/* FILE: cif_standard.C */ /* VERSION: 1.600 */ /* DATE: 2/3/2005 */ /* Comments and Questions to: sw-help@rcsb.rutgers.edu */ /* COPYRIGHT 1999-2005 Rutgers - The State University of New Jersey This software is provided WITHOUT WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OR ANY OTHER WARRANTY, EXPRESS OR IMPLIED. RUTGERS MAKE NO REPRESENTATION OR WARRANTY THAT THE SOFTWARE WILL NOT INFRINGE ANY PATENT, COPYRIGHT OR OTHER PROPRIETARY RIGHT. The user of this software shall indemnify, hold harmless and defend Rutgers, its governors, trustees, officers, employees, students, agents and the authors against any and all claims, suits, losses, liabilities, damages, costs, fees, and expenses including reasonable attorneys' fees resulting from or arising out of the use of this software. This indemnification shall include, but is not limited to, any and all claims alleging products liability. */ /* PDB SOFTWARE LICENSE AGREEMENT BY CLICKING THE ACCEPTANCE BUTTON OR INSTALLING OR USING THIS "SOFTWARE, THE INDIVIDUAL OR ENTITY LICENSING THE SOFTWARE ("LICENSEE") IS CONSENTING TO BE BOUND BY AND IS BECOMING A PARTY TO THIS AGREEMENT. IF LICENSEE DOES NOT AGREE TO ALL OF THE TERMS OF THIS AGREEMENT THE LICENSEE MUST NOT INSTALL OR USE THE SOFTWARE. 1. LICENSE AGREEMENT This is a license between you ("Licensee") and the Protein Data Bank (PDB) at Rutgers, The State University of New Jersey (hereafter referred to as "RUTGERS"). The software is owned by RUTGERS and protected by copyright laws, and some elements are protected by laws governing trademarks, trade dress and trade secrets, and may be protected by patent laws. 2. LICENSE GRANT RUTGERS grants you, and you hereby accept, non-exclusive, royalty-free perpetual license to install, use, modify, prepare derivative works, incorporate into other computer software, and distribute in binary and source code format, or any derivative work thereof, together with any associated media, printed materials, and on-line or electronic documentation (if any) provided by RUTGERS (collectively, the "SOFTWARE"), subject to the following terms and conditions: (i) any distribution of the SOFTWARE shall bind the receiver to the terms and conditions of this Agreement; (ii) any distribution of the SOFTWARE in modified form shall clearly state that the SOFTWARE has been modified from the version originally obtained from RUTGERS. 2. COPYRIGHT; RETENTION OF RIGHTS. The above license grant is conditioned on the following: (i) you must reproduce all copyright notices and other proprietary notices on any copies of the SOFTWARE and you must not remove such notices; (ii) in the event you compile the SOFTWARE, you will include the copyright notice with the binary in such a manner as to allow it to be easily viewable; (iii) if you incorporate the SOFTWARE into other code, you must provide notice that the code contains the SOFTWARE and include a copy of the copyright notices and other proprietary notices. All copies of the SOFTWARE shall be subject to the terms of this Agreement. 3. NO MAINTENANCE OR SUPPORT; TREATMENT OF ENHANCEMENTS RUTGERS is under no obligation whatsoever to: (i) provide maintenance or support for the SOFTWARE; or (ii) to notify you of bug fixes, patches, or upgrades to the features, functionality or performance of the SOFTWARE ("Enhancements") (if any), whether developed by RUTGERS or third parties. If, in its sole discretion, RUTGERS makes an Enhancement available to you and RUTGERS does not separately enter into a written license agreement with you relating to such bug fix, patch or upgrade, then it shall be deemed incorporated into the SOFTWARE and subject to this Agreement. You are under no obligation whatsoever to provide any Enhancements to RUTGERS or the public that you may develop over time; however, if you choose to provide your Enhancements to RUTGERS, or if you choose to otherwise publish or distribute your Enhancements, in source code form without contemporaneously requiring end users or RUTGERS to enter into a separate written license agreement for such Enhancements, then you hereby grant RUTGERS a non-exclusive, royalty-free perpetual license to install, use, modify, prepare derivative works, incorporate into the SOFTWARE or other computer software, distribute, and sublicense your Enhancements or derivative works thereof, in binary and source code form. 4. FEES. There is no license fee for the SOFTWARE. If Licensee wishes to receive the SOFTWARE on media, there may be a small charge for the media and for shipping and handling. Licensee is responsible for any and all taxes. 5. TERMINATION. Without prejudice to any other rights, Licensor may terminate this Agreement if Licensee breaches any of its terms and conditions. Upon termination, Licensee shall destroy all copies of the SOFTWARE. 6. PROPRIETARY RIGHTS. Title, ownership rights, and intellectual property rights in the Product shall remain with RUTGERS. Licensee acknowledges such ownership and intellectual property rights and will not take any action to jeopardize, limit or interfere in any manner with RUTGERS' ownership of or rights with respect to the SOFTWARE. The SOFTWARE is protected by copyright and other intellectual property laws and by international treaties. Title and related rights in the content accessed through the SOFTWARE is the property of the applicable content owner and is protected by applicable law. The license granted under this Agreement gives Licensee no rights to such content. 7. DISCLAIMER OF WARRANTY. THE SOFTWARE IS PROVIDED FREE OF CHARGE, AND, THEREFORE, ON AN "AS IS" BASIS, WITHOUT WARRANTY OF ANY KIND, INCLUDING WITHOUT LIMITATION THE WARRANTIES THAT IT IS FREE OF DEFECTS, MERCHANTABLE, FIT FOR A PARTICULAR PURPOSE OR NON-INFRINGING. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS BORNE BY LICENSEE. SHOULD THE SOFTWARE PROVE DEFECTIVE IN ANY RESPECT, THE LICENSEE AND NOT LICENSOR ASSUMES THE ENTIRE COST OF ANY SERVICE AND REPAIR. THIS DISCLAIMER OF WARRANTY CONSTITUTES AN ESSENTIAL PART OF THIS AGREEMENT. NO USE OF THE PRODUCT IS AUTHORIZED HEREUNDER EXCEPT UNDER THIS DISCLAIMER. 8. LIMITATION OF LIABILITY. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, IN NO EVENT WILL LICENSOR BE LIABLE FOR ANY INDIRECT, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OF OR INABILITY TO USE THE SOFTWARE, INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS OF GOODWILL, WORK STOPPAGE, COMPUTER FAILURE OR MALFUNCTION, OR ANY AND ALL OTHER COMMERCIAL DAMAGES OR LOSSES, EVEN IF ADVISED OF THE POSSIBILITY THEREOF. */ #include #include #include #include #include "DICFileObj.h" #include "ndb_misclib.h" #include "_maxit.h" #include "bond_angle_torsion.h" #include "erf_category.h" #include "erf_util.h" #include "util.h" #include "cif_standard.h" #define NS 10 #define dictxraybinfile "mmcif_xray.odb" #define dictnmrbinfile "mmcif_nmr.odb" static void remove_infinite(ISTable *Table, const int id_idx); static void remove_comma(ISTable *Table, const int id_idx); static void remove_k(ISTable *Table, const int col, const int row); static void update_index(ISTable *Table, const int idx, ISTable *Table1, const int idx1); static ISTable *get_extend_table(ISTable *Table, _PDB_INSERT_CODE *Miss_Man_I_cat); static int find_index(SCHEME_TABLE *scheme, const char *chnid, CifString &num); static void AddMandatoryItem(CifFileObj *in, const int experiment_type); static void GetKeywordFromItemName(char *keyword, const char *itemName); static void ConvertEscapedString(CifString& inString, CifString& outString); void _MAXIT::ndbcif_postprocessing_cif_standard(const int standardized, const int keep_contact_author) { ISTable *Table = NULL, *Table1 = NULL, *Table2 = NULL, *Table3 = NULL; CifString string, id, old_id, text, beg, end; char buffer[1000], chnid[10], *p, *q, buff[1000], buff1[1000]; char **categoryName = NULL, classification[4][20]; int i, j, k, l, m, id_idx, irow, rowNo, rowNo1, colNo, mon_id; int type_idx = -1, mon_idx = -1, name_idx = -1, syno_idx = -1, form_idx = -1, wght_idx = -1; int auth_seq_beg, auth_seq_end, seq_beg, seq_end, chn_idx = -1; int auth_beg_ins, auth_end_ins, idx[NUM_ITEM]; int found_gap, last, seq_num, align_id, align_id1, numCategory = 0; int ndb_chain_id, rowNo2, entity_id, pdb_id, index1, index2, index3, index4; int mol_id, num1, num2, num3, num4, len; ReVarCifArray *ColumnNames = NULL; double weight; ConnectFormat *drug = NULL; SCHEME_TABLE *scheme = NULL; _ATOM *atom = NULL, **at1 = NULL, **at2 = NULL, **at3 = NULL, **at4 = NULL; _RESIDUE *residue = NULL, *prev = NULL, *curr = NULL; _CHAIN *chain = NULL; Table = _CifObj->GetTablePtr(_StructureId, "computing"); Table1 = _CifObj->GetTablePtr(_StructureId, "software"); if (!is_empty_table(Table) && is_empty_table(Table1)) { _CifObj->DeleteTable(_StructureId, "software"); Table1 = new ISTable("software"); Table1->AddColumn("name"); Table1->AddColumn("classification"); Table1->AddColumn("version"); Table1->AddColumn("citation_id"); idx[0] = Table->GetColumnIndex("data_collection"); strcpy(classification[0], "data collection"); idx[1] = Table->GetColumnIndex("data_reduction"); strcpy(classification[1], "data reduction"); idx[2] = Table->GetColumnIndex("structure_solution"); strcpy(classification[2], "model building"); idx[3] = Table->GetColumnIndex("structure_refinement"); strcpy(classification[3], "refinement"); irow = 0; for (i = 0; i < 4; i++) { if (idx[i] < 0) continue; Table->GetCell(string, idx[i], 0); if (string == "" || string == "." || string == "?") continue; strcpy(buff, string.Text()); len = strlen(buff); for (j = 0; j < len; j++) { if (buff[j] == ';') buff[j] = ','; } p = strstr(buff, " AND "); if (p) { *(p + 1) = ' '; *(p + 2) = ','; *(p + 3) = ' '; } p = strstr(buff, " and "); if (p) { *(p + 1) = ' '; *(p + 2) = ','; *(p + 3) = ' '; } p = strstr(buff, " And "); if (p) { *(p + 1) = ' '; *(p + 2) = ','; *(p + 3) = ' '; } q = buff; while (q) { q = get_words_separate_by_delimit(q, buff1, ','); Table1->AddRow(); p = get_word(buff1, buffer); text.Copy(buffer); Table1->UpdateCell(text, 0, irow); text.Copy(classification[i]); Table1->UpdateCell(text, 1, irow); if (p) text.Copy(p); else text.Copy("."); Table1->UpdateCell(text, 2, irow); text.Copy("."); Table1->UpdateCell(text, 3, irow); irow++; } } _CifObj->WriteTable(Table1, _StructureId, "software"); } Table = _CifObj->GetTablePtr(_StructureId, "ndb_struct_na_base_pair"); if (!is_empty_table(Table)) { rowNo = Table->GetNumRows(); type_idx = Table->GetColumnIndex("hbond_type_28"); name_idx = Table->GetColumnIndex("hbond_type_12"); for (i = 0; i < rowNo; i++) { if (type_idx >= 0) { Table->GetCell(string, type_idx, i); for (j = 0; j < NUM_SAENGER; j++) { if (!strcasecmp(string.Text(), _saenger[j].name)) { sprintf(buffer, "%d", _saenger[j].index); string.Copy(buffer); Table->UpdateCell(string, type_idx, i); break; } } } if (name_idx >= 0) { Table->GetCell(string, name_idx, i); for (j = 0; j < NUM_WESTHOF; j++) { if (!strcasecmp(string.Text(), _westhof[j].name)) { sprintf(buffer, "%d", _westhof[j].index); string.Copy(buffer); Table->UpdateCell(string, name_idx, i); break; } } } } _CifObj->WriteTable(Table, _StructureId, "ndb_struct_na_base_pair"); } Table = _CifObj->GetTablePtr(_StructureId, "cell"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("Z_PDB"); get_value(string, Table, id_idx, 0); if (string == "" || string == "0") { string.Copy("1"); Table->UpdateCell(string, id_idx, 0); _CifObj->WriteTable(Table, _StructureId, "cell"); } } for (k = 0; k < NUM_DIFFRN; k++) { Table = _CifObj->GetTablePtr(_StructureId, diffrn_related_categories[k]); if (is_empty_table(Table)) continue; type_idx = Table->GetColumnIndex("diffrn_id"); if (type_idx < 0) continue; Table1 = _CifObj->GetTablePtr(_StructureId, "diffrn"); if (!Table1) { Table1 = new ISTable("diffrn"); Table1->AddColumn("id"); Table1->AddColumn("crystal_id"); } id_idx = Table1->GetColumnIndex("id"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "") { sprintf(buffer, "%d", i + 1); string.Copy(buffer); Table->UpdateCell(string, type_idx, i); } } update_index(Table, type_idx, Table1, id_idx); _CifObj->WriteTable(Table, _StructureId, diffrn_related_categories[k]); _CifObj->WriteTable(Table1, _StructureId, "diffrn"); } Table = _CifObj->GetTablePtr(_StructureId, "diffrn_radiation_wavelength"); if (is_empty_table(Table, "id")) _CifObj->DeleteTable(_StructureId, "diffrn_radiation_wavelength"); Table = _CifObj->GetTablePtr(_StructureId, "diffrn_source"); if (!is_empty_table(Table)) { Table1 = _CifObj->GetTablePtr(_StructureId, "diffrn_radiation_wavelength"); if (is_empty_table(Table1, "id")) { _CifObj->DeleteTable(_StructureId, "diffrn_radiation_wavelength"); Table1 = new ISTable("diffrn_radiation_wavelength"); Table1->AddColumn("id"); Table1->AddColumn("wavelength"); Table1->AddColumn("wt"); } id_idx = Table1->GetColumnIndex("id"); name_idx = Table1->GetColumnIndex("wavelength"); irow = Table1->GetNumRows(); mon_idx = Table->GetColumnIndex("rcsb_wavelength"); type_idx = Table->GetColumnIndex("rcsb_wavelength_list"); index1 = Table->GetColumnIndex("source"); index2 = Table->GetColumnIndex("type"); index3 = Table->GetColumnIndex("ndb_synchrotron_site"); index4 = Table->GetColumnIndex("ndb_synchrotron_beamline"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(text, Table, index1, i); get_value(beg, Table, index2, i); get_value(end, Table, index3, i); Table->GetCell(string, index4, i); if (!strcasecmp(text.Text(), "Synchrotron") && beg != "" && end == "" && (string == "" || string == "?" || string == ".")) { beg.ToUpper(); separation_string_base_word(beg.Text(), buffer, buff, "BEAMLINE"); string.Copy(buffer); Table->UpdateCell(string, index3, i); string.Copy(buff); Table->UpdateCell(string, index4, i); } get_value(string, Table, mon_idx, i); get_value(beg, Table, type_idx, i); if (string == "" && beg == "") continue; string += " "; string += beg; strcpy(buffer, string.Text()); len = strlen(buffer); for (j = 0; j < len; j++) { if (buffer[j] == ',' || buffer[j] == ';' || buffer[j] == '-') buffer[j] = ' '; } ndb_clean_string(buffer); p = buffer; while (p) { p = get_word(p, chnid); len = strlen(chnid); for (j = 0; j < len; j++) { if (chnid[j] != '.' && !isdigit(chnid[j])) break; } if (j < len) continue; irow = Table1->GetNumRows(); for (j = 0; j < irow; j++) { get_value(beg, Table1, id_idx, j); get_value(end, Table1, name_idx, j); if (beg == "1" && end == "" || end == chnid) break; } if (j < irow) { string.Copy(chnid); Table1->UpdateCell(string, name_idx, j); } else { Table1->AddRow(); sprintf(buff, "%d", irow+1); string.Copy(buff); Table1->UpdateCell(string, id_idx, irow); string.Copy(chnid); Table1->UpdateCell(string, name_idx, irow); } } } _CifObj->WriteTable(Table, _StructureId, "diffrn_source"); if (is_empty_table(Table1, "id")) // _CifObj->DeleteTable(_StructureId, "diffrn_radiation_wavelength"); delete Table1; else _CifObj->WriteTable(Table1, _StructureId, "diffrn_radiation_wavelength"); } Table = _CifObj->GetTablePtr(_StructureId, "diffrn_radiation"); if (!is_empty_table(Table)) { rowNo = Table->GetNumRows(); id_idx = Table->GetColumnIndex("wavelength_id"); if (id_idx >= 0) { Table1 = _CifObj->GetTablePtr(_StructureId, "diffrn_radiation_wavelength"); if (!Table1) { Table1 = new ISTable("diffrn_radiation_wavelength"); Table1->AddColumn("id"); Table1->AddColumn("wavelength"); Table1->AddColumn("wt"); } type_idx = Table1->GetColumnIndex("id"); update_index(Table, id_idx, Table1, type_idx); if (is_empty_table(Table1)) _CifObj->DeleteTable(_StructureId, "diffrn_radiation_wavelength"); else _CifObj->WriteTable(Table1, _StructureId, "diffrn_radiation_wavelength"); } id_idx = Table->GetColumnIndex("diffrn_id"); if (id_idx >= 0 && rowNo == 1) { Table->GetCell(string, id_idx, 0); if (string == "" || string == "?") { string.Copy("1"); Table->UpdateCell(string, id_idx, 0); } } id_idx = Table->GetColumnIndex("ndb_monochromatic_or_laue_m_l"); if (id_idx >= 0) { for (i = 0; i < rowNo; i++) { Table->GetCell(string, id_idx, i); if (string == "N" || string == "n") { string.Copy("M"); Table->UpdateCell(string, id_idx, i); } if (string == "Y" || string == "y") { string.Copy("L"); Table->UpdateCell(string, id_idx, i); } } } _CifObj->WriteTable(Table, _StructureId, "diffrn_radiation"); } for (k = 0; k < NUM_MISSING_M_I_C; k++) { Table = _CifObj->GetTablePtr(_StructureId, _Miss_Man_I_cat[k].category); if (is_empty_table(Table)) continue; Table1 = get_extend_table(Table, &_Miss_Man_I_cat[k]); if (Table1) { _CifObj->DeleteTable(_StructureId, _Miss_Man_I_cat[k].category); _CifObj->WriteTable(Table1, _StructureId, _Miss_Man_I_cat[k].category); } } for (k = 0; k < NUM_MISSIMG_V_I_C; k++) { Table = _CifObj->GetTablePtr(_StructureId, _Miss_Value_I_cat[k].category); if (is_empty_table(Table)) continue; for (i = 0; i < _Miss_Value_I_cat[k].num_items; i++) { idx[i] = Table->GetColumnIndex(_Miss_Value_I_cat[k].items[i]); } rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { for (j = 0; j < _Miss_Value_I_cat[k].num_items; j++) { Table->GetCell(string, idx[j], i); if (string == "?" || string == "") { string.Copy("."); Table->UpdateCell(string, idx[j], i); } } } _CifObj->WriteTable(Table, _StructureId, _Miss_Value_I_cat[k].category); } Table = _CifObj->GetTablePtr(_StructureId, "refine"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("ndb_ls_sigma_F"); if (id_idx >= 0) { Table->GetCell(string, id_idx, 0); p = strstr(string.Text(), "COMPLETENESS FOR RANGE (%) :"); if (p) { get_word(string.Text(), buffer); id.Copy(buffer); Table->UpdateCell(id, id_idx, 0); p += 28; q = new char[strlen(p)+1]; strcpy(q, p); ndb_clean_string(q); if (strcmp(q, "NULL") && strcmp(q, "NONE") && strcmp(q, "N/A") && strcmp(q, "")) { type_idx = Table->GetColumnIndex("ls_percent_reflns_obs"); Table->GetCell(id, type_idx, 0); if (id == "?" || id == "") { id.Copy(q); Table->UpdateCell(id, type_idx, 0); } } delete [] q; } } id_idx = Table->GetColumnIndex("B_iso_mean"); if (id_idx >= 0) { Table->GetCell(string, id_idx, 0); p = strstr(string.Text(), "OVERALL ANISOTROPIC B VALUE"); if (p) { get_word(string.Text(), buffer); id.Copy(buffer); Table->UpdateCell(id, id_idx, 0); } } id_idx = Table->GetColumnIndex("aniso_B[2][3]"); if (id_idx >= 0) { get_value(string, Table, id_idx, 0); p = strstr(string.Text(), "ESTIM"); if (p) { *p = '\0'; ndb_clean_string(string.Text()); Table->UpdateCell(string, id_idx, 0); } } id_idx = Table->GetColumnIndex("ls_d_res_low"); remove_infinite(Table, id_idx); id_idx = Table->GetColumnIndex("ls_number_reflns_obs"); remove_comma(Table, id_idx); id_idx = Table->GetColumnIndex("ls_number_reflns_all"); remove_comma(Table, id_idx); _CifObj->WriteTable(Table, _StructureId, "refine"); } Table = _CifObj->GetTablePtr(_StructureId, "refine_hist"); Table1 = _CifObj->GetTablePtr(_StructureId, "refine"); if (!is_empty_table(Table) && !is_empty_table(Table1)) { id_idx = Table1->GetColumnIndex("ls_d_res_high"); type_idx = Table1->GetColumnIndex("ls_d_res_low"); get_value(id, Table1, id_idx, 0); get_value(text, Table1, type_idx, 0); if (id != "") { id_idx = Table->GetColumnIndex("d_res_high"); Table->UpdateCell(id, id_idx, 0); } if (text != "") { type_idx = Table->GetColumnIndex("d_res_low"); Table->UpdateCell(text, type_idx, 0); } _CifObj->WriteTable(Table, _StructureId, "refine_hist"); } Table = _CifObj->GetTablePtr(_StructureId, "struct_ref"); Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq"); Table2 = _CifObj->GetTablePtr(_StructureId, "ndb_poly_seq_scheme"); if (!is_empty_table(Table) && !is_empty_table(Table1) && !is_empty_table(Table2)) { rowNo = Table->GetNumRows(); id_idx = Table->GetColumnIndex("id"); type_idx = Table->GetColumnIndex("entity_id"); rowNo1 = Table1->GetNumRows(); align_id = Table1->GetColumnIndex("ref_id"); ndb_chain_id = Table1->GetColumnIndex("ndb_chain_id"); rowNo2 = Table2->GetNumRows(); entity_id = Table2->GetColumnIndex("entity_id"); pdb_id = Table2->GetColumnIndex("pdb_id"); for (i = 0; i < rowNo; i++) { get_value(text, Table, id_idx, i); found_gap = 0; if (text != "") { for (j = 0; j < rowNo1; j++) { get_value(beg, Table1, align_id, j); if (beg == text) { get_value(old_id, Table1, ndb_chain_id, j); for (k = 0; k < rowNo2; k++) { get_value(end, Table2, pdb_id, k); if (old_id == end) { found_gap = 1; Table2->GetCell(string, entity_id, k); Table->UpdateCell(string, type_idx, i); break; } } } if (found_gap) break; } } if (!found_gap) { string.Copy("."); Table->UpdateCell(string, type_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "struct_ref"); } Table = _CifObj->GetTablePtr(_StructureId, "reflns"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("d_resolution_low"); remove_infinite(Table, id_idx); id_idx = Table->GetColumnIndex("number_obs"); remove_comma(Table, id_idx); id_idx = Table->GetColumnIndex("number_all"); remove_comma(Table, id_idx); _CifObj->WriteTable(Table, _StructureId, "reflns"); } Table = _CifObj->GetTablePtr(_StructureId, "rcsb_nmr_spectrometer"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("field_strength"); if (id_idx >= 0) { rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, id_idx, i); if (string != "") { q = new char[strlen(string.Text())+1]; strcpy(q, string.Text()); ndb_clean_string(q); string.Copy(q); Table->UpdateCell(string, id_idx, i); delete [] q; } } _CifObj->WriteTable(Table, _StructureId, "rcsb_nmr_spectrometer"); } } for (k = 0; k < NUM_SYMM_V_C; k++) { Table = _CifObj->GetTablePtr(_StructureId, _Symmetry_Value_Cat[k].category); if (is_empty_table(Table)) continue; for (i = 0; i < _Symmetry_Value_Cat[k].num_items; i++) { idx[i] = Table->GetColumnIndex(_Symmetry_Value_Cat[k].items[i]); } rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { for (j = 0; j < _Symmetry_Value_Cat[k].num_items; j++) { get_value(string, Table, idx[j], i); if (string != "") { strcpy(buffer, string.Text()); l = strlen(buffer); for (m = 0; m < l; m++) if (buffer[m] == '_') break; if (m == l) { for (m = 0; m < (l - 3); m++) chnid[m] = buffer[m]; chnid[m] = '_'; for (m = (l - 3); m < l; m++) chnid[m+1] = buffer[m]; chnid[m+1] = '\0'; string.Copy(chnid); Table->UpdateCell(string, idx[j], i); } } } } _CifObj->WriteTable(Table, _StructureId, _Symmetry_Value_Cat[k].category); } if (_num_mol) { _CifObj->DeleteTable(_StructureId, "chem_comp"); Table = new ISTable("chem_comp"); Table->AddColumn("id"); Table->AddColumn("type"); Table->AddColumn("mon_nstd_flag"); Table->AddColumn("name"); Table->AddColumn("ndb_synonyms"); Table->AddColumn("formula"); Table->AddColumn("formula_weight"); _CifObj->DeleteTable(_StructureId, "atom_type"); Table1 = new ISTable("atom_type"); Table1->AddColumn("symbol"); chain = _molecules[_rep_mol]->GetFirstChain(); while (chain) { residue = chain->GetFirstResidue(); while (residue) { atom = residue->GetFirstAtom(); while (atom) { rowNo = Table->GetNumRows(); rowNo1 = Table1->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, 0, i); if (string == atom->restype()) break; } if (i == rowNo) { Table->AddRow(); string.Copy(atom->restype()); Table->UpdateCell(string, 0, i); } for (i = 0; i < rowNo1; i++) { get_value(string, Table1, 0, i); if (string == atom->atom_type()) break; } if (i == rowNo1) { Table1->AddRow(); string.Copy(atom->atom_type()); Table1->UpdateCell(string, 0, i); } atom = residue->GetNextAtom(); } residue = chain->GetNextResidue(); } chain = _molecules[_rep_mol]->GetNextChain(); } _CifObj->WriteTable(Table1, _StructureId, "atom_type"); Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq_dif"); if (!is_empty_table(Table1)) { mon_id = Table1->GetColumnIndex("db_mon_id"); update_index(Table1, mon_id, Table, 0); } Table1 = _CifObj->GetTablePtr(_StructureId, "entity_poly_seq"); if (!is_empty_table(Table1)) { mon_id = Table1->GetColumnIndex("mon_id"); update_index(Table1, mon_id, Table, 0); } rowNo = Table->GetNumRows(); id_idx = Table->GetColumnIndex("id"); type_idx = Table->GetColumnIndex("type"); mon_idx = Table->GetColumnIndex("mon_nstd_flag"); name_idx = Table->GetColumnIndex("name"); syno_idx = Table->GetColumnIndex("ndb_synonyms"); form_idx = Table->GetColumnIndex("formula"); wght_idx = Table->GetColumnIndex("formula_weight"); for (i = 0; i < rowNo; i++) { get_value(string, Table, id_idx, i); drug = find_drug(string.Text()); if (!drug) { if (!strcmp(string.Text(), "UNL")) { string.Copy("UNKNOWN LIGAND"); Table->UpdateCell(string, name_idx, i); string.Copy("non-polymer"); Table->UpdateCell(string, type_idx, i); string.Copy("."); Table->UpdateCell(string, mon_idx, i); } continue; } weight = get_molecule_weight(string.Text()); string.Copy(drug->type()); string.ToLower(); if (string == "d-peptide linking") string.Copy("D-peptide