Description: switching from pcre3 to pcre2 This was tested by running pcreTest (in cpp/src/GsnapParseTest1.cc) successfully. Yet I did not compile the other files. Author: Pierre Gruet Bug-Debian: https://bugs.debian.org/1000043 Forwarded: https://github.com/CampagneLaboratory/goby3/issues/9 Last-Update: 2021-11-26 --- a/cpp/src/GsnapParseTest1.cc +++ b/cpp/src/GsnapParseTest1.cc @@ -20,6 +20,7 @@ #if HAVE_CONFIG_H #include #endif +#include #include #include @@ -29,11 +30,13 @@ #else #include #endif +#include #include "goby/C_Gsnap.h" #include "goby/C_Alignments.h" #include "goby/C_Reads.h" -#include +#define PCRE2_CODE_UNIT_WIDTH 8 +#include using namespace std; @@ -178,21 +181,53 @@ } void pcreTest() { - char *a = "s-reads:3.2..b:5.1,l_reads:3,c:3,d:5,a:0..b:2.8"; + string a = string("s-reads:3.2..b:5.1,l_reads:3,c:3,d:5,a:0..b:2.8"); string temp1; string da; string daClip; string daProb; string temp2; string std; string stdClip; - pcrecpp::StringPiece input(a); // Wrap in a StringPiece - pcrecpp::RE re("(([A-Za-z_\\-]+):(\\d+)\\.(\\d))|(([A-Za-z_\\-]+):(\\d))"); + //Setting up pcre2. + pcre2_code *re; + string sPattern = string("(([A-Za-z_\\-]+):(\\d+)\\.(\\d))|(([A-Za-z_\\-]+):(\\d))"); + PCRE2_SPTR pattern = (PCRE2_SPTR)(sPattern.c_str()); + PCRE2_SPTR subject = (PCRE2_SPTR)(a.c_str()); + int errornumber; + int rc; + PCRE2_SIZE erroroffset; + PCRE2_SIZE *ovector; + PCRE2_SIZE subject_length = (PCRE2_SIZE)(sPattern.length()); + pcre2_match_data *match_data; + re = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, 0, &errornumber, &erroroffset, NULL); + match_data = pcre2_match_data_create_from_pattern(re, NULL); + PCRE2_SIZE start_offset = 0; printf("pcre start\n"); - while (re.FindAndConsume(&input, - &temp1, - &da, &daClip, &daProb, - &temp2, - &std, &stdClip)) { + do { + rc = pcre2_match(re, subject, subject_length, start_offset, 0, match_data, NULL); + + if (rc != PCRE2_ERROR_NOMATCH) { + char** sStore = new char*[7]; + //Match succeeded. Converting strings to the needed types. + ovector = pcre2_get_ovector_pointer(match_data); + for (int i = 0 ; i < 7 ; i++) { + sStore[i] = new char[ovector[2 * i + 3] - ovector[2 * i + 2]]; + sprintf(sStore[i1], "%.*s", (int)(ovector[2 * i + 3] - ovector[2 * i + 2]), (char*)(subject + ovector[2 * i + 2])); + } + temp1 = string(sStore[0]); + da = string(sStore[1]); + daClip = string(sStore[2]); + daProb = string(sStore[3]); + temp2 = string(sStore[4]); + std = string(sStore[5]); + stdClip = string(sStore[6]); + for (int i = 0 ; i < 7 ; i++) { + delete[] sStore[i]; + } + delete[] sStore; + //Preparing next match. + start_offset = ovector[1]; + } if (da.length() > 0) { cout << "da=" << da << " "; @@ -203,7 +238,7 @@ cout << "stdClip=" << stdClip << endl; } - } + } while (rc != PCRE2_ERROR_NOMATCH); printf("done\n"); } --- a/cpp/src/goby/C_Alignments.cc +++ b/cpp/src/goby/C_Alignments.cc @@ -18,7 +18,8 @@ #include #include -#include +#define PCRE2_CODE_UNIT_WIDTH 8 +#include #include #include "Reads.h" @@ -872,8 +873,21 @@ } void applyCigar(CSamHelper *samHelper) { - pcrecpp::RE re("([0-9]+)([SMID])"); - pcrecpp::StringPiece input(samHelper->cpp_cigar->c_str()); + //Setting up pcre2. + pcre2_code *re; + string sPattern = string("([0-9]+)([SMID])"); + PCRE2_SPTR pattern = (PCRE2_SPTR)(sPattern.c_str()); + PCRE2_SPTR subject = (PCRE2_SPTR)(samHelper->cpp_cigar->c_str()); + int errornumber; + int rc; + PCRE2_SIZE erroroffset; + PCRE2_SIZE *ovector; + PCRE2_SIZE subject_length = (PCRE2_SIZE)(sPattern.length()); + pcre2_match_data *match_data; + re = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, 0, &errornumber, &erroroffset, NULL); + match_data = pcre2_match_data_create_from_pattern(re, NULL); + PCRE2_SIZE start_offset = 0; + debug (fprintf(stderr, ":: Applying cigar=%s\n", samHelper->cpp_cigar->c_str());) int length; char op; @@ -883,7 +897,25 @@ samHelper->numInsertions = 0; samHelper->numDeletions = 0; samHelper->numMisMatches = 0; - while (re.Consume(&input, &length, &op)) { + + do { + rc = pcre2_match(re, subject, subject_length, start_offset, 0, match_data, NULL); + + if (rc != PCRE2_ERROR_NOMATCH) { + //Match succeeded. Converting strings to the needed types. + ovector = pcre2_get_ovector_pointer(match_data); + char* sLength = new char[ovector[3] - ovector[2] + 1]; + char* sOp = new char[ovector[5] - ovector[4] + 1]; + sprintf(sLength, "%.*s", (int)(ovector[3] - ovector[2]), (char*)(subject + ovector[2])); + sprintf(sOp, "%.*s", (int)(ovector[5] - ovector[4]), (char*)(subject + ovector[4])); + length = stoi(string(sLength)); + op = string(sOp)[0]; + delete[] sLength; + delete[] sOp; + //Preparing next match. + start_offset = ovector[1]; + } + switch(op) { case 'S': // Soft clipping @@ -933,20 +965,48 @@ break; } startOfCigar = false; - } + } while (rc != PCRE2_ERROR_NOMATCH); } void applyMd(CSamHelper *samHelper) { // My RE is simplified from the SAM spec but performs the same task // the major difference being mine would allow 5ACG where the current // spec would require 5A0C0G0 (which mine will still work with just fine). - pcrecpp::RE re("([0-9]+|[ACGTN]|\\^[ACGTN]+)"); - pcrecpp::StringPiece input(samHelper->cpp_md->c_str()); + + //Setting up pcre2. + pcre2_code *re; + string sPattern = string("([0-9]+|[ACGTN]|\\^[ACGTN]+)"); + PCRE2_SPTR pattern = (PCRE2_SPTR)(sPattern.c_str()); + PCRE2_SPTR subject = (PCRE2_SPTR)(samHelper->cpp_md->c_str()); + int errornumber; + int rc; + PCRE2_SIZE erroroffset; + PCRE2_SIZE *ovector; + PCRE2_SIZE subject_length = (PCRE2_SIZE)(sPattern.length()); + pcre2_match_data *match_data; + re = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, 0, &errornumber, &erroroffset, NULL); + match_data = pcre2_match_data_create_from_pattern(re, NULL); + PCRE2_SIZE start_offset = 0; + debug (fprintf(stderr, ":: Applying md=%s\n", samHelper->cpp_md->c_str());) string mdPart; int position = samHelper->numLeftClipped; int i; - while (re.Consume(&input, &mdPart)) { + + do { + rc = pcre2_match(re, subject, subject_length, start_offset, 0, match_data, NULL); + + if (rc != PCRE2_ERROR_NOMATCH) { + //Match succeeded. + ovector = pcre2_get_ovector_pointer(match_data); + char* sMdPart = new char[ovector[3] - ovector[2] + 1]; + sprintf(sMdPart, "%.*s", (int)(ovector[3] - ovector[2]), (char*)(subject + ovector[2])); + mdPart = string(sMdPart); + delete[] sMdPart; + //Preparing next match. + start_offset = ovector[1]; + } + if (isdigit(mdPart[0])) { int length = atoi(mdPart.c_str()); position += length; @@ -962,7 +1022,7 @@ samHelper->numMisMatches++; } } - } + } while (rc != PCRE2_ERROR_NOMATCH); } void samHelper_constructRefAndQuery(CSamHelper *samHelper) { --- a/cpp/src/goby/C_Gsnap.cc +++ b/cpp/src/goby/C_Gsnap.cc @@ -27,7 +27,8 @@ #include "C_Gsnap_structs.h" #include "C_Alignments.h" #include "hash.h" -#include "pcrecpp.h" +#define PCRE2_CODE_UNIT_WIDTH 8 +#include /* * TODO: Make sure the fields I moved from segment to alignment really @@ -44,10 +45,6 @@ using namespace std; -using pcrecpp::StringPiece; -using pcrecpp::RE; -using pcrecpp::RE_Options; - #undef C_GSNAP_DEBUG #ifdef C_GSNAP_DEBUG #define debug(x) x @@ -359,13 +356,47 @@ string temp2; string std; string stdClip; - pcrecpp::StringPiece input(a); // Wrap in a StringPiece - pcrecpp::RE re("(([A-Za-z_\\-]+):(\\d+)\\.