//public static void Test_FindInFile_01(string file)
//{
// //RegexValuesList regexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("QuestionInfos/QuestionInfo"), compileRegex: true);
// RegexValuesList regexList = QuestionRun.GetQuestionRegexValuesList();
// foreach (string line in zFile.ReadLines(file))
// {
// string line2 = line.Trim();
// if (line2 == "")
// continue;
// Trace.WriteLine($"read : \"{line2}\"");
// FindText findText = regexList.Find(line2);
// if (findText.Found)
// {
// foreach (KeyValuePair<string, ZValue> namedValue in findText.matchValues.GetValues())
// Trace.WriteLine($" found : {namedValue.Key} = \"{namedValue.Value}\"");
// }
// else
// {
// Trace.WriteLine($" unknow text");
// }
// }
//}
public static void Test_Find_02(string text, RegexValuesList regexList)
{
Trace.WriteLine($"text : \"{text}\"");
FindText_v2 findText = regexList.Find(text);
if (findText.Success)
{
//MatchValues matchValues = findText.matchValues;
while (findText.Success)
{
Trace.WriteLine();
Trace.WriteLine($" found : index {findText.Match.Index} length {findText.Match.Length}");
foreach (KeyValuePair <string, ZValue> namedValue in findText.GetValues())
{
Trace.WriteLine($" : {namedValue.Key} = \"{namedValue.Value}\"");
}
//matchValues = matchValues.Next();
findText.Next();
}
}
else
{
Trace.WriteLine($" unknow text");
}
}
public static void Init()
{
InitImage(XmlConfig.CurrentConfig.GetElement("Image"));
_textInfoRegexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("RapideDdl/Parameters/TextInfos/TextInfo"));
_printTextValuesManager = new PrintTextValuesManager(new RegexValuesList(XmlConfig.CurrentConfig.GetElements("TextInfos/TextInfo")), _trimFunc1);
//InitImagesToSkip();
}
public static void Test_FindDate_01(RegexValuesList dateRegexList, string file)
{
file = zPath.Combine(GetDirectoryDate(), file);
string traceFile = zpath.PathSetFileNameWithoutExtension(file, zPath.GetFileNameWithoutExtension(file) + "_out");
Trace.WriteLine("Test_FindDate \"{0}\" (nb date regex {1})", file, dateRegexList.Count);
//Trace.CurrentTrace.DisableBaseLog();
Trace.CurrentTrace.DisableViewer = true;
//Trace.CurrentTrace.AddTraceFile(traceFile, LogOptions.RazLogFile);
Trace.CurrentTrace.AddOnWrite("Test_FindDate_01", WriteToFile.Create(traceFile, FileOption.RazFile).Write);
try
{
TraceRegexValuesList(dateRegexList);
Trace.WriteLine();
foreach (BsonDocument document in zMongo.BsonRead <BsonDocument>(file))
{
FindDate(dateRegexList, document["text"].AsString);
}
}
finally
{
//Trace.CurrentTrace.EnableBaseLog();
//Trace.CurrentTrace.RemoveTraceFile(traceFile);
Trace.CurrentTrace.RemoveOnWrite("Test_FindDate_01");
Trace.CurrentTrace.DisableViewer = false;
}
}
public static RegexValuesList CreateQuestionRegexValuesList()
{
RegexValuesList regexValuesList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("QuestionInfos_v2/QuestionInfo"), compileRegex: true);
//regexValuesList.Add(XmlConfig.CurrentConfig.GetElements("ResponseInfos/ResponseInfo"), compileRegex: true);
return(regexValuesList);
}
public static void Test_RegexValues(IEnumerable <XElement> elements, string input)
{
Trace.WriteLine("\"{0}\" : ", input);
RegexValuesList rvs = new RegexValuesList(elements);
//FindText_old findText = rvs.Find_old(input);
FindText findText = rvs.Find(input);
if (findText.Found)
{
//NamedValues<ZValue> values = findText.regexValues.GetValues_old();
NamedValues <ZValue> values = findText.matchValues.GetValues();
//Trace.WriteLine(" found \"{0}\" {1} values", findText.regexValues.Name, values.Count);
Trace.WriteLine(" found \"{0}\" {1} values", findText.matchValues.Name, values.Count);
foreach (KeyValuePair <string, ZValue> value in values)
{
if (value.Value is ZStringArray)
{
Trace.WriteLine(" value {0} = [{1}] {2}", value.Key, ((string[])value.Value).Length, ((string[])value.Value).zToStringValues(s => "\"" + s.zToStringOrNull() + "\""));
}
else
{
Trace.WriteLine(" value {0} = \"{1}\"", value.Key, value.Value.zToStringOrNull());
}
}
}
else
{
Trace.WriteLine(" not found");
}
Trace.WriteLine();
}
public static void Test2(RegexValuesList regexValuesList, string text, bool contiguous = false)