linking"); else if (string == "l-peptide linking") string.Copy("L-peptide linking"); else if (string == "d-peptide nh3 amino terminus") string.Copy("D-peptide NH3 amino terminus"); else if (string == "l-peptide nh3 amino terminus") string.Copy("L-peptide NH3 amino terminus"); else if (string == "d-peptide cooh carboxy terminus") string.Copy("D-peptide COOH carboxy terminus"); else if (string == "l-peptide cooh carboxy terminus") string.Copy("L-peptide COOH carboxy terminus"); else if (string == "dna linking") string.Copy("DNA linking"); else if (string == "rna linking") string.Copy("RNA linking"); else if (string == "d-saccharide") string.Copy("D-saccharide"); else if (string == "l-saccharide") string.Copy("L-saccharide"); Table->UpdateCell(string, type_idx, i); if (is_a_standard_residue(drug->drugname()) && strcmp(drug->drugname(), "N") && strcmp(drug->drugname(), "UNK")) string.Copy("y"); else if (drug->parent_residue() && strcmp(drug->parent_residue(), "")) string.Copy("n"); else string.Copy("."); Table->UpdateCell(string, mon_idx, i); if (drug->chemical_name()) { string.Copy(drug->chemical_name()); Table->UpdateCell(string, name_idx, i); } if (drug->synonym()) { string.Copy(drug->synonym()); Table->UpdateCell(string, syno_idx, i); } if (drug->formula()) { if (drug->formal_charge() && strcmp(drug->formal_charge(), "") && strcmp(drug->formal_charge(), "0")) { sprintf(buffer, "%s %s", drug->formula(), drug->formal_charge()); string.Copy(buffer); } else string.Copy(drug->formula()); Table->UpdateCell(string, form_idx, i); } if (weight > 0.0) { sprintf(buffer, "%.3f", weight); string.Copy(buffer); } else string.Copy(""); Table->UpdateCell(string, wght_idx, i); } _CifObj->WriteTable(Table, _StructureId, "chem_comp"); } Table = _CifObj->GetTablePtr(_StructureId, "entry"); if (is_empty_table(Table)) { if (!Table) { Table = new ISTable("entry"); Table->AddColumn("id"); Table->AddRow(); } id_idx = Table->GetColumnIndex("id"); string.Copy(_StructureId); Table->UpdateCell(string, id_idx, 0); _CifObj->WriteTable(Table, _StructureId, "entry"); } Table1 = _CifObj->GetTablePtr(_StructureId, "struct_biol_gen"); if (Table1) { type_idx = Table1->GetColumnIndex("biol_id"); Table = _CifObj->GetTablePtr(_StructureId, "struct_biol"); if (is_empty_table(Table) && Table) { _CifObj->DeleteTable(_StructureId, "struct_biol"); Table = NULL; } if (!Table) { Table = new ISTable("struct_biol"); Table->AddColumn("id"); } id_idx = Table->GetColumnIndex("id"); update_index(Table1, type_idx, Table, id_idx); _CifObj->WriteTable(Table, _StructureId, "struct_biol"); } Table1 = _CifObj->GetTablePtr(_StructureId, "database_PDB_remark"); if (!is_empty_table(Table1)) { rowNo = Table1->GetNumRows(); id_idx = Table1->GetColumnIndex("id"); type_idx = Table1->GetColumnIndex("text"); if (id_idx >= 0 && type_idx >= 0) { old_id.Copy("-1"); j = 0; for (i = 0; i < rowNo; i++) { Table1->GetCell(id, id_idx, i); if (id != old_id) old_id.Copy(id); else break; } if (i < rowNo) { Table = new ISTable("database_PDB_remark"); Table->AddColumn("id"); Table->AddColumn("text"); old_id.Copy("-1"); text.Copy(""); irow = 0; for (i = 0; i < rowNo; i++) { Table1->GetCell(id, id_idx, i); Table1->GetCell(string, type_idx, i); if (string == "?" || string == ".") string.Copy(""); if (id != old_id) { if (text != "") { Table->AddRow(); Table->UpdateCell(old_id, 0, irow); Table->UpdateCell(text, 1, irow); irow++; } text.Copy(string); old_id.Copy(id); } else { text += '\n'; text += string; } } if (text != "") { Table->AddRow(); Table->UpdateCell(old_id, 0, irow); Table->UpdateCell(text, 1, irow); } _CifObj->DeleteTable(_StructureId, "database_PDB_remark"); _CifObj->WriteTable(Table, _StructureId, "database_PDB_remark"); } } } for (k = 0; k < NUM_CAT_ITEM; k++) { Table = _CifObj->GetTablePtr(_StructureId, _cat_items[k].category); if (is_empty_table(Table)) continue; type_idx = Table->GetColumnIndex(_cat_items[k].item); if (type_idx < 0) continue; rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "") continue; strcpy(buffer, string.Text()); ndb_convert_time_format(NDB_FILE_FORMAT_NDB, buffer, buffer, 11); string.Copy(buffer); Table->UpdateCell(string, type_idx, i); } _CifObj->WriteTable(Table, _StructureId, _cat_items[k].category); } Table = _CifObj->GetTablePtr(_StructureId, "diffrn"); if (!is_empty_table(Table)) { rowNo = Table->GetNumRows(); id_idx = Table->GetColumnIndex("ambient_temp"); type_idx = Table->GetColumnIndex("crystal_id"); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "") { string.Copy("1"); Table->UpdateCell(string, type_idx, i); } remove_k(Table, id_idx, i); } _CifObj->WriteTable(Table, _StructureId, "diffrn"); Table1 = _CifObj->GetTablePtr(_StructureId, "exptl_crystal"); if (!Table1) { Table1 = new ISTable("exptl_crystal"); Table1->AddColumn("id"); } id_idx = Table1->GetColumnIndex("id"); update_index(Table, type_idx, Table1, id_idx); _CifObj->WriteTable(Table1, _StructureId, "exptl_crystal"); } Table = _CifObj->GetTablePtr(_StructureId, "diffrn_radiation_wavelength"); if (!is_empty_table(Table)) { name_idx = Table->GetColumnIndex("wavelength"); type_idx = Table->GetColumnIndex("wt"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { Table->GetCell(string, name_idx, i); if (string == "" || string == "?") { string.Copy("."); Table->UpdateCell(string, name_idx, i); } Table->GetCell(string, type_idx, i); if (string == "" || string == "?" || string == ".") { string.Copy("1.0"); Table->UpdateCell(string, type_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "diffrn_radiation_wavelength"); } Table = _CifObj->GetTablePtr(_StructureId, "rcsb_nmr_exptl_sample_conditions"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("temperature"); type_idx = Table->GetColumnIndex("pressure"); name_idx = Table->GetColumnIndex("pressure_units"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { remove_k(Table, id_idx, i); if (type_idx >= 0) { get_value(string, Table, type_idx, i); get_value(beg, Table, name_idx, i); if (string == "1 ATM") { string.Copy("1"); Table->UpdateCell(string, type_idx, i); string.Copy("ATM"); Table->UpdateCell(string, name_idx, i); } else { if (beg == "" && isxdigit(string.Text()[0]) && isxdigit(string.Text()[strlen(string.Text())-1])) { string.Copy("ATM"); Table->UpdateCell(string, name_idx, i); } } } } _CifObj->WriteTable(Table, _StructureId, "rcsb_nmr_exptl_sample_conditions"); } Table = _CifObj->GetTablePtr(_StructureId, "exptl_crystal_grow"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("temp"); type_idx = Table->GetColumnIndex("pH"); mon_idx = Table->GetColumnIndex("rcsb_pH_range"); name_idx = Table->GetColumnIndex("crystal_id"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { remove_k(Table, id_idx, i); get_value(string, Table, type_idx, i); if (string != "") { strcpy(buffer, string.Text()); len = strlen(buffer); for (j = 0; j < len; j++) { if (!isdigit(buffer[j]) && buffer[j] != '.') break; } if (j < len) { string.Copy(""); Table->UpdateCell(string, type_idx, i); string.Copy(buffer); Table->UpdateCell(string, mon_idx, i); } } get_value(string, Table, name_idx, i); if (string == "") { string.Copy("1"); Table->UpdateCell(string, name_idx, i); } get_value(string, Table, mon_idx, i); if (string == "") continue; strcpy(buffer, string.Text()); len = strlen(buffer); for (j = 0; j < len; j++) { if (!isdigit(buffer[j]) && buffer[j] != '.') break; } if (j == len) { string.Copy(""); Table->UpdateCell(string, mon_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "exptl_crystal_grow"); } Table = _CifObj->GetTablePtr(_StructureId, "exptl_crystal_grow_comp"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("crystal_id"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, id_idx, i); if (string == "") { string.Copy("1"); Table->UpdateCell(string, name_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "exptl_crystal_grow_comp"); } _CifObj->DeleteTable(_StructureId, "struct_mon_prot_cis"); Table = new