(keys[i]);
}
}
List <KeyValuePair <int, WordPattern> > metaHistogramSorted = (from metaHistoryEntry in metaHistogram orderby metaHistoryEntry.Key * metaHistoryEntry.Value.NumOfOccurences descending select metaHistoryEntry).ToList();
List <List <List <string> > > AllWordBlocks = new List <List <List <string> > >();
foreach (KeyValuePair <int, WordPattern> metaHistoryEntry in metaHistogramSorted)
{
// Get the blocks for all repetitive patterns that were found in the pdf
WordPattern wordPattern = metaHistoryEntry.Value;
List <List <int> > wordInces = new List <List <int> >();
foreach (string word in wordPattern.Words)
{
wordInces.Add(histogram[word]);
}
// Run over first word in best pattern series
List <List <string> > wordBlocks = new List <List <string> >();
for (int patternIndex = 0; patternIndex < wordInces[0].Count; patternIndex++)
{
List <string> wordBlock = new List <string>();
// Run over each text block between the first best word idex and the last best word index
int firstBlockIndex = wordInces[0][patternIndex];
int lastBlockIndex = wordInces[wordInces.Count - 1][patternIndex];
for (int wordBlockIndex = firstBlockIndex; wordBlockIndex < lastBlockIndex; wordBlockIndex++)
{
wordBlock.Add(blocks[wordBlockIndex]);
}
if (wordBlock.Count > 0)
{
wordBlocks.Add(wordBlock);
}
}
if (wordBlocks.Count > 0)
{
AllWordBlocks.Add(wordBlocks);
}
}
return(AllWordBlocks);
}
public static int GetWordCount(string content)
{
var wordsMatches = WordPattern.Matches(content);
return(wordsMatches.Count);
}
public void IsPatternMatch_EmptyStringPattern_ReturnsTrue()
{
Assert.IsTrue(WordPattern.IsPatternMatch("", ""));
}
public void IsPatternMatch_MatchingPattern_ReturnsTrue()
{
Assert.IsTrue(WordPattern.IsPatternMatch("abba", "dog cat cat dog"));
}
public void IsPatternMatch_NoPattern_ReturnsFalse()
{
Assert.IsFalse(WordPattern.IsPatternMatch("abba", "dog cat dog cat"));
}
