FileClassEnvironmentFactory.cpp File Reference

#include "indri/FileClassEnvironmentFactory.hpp"
#include "indri/DocumentIteratorFactory.hpp"
#include "indri/ParserFactory.hpp"
#include "indri/TokenizerFactory.hpp"
#include <ctype.h>
#include <map>
#include <vector>

Classes
struct	extension_conflations
struct	file_class_environment_spec
Functions
std::string	_canonicalExtension (const std::string &name)
void	copy_strings_to_vector (std::vector< std::string > &vec, const char **array)
void	copy_string_tuples_to_map (std::map< indri::parse::ConflationPattern *, std::string > &m, const char **array)
void	cleanup_conflations_map (std::map< indri::parse::ConflationPattern *, std::string > &conflations)
indri::parse::FileClassEnvironment *	build_file_class_environment (const file_class_environment_spec *spec)
indri::parse::FileClassEnvironment *	build_file_class_environment (const indri::parse::FileClassEnvironmentFactory::Specification *spec)
Variables
const char *	pdf_index_tags [] = { "title", "author", 0 }
const char *	pdf_metadata_tags [] = { "title", "author", 0 }
const char *	html_index_tags [] = { "title", "author", "h1", "h2", "h3", "h4", 0 }
const char *	html_metadata_tags [] = { "title", "author", 0 }
const char *	html_conflations [] = { "h1", NULL, NULL, "heading", "h2", NULL, NULL, "heading", "h3", NULL, NULL, "heading", "h4", NULL, NULL, "heading", 0, 0, 0, 0 }
const char *	trec_include_tags [] = { "text", "hl", "head", "headline", "title", "ttl", "dd", "date", "date_time", "lp", "leadpara", 0 }
const char *	trecalt_include_tags [] = { "text", 0 }
const char *	trecalt_index_tags [] = { "text", 0 }
const char *	trec_metadata_tags [] = { "docno", "title", "author", 0 }
const char *	trec_conflations [] = { "hl", NULL, NULL, "headline", "head", NULL, NULL, "headline", "ttl", NULL, NULL, "title", "dd", NULL, NULL, "date", "date_time", NULL, NULL, "date", 0, 0, 0, 0 }
const char *	trec_index_tags [] = { "author", "hl", "head", "headline", "title", "ttl", "dd", "date_time", "date", 0 }
const char *	html_exclude_tags [] = { "script", "style", 0}
const char *	_html [] = {"html", "htm", 0}
const char *	_txt [] = {"txt", "text", 0}
const char *	_doc [] = {"doc", "docx", 0}
const char *	_ppt [] = {"ppt", "pptx", 0}
extension_conflations	extensions []
file_class_environment_spec	environments []

---

Function Documentation

std::string _canonicalExtension (  const std::string &  name  )  [static]

indri::parse::FileClassEnvironment* build_file_class_environment (  const indri::parse::FileClassEnvironmentFactory::Specification *  spec  )

indri::parse::FileClassEnvironment* build_file_class_environment (  const file_class_environment_spec *  spec  )

void cleanup_conflations_map (  std::map< indri::parse::ConflationPattern *, std::string > &  conflations  )  [static]

void copy_string_tuples_to_map (  std::map< indri::parse::ConflationPattern *, std::string > &  m, const char **  array )  [static]

void copy_strings_to_vector (  std::vector< std::string > &  vec, const char **  array )  [static]

---

Variable Documentation

const char* _doc[] = {"doc", "docx", 0} [static]

const char* _html[] = {"html", "htm", 0} [static]

const char* _ppt[] = {"ppt", "pptx", 0} [static]

const char* _txt[] = {"txt", "text", 0} [static]

file_class_environment_spec environments[] [static]

extension_conflations extensions[] [static]

Initial value: { _html, _txt, _doc, _ppt, {0} }

const char* html_conflations[] = { "h1", NULL, NULL, "heading", "h2", NULL, NULL, "heading", "h3", NULL, NULL, "heading", "h4", NULL, NULL, "heading", 0, 0, 0, 0 } [static]

const char* html_exclude_tags[] = { "script", "style", 0} [static]

const char* html_index_tags[] = { "title", "author", "h1", "h2", "h3", "h4", 0 } [static]

const char* html_metadata_tags[] = { "title", "author", 0 } [static]

const char* pdf_index_tags[] = { "title", "author", 0 } [static]

const char* pdf_metadata_tags[] = { "title", "author", 0 } [static]

const char* trec_conflations[] = { "hl", NULL, NULL, "headline", "head", NULL, NULL, "headline", "ttl", NULL, NULL, "title", "dd", NULL, NULL, "date", "date_time", NULL, NULL, "date", 0, 0, 0, 0 } [static]

const char* trec_include_tags[] = { "text", "hl", "head", "headline", "title", "ttl", "dd", "date", "date_time", "lp", "leadpara", 0 } [static]

const char* trec_index_tags[] = { "author", "hl", "head", "headline", "title", "ttl", "dd", "date_time", "date", 0 } [static]

const char* trec_metadata_tags[] = { "docno", "title", "author", 0 } [static]

const char* trecalt_include_tags[] = { "text", 0 } [static]

const char* trecalt_index_tags[] = { "text", 0 } [static]

---