Initial commit

.gitignore

@@ -0,0 +1,8 @@
+cmake-build*
+.idea/
+**/build/
+**/dist/
+**/nbproject/
+/pgdumps/
+*~
+/tpctl.config

CMakeLists.txt

@@ -0,0 +1,86 @@
+cmake_minimum_required(VERSION 3.5)
+project(tpctools)
+
+set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake/Modules/")
+set(CMAKE_CXX_STANDARD 11)
+set(CMAKE_CXX_STANDARD_REQUIRED ON) #...is required...
+set(CMAKE_CXX_FLAGS "-DBOOST_NO_CXX11_SCOPED_ENUMS -w")
+
+set(CMAKE_STATIC_LIBRARY_PREFIX "") # avoid the prefix "lib"
+set(CMAKE_SHARED_LIBRARY_PREFIX "")
+
+SET(CMAKE_INSTALL_PREFIX /usr/local)
+SET(CMAKE_SKIP_BUILD_RPATH  FALSE)
+SET(CMAKE_BUILD_WITH_INSTALL_RPATH TRUE)
+set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/lib")
+set(CMAKE_INSTALL_RPATH_USE_LINK_PATH TRUE)
+
+find_package(APR REQUIRED)
+include_directories(${APR_INCLUDE_DIR})
+
+find_package(Lucene++ REQUIRED)
+include_directories(${Lucene++_INCLUDE_DIRS})
+
+find_package(Textpresso REQUIRED)
+include_directories(${Textpresso_INCLUDE_DIR})
+
+include_directories("/home/valerio/workspace/caltech/libtpc")
+
+#### Main projects ####
+
+add_executable(printbibfromnxmlorcasfile printbibfromnxmlorcasfile/main.cpp printbibfromnxmlorcasfile/cmdline.h)
+target_link_libraries(printbibfromnxmlorcasfile uima xerces-c boost_system boost_filesystem boost_iostreams
+        boost_regex)
+
+#### Subprojects ####
+
+add_executable(ppm2jpg ppm2jpg/main.cpp)
+target_link_libraries(ppm2jpg boost_system boost_filesystem boost_program_options pthread ${CImg_SYSTEM_LIBS})
+
+add_executable(cas2index cas2index/cas2index.cpp)
+target_link_libraries(cas2index ${Textpresso_LIBRARY} boost_filesystem boost_system boost_program_options lucene++)
+
+add_executable(updatecorpuscounter cas2index/update_corpus_counter.cpp)
+target_link_libraries(updatecorpuscounter ${Textpresso_LIBRARY} boost_filesystem boost_system boost_program_options
+        lucene++)
+
+add_executable(indexmerger cas2index/index_merger.cpp lucene/CaseSensitiveAnalyzer.cpp)
+target_link_libraries(indexmerger uima boost_filesystem boost_system lucene++)
+
+
+add_executable(articles2cas articles2cas/articles2cas.cpp articles2cas/Utils.h articles2cas/Utils.cpp)
+target_link_libraries(articles2cas ${Textpresso_LIBRARY} boost_filesystem boost_system boost_program_options boost_iostreams)
+
+add_executable(getbib getbib/getbib.cpp getbib/getbibUtils.h getbib/getbibUtils.cpp
+        TextpressoCentralGlobalDefinitions.h TextpressoCentralGlobals.h)
+target_link_libraries(getbib lucene++ xerces-c icuuc boost_system uima boost_filesystem boost_iostreams)
+
+add_executable(getbib4nxml getbib/getbib4nxml.cpp getbib/getbib4nxmlUtils.h getbib/getbib4nxmlUtils.cpp
+        TextpressoCentralGlobalDefinitions.h TextpressoCentralGlobals.h)
+target_link_libraries(getbib4nxml lucene++ xerces-c icuuc boost_system uima boost_filesystem boost_iostreams)
+
+add_executable(saveidstodb cas2index/saveidstodb.cpp)
+target_link_libraries(saveidstodb lucene++ boost_filesystem boost_system boost_program_options ${Textpresso_LIBRARY}
+        db_cxx db_stl)
+
+#### INSTALL ####
+
+install(TARGETS getbib getbib4nxml
+        RUNTIME DESTINATION bin
+        LIBRARY DESTINATION lib)
+
+install(TARGETS cas2index RUNTIME DESTINATION ${CMAKE_INSTALL_PREFIX}/bin)
+install(TARGETS saveidstodb RUNTIME DESTINATION ${CMAKE_INSTALL_PREFIX}/bin)
+install(TARGETS updatecorpuscounter RUNTIME DESTINATION ${CMAKE_INSTALL_PREFIX}/bin)
+install(PROGRAMS cas2index/create_index_multi_literatures.sh cas2index/create_single_index.sh
+        ppm2jpg/ppm2jpgCas.sh run_tpc_pipeline_incremental.sh
+        getpdfs/getpdfs.py getbibinfoforpdffromserver/download_pdfinfo.pl
+        getbibinfoforpdffromserver/extract_pdfbibinfo.pl
+        DESTINATION ${CMAKE_INSTALL_PREFIX}/bin)
+
+install(TARGETS articles2cas RUNTIME DESTINATION ${CMAKE_INSTALL_PREFIX}/bin)
+install(PROGRAMS articles2cas/convertallarticles2cas.sh
+        DESTINATION ${CMAKE_INSTALL_PREFIX}/bin)
+
+
+