{
Trace.WriteLine("Input : {0}", text);
//FindText findText = regexValuesList.Find(text);
FindText_v2 findText = regexValuesList.Find(text);
if (!findText.Success)
{
Trace.WriteLine("Result : not found");
return;
}
while (findText.Success)
{
Trace.Write($"found \"{findText.Text}\" position {findText.Match.Index + 1} length {findText.Match.Length}");
NamedValues <ZValue> values = findText.GetValues();
if (values.Count > 0)
{
Trace.Write(" values");
foreach (KeyValuePair <string, ZValue> value in findText.GetValues())
{
Trace.Write($" \"{value.Key}\" = \"{value.Value}\"");
}
}
if (findText.Attributes.Count > 0)
{
Trace.Write(" attributes");
foreach (KeyValuePair <string, string> attribute in findText.Attributes)
{
Trace.Write($" \"{attribute.Key}\" = \"{attribute.Value}\"");
}
}
Trace.WriteLine();
findText.FindNext(contiguous);
}
}
public static void Test(RegexValuesList regexValuesList, string text)
{
Trace.WriteLine("Input : {0}", text);
FindText findText = regexValuesList.Find(text);
if (!findText.Found)
Trace.WriteLine("Result : not found");
else
{
Trace.WriteLine("Result : found");
Trace.WriteLine(" Text : \"{0}\"", findText.Text);
Trace.WriteLine(" Key : \"{0}\"", findText.matchValues.Key);
Trace.WriteLine(" Name : \"{0}\"", findText.matchValues.Name);
Trace.WriteLine(" Pattern : \"{0}\"", findText.matchValues.Pattern);
Trace.WriteLine(" Attributes : {0}", findText.matchValues.Attributes.Count);
foreach (KeyValuePair<string, string> attribute in findText.matchValues.Attributes)
Trace.WriteLine(" Attribute : \"{0}\" = \"{1}\"", attribute.Key, attribute.Value);
NamedValues<ZValue> values = findText.matchValues.GetValues();
Trace.WriteLine(" GetValues() : {0}", findText.matchValues.Attributes.Count);
foreach (KeyValuePair<string, ZValue> value in values)
Trace.WriteLine(" Value : \"{0}\" = \"{1}\"", value.Key, value.Value);
values = findText.matchValues.GetAllValues();
Trace.WriteLine(" GetAllValues() : {0}", findText.matchValues.Attributes.Count);
foreach (KeyValuePair<string, ZValue> value in values)
Trace.WriteLine(" AllValue : \"{0}\" = {1}", value.Key, value.Value);
}
}
public static void FindDate(RegexValuesList dateRegexList, string title)
{
bool found = false;
foreach (RegexValues rv in dateRegexList.Values)
{
//rv.Match_old(title);
MatchValues matchValues = rv.Match(title);
//if (rv.Success_old)
if (matchValues.Success)
{
//NamedValues<ZValue> values = rv.GetValues_old();
NamedValues <ZValue> values = matchValues.GetValues();
Date date;
DateType dateType;
//Trace.Write(" date ");
//found = zdate.TryCreateDate(values, out date);
//if (found)
// Trace.Write("{0:dd-MM-yyyy}", date);
//else
// Trace.Write("not found ");
//Trace.Write(" ");
//values.zTrace();
//Trace.WriteLine(" found date : {0}", rv.MatchValue_old);
Trace.WriteLine(" found date : {0}", matchValues.Match.Value);
//Trace.WriteLine(" remain text : {0}", rv.MatchReplace_old("_"));
Trace.WriteLine(" remain text : {0}", matchValues.Replace("_"));
Trace.Write(" date : ");
found = zdate.TryCreateDate(values, out date, out dateType);
if (found)
{
Trace.Write("{0:dd-MM-yyyy} type {1}", date, dateType);
}
else
{
Trace.Write("not found ");
}
Trace.WriteLine();
Trace.Write(" values : ");
values.zTrace();
Trace.WriteLine();
if (found)
{
break;
}
}
}
if (!found)
{
Trace.WriteLine(" date not found ");
}
}
public static RegexValuesList CreateAnkiRegexValuesList()
{
string configFile = @"c:\pib\drive\google\dev\.net\Apps\tools\anki\anki.config.xml";
XmlConfig config = new XmlConfig(configFile);
RegexValuesList regexValuesList = new RegexValuesList(config.GetElements("QuestionInfos/QuestionInfo"), compileRegex: true);
//regexValuesList.Add(config.GetElements("ResponseInfos/ResponseInfo"), compileRegex: true);
return(regexValuesList);
}
private static void TraceRegexValuesList(RegexValuesList regexList)
{
int i = 1;
foreach (RegexValues regex in regexList.Values)
{
Trace.WriteLine("pattern {0} \"{1}\"", i++, regex.Pattern);
}
}
public static NamedValues <ZValue> Test_Find_01(string text, RegexValuesList regexList)
{
Trace.WriteLine($"text : \"{text}\"");