(\\d+))|(([A-Za-z_\\-]+):(\\d+))"); + //Setting up pcre2. + pcre2_code *re; + string sPattern = string("(([A-Za-z_\\-]+):(\\d+)\\.(\\d+))|(([A-Za-z_\\-]+):(\\d+))"); + PCRE2_SPTR pattern = (PCRE2_SPTR)(sPattern.c_str()); + PCRE2_SPTR subject = (PCRE2_SPTR)(a); + int errornumber; + int rc; + PCRE2_SIZE erroroffset; + PCRE2_SIZE *ovector; + PCRE2_SIZE subject_length = (PCRE2_SIZE)(sPattern.length()); + pcre2_match_data *match_data; + re = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, 0, &errornumber, &erroroffset, NULL); + match_data = pcre2_match_data_create_from_pattern(re, NULL); + PCRE2_SIZE start_offset = 0; + + do { + rc = pcre2_match(re, subject, subject_length, start_offset, 0, match_data, NULL); + + if (rc != PCRE2_ERROR_NOMATCH) { + char** sStore = new char*[7]; + //Match succeeded. Converting strings to the needed types. + ovector = pcre2_get_ovector_pointer(match_data); + for (int i = 0 ; i < 7 ; i++) { + sStore[i] = new char[ovector[2 * i + 3] - ovector[2 * i + 2]]; + sprintf(sStore[i1], "%.*s", (int)(ovector[2 * i + 3] - ovector[2 * i + 2]), (char*)(subject + ovector[2 * i + 2])); + } + temp1 = string(sStore[0]); + da = string(sStore[1]); + daClip = string(sStore[2]); + daProb = string(sStore[3]); + temp2 = string(sStore[4]); + std = string(sStore[5]); + stdClip = string(sStore[6]); + for (int i = 0 ; i < 7 ; i++) { + delete[] sStore[i]; + } + delete[] sStore; + //Preparing next match. + start_offset = ovector[1]; + } - while (re.FindAndConsume(&input, &temp1, - &da, &daClip, &daProb, - &temp2, - &std, &stdClip)) { if (da.length() > 0) { if (da.length() > 12 && da.substr(0, 12) == "splice_dist_") { // Remove the _1, _2 suffix @@ -382,7 +413,7 @@ keys.push_back(std); result[std] = stdClip; } - } + } while (rc != PCRE2_ERROR_NOMATCH); } return pair, LIBGOBY_HASH_MAP >(keys, result); --- a/cpp/configure.ac +++ b/cpp/configure.ac @@ -63,9 +63,9 @@ # use pkg-config to make sure that protobuf is available # see http://code.google.com/p/protobuf/ PKG_CHECK_MODULES([PROTOBUF], [protobuf >= 2.4.1]) -# use pkg-config to make sure that PCRE is available +# use pkg-config to make sure that PCRE2 is available # see http://www.pcre.org/ -PKG_CHECK_MODULES([LIBPCRECPP], [libpcrecpp >= 8.10]) +PKG_CHECK_MODULES([LIBPCRE2], [libpcre2-8 >= 10.39]) AC_CONFIG_FILES([Makefile src/Makefile goby.pc]) AC_OUTPUT --- a/cpp/goby.pc.in +++ b/cpp/goby.pc.in @@ -6,5 +6,5 @@ Name: Goby Description: API for reading binary data files created using the Goby next-gen data management framework Version: @VERSION@ -Libs: -L${libdir} -lgoby @PROTOBUF_LIBS@ @LIBPCRECPP_LIBS@ -Cflags: -I${includedir} @PROTOBUF_CFLAGS@ @LIBPCRECPP_CFLAGS@ +Libs: -L${libdir} -lgoby @PROTOBUF_LIBS@ @LIBPCRE2_LIBS@ +Cflags: -I${includedir} @PROTOBUF_CFLAGS@ @LIBPCRE2_CFLAGS@ --- a/cpp/src/Makefile.am +++ b/cpp/src/Makefile.am @@ -22,10 +22,10 @@ BUILT_SOURCES = goby.timestamp -AM_CXXFLAGS = @PROTOBUF_CFLAGS@ @LIBPCRECPP_CFLAGS@ ${BOOST_CPPFLAGS} -AM_CFLAGS = @PROTOBUF_CFLAGS@ @LIBPCRECPP_CFLAGS@ ${BOOST_CPPFLAGS} +AM_CXXFLAGS = @PROTOBUF_CFLAGS@ @LIBPCRE2_CFLAGS@ ${BOOST_CPPFLAGS} +AM_CFLAGS = @PROTOBUF_CFLAGS@ @LIBPCRE2_CFLAGS@ ${BOOST_CPPFLAGS} -libgoby_la_LIBADD = @PROTOBUF_LIBS@ @LIBPCRECPP_LIBS@ +libgoby_la_LIBADD = @PROTOBUF_LIBS@ @LIBPCRE2_LIBS@ libgoby_la_LDFLAGS = -version-info @version_info@ -export-dynamic # -no-undefined libgoby_la_SOURCES = \ goby/ICBFileInputStream.cc \