TextpressoCentralGlobalDefinitions.h

@@ -0,0 +1,52 @@
+// Global file containing all global definitions.
+
+#ifndef TEXTPRESSOCENTRALGLOBALDEFINITIONS_H
+#define TEXTPRESSOCENTRALGLOBALDEFINITIONS_H
+
+// Are these definitions really global? Otherwise move them back to their local project.
+#define PDF2TPCASDESCRIPTOR "/usr/local/uima_descriptors/TpTokenizer.xml"
+#define XML2TPCASDESCRIPTOR "/usr/local/uima_descriptors/TxTokenizer.xml"
+#define TPCAS2LINDEXDESCRIPTOR "/usr/local/uima_descriptors/Tpcas2Lindex.xml"
+
+#define AVAILABLELITERATUREFILE "/usr/local/textpresso/luceneindex/subindex.config"
+#define USERUPLOADROOTDIR "/usr/local/textpresso/useruploads"
+
+#define PGONTOLOGYBROWSER "dbname=www-data"
+#define PGONTOLOGYBROWSWERCOLUMNS "ontologybrowsercolumnnames"
+#define PGONTOLOGY "dbname=www-data"
+#define PGONTOLOGYTABLENAME "tpontology"
+#define PGONTOLOGYTMPTABLENAME "tmptpontology"
+#define PGCURATION "dbname=www-data"
+#define PGCURATIONTABLENAME "tpcuration"
+#define PCRELATIONSTABLENAME "pcrelations"
+#define PCRELATIONSTMPTABLENAME "tmppcrelations"
+#define PADCRELATIONSTABLENAME "padcrelations"
+#define PADCRELATIONSTMPTABLENAME "tmppadcrelations"
+#define STOPWORDTABLENAME "stopwords"
+#define PGLITERATURE "dbname=www-data"
+#define PGLITPREFTABLENAME "literaturepreference"
+#define PGPRELOADEDCATEGORIES "dbname=www-data"
+#define PGPRELOADEDCATTABLENAME "preloadedcategories"
+#define PGTIPOFDAY "dbname=www-data"
+#define PGTIPOFDAYTABLENAME "tipoftheday"
+#define PGCURATIONFIELDS "dbname=www-data"
+#define PGCURATIONFIELDSTABLENAME "curationfields"
+#define PGCURATIONFORMS "dbname=www-data"
+#define PGCURATIONFORMSTABLENAME "curationforms"
+#define PGCURATIONDATAFROMVIEWER "dbname=www-data"
+#define PGCURATIONDATAFROMVIEWERTABLENAME "curationdatafromviewer"
+#define PGLISTOFONTOLOGIES "dbname=www-data"
+#define PGLISTOFONTOLOGIESTABLENAME "listofontologies"
+#define PGPREPOPULATION "dbname=www-data"
+#define PGPREPOPULATIONTABLENAME "prepopulation"
+#define AUTHIDENTITIES "dbname=www-data"
+#define AUTHIDENTITIESTABLENAME "auth_identity"
+#define PGLITERATUREPERMISSION "dbname=www-data"
+#define PGLITERATUREPERMISSIONTABLENAME "literaturepermissions"
+#define PGCUSTOMCOLORS "dbname=www-data"
+#define PGCUSTOMCOLORSTABLENAME "customcolor"
+#define PGDIALOGPREFERENCES "dbname=www-data"
+#define PGDIALOGPREFERENCESTABLENAME "dialogpreferencestable"
+#define SENTENCE_SEARCH_MAX_NUM_DISPLAY_WORDS 50
+
+#endif

TextpressoCentralGlobals.h

@@ -0,0 +1,41 @@
+// Global file containing all global definitions.
+
+#ifndef TEXTPRESSOCENTRALGLOBALS_H
+#define TEXTPRESSOCENTRALGLOBALS_H
+
+
+// Are these definitions really global? Otherwise move them back to their local project. 
+
+#include "TextpressoCentralGlobalDefinitions.h"
+
+#include <uima/api.hpp>
+
+// If a composite delimiter exists, then there cannot be another delimiter
+// that is a subset of that composite token delimiter. Decompose it accordingly.
+// This applies to token and sentence delimiter
+UnicodeString G_initT[] = {
+    " ", "\n", "\t", "'", "\"",
+    "/", "—", "(", ")", "[",
+    "]", "{", "}", ":", ". ",
+    "; ", ", ", "! ", "? "
+};
+
+const int G_initT_No = 19;
+UnicodeString G_initS[] = {
+    ".\n", "!\n", "?\n", ". ", "! ", "? ",
+    ".\t", "!\t", "?\t", ".<", "!<", "?<"
+};
+const int G_initS_No = 12;
+UnicodeString G_initP[] = {"<_pdf _image", "<_pdf _sbr", "<_pdf _hbr",
+    "<_pdf _fsc", "<_pdf _fnc", "<_pdf _ydiff", "<_pdf _cr", "<_pdf _page"};
+const int G_initP_No = 8;
+const std::string ServerNames[] = {"http://goldturtle.caltech.edu/cgi-bin/ReceivePost.cgi",
+    "http://go-genkisugi.rhcloud.com/capella", "http://localhost/cgi-bin/ReceivePost.cgi"};
+const int ServerNames_No = 3;
+
+//const std::string G_pdftagstart("<_pdf ");
+//const std::string G_pdftagend("/>");
+const UnicodeString usG_pdftagstart("<_pdf ");
+const UnicodeString usG_pdftagend("/>");
+
+#endif