//RegexValuesList regexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("QuestionInfos/QuestionInfo"), compileRegex: true);
//FindText findText = QuestionRun.GetQuestionRegexValuesList().Find(text);
FindText_v2 findText = regexList.Find(text);
Trace.WriteLine($"found : {findText.Success}");
return(findText.GetAllValues());
}
public static PrintFileManager_v2 CreatePrintFileManager_v2(UncompressQueueManager uncompressManager = null, bool simulate = false, bool moveFiles = false, bool moveInfoFiles = false)
{
DownloadAutomateManagerCreator downloadAutomateManagerCreator = GetDownloadAutomateManagerCreator();
if (uncompressManager == null)
uncompressManager = downloadAutomateManagerCreator.CreateUncompressManager();
RegexValuesList bonusDirectories = new RegexValuesList(XmlConfig.CurrentConfig.GetConfig("PrintList2Config").GetElements("BonusDirectories/Directory"), compileRegex: true);
PrintFileManager_v2 printFileManager = new PrintFileManager_v2();
printFileManager.Simulate = simulate;
printFileManager.MoveFiles = moveFiles;
printFileManager.MoveInfoFiles = moveInfoFiles;
printFileManager.UncompressManager = uncompressManager;
printFileManager.BonusDirectories = bonusDirectories;
return printFileManager;
}
//public static void Test_SelectPostManager_01(string query, int limit = 0)
//{
// SelectPostManager selectPostManager = new SelectPostManager(XmlConfig.CurrentConfig.GetElement("SelectPosts"));
// MongoCursor<BsonDocument> cursor = pb.Data.Mongo.MongoCommand.Find("dl", "RapideDdl_Detail2", query, fields: "{ 'download.title': 1 }", limit: limit, sort: "{ 'download.creationDate': -1 }");
// foreach (BsonDocument document in cursor)
// {
// string title = document["download"]["title"].AsString;
// //Trace.Write("FindDate : {0,-100}", "\"" + title + "\"");
// Trace.WriteLine("FindDate : \"{0}\"", title);
// selectPostManager.FindDate(title);
// }
//}
public static void Test_FindDate_01(string query, int limit = 0)
{
//SelectPostManager selectPostManager = new SelectPostManager(XmlConfig.CurrentConfig.GetElement("SelectPosts"));
// zElements("Dates/Date")
RegexValuesList dateRegexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElement("SelectPosts").zXPathElements("Dates/Date"), compileRegex: true);
MongoCursor<BsonDocument> cursor = pb.Data.Mongo.MongoCommand.Find("dl", "RapideDdl_Detail2", query, fields: "{ 'download.title': 1 }", limit: limit, sort: "{ 'download.creationDate': -1 }");
foreach (BsonDocument document in cursor)
{
string title = document["download"]["title"].AsString;
//Trace.Write("FindDate : {0,-100}", "\"" + title + "\"");
Trace.WriteLine("FindDate : \"{0}\"", title);
FindDate(dateRegexList, title);
Trace.WriteLine();
}
}
public static void FindDate(RegexValuesList dateRegexList, string title)
{
bool found = false;
foreach (RegexValues rv in dateRegexList.Values)
{
//rv.Match_old(title);
MatchValues matchValues = rv.Match(title);
//if (rv.Success_old)
if (matchValues.Success)
{
//NamedValues<ZValue> values = rv.GetValues_old();
NamedValues<ZValue> values = matchValues.GetValues();
Date date;
DateType dateType;
//Trace.Write(" date ");
//found = zdate.TryCreateDate(values, out date);
//if (found)
// Trace.Write("{0:dd-MM-yyyy}", date);
//else
// Trace.Write("not found ");
//Trace.Write(" ");
//values.zTrace();
//Trace.WriteLine(" found date : {0}", rv.MatchValue_old);
Trace.WriteLine(" found date : {0}", matchValues.Match.Value);
//Trace.WriteLine(" remain text : {0}", rv.MatchReplace_old("_"));
Trace.WriteLine(" remain text : {0}", matchValues.Replace("_"));
Trace.Write(" date : ");
found = zdate.TryCreateDate(values, out date, out dateType);
if (found)
Trace.Write("{0:dd-MM-yyyy} type {1}", date, dateType);
else
Trace.Write("not found ");
Trace.WriteLine();
Trace.Write(" values : ");
values.zTrace();
Trace.WriteLine();
if (found)
break;
}
}
if (!found)
Trace.WriteLine(" date not found ");
}
//public static void Test_SelectPostManager_01(string query, int limit = 0)
//{
// SelectPostManager selectPostManager = new SelectPostManager(XmlConfig.CurrentConfig.GetElement("SelectPosts"));
