sequence_set.cpp

ncbi源码

            if (!registers) throw "can't allocate registers";
            patternBuffer->regs_allocated = REGS_UNALLOCATED;
        }

        // update pattern buffer if not the same pattern as before
        static string previousPrositePattern;
        static int nGroups;
        int i;
        if (prositePattern != previousPrositePattern) {

            // convert from ProSite syntax
            string regexPattern;
            if (!Prosite2Regex(prositePattern, &regexPattern, &nGroups))
                throw "error converting ProSite to regex syntax";

            // create pattern buffer
            TRACEMSG("compiling pattern '" << regexPattern << "'");
            const char *error = re_compile_pattern(regexPattern.c_str(), regexPattern.size(), patternBuffer);
            if (error) throw error;

            // optimize pattern buffer
            int err = re_compile_fastmap(patternBuffer);
            if (err) throw "re_compile_fastmap internal error";

            previousPrositePattern = prositePattern;
        }

        // do the search, finding all non-overlapping matches
        int start = 0;
        while (start < Length()) {

            int result = re_search(patternBuffer, sequenceString.c_str(),
                Length(), start, Length(), registers);
            if (result == -1)
                break;
            else if (result == -2)
                throw "re_search internal error";

            // re_search gives the longest hit, but we want the shortest; so try to find the
            // shortest hit within the hit already found by limiting the length of the string
            // allowed to be included in the search. (This isn't very efficient! but
            // the regex API doesn't have an option for finding the shortest hit...)
            int stringSize = start + 1;
            while (stringSize <= Length()) {
                result = re_search(patternBuffer, sequenceString.c_str(),
                    stringSize, start, stringSize, registers);
                if (result >= 0) break;
                ++stringSize;
            }

            // parse the match registers, highlight ranges
//            TESTMSG("found match starting at " << identifier->ToString() << " loc " << result+1);
            int lastMatched = result;
            for (i=1; i<registers->num_regs; ++i) {
                int from = registers->start[i], to = registers->end[i] - 1;
                if (from >= 0 && to >= 0) {
                    if (to > lastMatched) lastMatched = to;

                    // highlight this ranage
//                    TESTMSG("register " << i << ": from " << from+1 << " to " << to+1);
                    GlobalMessenger()->AddHighlights(this, from, to);
                }
            }

            start = lastMatched + 1;
        }

    } catch (const char *err) {
        ERRORMSG("Sequence::HighlightPattern() - " << err);
        retval = false;
    }

    // cleanup
    re_set_syntax(oldSyntax);

    return retval;
}

END_SCOPE(Cn3D)