ISTable("struct_mon_prot_cis"); Table->AddColumn("id"); Table->AddColumn("label_comp_id"); Table->AddColumn("label_seq_id"); Table->AddColumn("label_asym_id"); Table->AddColumn("label_alt_id"); Table->AddColumn("ndb_label_ins_code"); Table->AddColumn("auth_comp_id"); Table->AddColumn("auth_seq_id"); Table->AddColumn("auth_asym_id"); Table->AddColumn("ndb_auth_ins_code"); Table->AddColumn("ndb_label_comp_id_2"); Table->AddColumn("ndb_label_seq_id_2"); Table->AddColumn("ndb_label_asym_id_2"); Table->AddColumn("ndb_label_ins_code_2"); Table->AddColumn("ndb_auth_comp_id_2"); Table->AddColumn("ndb_auth_seq_id_2"); Table->AddColumn("ndb_auth_asym_id_2"); Table->AddColumn("ndb_auth_ins_code_2"); Table->AddColumn("ndb_model_num"); Table->AddColumn("ndb_omega_angle"); irow = 0; for (i = 0; i < _num_mol; i++) { mol_id = _molecules[i]->Mol_ID(); chain = _molecules[i]->GetFirstChain(); while (chain) { if (chain->chain_type() != ATOMP_TOKEN && chain->chain_type() != ATOMN_TOKEN) { chain = _molecules[i]->GetNextChain(); continue; } prev = NULL; curr = chain->GetFirstResidue(); while (curr) { if (prev && curr) { at1 = prev->find_atom("CA", num1); at2 = prev->find_atom("C", num2); at3 = curr->find_atom("N", num3); at4 = curr->find_atom("CA", num4); if (at1 && at2 && at3 && at4 && is_a_bond(at2[0], at3[0])) { weight = cal_torsion(at1[0], at2[0], at3[0], at4[0]); weight = mod(weight, -180, 180); if (fabs(weight) <= 30.0) { Table->AddRow(); sprintf(buffer, "%d", irow+1); string.Copy(buffer); Table->UpdateCell(string, 0, irow); string.Copy(prev->ResName()); Table->UpdateCell(string, 1, irow); Table->UpdateCell(string, 6, irow); sprintf(buffer, "%d", prev->GetFirstAtom()->resnum()); string.Copy(buffer); Table->UpdateCell(string, 2, irow); string.Copy(chain->ChainID()); Table->UpdateCell(string, 3, irow); Table->UpdateCell(string, 12, irow); string.Copy("."); Table->UpdateCell(string, 4, irow); string.Copy(prev->GetFirstAtom()->ins_code()); Table->UpdateCell(string, 5, irow); Table->UpdateCell(string, 9, irow); sprintf(buffer, "%d", prev->GetFirstAtom()->pdb_resnum()); string.Copy(buffer); Table->UpdateCell(string, 7, irow); string.Copy(chain->PDB_ChainID()); Table->UpdateCell(string, 8, irow); Table->UpdateCell(string, 16, irow); string.Copy(curr->ResName()); Table->UpdateCell(string, 10, irow); Table->UpdateCell(string, 14, irow); sprintf(buffer, "%d", curr->GetFirstAtom()->resnum()); string.Copy(buffer); Table->UpdateCell(string, 11, irow); string.Copy(curr->GetFirstAtom()->ins_code()); Table->UpdateCell(string, 13, irow); Table->UpdateCell(string, 17, irow); sprintf(buffer, "%d", curr->GetFirstAtom()->pdb_resnum()); string.Copy(buffer); Table->UpdateCell(string, 15, irow); sprintf(buffer, "%d", mol_id); string.Copy(buffer); Table->UpdateCell(string, 18, irow); sprintf(buffer, "%.2f", weight); string.Copy(buffer); Table->UpdateCell(string, 19, irow); irow++; } } if (at1) delete [] at1; if (at2) delete [] at2; if (at3) delete [] at3; if (at4) delete [] at4; } prev = curr; curr = chain->GetNextResidue(); } chain = _molecules[i]->GetNextChain(); } } _CifObj->WriteTable(Table, _StructureId, "struct_mon_prot_cis"); for (k = 0; k < NUM_STRUCT_SITE; k++) { Table = _CifObj->GetTablePtr(_StructureId, _struct_sites[k].category); if (is_empty_table(Table)) continue; type_idx = Table->GetColumnIndex(_struct_sites[k].item); if (type_idx < 0) continue; Table1 = _CifObj->GetTablePtr(_StructureId, "struct_site"); if (is_empty_table(Table1) && Table1) { _CifObj->DeleteTable(_StructureId, "struct_site"); Table1 = NULL; } if (!Table1) { Table1 = new ISTable("struct_site"); Table1->AddColumn("id"); Table1->AddColumn("details"); } id_idx = Table1->GetColumnIndex("id"); update_index(Table, type_idx, Table1, id_idx); _CifObj->WriteTable(Table1, _StructureId, "struct_site"); } Table = _CifObj->GetTablePtr(_StructureId, "refine_B_iso"); if (!is_empty_table(Table)) { rowNo = Table->GetNumRows(); type_idx = Table->GetColumnIndex("treatment"); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "I") { string.Copy("isotropic"); Table->UpdateCell(string, type_idx, i); } else if (string == "A") { string.Copy("anisotropic"); Table->UpdateCell(string, type_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "refine_B_iso"); } Table = _CifObj->GetTablePtr(_StructureId, "refine_occupancy"); if (!is_empty_table(Table)) { rowNo = Table->GetNumRows(); type_idx = Table->GetColumnIndex("treatment"); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "OF") { string.Copy("fix"); Table->UpdateCell(string, type_idx, i); } else if (string == "OR") { string.Copy("ref"); Table->UpdateCell(string, type_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "refine_occupancy"); } Table3 = _CifObj->GetTablePtr(_StructureId, "ndb_poly_seq_scheme"); if (is_empty_table(Table3)) ndb_log_message(NDB_MSG_ERR, "%s: empty ndb_poly_seq_scheme category\n", _StructureId); else { char **tmp; PAIR_INDEX **tmp_index; CIF_PDB_PAIR **tmp_pair; scheme = new SCHEME_TABLE; scheme->num = 0; scheme->chnid = NULL; scheme->index = NULL; rowNo = Table3->GetNumRows(); chn_idx = Table3->GetColumnIndex("pdb_id"); seq_beg = Table3->GetColumnIndex("seq_id"); seq_end = Table3->GetColumnIndex("pdb_num"); strcpy(buffer, "XXX"); j = 0; for (i = 0; i < rowNo; i++) { get_value(string, Table3, chn_idx, i); strcpy(chnid, string.Text()); get_value(beg, Table3, seq_beg, i); get_value(end, Table3, seq_end, i); if (strcmp(buffer, chnid)) { strcpy(buffer, chnid); j = scheme->num; scheme->num++; tmp = scheme->chnid; scheme->chnid = new char*[scheme->num]; for (k = 0; k < j; k++) scheme->chnid[k] = tmp[k]; delete [] tmp; scheme->chnid[j] = new char[5]; strcpy(scheme->chnid[j], chnid); tmp_index = scheme->index; scheme->index = new PAIR_INDEX*[scheme->num]; for (k = 0; k < j; k++) scheme->index[k] = tmp_index[k]; delete [] tmp_index; scheme->index[j] = new PAIR_INDEX; scheme->index[j]->length = 0; scheme->index[j]->pair = NULL; } scheme->index[j]->length++; tmp_pair = scheme->index[j]->pair; scheme->index[j]->pair = new CIF_PDB_PAIR*[scheme->index[j]->length]; for (k = 0; k < scheme->index[j]->length - 1; k++) scheme->index[j]->pair[k] = tmp_pair[k]; delete [] tmp_pair; scheme->index[j]->pair[scheme->index[j]->length-1] = new CIF_PDB_PAIR; strcpy(scheme->index[j]->pair[scheme->index[j]->length-1]->cif, beg.Text()); strcpy(scheme->index[j]->pair[scheme->index[j]->length-1]->pdb, end.Text()); } for (i = 0; i < scheme->num; i++) { found_gap = 0; last = -1; for (j = 0; j < scheme->index[i]->length; j++) { if (!strcmp(scheme->index[i]->pair[j]->pdb, "") || !strcmp(scheme->index[i]->pair[j]->pdb, "?")) found_gap = 1; else { last = j; if (found_gap) { seq_num = atoi(scheme->index[i]->pair[j]->pdb); for (k = j - 1; k >= 0; k--) { if (strcmp(scheme->index[i]->pair[k]->pdb, "") && strcmp(scheme->index[i]->pair[k]->pdb, "?")) break; seq_num--; sprintf(scheme->index[i]->pair[k]->pdb, "%d", seq_num); } found_gap = 0; } } } if (found_gap && last >= 0) { seq_num = atoi(scheme->index[i]->pair[last]->pdb); for (k = last + 1; k < scheme->index[i]->length; k++) { if (strcmp(scheme->index[i]->pair[k]->pdb, "") && strcmp(scheme->index[i]->pair[k]->pdb, "?")) break; seq_num++; sprintf(scheme->index[i]->pair[k]->pdb, "%d", seq_num); } } } /* if (standardized) { rowNo = Table3->GetNumRows(); colNo = Table3->GetNumColumns(); for (i = 0; i < rowNo; i++) { for (j = 0; j < colNo; j++) { Table3->GetCell(string, j, i); if (string == "...") { string.Copy("n/a"); Table3->UpdateCell(string, j, i); } } } _CifObj->WriteTable(Table3, _StructureId, "ndb_poly_seq_scheme"); } */ } Table = _CifObj->GetTablePtr(_StructureId, "struct_ref"); Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq"); if (!is_empty_table(Table) && !is_empty_table(Table1) && scheme) { Table2 = get_extend_table(Table1, &_Struct_Ref_Seq); if (Table2) { _CifObj->DeleteTable(_StructureId, _Struct_Ref_Seq.category); _CifObj->WriteTable(Table2, _StructureId, _Struct_Ref_Seq.category); Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq"); } auth_seq_beg = Table1->GetColumnIndex("ndb_auth_seq_align_beg"); auth_beg_ins = Table1->GetColumnIndex("ndb_seq_align_beg_ins_code"); auth_seq_end = Table1->GetColumnIndex("ndb_auth_seq_align_end"); auth_end_ins = Table1->GetColumnIndex("ndb_seq_align_end_ins_code"); seq_beg = Table1->GetColumnIndex("seq_align_beg"); seq_end = Table1->GetColumnIndex("seq_align_end"); chn_idx = Table1->GetColumnIndex("ndb_chain_id"); rowNo = Table1->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(beg, Table1, seq_beg, i); Table1->UpdateCell(beg, auth_seq_beg, i); get_value(string, Table1, auth_beg_ins, i); if (string != "") beg += string; get_value(end, Table1, seq_end, i); Table1->UpdateCell(end, auth_seq_end, i); get_value(string, Table1, auth_end_ins, i); if (string != "") end += string; get_value(string, Table1, chn_idx, i); if (beg != "" && find_index(scheme, string.Text(), beg) && end != "" && find_index(scheme, string.Text(), end)) { Table1->UpdateCell(beg, seq_beg, i); Table1->UpdateCell(end, seq_end, i); } else { ndb_log_message(NDB_MSG_ERR, "%s: can't find CIF index for %s %s %s\n", _StructureId, string.Text(), beg.Text(), end.Text()); } } _CifObj->WriteTable(Table1, _StructureId, "struct_ref_seq"); } Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq_dif"); if (!is_empty_table(Table) && !is_empty_table(Table1) && scheme) { Table2 = get_extend_table(Table1, &_Struct_Ref_Seq_Dif); if (Table2) { _CifObj->DeleteTable(_StructureId, _Struct_Ref_Seq_Dif.category); _CifObj->WriteTable(Table2, _StructureId, _Struct_Ref_Seq_Dif.category); Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq_dif"); } auth_seq_beg = Table1->GetColumnIndex("ndb_auth_seq_num"); seq_beg = Table1->GetColumnIndex("seq_num"); chn_idx = Table1->GetColumnIndex("ndb_pdb_chain_id"); rowNo = Table1->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(beg, Table1, seq_beg, i); Table1->UpdateCell(beg, auth_seq_beg, i); if (beg == "") continue; get_value(string, Table1, chn_idx, i); if (find_index(scheme, string.Text(), beg)) Table1->UpdateCell(beg, seq_beg, i); else ndb_log_message(NDB_MSG_ERR, "%s: can't find CIF index for %s %s\n", _StructureId, string.Text(), beg.Text()); } _CifObj->WriteTable(Table1, _StructureId, "struct_ref_seq_dif"); } Table = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq"); Table1 = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq_dif"); if (!is_empty_table(Table) && !is_empty_table(Table1)) { rowNo = Table->GetNumRows(); rowNo1 = Table1->GetNumRows(); align_id = Table->GetColumnIndex("align_id"); align_id1 = Table1->GetColumnIndex("align_id"); chn_idx = Table->GetColumnIndex("ndb_chain_id"); id_idx = Table1->GetColumnIndex("ndb_pdb_chain_id"); for (i = 0; i < rowNo1; i++) { Table1->GetCell(id, id_idx, i); for (j = 0; j < rowNo; j++) { Table->GetCell(string, chn_idx, j); if (string == id) { Table->GetCell(string, align_id, j); Table1->UpdateCell(string, align_id1, i); break; } } } _CifObj->WriteTable(Table1, _StructureId, "struct_ref_seq_dif"); } if (scheme) { if (scheme->chnid) { for (i = 0; i < scheme->num; i++) if (scheme->chnid[i]) delete [] scheme->chnid[i]; delete [] scheme->chnid; } if (scheme->index) { for (i = 0; i < scheme->num; i++) { if (scheme->index[i]) { if (scheme->index[i]->pair) { for (j = 0; j < scheme->index[i]->length; j++) { if (scheme->index[i]->pair[j]) delete scheme->index[i]->pair[j]; } delete [] scheme->index[i]->pair; } delete scheme->index[i]; } } delete [] scheme->index; } delete scheme; scheme = NULL; } Table = _CifObj->GetTablePtr(_StructureId, "database_PDB_rev"); if (!is_empty_table(Table)) { type_idx = Table->GetColumnIndex("status"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "") continue; string.ToLower(); if (string == "y") { string.Copy("full release"); Table->UpdateCell(string, type_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "database_PDB_rev"); } for (k = 0; k < NUM_CAT_ITEM_TO_LOWER; k++) { Table = _CifObj->GetTablePtr(_StructureId, _cat_items_to_lower[k].category); if (is_empty_table(Table)) continue; type_idx = Table->GetColumnIndex(_cat_items_to_lower[k].item); if (type_idx < 0) continue; rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string == "") continue; string.ToLower(); Table->UpdateCell(string, type_idx, i); } _CifObj->WriteTable(Table, _StructureId, _cat_items_to_lower[k].category); } Table = _CifObj->GetTablePtr(_StructureId, "refine_ls_shell"); if (!is_empty_table(Table)) { rowNo = Table->GetNumRows(); type_idx = Table->GetColumnIndex("ndb_total_number_of_bins_used"); id_idx = Table->GetColumnIndex("number_reflns_R_free"); text.Clear(); for (i = 0; i < rowNo; i++) { get_value(string, Table, type_idx, i); if (string != "") { last = atoi(string.Text()); sprintf(buffer, "%d", last); text.Copy(buffer); Table->UpdateCell(text, type_idx, i); } else if (text != "") Table->UpdateCell(text, type_idx, i); get_value(string, Table, id_idx, i); if (string != "") { last = atoi(string.Text()); sprintf(buffer, "%d", last); string.Copy(buffer); Table->UpdateCell(string, id_idx, i); } } _CifObj->WriteTable(Table, _StructureId, "refine_ls_shell"); } Table = _CifObj->GetTablePtr(_StructureId, "database_PDB_caveat"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("id"); rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { sprintf(buffer, "%d", i + 1); string.Copy(buffer); Table->UpdateCell(string, id_idx, i); } _CifObj->WriteTable(Table, _StructureId, "database_PDB_caveat"); } for (i = 0; i < NUM_PDB_INSERT_CODE; i++) { Table = _CifObj->GetTablePtr(_StructureId, _pdb_insert_code[i].category); if (is_empty_table(Table)) continue; rowNo = Table->GetNumRows(); for (j = 0; j < _pdb_insert_code[i].num_items; j++) { auth_seq_beg = Table->GetColumnIndex(_pdb_insert_code[i].items[2 * j]); auth_beg_ins = Table->GetColumnIndex(_pdb_insert_code[i].items[2*j+1]); if (auth_seq_beg >= 0 && auth_beg_ins >= 0) { for (k = 0; k < rowNo; k++) { Table->GetCell(string, auth_seq_beg, k); Table->GetCell(id, auth_beg_ins, k); if (string != "" && string != "." && string != "?" && id != "" && id != "." && id != "?") { string += id; Table->UpdateCell(string, auth_seq_beg, k); } } } } _CifObj->WriteTable(Table, _StructureId, _pdb_insert_code[i].category); } for (i = 0; i < NUM_TABLELIST; i++) { Table = _CifObj->GetTablePtr(_StructureId, _table_list[i]); if (!Table) continue; ColumnNames = Table->GetColumnNames(); colNo = Table->GetNumColumns(); rowNo = Table->GetNumRows(); for (j = 0; j < rowNo; j++) { if (is_empty_row(Table, ColumnNames, colNo, j, "entry_id")) { /* Table->RemoveRow(j); j--; rowNo--; */ Table->DeleteRow(j); } } Table->CompressTable(); if (ColumnNames) delete ColumnNames; rowNo = Table->GetNumRows(); if (!rowNo) _CifObj->DeleteTable(_StructureId, _table_list[i]); else _CifObj->WriteTable(Table, _StructureId, _table_list[i]); } for (i = 0; i < NUM_EM_CATEGORYS; i++) { Table = _CifObj->GetTablePtr(_StructureId, em_categories[i].category); if (is_empty_table(Table)) continue; rowNo = Table->GetNumRows(); for (j = 0; j < em_categories[i].NumItems; j++) { id_idx = Table->GetColumnIndex(em_categories[i].items[j].item); if (id_idx < 0) continue; for (k = 0; k < rowNo; k++) { Table->GetCell(string, id_idx, k); if (string != "?" && string != "" && string != ".") continue; if (em_categories[i].items[j].is_entry_id == 1) string.Copy(_StructureId); else if (em_categories[i].items[j].is_entry_id == 2) string.Copy("1"); else { sprintf(buffer, "%d", k + 1); string.Copy(buffer); } Table->UpdateCell(string, id_idx, k); } } _CifObj->WriteTable(Table, _StructureId, em_categories[i].category); } Table = _CifObj->GetTablePtr(_StructureId, "struct_ref_seq_dif"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("rcsb_ordinal"); if (id_idx < 0) { rowNo = Table->GetNumRows(); colNo = Table->GetNumColumns(); ColumnNames = Table->GetColumnNames(); Table1 = new ISTable("struct_ref_seq_dif"); for (i = 0; i < colNo; i++) { Table1->AddColumn((*ColumnNames)[i].Text()); } if (ColumnNames) delete ColumnNames; Table1->AddColumn("rcsb_ordinal"); id_idx = Table1->GetColumnIndex("rcsb_ordinal"); for (i = 0; i < rowNo; i++) { Table1->AddRow(); for (j = 0; j < colNo; j++) { Table->GetCell(string, j, i); Table1->UpdateCell(string, j, i); } sprintf(buffer, "%d", i+1); string.Copy(buffer); Table1->UpdateCell(string, id_idx, i); } _CifObj->DeleteTable(_StructureId, "struct_ref_seq_dif"); _CifObj->WriteTable(Table1, _StructureId, "struct_ref_seq_dif"); } } Table = _CifObj->GetTablePtr(_StructureId, "citation_author"); if (!is_empty_table(Table)) { id_idx = Table->GetColumnIndex("ordinal"); rowNo = Table->GetNumRows(); if (id_idx < 0) { colNo = Table->GetNumColumns(); ColumnNames = Table->GetColumnNames(); Table1 = new