// MongoCursor<BsonDocument> cursor = pb.Data.Mongo.MongoCommand.Find("dl", "RapideDdl_Detail2", query, fields: "{ 'download.title': 1 }", limit: limit, sort: "{ 'download.creationDate': -1 }");
// foreach (BsonDocument document in cursor)
// {
// string title = document["download"]["title"].AsString;
// //Trace.Write("FindDate : {0,-100}", "\"" + title + "\"");
// Trace.WriteLine("FindDate : \"{0}\"", title);
// selectPostManager.FindDate(title);
// }
//}
public static void Test_FindDate_01(string query, int limit = 0)
{
//SelectPostManager selectPostManager = new SelectPostManager(XmlConfig.CurrentConfig.GetElement("SelectPosts"));
// zElements("Dates/Date")
RegexValuesList dateRegexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElement("SelectPosts").zXPathElements("Dates/Date"), compileRegex: true);
MongoCursor <BsonDocument> cursor = pb.Data.Mongo.MongoCommand.Find("dl", "RapideDdl_Detail2", query, fields: "{ 'download.title': 1 }", limit: limit, sort: "{ 'download.creationDate': -1 }");
foreach (BsonDocument document in cursor)
{
string title = document["download"]["title"].AsString;
//Trace.Write("FindDate : {0,-100}", "\"" + title + "\"");
Trace.WriteLine("FindDate : \"{0}\"", title);
FindDate(dateRegexList, title);
Trace.WriteLine();
}
}
public static PrintFileManager_v2 CreatePrintFileManager_v2(UncompressQueueManager uncompressManager = null, bool simulate = false, bool moveFiles = false, bool moveInfoFiles = false)
{
DownloadAutomateManagerCreator downloadAutomateManagerCreator = GetDownloadAutomateManagerCreator();
if (uncompressManager == null)
{
uncompressManager = downloadAutomateManagerCreator.CreateUncompressManager();
}
RegexValuesList bonusDirectories = new RegexValuesList(XmlConfig.CurrentConfig.GetConfig("PrintList2Config").GetElements("BonusDirectories/Directory"), compileRegex: true);
PrintFileManager_v2 printFileManager = new PrintFileManager_v2();
printFileManager.Simulate = simulate;
printFileManager.MoveFiles = moveFiles;
printFileManager.MoveInfoFiles = moveInfoFiles;
printFileManager.UncompressManager = uncompressManager;
printFileManager.BonusDirectories = bonusDirectories;
return(printFileManager);
}
public static void Test_FindInFile_01(string file, RegexValuesList regexList)
{
//RegexValuesList regexList = QuestionRun.GetResponseRegexValuesList();
int lineNumber = 0;
foreach (string line in zFile.ReadLines(file))
{
//string line2 = line.Trim();
//if (line2 == "")
// continue;
lineNumber++;
if (line == "")
{
continue;
}
Trace.WriteLine();
Trace.WriteLine($"read : line {lineNumber} \"{line}\"");
FindText_v2 findText = regexList.Find(line);
if (findText.Success)
{
//MatchValues matchValues = findText.matchValues;
while (findText.Success)
{
Trace.WriteLine();
Trace.WriteLine($" found : index {findText.Match.Index} length {findText.Match.Length}");
foreach (KeyValuePair <string, ZValue> namedValue in findText.GetValues())
{
Trace.WriteLine($" : {namedValue.Key} = \"{namedValue.Value}\"");
}
//matchValues = matchValues.Next();
findText.Next();
}
}
else
{
Trace.WriteLine($" unknow text");
}
}
}
public static void Test(RegexValuesList regexValuesList, string text)
{
Trace.WriteLine("Input : {0}", text);
//FindText findText = regexValuesList.Find(text);
FindText_v2 findText = regexValuesList.Find(text);
if (!findText.Success)
{
Trace.WriteLine("Result : not found");
}
else
{
Trace.WriteLine("Result : found");
Trace.WriteLine(" Text : \"{0}\"", findText.Text);
Trace.WriteLine(" Key : \"{0}\"", findText.Key);
Trace.WriteLine(" Name : \"{0}\"", findText.Name);
Trace.WriteLine(" Pattern : \"{0}\"", findText.Pattern);
Trace.WriteLine(" Attributes : {0}", findText.Attributes.Count);
foreach (KeyValuePair <string, string> attribute in findText.Attributes)
{
Trace.WriteLine(" Attribute : \"{0}\" = \"{1}\"", attribute.Key, attribute.Value);
}
NamedValues <ZValue> values = findText.GetValues();
Trace.WriteLine(" GetValues() : {0}", values.Count);
foreach (KeyValuePair <string, ZValue> value in values)
{
Trace.WriteLine(" Value : \"{0}\" = \"{1}\"", value.Key, value.Value);
}
values = findText.GetAllValues();
