/* FILE: pdb_to_ndb.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 "_maxit.h" #include "erf_util.h" #include "ndb_misclib.h" #include "util.h" #include "remark_100s.h" #include "remark_200s.h" #include "remark_300s.h" #include "remark_general.h" #include "remark_3_prolsqs.h" #include "remark_3_refmac5.h" #include "remark_3_shelxl.h" #include "remark_3_tnt.h" #include "remark_3_xplors.h" #include "remark_cryoem.h" #include "remark_solution.h" #include "remark_olds.h" #define NUM_REMARK 17 static int remark_nums[NUM_REMARK] = { 2, 3, 18, 200, 210, 215, 220, 225, 230, 240, 245, 250, 265, 280, 300, 350, 800 }; static REMARKS *old_remarks[8] = { Remark_3_xplor_old, Remark_3_xplor_old, Remark_3_nuclsq_old, Remark_3_prolsq_old, Remark_3_prolsq_old, Remark_3_prolsq_old, Remark_3_tnt_old, Remark_3_shelx_old }; static int num_old_remarks[8] = { Num_Remark_3_xplor_old, Num_Remark_3_xplor_old, Num_Remark_3_nuclsq_old, Num_Remark_3_prolsq_old, Num_Remark_3_prolsq_old, Num_Remark_3_prolsq_old, Num_Remark_3_tnt_old, Num_Remark_3_shelx_old }; static int find_a_match(const char *context, REMARKS *remark); static void clean_middle_space(char *string); static int string_compare(const char *remark, const char *string, const int pos, const int len); static void get_remark_answer(char *answer, One_Card_Info_Format *remarkCard, REMARKS *remark, int field_no); static void get_name_and_version(char *name, char *version, char *program); static void clean_author_name(char *authors, char *name); void _MAXIT::pdb_to_ndb_general(const char *ndb_sid, const int is_encapsulated) { pdb_to_ndb_proc_header(ndb_sid); pdb_to_ndb_proc_compnd_and_source(is_encapsulated); pdb_to_ndb_proc_remarks(); pdb_to_ndb_postprocessing(); } void _MAXIT::pdb_to_ndb_proc_header(const char *ndb_sid) { char buffer[50]; One_Card_Info_Format *pCard = NULL, *qCard = NULL; pCard = _CardInfo[HEADER_TOKEN-1].Head(); if (pCard == NULL) pCard = _CardInfo[HEADER_TOKEN-1]. add_a_new_card(NdbToken[HEADER_TOKEN-1], 0); if (_num_mol && !strcmp(pCard->Field(1), "")) { if (_molecules[0]->type() == ATOMN_TYPE_DNA_RNA) { pCard->AddElementToField(NDB_HYBRID_TEXT, 1); } else if (_molecules[0]->type() == ATOM_TYPE_TRNA) { pCard->AddElementToField(NDB_T_RNA_TEXT, 1); } else if (_molecules[0]->type() & ATOMN_TYPE_RNA_ONLY) { if (_molecules[0]->type() & ATOM_TYPE_PROTEIN) { sprintf(buffer, "PROTEIN/%s", NDB_RNA_TEXT); pCard->AddElementToField(buffer, 1); } else pCard->AddElementToField(NDB_RNA_TEXT, 1); } else if (_molecules[0]->type() & ATOM_TYPE_DNA_ONLY) { if (_molecules[0]->type() & ATOM_TYPE_PROTEIN) { sprintf(buffer, "PROTEIN/%s", NDB_DNA_TEXT); pCard->AddElementToField(buffer, 1); } else pCard->AddElementToField(NDB_DNA_TEXT, 1); } } if (!strcmp(pCard->Field(1), "DEOXYRIBONUCLEIC ACID")) pCard->AddElementToField(NDB_DNA_TEXT, 1); else if (!strcmp(pCard->Field(1), "RIBONUCLEIC ACID")) pCard->AddElementToField(NDB_RNA_TEXT, 1); else if (!strcmp(pCard->Field(1), "AMINO-ACID TRANSPORT")) pCard->AddElementToField(NDB_T_RNA_TEXT, 1); if (strcmp(pCard->Field(3), "")) { qCard = _CardInfo[PDBFIL_TOKEN-1].Head(); if (qCard == NULL) qCard = _CardInfo[PDBFIL_TOKEN-1]. add_a_new_card(NdbToken[PDBFIL_TOKEN-1], 0); qCard->AddElementToField(pCard->Field(3), 1); } qCard = _CardInfo[NDBFIL_TOKEN-1].Head(); if (qCard == NULL) qCard = _CardInfo[NDBFIL_TOKEN-1]. add_a_new_card(NdbToken[NDBFIL_TOKEN-1], 0); if (!strcmp(qCard->Field(1), "")) { if (ndb_sid != NULL && strcmp(ndb_sid, "")) { qCard->AddElementToField(ndb_sid, 1); pCard->AddElementToField(ndb_sid, 3); } else qCard->AddElementToField(pCard->Field(3), 1); } else pCard->AddElementToField(qCard->Field(1), 3); qCard->UpperCaseField(1); qCard = _CardInfo[STATUS_TOKEN-1].Head(); if (qCard == NULL) qCard = _CardInfo[STATUS_TOKEN-1]. add_a_new_card(NdbToken[STATUS_TOKEN-1], 0); if (strcmp(pCard->Field(2), "") && !strcmp(qCard->Field(6), "")) qCard->AddElementToField(pCard->Field(2), 6); } void _MAXIT::pdb_to_ndb_proc_compnd_and_source(const int is_encapsulated) { typedef struct { int mol_id; int num_entity; int *entity; char *molecule; char *chains; char *fragment; char *synonym; char *ec_number; char *engineered; char *mutation; char *biol_unit; char *details; } MOL_COMPND; MOL_COMPND *mol_compnd = NULL; int n_mol = 0, found, i, j, k, len, entity; int nsource_Nat = 0, nsource_Gen = 0, nsource_Syn = 0; int mol_id = -1, nsource = 0; PDB_Source_Format *PdbSource = NULL; char source[800], match[100], pdb_chnid[10], *p; One_Card_Info_Format *pCard = NULL, *qCard = NULL, *rCard = NULL; pCard = _CardInfo[COMPND_TOKEN-1].Head(); if (!pCard) return; while (pCard) { qCard = _CardInfo[PCOMPN_TOKEN-1]. add_a_new_card(NdbToken[PCOMPN_TOKEN-1], 0); for (i = 1; i < NdbToken[PCOMPN_TOKEN-1].NumField; i++) { qCard->AddElementToField(pCard->Field(i), i); } if (!strncmp(pCard->Field(2), "MOL_ID:", 7)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); } else { n_mol++; mol_compnd = (MOL_COMPND *) realloc(mol_compnd, n_mol * sizeof(MOL_COMPND)); } mol_compnd[n_mol-1].mol_id = atoi(&(pCard->Field(2)[7])); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } else if (!strncmp(pCard->Field(2), "MOLECULE:", 