/*
* ---------------------------------------------------------------------------
* $Log: sequence_set.cpp,v $
* Revision 1000.4  2004/06/01 18:29:07  gouriano
* PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.68
*
* Revision 1.68  2004/05/21 21:41:39  gorelenk
* Added PCH ncbi_pch.hpp
*
* Revision 1.67  2004/05/21 17:29:51  thiessen
* allow conversion of mime to cdd data
*
* Revision 1.66  2004/03/15 18:27:12  thiessen
* prefer prefix vs. postfix ++/-- operators
*
* Revision 1.65  2004/03/01 22:56:09  thiessen
* convert 'T' to 'U' for RNA with IUPACna
*
* Revision 1.64  2004/02/19 17:05:06  thiessen
* remove cn3d/ from include paths; add pragma to disable annoying msvc warning
*
* Revision 1.63  2004/01/05 17:09:16  thiessen
* abort import and warn if same accession different gi
*
* Revision 1.62  2003/11/26 20:37:54  thiessen
* prefer gi for URLs
*
* Revision 1.61  2003/11/20 22:08:49  thiessen
* update Entrez url
*
* Revision 1.60  2003/09/03 18:14:01  thiessen
* hopefully fix Seq-id issues
*
* Revision 1.59  2003/08/30 14:01:15  thiessen
* use existing CSeq_id instead of creating new one
*
* Revision 1.58  2003/07/14 18:37:08  thiessen
* change GetUngappedAlignedBlocks() param types; other syntax changes
*
* Revision 1.57  2003/03/13 14:26:18  thiessen
* add file_messaging module; split cn3d_main_wxwin into cn3d_app, cn3d_glcanvas, structure_window, cn3d_tools
*
* Revision 1.56  2003/02/03 19:20:05  thiessen
* format changes: move CVS Log to bottom of file, remove std:: from .cpp files, and use new diagnostic macros
*
* Revision 1.55  2002/12/12 15:04:11  thiessen
* fix for report launch on nucleotides
*
* Revision 1.54  2002/12/09 16:25:04  thiessen
* allow negative score threshholds
*
* Revision 1.53  2002/11/22 19:54:29  thiessen
* fixes for wxMac/OSX
*
* Revision 1.52  2002/11/19 21:19:44  thiessen
* more const changes for objects; fix user vs default style bug
*
* Revision 1.51  2002/11/06 00:18:10  thiessen
* fixes for new CRef/const rules in objects
*
* Revision 1.50  2002/10/11 17:21:39  thiessen
* initial Mac OSX build
*
* Revision 1.49  2002/09/13 14:21:45  thiessen
* finish hooking up browser launch on unix
*
* Revision 1.48  2002/09/06 18:56:48  thiessen
* add taxonomy to description
*
* Revision 1.47  2002/09/05 17:14:14  thiessen
* create new netscape window on unix
*
* Revision 1.46  2002/07/12 13:24:10  thiessen
* fixes for PSSM creation to agree with cddumper/RPSBLAST
*
* Revision 1.45  2002/01/24 20:08:17  thiessen
* fix local id problem
*
* Revision 1.44  2002/01/19 02:34:46  thiessen
* fixes for changes in asn serialization API
*
* Revision 1.43  2002/01/11 15:49:01  thiessen
* update for Mac CW7
*
* Revision 1.42  2001/11/30 14:59:55  thiessen
* add netscape launch for Mac
*
* Revision 1.41  2001/11/30 14:02:05  thiessen
* progress on sequence imports to single structures
*
* Revision 1.40  2001/11/27 16:26:09  thiessen
* major update to data management system
*
* Revision 1.39  2001/09/27 15:37:59  thiessen
* decouple sequence import and BLAST
*
* Revision 1.38  2001/08/21 01:10:31  thiessen
* fix lit. launch for nucleotides
*
* Revision 1.37  2001/08/09 19:07:13  thiessen
* add temperature and hydrophobicity coloring
*
* Revision 1.36  2001/07/24 15:02:59  thiessen
* use ProSite syntax for pattern searches
*
* Revision 1.35  2001/07/23 20:09:23  thiessen
* add regex pattern search
*
* Revision 1.34  2001/07/19 19:14:38  thiessen
* working CDD alignment annotator ; misc tweaks
*
* Revision 1.33  2001/07/10 16:39:55  thiessen
* change selection control keys; add CDD name/notes dialogs
*
* Revision 1.32  2001/06/21 02:02:34  thiessen
* major update to molecule identification and highlighting ; add toggle highlight (via alt)
*
* Revision 1.31  2001/06/18 21:48:27  thiessen
* add [PACC] to Entrez query for PDB id's
*
* Revision 1.30  2001/06/02 17:22:46  thiessen
* fixes for GCC
*
* Revision 1.29  2001/05/31 18:47:09  thiessen
* add preliminary style dialog; remove LIST_TYPE; add thread single and delete all; misc tweaks
*
* Revision 1.28  2001/05/31 14:32:03  thiessen
* better netscape launch for unix
*
* Revision 1.27  2001/05/25 01:38:16  thiessen
* minor fixes for compiling on SGI
*
* Revision 1.26  2001/05/24 13:32:32  thiessen
* further tweaks for GTK
*
* Revision 1.25  2001/05/15 23:48:37  thiessen
* minor adjustments to compile under Solaris/wxGTK
*
* Revision 1.24  2001/05/02 16:35:15  thiessen
* launch entrez web page on sequence identifier
*
* Revision 1.23  2001/04/20 18:03:22  thiessen
* add ncbistdaa parsing
*
* Revision 1.22  2001/04/18 15:46:53  thiessen
* show description, length, and PDB numbering in status line
*
* Revision 1.21  2001/03/28 23:02:17  thiessen
* first working full threading
*
* Revision 1.20  2001/02/16 00:40:01  thiessen
* remove unused sequences from asn data
*
* Revision 1.19  2001/02/13 01:03:57  thiessen
* backward-compatible domain ID's in output; add ability to delete rows
*
* Revision 1.18  2001/02/08 23:01:50  thiessen
* hook up C-toolkit stuff for threading; working PSSM calculation
*
* Revision 1.17  2001/01/25 20:21:18  thiessen
* fix ostrstream memory leaks
*
* Revision 1.16  2001/01/09 21:45:00  thiessen
* always use pdbID as title if known
*
* Revision 1.15  2001/01/04 18:20:52  thiessen
* deal with accession seq-id
*
* Revision 1.14  2000/12/29 19:23:39  thiessen
* save row order
*
* Revision 1.13  2000/12/28 20:43:09  vakatov
* Do not use "set" as identifier.
* Do not include <strstream>. Use "CNcbiOstrstream" instead of "ostrstream".
*
* Revision 1.12  2000/12/21 23:42:15  thiessen
* load structures from cdd's
*
* Revision 1.11  2000/12/20 23:47:48  thiessen
* load CDD's
*
* Revision 1.10  2000/12/15 15:51:47  thiessen
* show/hide system installed
*
* Revision 1.9  2000/11/17 19:48:13  thiessen
* working show/hide alignment row
*
* Revision 1.8  2000/11/11 21:15:54  thiessen
* create Seq-annot from BlockMultipleAlignment
*
* Revision 1.7  2000/09/15 19:24:22  thiessen
* allow repeated structures w/o different local id
*
* Revision 1.6  2000/09/03 18:46:49  thiessen
* working generalized sequence viewer
*
* Revision 1.5  2000/08/30 23:46:27  thiessen
* working alignment display
*
* Revision 1.4  2000/08/30 19:48:41  thiessen
* working sequence window
*
* Revision 1.3  2000/08/28 23:47:19  thiessen
* functional denseg and dendiag alignment parsing
*
* Revision 1.2  2000/08/28 18:52:42  thiessen
* start unpacking alignments
*
* Revision 1.1  2000/08/27 18:52:22  thiessen
* extract sequence information
*
*/