Trace.WriteLine(" GetAllValues() : {0}", values.Count);
foreach (KeyValuePair <string, ZValue> value in values)
{
Trace.WriteLine(" AllValue : \"{0}\" = {1}", value.Key, value.Value);
}
}
}
public static void Init(XElement xe)
{
_xmlDirectory = xe.zXPathValue("CacheDirectory");
_xmlImageDirectory = xe.zXPathValue("ImageCacheDirectory", _xmlImageDirectory);
_textInfoRegexList = new RegexValuesList(xe.XPathSelectElements("TextInfo"));
}
//private bool _extractValuesFromText = true;
public PrintTextValuesManager(RegexValuesList textInfoRegexList, Func<string, string> trim = null)
{
_textInfoRegexList = textInfoRegexList;
_trim = trim;
}
private void init(XElement xelement)
{
_directory = xelement.zXPathValue("Directory");
Dictionary <string, RegexValuesModel> regexModels = new Dictionary <string, RegexValuesModel>();
// zElements("FilenamesModel/FilenameModel")
foreach (XElement xe in xelement.zXPathElements("FilenamesModel/FilenameModel"))
{
RegexValuesModel rvm = new RegexValuesModel(xe);
regexModels.Add(rvm.key, rvm);
}
_regexModels = new Dictionary <string, RegexValuesModel>();
// zElements("FilenameModels/FilenameModel")
foreach (XElement xe in xelement.zXPathElements("FilenameModels/FilenameModel"))
{
RegexValuesModel rvm = new RegexValuesModel(xe);
_regexModels.Add(rvm.key, rvm);
}
// zElements("FilenameDates/FilenameDate")
_dateRegexList = new RegexValuesList(xelement.zXPathElements("FilenameDates/FilenameDate"));
//if (_trace)
// Trace.CurrentTrace.WriteLine("_dateRegexList {0}", _dateRegexList.Count);
_printRegexList = new RegexValuesList();
_printRegexList2 = new RegexValuesList();
_prints = new Dictionary <string, Print1>();
// zElements("Prints/Print")
foreach (XElement xe in xelement.zXPathElements("Prints/Print"))
{
Print1 print = null;
switch (xe.zXPathValue("Class"))
{
case "LeMonde":
print = new LeMonde(xe, _directory, _regexModels);
break;
case "LeParisien":
print = new LeParisien(xe, _directory, _regexModels);
break;
case "LeVifExpress":
print = new LeVifExpress(xe, _directory, _regexModels);
break;
default:
print = new Print1(xe, _directory, _regexModels);
break;
}
string name = print.Name;
_prints.Add(name, print);
int n = 1;
// zElements("Filenames/Filename")
foreach (XElement xe2 in xe.zXPathElements("Filenames/Filename"))
{
string key = name + n++.ToString();
//string name = xe2.zExplicitAttributeValue("name");
//string model = xe2.zExplicitAttributeValue("model");
string model = xe2.zAttribValue("model");
if (model != null)
{
if (!regexModels.ContainsKey(model))
{
throw new PBException("unknow filename model \"{0}\"", model);
}
RegexValuesModel rvm = regexModels[model];
//Dictionary<string, string> attribs = (from xa in xe2.Attributes() where xa.Name.ToString().StartsWith("v_") select xa).zAttribs();
Dictionary <string, string> attribs = new Dictionary <string, string>();
attribs.zAdd(from xa in xe2.Attributes() where xa.Name.ToString().StartsWith("v_") select xa);
string pattern = rvm.pattern.zSetTextVariables(attribs, true);
//Trace.CurrentTrace.WriteLine("\"{0}\" - \"{1}\"", rvm.pattern, pattern);
//string values = xe2.zAttribValue("values");
//if (values != null)
//{
// if (rvm.values != null && rvm.values != "")
// values = rvm.values + ", " + values;
//}
//else
// values = rvm.values;
string values = rvm.values;
if (values != null)
{
values = values.zSetTextVariables(attribs, true);
}
_printRegexList.Add(key, new RegexValues(key, name, pattern, rvm.options, values));
}
else
{
string regex = xe2.zExplicitAttribValue("regex");
string values = xe2.zExplicitAttribValue("values");
string options = xe2.zAttribValue("options");
_printRegexList.Add(key, new RegexValues(key, name, regex, options, values));
}
}
n = 1;
// zElements("Filenames2/Filename")
foreach (XElement xe2 in xe.zXPathElements("Filenames2/Filename"))
{
string key = name + n++.ToString();
string regex = xe2.zExplicitAttribValue("regex");
string values = xe2.zAttribValue("values");
string options = xe2.zAttribValue("options");
_printRegexList2.Add(key, new RegexValues(key, name, regex, options, values));
}
}
}
private void init(XElement xelement)