9)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[9])); mol_compnd[n_mol-1].molecule = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].molecule, &(pCard->Field(2)[9])); len = strlen(mol_compnd[n_mol-1].molecule); if (mol_compnd[n_mol-1].molecule[len-1] == ';') mol_compnd[n_mol-1].molecule[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].molecule); } else if (!strncmp(pCard->Field(2), "CHAIN:", 6)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[6])); mol_compnd[n_mol-1].chains = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].chains, &(pCard->Field(2)[6])); len = strlen(mol_compnd[n_mol-1].chains); if (mol_compnd[n_mol-1].chains[len-1] == ';') mol_compnd[n_mol-1].chains[len-1] = '\0'; len = strlen(mol_compnd[n_mol-1].chains); for (i = 0; i < len; i++) if (mol_compnd[n_mol-1].chains[i] == ',') mol_compnd[n_mol-1].chains[i] = ' '; ndb_clean_string(mol_compnd[n_mol-1].chains); if (!strcmp(mol_compnd[n_mol-1].chains, "NULL") || !strcmp(mol_compnd[n_mol-1].chains, "N/A") || !strcmp(mol_compnd[n_mol-1].chains, "NONE")) mol_compnd[n_mol-1].chains[0] = '\0'; } else if (!strncmp(pCard->Field(2), "FRAGMENT:", 9)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[9])); mol_compnd[n_mol-1].fragment = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].fragment, &(pCard->Field(2)[9])); len = strlen(mol_compnd[n_mol-1].fragment); if (mol_compnd[n_mol-1].fragment[len-1] == ';') mol_compnd[n_mol-1].fragment[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].fragment); } else if (!strncmp(pCard->Field(2), "SYNONYM:", 8)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[8])); mol_compnd[n_mol-1].synonym = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].synonym, &(pCard->Field(2)[8])); len = strlen(mol_compnd[n_mol-1].synonym); if (mol_compnd[n_mol-1].synonym[len-1] == ';') mol_compnd[n_mol-1].synonym[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].synonym); } else if (!strncmp(pCard->Field(2), "EC:", 3)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[3])); mol_compnd[n_mol-1].ec_number = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].ec_number, &(pCard->Field(2)[3])); len = strlen(mol_compnd[n_mol-1].ec_number); if (mol_compnd[n_mol-1].ec_number[len-1] == ';') mol_compnd[n_mol-1].ec_number[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].ec_number); } else if (!strncmp(pCard->Field(2), "ENGINEERED:", 11)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[11])); mol_compnd[n_mol-1].engineered = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].engineered, &(pCard->Field(2)[11])); len = strlen(mol_compnd[n_mol-1].engineered); if (mol_compnd[n_mol-1].engineered[len-1] == ';') mol_compnd[n_mol-1].engineered[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].engineered); } else if (!strncmp(pCard->Field(2), "MUTATION:", 9)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[9])); mol_compnd[n_mol-1].mutation = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].mutation, &(pCard->Field(2)[9])); len = strlen(mol_compnd[n_mol-1].mutation); if (mol_compnd[n_mol-1].mutation[len-1] == ';') mol_compnd[n_mol-1].mutation[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].mutation); } else if (!strncmp(pCard->Field(2), "BIOLOGICAL_UNIT:", 16)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[16])); mol_compnd[n_mol-1].biol_unit = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].biol_unit, &(pCard->Field(2)[16])); len = strlen(mol_compnd[n_mol-1].biol_unit); if (mol_compnd[n_mol-1].biol_unit[len-1] == ';') mol_compnd[n_mol-1].biol_unit[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].biol_unit); } else if (!strncmp(pCard->Field(2), "OTHER_DETAILS:", 14)) { if (mol_compnd == NULL) { n_mol = 1; mol_compnd = (MOL_COMPND *) malloc(sizeof(MOL_COMPND)); mol_compnd[n_mol-1].num_entity = 0; mol_compnd[n_mol-1].entity = NULL; mol_compnd[n_mol-1].molecule = NULL; mol_compnd[n_mol-1].chains = NULL; mol_compnd[n_mol-1].fragment = NULL; mol_compnd[n_mol-1].synonym = NULL; mol_compnd[n_mol-1].ec_number = NULL; mol_compnd[n_mol-1].engineered = NULL; mol_compnd[n_mol-1].mutation = NULL; mol_compnd[n_mol-1].biol_unit = NULL; mol_compnd[n_mol-1].details = NULL; } len = strlen(&(pCard->Field(2)[14])); mol_compnd[n_mol-1].details = (char *) calloc(len + 1, sizeof(char)); strcpy(mol_compnd[n_mol-1].details, &(pCard->Field(2)[14])); len = strlen(mol_compnd[n_mol-1].details); if (mol_compnd[n_mol-1].details[len-1] == ';') mol_compnd[n_mol-1].details[len-1] = '\0'; ndb_clean_string(mol_compnd[n_mol-1].details); } pCard = pCard->Next(); } pCard = _CardInfo[COMPND_TOKEN-1].Head(); while (pCard) pCard = _CardInfo[COMPND_TOKEN-1].delete_a_card(pCard); for (i = 0; i < n_mol; i++) { if (mol_compnd[i].chains == NULL) { pdb_chnid[0] = '\0'; entity = get_entity_index(pdb_chnid, is_encapsulated); if (entity != -1) { mol_compnd[i].num_entity = 1; mol_compnd[i].entity = (int *) malloc(sizeof(int)); mol_compnd[i].entity[0] = entity; } } else { p = mol_compnd[i].chains; while (p) { p = get_word(p, pdb_chnid); if (!strcmp(pdb_chnid, "NULL") || !strcmp(pdb_chnid, "N/A") || !strcmp(pdb_chnid, "NONE")) pdb_chnid[0] = '\0'; entity = get_entity_index(pdb_chnid, is_encapsulated); if (entity != -1) { found = 0; for (j = 0; j < mol_compnd[i].num_entity; j++) { if (entity == mol_compnd[i].entity[j]) { found = 1; break; } } if (!found) { mol_compnd[i].num_entity++; mol_compnd[i].entity = (int *) realloc(mol_compnd[i].entity, mol_compnd[i].num_entity * sizeof(int)); mol_compnd[i].entity[mol_compnd[i].num_entity-1] = entity; } } } } for (j = 0; j < mol_compnd[i].num_entity; j++) { if (is_entity_a_protein(mol_compnd[i].entity[j])) { if (mol_compnd[i].molecule) { insert_a_card(mol_compnd[i].entity[j], PROTEI_TOKEN-1, 3, mol_compnd[i].molecule); } strcpy(source, "nat"); if (mol_compnd[i].engineered && !strcmp(mol_compnd[i].engineered, "YES")) strcpy(source, "man"); pCard = _CardInfo[ENTITY_TOKEN-1].Head(); while (pCard) { if (atoi(pCard->Field(1)) == mol_compnd[i].entity[j]) { if (strcasecmp(pCard->Field(3), "nat") && !strcasecmp(source, "nat") || !strcasecmp(pCard->Field(3), "nat") && strcasecmp(source, "nat")) pCard->AddElementToField(source, 3); break; } pCard = pCard->Next(); } } if (mol_compnd[i].molecule) insert_a_card(mol_compnd[i].entity[j], ENTDES_TOKEN-1, 3, mol_compnd[i].molecule); if (mol_compnd[i].fragment) insert_a_card(mol_compnd[i].entity[j], FRAGMT_TOKEN-1, 3, mol_compnd[i].fragment); if (mol_compnd[i].synonym) insert_a_card(mol_compnd[i].entity[j], ENTNAM_TOKEN-1, 3, mol_compnd[i].synonym); if (mol_compnd[i].ec_number) insert_a_card(mol_compnd[i].entity[j], ECNUM_TOKEN-1, 2, mol_compnd[i].ec_number); if (mol_compnd[i].mutation) insert_a_card(mol_compnd[i].entity[j], ENTMUT_TOKEN-1, 3, mol_compnd[i].mutation); if (mol_compnd[i].biol_unit) insert_a_card(mol_compnd[i].entity[j], BIOUNI_TOKEN-1, 3, mol_compnd[i].biol_unit); if (mol_compnd[i].details) insert_a_card(mol_compnd[i].entity[j], ENTAIL_TOKEN-1, 3, mol_compnd[i].details); } } while (PdbSource_Nat[nsource_Nat].pdbname && strcmp(PdbSource_Nat[nsource_Nat].pdbname, "")) { if (PdbSource_Nat[nsource_Nat].NdbTokenId <= 0) PdbSource_Nat[nsource_Nat].NdbTokenId = find_card_id(PdbSource_Nat[nsource_Nat].NdbTokenName)+1; nsource_Nat++; } while (PdbSource_Gen[nsource_Gen].pdbname && strcmp(PdbSource_Gen[nsource_Gen].pdbname, "")) { if (PdbSource_Gen[nsource_Gen].NdbTokenId <= 0) PdbSource_Gen[nsource_Gen].NdbTokenId = find_card_id(PdbSource_Gen[nsource_Gen].NdbTokenName)+1; nsource_Gen++; } while (PdbSource_Syn[nsource_Syn].pdbname && strcmp(PdbSource_Syn[nsource_Syn].pdbname, "")) { if (PdbSource_Syn[nsource_Syn].NdbTokenId <= 0) PdbSource_Syn[nsource_Syn].NdbTokenId = find_card_id(PdbSource_Syn[nsource_Syn].NdbTokenName)+1; nsource_Syn++; } pCard = _CardInfo[SOURCE_TOKEN-1].Head(); while (pCard) { qCard = _CardInfo[PSOURC_TOKEN-1]. add_a_new_card(NdbToken[PSOURC_TOKEN-1], 0); for (i = 1; i < NdbToken[PSOURC_TOKEN-1].NumField; i++) { qCard->AddElementToField(pCard->Field(i), i); } if (!strncmp(pCard->Field(2), "MOL_ID:", 7)) { mol_id = atoi(&(pCard->Field(2)[7])); PdbSource = NULL; nsource = 0; for (i = 0; i < n_mol; i++) { if (mol_compnd[i].mol_id == mol_id) { mol_id = i; if (mol_compnd[i].engineered && !strcmp(mol_compnd[i].engineered, "YES")) { PdbSource = PdbSource_Gen; nsource = nsource_Gen; } else { PdbSource = PdbSource_Nat; nsource = nsource_Nat; } break; } } } else if (!strncmp(pCard->Field(2), "SYNTHETIC:", 10)) { PdbSource = PdbSource_Syn; nsource = nsource_Syn; if (mol_id == -1 && n_mol) mol_id = 0; if (mol_id >= 0) { for (k = 0; k < mol_compnd[mol_id].num_entity; k++) { rCard = _CardInfo[ENTITY_TOKEN-1].Head(); while (rCard) { if (atoi(rCard->Field(1)) == mol_compnd[mol_id].entity[k]) { rCard->AddElementToField("syn", 3); break; } rCard = rCard->Next(); } } } } else { if (mol_id == -1 && n_mol) { if (mol_compnd[0].engineered && !strcmp(mol_compnd[0].engineered, "YES")) { PdbSource = PdbSource_Gen; nsource = nsource_Gen; } else { PdbSource = PdbSource_Nat; nsource = nsource_Nat; } mol_id = 0; } if (mol_id >= 0) { if (!strncmp(pCard->Field(2), "SYNTHETIC: YES", 14)) { for (k = 0; k < mol_compnd[mol_id].num_entity; k++) { rCard = _CardInfo[ENTITY_TOKEN-1].Head(); while (rCard) { if (atoi(rCard->Field(1)) == mol_compnd[mol_id].entity[k]) { rCard->AddElementToField("syn", 3); break; } rCard = rCard->Next(); } } } else { for (i = 0; i < nsource; i++) { sprintf(match, "%s: ", PdbSource[i].pdbname); len = strlen(match); if (!strncmp(pCard->Field(2), match, len) && PdbSource[i].NdbTokenId > 0 && PdbSource[i].NdbFieldNo > 0) { strcpy(source, &(pCard->Field(2)[len])); ndb_clean_string(source); len = strlen(source); if (len && source[len - 1] == ';') source[len-1] = '\0'; if (source[0] == '\0' || !strcmp(source, "NULL") || !strcmp(source, "N/A") || !strcmp(source, "NONE")) break; for (k = 0; k < mol_compnd[mol_id].num_entity; k++) { insert_a_card(mol_compnd[mol_id].entity[k], PdbSource[i].NdbTokenId - 1, PdbSource[i].NdbFieldNo - 1, source); } } } } } } pCard = pCard->Next(); } pCard = _CardInfo[SOURCE_TOKEN-1].Head(); while (pCard) pCard = _CardInfo[SOURCE_TOKEN-1].delete_a_card(pCard); for (i = 0; i < n_mol; i++) { if (mol_compnd[i].entity) free((void *) mol_compnd[i].entity); if (mol_compnd[i].molecule) free((void *) mol_compnd[i].molecule); if (mol_compnd[i].chains) free((void *) mol_compnd[i].chains); if (mol_compnd[i].fragment) free((void *) mol_compnd[i].fragment); if (mol_compnd[i].synonym) free((void *) mol_compnd[i].synonym); if (mol_compnd[i].ec_number) free((void *) mol_compnd[i].ec_number); if (mol_compnd[i].engineered) free((void *) mol_compnd[i].engineered); if (mol_compnd[i].mutation) free((void *) mol_compnd[i].mutation); if (mol_compnd[i].biol_unit) free((void *) mol_compnd[i].biol_unit); if (mol_compnd[i].details) free((void *) mol_compnd[i].details); } if (mol_compnd) free ((void *) mol_compnd); } void _MAXIT::pdb_to_ndb_proc_remarks() { int i, j, k, group_index = 0, nRemarkCard = 0, num_block = 0, *block = NULL; int num_remark = 0, num_old_remark = 0, found = 0, found_any = 0, *used_index = NULL; int SOLEXP_TOKEN, SOLMDL_TOKEN; One_Card_Info_Format *pCard = NULL, *rCard = NULL, **pRemarkCard = NULL; REMARKS *old_remark = NULL; GROUP_REMARKS *group_remark = NULL; SOLEXP_TOKEN = find_card_id("SOLEXP"); SOLMDL_TOKEN = find_card_id("SOLMDL"); pCard = _CardInfo[REMARK_TOKEN-1].Head(); while (pCard) { rCard = _CardInfo[PREMRK_TOKEN-1]. add_a_new_card(NdbToken[PREMRK_TOKEN-1], 0); rCard->AddElementToField(pCard->Field(1), 1); rCard->AddElementToField(pCard->Field(2), 2); pCard = pCard->Next(); } for (i = 0; i < NUM_REMARK; i++) { pRemarkCard = get_remarks(remark_nums[i], nRemarkCard); if (!nRemarkCard) continue; if (remark_nums[i] == 2) pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, 1, &Remark_2, 0); else if (remark_nums[i] == 3) { if (_experiment_type & EXPERIMENT_TYPE_NMR || _experiment_type & EXPERIMENT_TYPE_MODEL) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_3_NMR, Remark_3_NMR, 0); } else if (_experiment_type & EXPERIMENT_TYPE_CRYO_EM) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_3_Cryo_Em, Remark_3_Cryo_Em, 0); } else if (_experiment_type & EXPERIMENT_TYPE_SOLN_SCT) { block = get_blocks_of_remarks(nRemarkCard, num_block, pRemarkCard); if (num_block) { k = 0; for (j = 0; j < num_block; j++) { num_remark = block[2 * j + 1] - block[2 * j] + 1; found = 0; while (k < Remark_3_Solution.Remarks_No) { found = pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[2 * j]], Remark_3_Solution.num_remarks[k], Remark_3_Solution.remarks[k], Remark_3_Solution.repeat[k]); if (found) break; k++; } if (found) { if (k < Remark_3_Solution.Remarks_No && !Remark_3_Solution.repeat[k]) k++; } else k = 0; } } if (block) { delete [] block; block = NULL; } } else { block = get_blocks_of_remarks(nRemarkCard, num_block, pRemarkCard); if (num_block) { num_remark = block[1] - block[0] + 1; if (!pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[0]], Num_Remark_3_1, Remark_3_1, 0)) { if (!pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[0]], Num_Remark_3_2, Remark_3_2, 0)) pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[0]], Num_Remark_3_3, Remark_3_3, 0); } group_index = 0; rCard = _CardInfo[REFMET_TOKEN-1].Head(); if (rCard) { if (!strncmp(rCard->Field(3), "CNS", 3) || !strncmp(rCard->Field(3), "CNX", 3)) group_index = 1; else if (!strncmp(rCard->Field(3),"NUCLSQ", 6)) group_index = 2; else if (!strncmp(rCard->Field(3), "PROLSQ", 6) || !strncmp(rCard->Field(3), "PROFFT", 6) || !strncmp(rCard->Field(3), "CCP4", 4) || !strncmp(rCard->Field(3), "PROTEIN", 7) || !strncmp(rCard->Field(3), "RESTRAIN", 8) || !strncmp(rCard->Field(3), "GPRLSA", 6)) group_index = 3; else if (!strncmp(rCard->Field(3), "REFMAC 5", 8)) group_index = 5; else if (!strncmp(rCard->Field(3), "REFMAC", 6)) group_index = 4; else if (!strncmp(rCard->Field(3),"TNT", 3)) group_index = 6; else if (!strncmp(rCard->Field(3),"SHELX", 5)) group_index = 7; } group_remark = group_remarks_in[group_index]; num_old_remark = num_old_remarks[group_index]; old_remark = old_remarks[group_index]; k = 0; found_any = 0; used_index = new int[group_remark->Remarks_No]; for (j = 0; j < group_remark->Remarks_No; j++) used_index[j] = 0; for (j = 1; j < num_block; j++) { num_remark = block[2 * j + 1] - block[2 * j] + 1; found = 0; while (k < group_remark->Remarks_No) { found = pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[2 * j]], group_remark->num_remarks[k], group_remark->remarks[k], group_remark->repeat[k]); if (found) { used_index[k] = 1; break; } k++; } if (found) { found_any = 1; for (k = 0; k < group_remark->Remarks_No; k++) { if (!used_index[k]) break; } /* if (k < group_remark->Remarks_No && !group_remark->repeat[k]) k++; */ } else k = 0; } delete [] used_index; if (!found_any) { for (j = 0; j < nRemarkCard; j++) { if (pdb_to_ndb_proc_remark_special(pRemarkCard[j], num_old_remark, old_remark)) found_any = 1; } } if (!found_any) { FILE *fp = fopen("MAXIT.ERROR", "a"); fprintf(fp, "%s: maxit failed to pick up the correct value in REMARK 3\n", _StructureId); for (j = 0; j < nRemarkCard; j++) { fprintf(fp, "\tREMARK 3 %s\n", pRemarkCard[j]->Field(2)); } fprintf(fp, "\n"); fclose (fp); } } if (block) { delete [] block; block = NULL; } } } else if (remark_nums[i] == 18) pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_18, Remark_18, 0); else if (remark_nums[i] == 200) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_200, Remark_200, 0); if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_XRAY; } else if (remark_nums[i] == 210) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_210, Remark_210, 0); if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_NMR; } else if (remark_nums[i] == 215) { if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_NMR; } else if (remark_nums[i] == 220) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_220, Remark_220, 0); if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_MODEL; } else if (remark_nums[i] == 225) { if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_MODEL; } else if (remark_nums[i] == 230) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_230, Remark_230, 0); if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_NEUTRON; } else if (remark_nums[i] == 240) { pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_240, Remark_240, 0); if (_experiment_type == EXPERIMENT_TYPE_BASIC) _experiment_type = EXPERIMENT_TYPE_ELECTRON; } else if (remark_nums[i] == 245) { if (_experiment_type == EXPERIMENT_TYPE_CRYO_EM) pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_245, Remark_245, 0); } else if (remark_nums[i] == 250) pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_250, Remark_250, 0); else if (remark_nums[i] == 265) { if (_experiment_type == EXPERIMENT_TYPE_SOLN_SCT) { block = get_blocks_of_remarks(nRemarkCard, num_block, pRemarkCard); if (num_block) { k = 0; for (j = 0; j < num_block; j++) { num_remark = block[2 * j + 1] - block[2 * j] + 1; found = 0; while (k < Remark_265.Remarks_No) { found = pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[2 * j]], Remark_265.num_remarks[k], Remark_265.remarks[k], Remark_265.repeat[k]); if (found) break; k++; } if (found) { if (k < Remark_265.Remarks_No && !Remark_265.repeat[k]) k++; } else k = 0; } } if (block) { delete [] block; block = NULL; } num_block = 0; if (SOLEXP_TOKEN >= 0) { pCard = _CardInfo[SOLEXP_TOKEN].Head(); while (pCard) { pCard->AddElementToField(++num_block, 1); rCard = pCard->Next(); if (rCard && !strcmp(rCard->Field(2), "") && strcmp(pCard->Field(2), "")) rCard->AddElementToField(pCard->Field(2), 2); pCard = pCard->Next(); } } if (SOLMDL_TOKEN >= 0 && SOLEXP_TOKEN >= 0) { rCard = _CardInfo[SOLMDL_TOKEN].Head(); j = 0; while (rCard) { pCard = _CardInfo[SOLEXP_TOKEN]. add_a_new_card(NdbToken[SOLEXP_TOKEN], 0); pCard->AddElementToField("THEORETICAL MODELLING", 2); pCard->AddElementToField(++num_block, 1); rCard->AddElementToField(num_block, 9); rCard->AddElementToField(++j, 10); rCard = rCard->Next(); } } } } else if (remark_nums[i] == 280) pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_280, Remark_280, 0); else if (remark_nums[i] == 300) { if (_experiment_type == EXPERIMENT_TYPE_CRYO_EM) { block = get_blocks_of_remarks(nRemarkCard, num_block, pRemarkCard); if (num_block) { k = 0; for (j = 1; j < num_block; j++) { num_remark = block[2 * j + 1] - block[2 * j] + 1; found = 0; while (k < Remark_Cryo_Em_300.Remarks_No) { found = pdb_to_ndb_proc_remark(num_remark, &pRemarkCard[block[2 * j]], Remark_Cryo_Em_300.num_remarks[k], Remark_Cryo_Em_300.remarks[k], Remark_Cryo_Em_300.repeat[k]); if (found) break; k++; } if (found) { if (k < Remark_Cryo_Em_300.Remarks_No && !Remark_Cryo_Em_300.repeat[k]) k++; } else k = 0; } } if (block) { delete [] block; block = NULL; } } } else if (remark_nums[i] == 800) pdb_to_ndb_proc_remark(nRemarkCard, pRemarkCard, Num_Remark_800, Remark_800, 0); if (pRemarkCard) delete [] pRemarkCard; } pCard = _CardInfo[REMARK_TOKEN-1].Head(); while (pCard) pCard = _CardInfo[REMARK_TOKEN-1].delete_a_card(pCard); if (_CardInfo[EXPDTA_TOKEN-1].Head() == NULL) { pCard = _CardInfo[EXPDTA_TOKEN-1].add_a_new_card(NdbToken[EXPDTA_TOKEN-1], 0); if (_experiment_type == EXPERIMENT_TYPE_XRAY) pCard->AddElementToField("X-RAY DIFFRACTION", 2); else if (_experiment_type == EXPERIMENT_TYPE_NMR) pCard->AddElementToField("NMR", 2); else if (_experiment_type == EXPERIMENT_TYPE_MODEL) pCard->AddElementToField("THEORETICAL MODEL", 2); else if (_experiment_type == EXPERIMENT_TYPE_NEUTRON) pCard->AddElementToField("NEUTRON DIFFRACTION", 2); else if (_experiment_type == EXPERIMENT_TYPE_ELECTRON) pCard->AddElementToField("ELECTRON DIFFRACTION", 2); else pCard->AddElementToField("X-RAY DIFFRACTION", 2); } if (_experiment_type == EXPERIMENT_TYPE_NMR) pdb_to_ndb_postprocessing_remarks(); } One_Card_Info_Format** _MAXIT::get_remarks(const int remarkNo, int &nRemarkCard) { int i; One_Card_Info_Format *pCard = NULL, **pRemarkCard = NULL, **tmp = NULL; nRemarkCard = 0; pCard = _CardInfo[REMARK_TOKEN-1].Head(); while (pCard) { if (atoi(pCard->Field(1)) == remarkNo) { tmp = pRemarkCard; pRemarkCard = new One_Card_Info_Format*[nRemarkCard+1]; for (i = 0; i < nRemarkCard; i++) pRemarkCard[i] = tmp[i]; pRemarkCard[nRemarkCard] = pCard; if (tmp) delete [] tmp; nRemarkCard++; } pCard = pCard->Next(); } return pRemarkCard; } int* _MAXIT::get_blocks_of_remarks(const int nRemarkCard, int &num_block, One_Card_Info_Format **pRemarkCard) { int i, j, *index = NULL, flag = 0; num_block = 0; for (i = 0; i < nRemarkCard; i++) { if (strcmp(pRemarkCard[i]->Field(2), "")) flag = 1; else { if (flag) num_block++; flag = 0; } } if (flag) num_block++; if (num_block) { index = new int[2 * num_block]; j = 0; flag = 0; for (i = 0; i < nRemarkCard; i++) { if (strcmp(pRemarkCard[i]->Field(2), "")) { if (!flag) index[2 * j] = i; index[2 * j + 1] = i; flag = 1; } else { if (flag) j++; flag = 0; } } } return index; } void _MAXIT::pdb_to_ndb_postprocessing() { int i = 0, jrnl_id = -1, RNOCUT_TOKEN = -1; char temp_string[800]; One_Card_Info_Format *pJrnlCard = NULL, *pCard = NULL; One_Card_Info_Format *qCard = NULL, *rCard = NULL, *tCard = NULL; jrnl_id = find_jrnl_id("REF"); pJrnlCard = _JrnlCardInfo[jrnl_id].Head(); while (pJrnlCard) { pJrnlCard->MixCaseField(4); pJrnlCard = pJrnlCard->Next(); } jrnl_id = find_jrnl_id("AUTH"); pJrnlCard = _JrnlCardInfo[jrnl_id].Head(); while (pJrnlCard) { pJrnlCard->MixCaseField(4); pJrnlCard = pJrnlCard->Next(); } jrnl_id = find_jrnl_id("TITL"); pJrnlCard = _JrnlCardInfo[jrnl_id].Head(); while (pJrnlCard) { pJrnlCard->MixCaseField(4); pJrnlCard = pJrnlCard->Next(); } jrnl_id = find_jrnl_id("PUBL"); pJrnlCard = _JrnlCardInfo[jrnl_id].Head(); while (pJrnlCard) { pJrnlCard->MixCaseField(4); pJrnlCard = pJrnlCard->Next(); } jrnl_id = find_jrnl_id("EDIT"); pJrnlCard = _JrnlCardInfo[jrnl_id].Head(); while (pJrnlCard) { pJrnlCard->MixCaseField(4); pJrnlCard = pJrnlCard->Next(); } pJrnlCard = _CardInfo[AUTHOR_TOKEN-1].Head(); while (pJrnlCard) { pJrnlCard->MixCaseField(2); pJrnlCard = pJrnlCard->Next(); } RNOCUT_TOKEN = find_card_id("RNOCUT"); pCard = _CardInfo[REFMET_TOKEN-1].Head(); if (pCard) { if (!strncasecmp(pCard->Field(3), "SHELX", 5)) { qCard = _CardInfo[RFACTR_TOKEN-1].Head(); if (qCard) { if ((rCard = _CardInfo[RNOCUT_TOKEN].Head())) { qCard->AddElementToField(rCard->Field(1), 8); qCard->AddElementToField(rCard->Field(2), 7); tCard = _CardInfo[FREERF_TOKEN-1].Head(); if (tCard == NULL) tCard = _CardInfo[FREERF_TOKEN-1]. add_a_new_card(NdbToken[FREERF_TOKEN-1], 0); tCard->AddElementToField(rCard->Field(3), 1); tCard->AddElementToField(rCard->Field(4), 3); tCard->AddElementToField(rCard->Field(5), 4); } } else { if ((rCard = _CardInfo[RNOCUT_TOKEN].Head())) { qCard = _CardInfo[RFACTR_TOKEN-1]. add_a_new_card(NdbToken[RFACTR_TOKEN-1], 0); qCard->AddElementToField("0.000", 4); qCard->AddElementToField(rCard->Field(1), 8); qCard->AddElementToField(rCard->Field(2), 7); tCard = _CardInfo[FREERF_TOKEN-1].Head(); if (tCard == NULL) tCard = _CardInfo[FREERF_TOKEN-1]. add_a_new_card(NdbToken[FREERF_TOKEN-1], 0); tCard->AddElementToField(rCard->Field(3), 1); tCard->AddElementToField(rCard->Field(4), 3); tCard->AddElementToField(rCard->Field(5), 4); } } } else if ((!strncasecmp(pCard->Field(3), "TNT", 3) || !strncasecmp(pCard->Field(3), "NUCLSQ", 6) || !strncasecmp(pCard->Field(3), "PROLSQ", 6)) && (rCard = _CardInfo[RNOCUT_TOKEN].Head())) { qCard = _CardInfo[RFACTR_TOKEN-1].Head(); if (qCard) { if (!strcmp(qCard->Field(4),"") || strcmp(qCard->Field(4),"") && fabs(atof(qCard->Field(4))) < 0.1) { if (!strcmp(qCard->Field(8), "")) qCard->AddElementToField(rCard->Field(1), 8); } } else { qCard = _CardInfo[RFACTR_TOKEN-1]. add_a_new_card(NdbToken[RFACTR_TOKEN-1], 0); qCard->AddElementToField("0.000", 4); qCard->AddElementToField(rCard->Field(1), 8); } } else if (!strncasecmp(pCard->Field(3), "CNS", 3) || !strncasecmp(pCard->Field(3), "X-PLOR", 6)) { qCard = _CardInfo[RFACTR_TOKEN-1].Head(); if (qCard && strcmp(qCard->Field(10), "")) qCard->AddElementToField(qCard->Field(10), 7); } } update_tokens(NDB_FILE_FORMAT_NDB); pCard = _CardInfo[XFILES_TOKEN-1].Head(); while (pCard) { for (i = 2; i < pCard->NumField(); i++) { if (strcmp(pCard->Field(i), "")) break; } if (i < pCard->NumField()) pCard = pCard->Next(); else pCard = _CardInfo[XFILES_TOKEN-1].delete_a_card(pCard); } pCard = _CardInfo[DTMETH_TOKEN-1].Head(); if (pCard && !strcmp(pCard->Field(1), "Y")) { rCard = _CardInfo[DTMEAS_TOKEN-1].Head(); if (rCard == NULL) rCard = _CardInfo[DTMEAS_TOKEN-1]. add_a_new_card(NdbToken[DTMEAS_TOKEN-1], 0); rCard->AddElementToField(pCard->Field(4), 2); pCard->AddElementToField("SYNCHROTRON", 4); temp_string[0] = '\0'; if (strcmp(rCard->Field(2), "") && strcmp(rCard->Field(3), "")) sprintf(temp_string, "%s BEAMLINE %s", rCard->Field(2), rCard->Field(3)); else if (strcmp(rCard->Field(2), "")) sprintf(temp_string, "%s", rCard->Field(2)); else if (strcmp(rCard->Field(3), "")) sprintf(temp_string, "BEAMLINE %s", rCard->Field(3)); ndb_clean_string(temp_string); if (temp_string[0] != '\0') { rCard = _CardInfo[RADIAT_TOKEN-1].Head(); if (rCard == NULL) rCard = _CardInfo[RADIAT_TOKEN-1]. add_a_new_card(NdbToken[RADIAT_TOKEN-1], 0); rCard->AddElementToField(temp_string, 5); } } pCard = _CardInfo[DTWAVE_TOKEN-1].Head(); while (pCard) { pCard->AddElementToField(1, 1); pCard->AddElementToField(1, 2); pCard = pCard->Next(); } pCard = _CardInfo[WAVLEN_TOKEN-1].Head(); if (pCard && (strstr(pCard->Field(2), ",") || strstr(pCard->Field(2), ";"))) { pCard->AddElementToField(pCard->Field(2), 3); pCard->AddElementToField("", 2); } RNOCUT_TOKEN = find_card_id("TLSRNG"); int