ISTable("citation_author"); for (i = 0; i < colNo; i++) { Table1->AddColumn((*ColumnNames)[i].Text()); } if (ColumnNames) delete ColumnNames; Table1->AddColumn("ordinal"); id_idx = Table1->GetColumnIndex("ordinal"); for (i = 0; i < rowNo; i++) { Table1->AddRow(); for (j = 0; j < colNo; j++) { Table->GetCell(string, j, i); Table1->UpdateCell(string, j, i); } } _CifObj->DeleteTable(_StructureId, "citation_author"); Table = Table1; } for (i = 0; i < rowNo; i++) { sprintf(buffer, "%d", i+1); string.Copy(buffer); Table->UpdateCell(string, id_idx, i); } _CifObj->WriteTable(Table, _StructureId, "citation_author"); } if (standardized) { AddMandatoryItem(_CifObj, _experiment_type); _CifObj->DeleteTable(_StructureId, "audit_conform"); Table = new ISTable("audit_conform"); Table->AddColumn("dict_name"); Table->AddColumn("dict_version"); Table->AddColumn("dict_location"); Table->AddRow(); string.Copy("cif_mm.dic"); Table->UpdateCell(string, 0, 0); string.Copy("2.0.03"); Table->UpdateCell(string, 1, 0); string.Copy("http://pdb.rutgers.edu/mmcif/dictionaries/ascii/cif_mm.dic"); Table->UpdateCell(string, 2, 0); Table->AddRow(); string.Copy("cif_pdbx.dic"); Table->UpdateCell(string, 0, 1); string.Copy("0.8.07"); Table->UpdateCell(string, 1, 1); string.Copy("http://pdb.rutgers.edu/mmcif/dictionaries/ascii/pdbx_exchange.dic"); Table->UpdateCell(string, 2, 1); _CifObj->WriteTable(Table, _StructureId, "audit_conform"); _CifObj->DeleteTable(_StructureId, "audit"); Table = new ISTable("audit"); Table->AddColumn("revision_id"); Table->AddColumn("creation_date"); Table->AddColumn("update_record"); Table1 = _CifObj->GetTablePtr(_StructureId, "database_PDB_rev"); if (is_empty_table(Table1)) { Table->AddRow(); id.Copy("1"); Table->UpdateCell(id, 0, 0); get_date(buffer); string.Copy(buffer); Table->UpdateCell(string, 1, 0); string.Copy("initial release"); Table->UpdateCell(string, 2, 0); } else { rowNo = Table1->GetNumRows(); id_idx = Table1->GetColumnIndex("num"); mon_id = Table1->GetColumnIndex("date"); rowNo1 = 0; Table2 = _CifObj->GetTablePtr(_StructureId, "database_PDB_rev_record"); if (!is_empty_table(Table2)) { name_idx = Table2->GetColumnIndex("rev_num"); type_idx = Table2->GetColumnIndex("type"); rowNo1 = Table2->GetNumRows(); } for (i = 0; i < rowNo; i++) { Table->AddRow(); Table1->GetCell(id, id_idx, i); Table->UpdateCell(id, 0, i); Table1->GetCell(string, mon_id, i); Table->UpdateCell(string, 1, i); if (!i) string.Copy("initial release"); else { strcpy(buffer, "adjusted to"); for (j = 0; j < rowNo1; j++) { Table2->GetCell(string, name_idx, j); if (string == id) { Table2->GetCell(string, type_idx, j); if (string == "HEADER" || string == "KEYWDS") { if (!strstr(buffer, " struct_keywords category,")) strcat(buffer, " struct_keywords category,"); } else if (string == "TITLE") { if (!strstr(buffer, " struct category,")) strcat(buffer, " struct category,"); } else if (string == "OBSLTE" || string == "SPRSDE") { if (!strstr(buffer, " pdbx_database_PDB_obs_spr category,")) strcat(buffer, " pdbx_database_PDB_obs_spr category,"); } else if (string == "SOURCE" || string == "COMPND") { if (!strstr(buffer, " source and compound information,")) strcat(buffer, " source and compound information,"); } else if (string == "JRNL") { if (!strstr(buffer, " citation information,")) strcat(buffer, " citation information,"); } else if (string == "REMARK") { if (!strstr(buffer, " citation information,")) strcat(buffer, " citation information,"); } else if (string == "DBREF" || string == "SEQADV" || string == "SEQRES" || string == "MODRES") { if (!strstr(buffer, " sequence information,")) strcat(buffer, " sequence information,"); } else if (string == "HET" || string == "HETNAM" || string == "HETSYN" || string == "FORMUL") { if (!strstr(buffer, " chem_comp category,")) strcat(buffer, " chem_comp category,"); } else if (string == "HELIX" || string == "SHEET" || string == "TURN") { if (!strstr(buffer, " secondary structure information,")) strcat(buffer, " secondary structure information,"); } else if (string == "SSBOND" || string == "LINK" || string == "CONECT" || string == "HYDBND" || string == "SLTBRG") { if (!strstr(buffer, " connectivity information,")) strcat(buffer, " connectivity information,"); } else if (string == "CRYST1" || string == "ORIGX" || string == "SCALE" || string == "MTRIX" || string == "TVECT") { if (!strstr(buffer, " crystallographic transformation,")) strcat(buffer, " crystallographic transformation,"); } else if (string == "ATOM" || "HETATM") { if (!strstr(buffer, " coordinates,")) strcat(buffer, " coordinates,"); } } } if (!strcmp(buffer, "adjusted to")) strcpy(buffer, ""); else buffer[strlen(buffer)-1] = '.'; string.Copy(buffer); } Table->UpdateCell(string, 2, i); } } _CifObj->WriteTable(Table, _StructureId, "audit"); _CifObj->DeleteTable(_StructureId, "pdbx_audit"); Table = new ISTable("pdbx_audit"); Table->AddColumn("entry_id"); Table->AddColumn("current_version"); Table->AddRow(); string.Copy(_StructureId); Table->UpdateCell(string, 0, 0); Table->UpdateCell(id, 1, 0); _CifObj->WriteTable(Table, _StructureId, "pdbx_audit"); _CifObj->DeleteTable(_StructureId, "ndb_database_proc"); if (!keep_contact_author) _CifObj->DeleteTable(_StructureId, "audit_contact_author"); _CifObj->DeleteTable(_StructureId, "ndb_database_remark"); _CifObj->DeleteTable(_StructureId, "chem_comp_atom"); categoryName = _CifObj->GetTableNames(numCategory); for (i = 0; i < numCategory; i++) { Table = _CifObj->GetTablePtr(_StructureId, categoryName[i]); if (Table) { if (is_empty_table(Table)) _CifObj->DeleteTable(_StructureId, categoryName[i]); else if (!strcasecmp(categoryName[i], "entity_name_com") || !strcasecmp(categoryName[i], "entity_keywords") || !strcasecmp(categoryName[i], "entity_name_sys")) { rowNo = Table->GetNumRows(); colNo = Table->GetNumColumns(); id_idx = Table->GetColumnIndex("entity_id"); for (j = 0; j < rowNo; j++) { found_gap = 0; for (k = 0; k < colNo; k++) { Table->GetCell(string, k, j); if (string != "?" && string != "." && string != "" && string != "@") { if (k == id_idx) continue; found_gap = 1; break; } } if (found_gap) break; } if (j == rowNo) _CifObj->DeleteTable(_StructureId, categoryName[i]); } } delete categoryName[i]; } if (categoryName) delete categoryName; } categoryName = _CifObj->GetTableNames(numCategory); for (i = 0; i < numCategory; i++) { Table = _CifObj->GetTablePtr(_StructureId, categoryName[i]); if (!is_empty_table(Table) && strcasecmp(categoryName[i], "ndb_pdb_compnd") && strcasecmp(categoryName[i], "ndb_pdb_source") && strcasecmp(categoryName[i], "database_PDB_remark") && strcasecmp(categoryName[i], "ndb_database_remark") && strcasecmp(categoryName[i], "atom_site") && strcasecmp(categoryName[i], "struct_ref") && strcasecmp(categoryName[i], "entity_poly") && strcasecmp(categoryName[i], "atom_site_anisotrop") && strcasecmp(categoryName[i], "ndb_original_ndb_coordinates") && strcasecmp(categoryName[i], "ndb_original_pdb_coordinates") && strcasecmp(categoryName[i], "rcsb_database_PDB_remark")) { colNo = Table->GetNumColumns(); rowNo = Table->GetNumRows(); for (j = 0; j < rowNo; j++) { for (k = 0; k < colNo; k++) { Table->GetCell(string, k, j); ndb_strip_leading_blanks(string.Text()); ndb_strip_trailing_blanks(string.Text()); Table->UpdateCell(string, k, j); } } _CifObj->WriteTable(Table, _StructureId, categoryName[i]); } delete [] categoryName[i]; } if (categoryName) delete [] categoryName; } static void remove_infinite(ISTable *Table, const int id_idx) { CifString cifstring; get_value(cifstring, Table, id_idx, 0); if (strstr(cifstring.Text(), "INFINI")) { cifstring.Copy(""); Table->UpdateCell(cifstring, id_idx, 0); } } static void remove_comma(ISTable *Table, const int id_idx) { unsigned int i, j; char buffer[20]; CifString cifstring; get_value(cifstring, Table, id_idx, 0); if (strstr(cifstring.Text(), ",")) { j = 0; for (i = 0; i < strlen(cifstring.Text()); i++) { if (isdigit(cifstring.Text()[i])) { buffer[j] = cifstring.Text()[i]; j++; } } buffer[j] = '\0'; cifstring.Copy(buffer); Table->UpdateCell(cifstring, id_idx, 0); } } static void remove_k(ISTable *Table, const int col, const int row) { char buffer[100], *p; CifString cifstring; get_value(cifstring, Table, col, row); if (cifstring != "") { strcpy(buffer, cifstring.Text()); p = strstr(buffer, "K"); if (p) *p = '\0'; p = strstr(buffer, "k"); if (p) *p = '\0'; cifstring.Copy(buffer); Table->UpdateCell(cifstring, col, row); } } static void update_index(ISTable *Table, const int idx, ISTable *Table1, const int idx1) { int i, j, rowNo, irow; CifString cifstring, beg; rowNo = Table->GetNumRows(); for (i = 0; i < rowNo; i++) { get_value(cifstring, Table, idx, i); if (cifstring == "") continue; irow = Table1->GetNumRows(); for (j = 0; j < irow; j++) { get_value(beg, Table1, idx1, j); if (beg == cifstring) break; if (beg == "") { Table1->UpdateCell(cifstring, idx1, j); break; } } if (j < irow) continue; Table1->AddRow(); Table1->UpdateCell(cifstring, idx1, irow); } } static ISTable *get_extend_table(ISTable *Table, _PDB_INSERT_CODE *Miss_Man_I_cat) { int i, j, colNo, rowNo, idx[NUM_ITEM]; CifString cifstring; ISTable *Table1 = NULL; ReVarCifArray *ColumnNames = NULL; for (i = 0; i < Miss_Man_I_cat->num_items; i++) { idx[i] = Table->GetColumnIndex(Miss_Man_I_cat->items[i]); } for (i = 0; i < Miss_Man_I_cat->num_items; i++) { if (idx[i] < 0) break; } if (i == Miss_Man_I_cat->num_items) return NULL; ColumnNames = Table->GetColumnNames(); colNo = Table->GetNumColumns(); rowNo = Table->GetNumRows(); Table1 = new ISTable(Miss_Man_I_cat->category); for (i = 0; i < colNo; i++) Table1->AddColumn((*ColumnNames)[i].Text()); if (ColumnNames) delete ColumnNames; for (i = 0; i < Miss_Man_I_cat->num_items; i++) { if (idx[i] < 0) Table1->AddColumn(Miss_Man_I_cat->items[i]); } for (i = 0; i < rowNo; i++) { Table1->AddRow(); for (j = 0; j < colNo; j++) { Table->GetCell(cifstring, j, i); Table1->UpdateCell(cifstring, j, i); } } return Table1; } static int find_index(SCHEME_TABLE *scheme, const char *chnid, CifString &num) { int i, j; for (i = 0; i < scheme->num; i++) { if (!strcmp(chnid, scheme->chnid[i])) { for (j = 0; j < scheme->index[i]->length; j++) { if (num == scheme->index[i]->pair[j]->pdb) { num.Copy(scheme->index[i]->pair[j]->cif); return 1; } } } } return 0; } static void AddMandatoryItem(CifFileObj *in, const int experiment_type) { int i, j, k, l, m, refItemIndex = -1, numBlocks, numTables, errCode; int colNo, rowNo, is_changed = 0, ItemTypeIndex, ItemTypeLIndex; int iOut, errCodeRegex, icmp = -1, ret, len; char binaryfile[300], **BlockNames = NULL, **TableNames = NULL, keyword[200]; char *p = NULL, *pm = NULL; CifString cifstring, cell, typeCode, patternString; DICFileObj *dict = NULL; ISTable *refItem = NULL, *Category = NULL, *Table = NULL, *ItemType = NULL, *ItemTypeL = NULL; regex_t preg; regmatch_t pmatch[NS]; ReVarCifArray refItemList, refItemTarget, MissingItems, *ColumnNames = NULL; ReVarCifArray ItemTypeLList, *ItemNames = NULL, target, nameList, ItemTypeLTarget; ReVarPCifArray *OutList = NULL; struct stat stattext; if (experiment_type == EXPERIMENT_TYPE_NMR || experiment_type == EXPERIMENT_TYPE_MODEL) sprintf(binaryfile, "%s/data/binary/%s", root_dir, dictnmrbinfile); else sprintf(binaryfile, "%s/data/binary/%s", root_dir, dictxraybinfile); if (stat(binaryfile, &stattext) != -1) dict = new DICFileObj(binaryfile, READ_MODE); if (dict == NULL) return; nameList.Clear(); nameList.Add("name"); refItem = dict->GetTablePtr("item"); refItemList.Clear(); refItemList.Add("category_id"); refItemList.Add("mandatory_code"); refItem->CreateIndex("index0", refItemList); refItemIndex = refItem->GetColumnIndex("name"); ItemType = dict->GetTablePtr("item_type"); ItemTypeIndex = ItemType->GetColumnIndex("code"); ItemTypeL = dict->GetTablePtr("item_type_list"); ItemTypeLList.Clear(); ItemTypeLList.Add("code"); ItemTypeLIndex = ItemTypeL->GetColumnIndex("construct"); BlockNames = in->GetBlockNames(numBlocks); for (i = 0; i < numBlocks; i++) { TableNames = in->GetTableNames(BlockNames[i],numTables); for (j = 0; j < numTables; j++) { if (!strcmp(TableNames[j], "atom_site") || !strcmp(TableNames[j], "citation") || !strcmp(TableNames[j], "atom_site_anisotrop")) { delete [] TableNames[j]; continue; } Category = in->GetTablePtr(BlockNames[i], TableNames[j]); if (is_empty_table(Category)) { delete [] TableNames[j]; continue; } refItemTarget.Clear(); refItemTarget.Add(TableNames[j]); refItemTarget.Add("yes"); OutList = refItem->Search(refItemTarget, refItemList, errCode); MissingItems.Clear(); if (errCode >= 0) { for (k = 0; k < (int) OutList->Length(); k++) { refItem->GetCell(cell, refItemIndex, (*OutList)[k]); GetKeywordFromItemName(keyword,cell.Text()); errCode = Category->GetColumnIndex(keyword); if (errCode < 0) { cifstring.Copy(keyword); MissingItems.Add(cifstring); } } } if (OutList) delete OutList; OutList = NULL; is_changed = 0; if (MissingItems.Length() > 0) { Table = Category; ColumnNames = Table->GetColumnNames(); colNo = Table->GetNumColumns(); rowNo = Table->GetNumRows(); Category = new ISTable(TableNames[j]); for (l = 0; l < colNo; l++) Category->AddColumn((*ColumnNames)[l].Text()); if (ColumnNames) delete ColumnNames; for (k = 0; k < (int) MissingItems.Length(); k++) Category->AddColumn(MissingItems[k].Text()); for (l = 0; l < rowNo; l++) { Category->AddRow(); for (m = 0; m < colNo; m++) { Table->GetCell(cifstring, m, l); Category->UpdateCell(cifstring, m, l); } } for (k = 0; k < (int) MissingItems.Length(); k++) { if (MissingItems[k] == "entry_id") cifstring.Copy(BlockNames[i]); else cifstring.Copy("."); for (l = 0; l < rowNo; l++) { Category->UpdateCell(cifstring, colNo + k, l); } } in->DeleteTable(BlockNames[i], TableNames[j]); is_changed = 1; } ItemNames = Category->GetItemNames(); for (k = 0; k < (int) ItemNames->Length(); k++) { target.Clear(); target.Add((*ItemNames)[k]); typeCode.Clear(); iOut = ItemType->FindFirst(target, nameList, errCodeRegex); if (errCodeRegex >= 0) ItemType->GetCell(typeCode, ItemTypeIndex, iOut); if (typeCode != "float" && typeCode != "int") continue; ItemTypeLTarget.Clear(); ItemTypeLTarget.Add(typeCode); iOut = ItemTypeL->FindFirst(ItemTypeLTarget, ItemTypeLList, errCode); ItemTypeL->GetCell(cell, ItemTypeLIndex, iOut); ConvertEscapedString(cell, patternString); icmp = regcomp(&preg, patternString.Text(), REG_EXTENDED); regfree(&preg); if (icmp != 0) continue; rowNo = Category->GetNumRows(); for (l = 0; l < rowNo; l++){ Category->GetCell(cifstring, k, l); if (cifstring == "." || cifstring == "?" || cifstring == "") continue; regcomp(&preg, patternString.Text(), REG_EXTENDED); ret = regexec(&preg, cifstring.Text(), NS, pmatch, 0); regfree(&preg); if (ret != 0) { cifstring.Copy("."); Category->UpdateCell(cifstring, k, l); is_changed = 1; } else { len = pmatch[0].rm_eo - pmatch[0].rm_so; if (len != (int) strlen(cifstring.Text())) { p = (char*)malloc(pmatch[0].rm_eo+1); strncpy(p, cifstring.Text(), pmatch[0].rm_eo); p[pmatch[0].rm_eo]='\0'; pm=p+pmatch[0].rm_so; cifstring.Copy(pm); Category->UpdateCell(cifstring, k, l); free (p); is_changed = 1; } } } } if (ItemNames) delete ItemNames; if (is_changed) in->WriteTable(Category, BlockNames[i], TableNames[j]); if (TableNames[j]) delete[] TableNames[j]; } if (TableNames) delete[] TableNames; if (BlockNames[i]) delete[] BlockNames[i]; } if (BlockNames) delete[] BlockNames; delete dict; } static void GetKeywordFromItemName(char *keyword, const char *itemName) { int i, j, k, ilen; strcpy(keyword, ""); if (!itemName || (ilen = strlen(itemName)) < 3) return; for (k = 1; k < ilen; k++) { if (itemName[k] == '.') break; } if (k == 1 || k == ilen) return; j = 0; for (i = k + 1; i < ilen; i++) { keyword[j] = itemName[i]; j++; } keyword[j] = '\0'; } static void ConvertEscapedString(CifString& inString, CifString& outString) { int tmpPos = 0; unsigned int posI; char* tmpStringP = NULL; tmpStringP = (char*) malloc(inString.Length() + 1); for (posI = 0; posI < inString.Length(); tmpPos++) { if ((inString.Text())[posI] == '\\') { switch ((inString.Text())[posI + 1]) { case '\\': tmpStringP[tmpPos] = '\\'; posI += 2; break; case 't': tmpStringP[tmpPos] = '\t'; posI += 2; break; case 'n': tmpStringP[tmpPos] = '\n'; posI += 2; break; default: tmpStringP[tmpPos] = (inString.Text())[posI]; posI++; break; } } else { tmpStringP[tmpPos] = (inString.Text())[posI]; posI++; } } tmpStringP[tmpPos] = '\0'; outString.Copy(tmpStringP); free (tmpStringP); }