{
_directory = xelement.zXPathValue("Directory");
Dictionary<string, RegexValuesModel> regexModels = new Dictionary<string, RegexValuesModel>();
// zElements("FilenamesModel/FilenameModel")
foreach (XElement xe in xelement.zXPathElements("FilenamesModel/FilenameModel"))
{
RegexValuesModel rvm = new RegexValuesModel(xe);
regexModels.Add(rvm.key, rvm);
}
_regexModels = new Dictionary<string, RegexValuesModel>();
// zElements("FilenameModels/FilenameModel")
foreach (XElement xe in xelement.zXPathElements("FilenameModels/FilenameModel"))
{
RegexValuesModel rvm = new RegexValuesModel(xe);
_regexModels.Add(rvm.key, rvm);
}
// zElements("FilenameDates/FilenameDate")
_dateRegexList = new RegexValuesList(xelement.zXPathElements("FilenameDates/FilenameDate"));
//if (_trace)
// Trace.CurrentTrace.WriteLine("_dateRegexList {0}", _dateRegexList.Count);
_printRegexList = new RegexValuesList();
_printRegexList2 = new RegexValuesList();
_prints = new Dictionary<string, Print1>();
// zElements("Prints/Print")
foreach (XElement xe in xelement.zXPathElements("Prints/Print"))
{
Print1 print = null;
switch (xe.zXPathValue("Class"))
{
case "LeMonde":
print = new LeMonde(xe, _directory, _regexModels);
break;
case "LeParisien":
print = new LeParisien(xe, _directory, _regexModels);
break;
case "LeVifExpress":
print = new LeVifExpress(xe, _directory, _regexModels);
break;
default:
print = new Print1(xe, _directory, _regexModels);
break;
}
string name = print.Name;
_prints.Add(name, print);
int n = 1;
// zElements("Filenames/Filename")
foreach (XElement xe2 in xe.zXPathElements("Filenames/Filename"))
{
string key = name + n++.ToString();
//string name = xe2.zExplicitAttributeValue("name");
//string model = xe2.zExplicitAttributeValue("model");
string model = xe2.zAttribValue("model");
if (model != null)
{
if (!regexModels.ContainsKey(model))
throw new PBException("unknow filename model \"{0}\"", model);
RegexValuesModel rvm = regexModels[model];
//Dictionary<string, string> attribs = (from xa in xe2.Attributes() where xa.Name.ToString().StartsWith("v_") select xa).zAttribs();
Dictionary<string, string> attribs = new Dictionary<string, string>();
attribs.zAdd(from xa in xe2.Attributes() where xa.Name.ToString().StartsWith("v_") select xa);
string pattern = rvm.pattern.zSetTextVariables(attribs, true);
//Trace.CurrentTrace.WriteLine("\"{0}\" - \"{1}\"", rvm.pattern, pattern);
//string values = xe2.zAttribValue("values");
//if (values != null)
//{
// if (rvm.values != null && rvm.values != "")
// values = rvm.values + ", " + values;
//}
//else
// values = rvm.values;
string values = rvm.values;
if (values != null)
values = values.zSetTextVariables(attribs, true);
_printRegexList.Add(key, new RegexValues(key, name, pattern, rvm.options, values));
}
else
{
string regex = xe2.zExplicitAttribValue("regex");
string values = xe2.zExplicitAttribValue("values");
string options = xe2.zAttribValue("options");
_printRegexList.Add(key, new RegexValues(key, name, regex, options, values));
}
}
n = 1;
// zElements("Filenames2/Filename")
foreach (XElement xe2 in xe.zXPathElements("Filenames2/Filename"))
{
string key = name + n++.ToString();
string regex = xe2.zExplicitAttribValue("regex");
string values = xe2.zAttribValue("values");
string options = xe2.zAttribValue("options");
_printRegexList2.Add(key, new RegexValues(key, name, regex, options, values));
}
}
}
public static IEnumerable <Response> Read(string file, RegexValuesList regexList, string baseDirectory = null)
{
return(new ResponseReader {
_file = file, _sourceFile = GetSubPath(file, baseDirectory), _regexList = regexList
}._Read());
}
public FindDayManager(IEnumerable<XElement> days, bool compileRegex = false)
{
_dayRegexList = new RegexValuesList(days, compileRegex: compileRegex);
}
private void init(XElement xelement)
{
_directory = xelement.zXPathValue("Directory");
_regexModels = new Dictionary<string, RegexValuesModel>();
// zElements("FilenameInfos/FilenameModel")
foreach (XElement xe in xelement.zXPathElements("FilenameInfos/FilenameModel"))
{
RegexValuesModel rvm = new RegexValuesModel(xe);
_regexModels.Add(rvm.key, rvm);
}
// zElements("FilenameInfos/FilenameDate")
_dateRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameDate"));
//XElement xe3 = xelement.zXPathElement("FilenameInfos/FilenameDay");
//if (xe3 != null)
// _dayRegex = new RegexValues(xe3);
// zElements("FilenameInfos/FilenameDay")
_dayRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameDay"));
// zElements("FilenameInfos/FilenameNumber")
_numberRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameNumber"));
//xe3 = xelement.zXPathElement("FilenameInfos/FilenameSpecial");
//if (xe3 != null)
// _specialRegex = new RegexValues(xe3);
// zElements("FilenameInfos/FilenameSpecial")
_specialRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameSpecial"));
_printRegexList = new RegexValuesList();
_prints = new Dictionary<string, Print>();
// zElements("Prints/Print")
foreach (XElement xe in xelement.zXPathElements("Prints/Print"))
{
Print print = null;
switch (xe.zXPathValue("Class"))
{
case "LeMonde":
print = new PrintLeMonde(xe, _directory, _regexModels);
break;
case "LeParisien":
print = new PrintLeParisien(xe, _directory, _regexModels);
break;
case "LExpress":
print = new PrintLExpress(xe, _directory, _regexModels);
break;
case "LeVifExpress":
print = new PrintLeVifExpress(xe, _directory, _regexModels);
break;
default:
print = new Print(xe, _directory, _regexModels);
break;
}
string name = print.Name;
_prints.Add(name, print);
int n = 1;
// zElements("Filename")
foreach (XElement xe2 in xe.zXPathElements("Filename"))
{
string key = name + n++.ToString();
string regex = xe2.zExplicitAttribValue("regex");
string values = xe2.zAttribValue("values");
string options = xe2.zAttribValue("options");
_printRegexList.Add(key, new RegexValues(key, name, regex, options, values));
}
}
}
public FindDateManager(IEnumerable<XElement> dates, IEnumerable<XElement> digitDates = null, bool compileRegex = false)
{
_dateRegexList = new RegexValuesList(dates, compileRegex: compileRegex);
if (digitDates != null)
_digitDateRegexList = new RegexValuesList(digitDates, compileRegex: compileRegex);
}
public static void Init()
{
InitImage(XmlConfig.CurrentConfig.GetElement("Image"));
_textInfoRegexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("RapideDdl/Parameters/TextInfos/TextInfo"));
_printTextValuesManager = new PrintTextValuesManager(new RegexValuesList(XmlConfig.CurrentConfig.GetElements("TextInfos/TextInfo")), _trimFunc1);
//InitImagesToSkip();
}
//private bool _extractValuesFromText = true;
public PrintTextValuesManager(RegexValuesList textInfoRegexList, Func <string, string> trim = null)
{
_textInfoRegexList = textInfoRegexList;
_trim = trim;
}
private void init(XElement xelement)
{
_directory = xelement.zXPathValue("Directory");
_regexModels = new Dictionary <string, RegexValuesModel>();
// zElements("FilenameInfos/FilenameModel")
foreach (XElement xe in xelement.zXPathElements("FilenameInfos/FilenameModel"))
{
RegexValuesModel rvm = new RegexValuesModel(xe);
_regexModels.Add(rvm.key, rvm);
}
// zElements("FilenameInfos/FilenameDate")
_dateRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameDate"));
//XElement xe3 = xelement.zXPathElement("FilenameInfos/FilenameDay");
//if (xe3 != null)
// _dayRegex = new RegexValues(xe3);
// zElements("FilenameInfos/FilenameDay")
_dayRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameDay"));
// zElements("FilenameInfos/FilenameNumber")
_numberRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameNumber"));
//xe3 = xelement.zXPathElement("FilenameInfos/FilenameSpecial");
//if (xe3 != null)
// _specialRegex = new RegexValues(xe3);
// zElements("FilenameInfos/FilenameSpecial")
_specialRegexList = new RegexValuesList(xelement.zXPathElements("FilenameInfos/FilenameSpecial"));
_printRegexList = new RegexValuesList();
_prints = new Dictionary <string, Print>();
// zElements("Prints/Print")
foreach (XElement xe in xelement.zXPathElements("Prints/Print"))
{
Print print = null;
switch (xe.zXPathValue("Class"))
{
case "LeMonde":
print = new PrintLeMonde(xe, _directory, _regexModels);
break;
case "LeParisien":
print = new PrintLeParisien(xe, _directory, _regexModels);
break;
case "LExpress":
print = new PrintLExpress(xe, _directory, _regexModels);
break;
case "LeVifExpress":
print = new PrintLeVifExpress(xe, _directory, _regexModels);
break;
default:
print = new Print(xe, _directory, _regexModels);
break;
}
string name = print.Name;
_prints.Add(name, print);
int n = 1;
// zElements("Filename")
foreach (XElement xe2 in xe.zXPathElements("Filename"))
{
string key = name + n++.ToString();
string regex = xe2.zExplicitAttribValue("regex");
string values = xe2.zAttribValue("values");
string options = xe2.zAttribValue("options");
_printRegexList.Add(key, new RegexValues(key, name, regex, options, values));
}
}
}
private static void TraceRegexValuesList(RegexValuesList regexList)
{
int i = 1;
foreach (RegexValues regex in regexList.Values)
{
Trace.WriteLine("pattern {0} \"{1}\"", i++, regex.Pattern);
}
}
public FindDateManager_v1(IEnumerable<XElement> xelements, bool compileRegex = false)
{
_dateRegexList = new RegexValuesList(xelements, compileRegex: compileRegex);
}
public static void Init()
{
//TelechargementPlus_LoadHeader.ClassInit(XmlConfig.CurrentConfig.GetElement("TelechargementPlus/Header"));
_textInfoRegexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("TelechargementPlus/TextInfo"));
InitImagesToSkip();
}
public static void Test_FindDate_01(RegexValuesList dateRegexList, string file)
{
file = zPath.Combine(GetDirectoryDate(), file);
string traceFile = zpath.PathSetFileNameWithoutExtension(file, zPath.GetFileNameWithoutExtension(file) + "_out");
Trace.WriteLine("Test_FindDate \"{0}\" (nb date regex {1})", file, dateRegexList.Count);
//Trace.CurrentTrace.DisableBaseLog();
Trace.CurrentTrace.DisableViewer = true;
//Trace.CurrentTrace.AddTraceFile(traceFile, LogOptions.RazLogFile);
Trace.CurrentTrace.AddOnWrite("Test_FindDate_01", WriteToFile.Create(traceFile, FileOption.RazFile).Write);
try
{
TraceRegexValuesList(dateRegexList);
Trace.WriteLine();
foreach (BsonDocument document in zmongo.BsonRead<BsonDocument>(file))
FindDate(dateRegexList, document["text"].AsString);
}
finally
{
//Trace.CurrentTrace.EnableBaseLog();
//Trace.CurrentTrace.RemoveTraceFile(traceFile);
Trace.CurrentTrace.RemoveOnWrite("Test_FindDate_01");
Trace.CurrentTrace.DisableViewer = false;
}
}
public static void Init()
{
//TelechargementPlus_LoadHeader.ClassInit(XmlConfig.CurrentConfig.GetElement("TelechargementPlus/Header"));
_textInfoRegexList = new RegexValuesList(XmlConfig.CurrentConfig.GetElements("TelechargementPlus/TextInfo"));
InitImagesToSkip();
}
public FindNumberManager(IEnumerable<XElement> xelements, bool compileRegex = false)
{
_numberRegexList = new RegexValuesList(xelements, compileRegex: compileRegex);
}
public static void Test_RegexValues(IEnumerable<XElement> elements, string input)
{
Trace.WriteLine("\"{0}\" : ", input);
RegexValuesList rvs = new RegexValuesList(elements);
//FindText_old findText = rvs.Find_old(input);
FindText findText = rvs.Find(input);
if (findText.Found)
{
//NamedValues<ZValue> values = findText.regexValues.GetValues_old();
NamedValues<ZValue> values = findText.matchValues.GetValues();
//Trace.WriteLine(" found \"{0}\" {1} values", findText.regexValues.Name, values.Count);
Trace.WriteLine(" found \"{0}\" {1} values", findText.matchValues.Name, values.Count);
foreach (KeyValuePair<string, ZValue> value in values)
{
if (value.Value is ZStringArray)
Trace.WriteLine(" value {0} = [{1}] {2}", value.Key, ((string[])value.Value).Length, ((string[])value.Value).zToStringValues(s => "\"" + s.zToStringOrNull() + "\""));
else
Trace.WriteLine(" value {0} = \"{1}\"", value.Key, value.Value.zToStringOrNull());
}
}
else
Trace.WriteLine(" not found");
Trace.WriteLine();
}
public FindDayManager(IEnumerable <XElement> days, bool compileRegex = false)
{
_dayRegexList = new RegexValuesList(days, compileRegex: compileRegex);
}
// RegexValuesList regexList
//public TextDataReader(IEnumerable<string> files)
//{
// //_regexList = regexList;
// _files = files;
//}
public void SetRegexList(RegexValuesList regexList)
{
_regexList = regexList;
}