count = 1; pCard = _CardInfo[RNOCUT_TOKEN].Head(); while (pCard) { pCard->AddElementToField(count, 6); count++; pCard = pCard->Next(); } RNOCUT_TOKEN = find_card_id("TLSGRO"); pCard = _CardInfo[RNOCUT_TOKEN].Head(); while (pCard) { pCard->AddElementToField("refined", 6); pCard = pCard->Next(); } } int _MAXIT::get_entity_index(const char *pdb_chnid, const int is_encapsulated) { if (is_encapsulated) { if (_CifObj) { int i, rowNo, entity_id, chain_id; CifString string; ISTable *Table = _CifObj->GetTablePtr(_StructureId, "entity_poly"); if (is_empty_table(Table)) return -1; rowNo = Table->GetNumRows(); entity_id = Table->GetColumnIndex("entity_id"); chain_id = Table->GetColumnIndex("ndb_chain_id"); for (i = 0; i < rowNo; i++) { Table->GetCell(string, chain_id, i); if ((!strcmp(pdb_chnid, "") && (string == "" || string == "?" || string == ".")) || strstr(string.Text(), pdb_chnid)) { Table->GetCell(string, entity_id, i); return (atoi(string.Text())); } } } return -1; } else { if (_num_mol == 0) return -1; _CHAIN *chain = _molecules[0]->GetFirstChain(); while (chain) { if (chain->chain_type() != ATOMN_TOKEN && chain->chain_type() != ATOMP_TOKEN) { chain = _molecules[0]->GetNextChain(); continue; } if (!strcmp(chain->PDB_ChainID(), pdb_chnid)) return chain->entity_id(); chain = _molecules[0]->GetNextChain(); } return -1; } } int _MAXIT::is_entity_a_protein(const int entity_id) { if (_num_mol == 0) return 0; _CHAIN *chain = _molecules[0]->GetFirstChain(); while (chain) { if (chain->entity_id() == entity_id) { if (chain->chain_type() == ATOMP_TOKEN) return 1; else return 0; } chain = _molecules[0]->GetNextChain(); } return 0; } void _MAXIT::insert_a_card(const int entity_id, const int card_id, const int fieldno, char *value) { One_Card_Info_Format *pCard = _CardInfo[card_id].Head(); while (pCard) { if (atoi(pCard->Field(1)) == entity_id) break; pCard = pCard->Next(); } if (pCard == NULL) pCard = _CardInfo[card_id].add_a_new_card(NdbToken[card_id], 0); pCard->AddElementToField(entity_id, 1); pCard->AddElementToField(value, fieldno); } int _MAXIT::pdb_to_ndb_proc_remark(const int nRemarkCard, One_Card_Info_Format **pRemarkCard, const int remark_num, REMARKS *remark, const int repeat) { int i, j, k, count = 0, card_id, field_no, *index = 0; char temp_string[MxStrLen], temp_string1[MxStrLen]; double temp_double; BTreeObj<_IIndex> OrderIndex; _IIndex idx; One_Card_Info_Format *newCard = NULL, *pCard = NULL; index = new int[nRemarkCard]; for (i = 0; i < nRemarkCard; i++) index[i] = 0; j = 0; for (i = 0; i < nRemarkCard; i++) { while (j < remark_num) { if (find_a_match(pRemarkCard[i]->Field(2), &remark[j])) { index[i] = j + 1; count++; break; } j++; } if (index[i]) { // if (repeat == -1) j = 0; if (repeat) j = 0; else if (j < remark_num && !remark[j].Multiple_Treatment) j++; } else j = 0; } card_id = 0; for (i = 0; i < remark_num; i++) { if (remark[i].NumField > 1) card_id++; else if (strcmp(remark[i].FieldList[0].TokenName, "") || strcmp(remark[i].FieldList[0].Text, "")) card_id++; } if (!count || (count < nRemarkCard && nRemarkCard < 8 || count < 4 * nRemarkCard / 5 && nRemarkCard >= 8) && count < 2 * card_id / 3) { delete [] index; return 0; } OrderIndex.Clear(); newCard = NULL; card_id = -1; field_no = -1; for (k = 0; k < nRemarkCard; k++) { if (index[k]) { newCard = NULL; card_id = -1; field_no = -1; j = index[k] - 1; for (i = 0; i < remark[j].NumField; i++) { card_id = find_card_id(remark[j].FieldList[i].TokenName); field_no = remark[j].FieldList[i].FieldId - 1; temp_string[0] = '\0'; if (strcmp(remark[j].FieldList[i].TokenName,"") && remark[j].FieldList[i].FieldId) { if ((int) strlen(pRemarkCard[k]->Field(2)) > remark[j].FieldList[i].FieldStCol - remark[j].FieldList[0].FieldStCol) get_remark_answer(temp_string, pRemarkCard[k], &remark[j], i); } ndb_clean_string(temp_string); ndb_set_string_to_upper(temp_string); if (strcmp(temp_string, "NULL") && strcmp(temp_string, "NONE") && strcmp(temp_string, "N/A") && strcmp(temp_string,"") && card_id != -1 && strcmp(remark[j].FieldList[i].TokenName,"") && remark[j].FieldList[i].FieldId) { if (!newCard) { if (remark[j].Multiple_Treatment < 0) { newCard = _CardInfo[card_id]. add_a_new_card(NdbToken[card_id], 0); pCard = NULL; if (!strcmp(remark[j].FieldList[i].TokenName, "CCPNCS")) pCard = _CardInfo[NCSGRO_TOKEN-1].Tail(); else if (!strcmp(remark[j].FieldList[i].TokenName, "TLSRNG")) pCard = _CardInfo[TLSGRO_TOKEN-1].Tail(); if (pCard) newCard->AddElementToField(pCard->Field(1), 1); } else if (repeat > 0) { idx.set_value(card_id); if (OrderIndex.Seek(idx) < 0) { OrderIndex.Add(idx); newCard = _CardInfo[card_id]. add_a_new_card(NdbToken[card_id], 0); if (!strcmp(remark[j].FieldList[i].TokenName, "TTENSR") || !strcmp(remark[j].FieldList[i].TokenName, "LTENSR") || !strcmp(remark[j].FieldList[i].TokenName, "STENSR")) { pCard = _CardInfo[TLSGRO_TOKEN-1].Tail(); if (pCard) newCard->AddElementToField(pCard->Field(1), 1); } } else if ((!strcmp(remark[j].FieldList[i].TokenName, "ASSEMB") || !strcmp(remark[j].FieldList[i].TokenName, "VIRPAR")) && field_no == 1) newCard = _CardInfo[card_id]. add_a_new_card(NdbToken[card_id], 0); newCard = _CardInfo[card_id].Tail(); } else if (remark[j].Multiple_Treatment == remark[j].FieldList[i].FieldId) { newCard = _CardInfo[card_id].Head(); while (newCard) { if (!strcmp(newCard->Field(field_no), temp_string)) break; newCard = newCard->Next(); } if (card_id == XFILES_TOKEN-1 && field_no == 1 && !strcmp(temp_string, "1")) newCard = _CardInfo[card_id].Head(); } else newCard = _CardInfo[card_id].Head(); if (!newCard) newCard = _CardInfo[card_id]. add_a_new_card(NdbToken[card_id], 0); } if (remark[j].Special_Treatment == PDB_REMARK_R_VALUE) { temp_double = atof(temp_string) * 100.0; sprintf(temp_string, "%lf", temp_double); newCard->AddElementToField(temp_string, field_no); } else newCard->AddElementToField(temp_string, field_no); } } } else if (card_id >= 0 && field_no > 0) { strcpy(temp_string1, pRemarkCard[k]->Field(2)); if (strstr(temp_string1, "):") || strstr(temp_string1, " :")) continue; ndb_clean_string(temp_string1); if (strcmp(temp_string1, "")) { if (!newCard) newCard = _CardInfo[card_id]. add_a_new_card(NdbToken[card_id], 0); if (strcmp(newCard->Field(field_no), "")) { newCard->AttachToField(" ", field_no); newCard->AttachToField(temp_string1, field_no); } else newCard->AddElementToField(temp_string1, field_no); } } } delete [] index; return 1; } int _MAXIT::pdb_to_ndb_proc_remark_special(One_Card_Info_Format *remarkCard, const int remark_no, REMARKS *remark) { if (remarkCard == NULL) return 1; int i, j, label, len, pos, card_id, field_no; char temp_string[80], temp_string1[80]; double temp_double; One_Card_Info_Format *newCard = NULL; for (j = 0; j < remark_no; j++) { strcpy(temp_string1, remarkCard->Field(2)); ndb_set_string_to_upper(temp_string1); ndb_strip_leading_blanks(temp_string1); label = 0; for (i = 0; i < remark[j].NumField; i++) { if (!strcmp(remark[j].FieldList[i].TokenName, "")) { strcpy(temp_string, remark[j].FieldList[i].Text); ndb_set_string_to_upper(temp_string); ndb_strip_leading_blanks(temp_string); len = strlen(temp_string); pos = remark[j].FieldList[i].FieldStCol - remark[j].FieldList[0].FieldStCol; if (string_compare(&temp_string1[pos], temp_string, pos, len)) { label = 1; break; } } } if (label) continue; for (i = 0; i < remark[j].NumField; i++) { card_id = find_card_id(remark[j].FieldList[i].TokenName); field_no = remark[j].FieldList[i].FieldId -1; temp_string[0] = '\0'; if (strcmp(remark[j].FieldList[i].TokenName, "") && remark[j].FieldList[i].FieldId) { get_remark_answer(temp_string, remarkCard, &remark[j], i); } ndb_clean_string(temp_string); ndb_set_string_to_upper(temp_string); if (strcmp(temp_string, "NULL") && strcmp(temp_string, "NONE") && strcmp(temp_string, "N/A") && strcmp(temp_string,"") && card_id != -1 && strcmp(remark[j].FieldList[i].TokenName,"") && remark[j].FieldList[i].FieldId) { newCard = _CardInfo[card_id].Head(); if (newCard == NULL) newCard = _CardInfo[card_id]. add_a_new_card(NdbToken[card_id], 0); if (remark[j].Special_Treatment == PDB_REMARK_R_VALUE) { temp_double = atof(temp_string) * 100.0; sprintf(temp_string, "%lf", temp_double); newCard->AddElementToField(temp_string, field_no); } else newCard->AddElementToField(temp_string, field_no); } } return 1; } return 0; } void _MAXIT::pdb_to_ndb_postprocessing_remarks() { int i, j, serial_no; char program[1000], authors[200], version[80], name[80], *p, *a; One_Card_Info_Format *pCard = NULL, *qCard = NULL; pCard = _CardInfo[REFMET_TOKEN-1].Head(); if (pCard) { p = pCard->Field(3); a = pCard->Field(4); while (p && a) { p = get_words_separate_by_AND(p, program); a = get_words_separate_by_AND(a, authors); get_name_and_version(name, version, program); clean_author_name(authors, name); if (strcmp(name, "") || strcmp(authors, "")) { qCard = _CardInfo[SFTWAR_TOKEN-1]. add_a_new_card(NdbToken[SFTWAR_TOKEN-1], 0); qCard->AddElementToField("refinement", 2); if (strcmp(name, "")) qCard->AddElementToField(name, 3); if (strcmp(version, "")) qCard->AddElementToField(version, 4); if (strcmp(authors, "")) qCard->AddElementToField(authors, 5); } } while (pCard) pCard = _CardInfo[REFMET_TOKEN-1].delete_a_card(pCard); } pCard = _CardInfo[NMRECD_TOKEN-1].Head(); if (pCard) { p = pCard->Field(2); serial_no = 1; while (p) { p = get_words(p, program); One_Card_Info_Format *qCard = _CardInfo[NMREXP_TOKEN-1]. add_a_new_card(NdbToken[NMREXP_TOKEN-1], 0); qCard->AddElementToField(serial_no, 1); qCard->AddElementToField(1, 2); qCard->AddElementToField(program, 3); qCard->AddElementToField(1, 4); serial_no++; } while (pCard) pCard = _CardInfo[NMRECD_TOKEN-1].delete_a_card(pCard); } pCard = _CardInfo[NMRSFT_TOKEN-1].Head(); if (pCard) { p = pCard->Field(2); while (p) { p = get_words(p, program); get_name_and_version(name, version, program); if (strcmp(name, "")) { One_Card_Info_Format *qCard = _CardInfo[SFTWAR_TOKEN-1]. add_a_new_card(NdbToken[SFTWAR_TOKEN-1], 0); qCard->AddElementToField("structure solution", 2); qCard->AddElementToField(name, 3); if (strcmp(version, "")) qCard->AddElementToField(version, 4); } } while (pCard) pCard = _CardInfo[NMRSFT_TOKEN-1].delete_a_card(pCard); } pCard = _CardInfo[NMRSPM_TOKEN-1].Head(); if (pCard) { typedef struct { char buf[100]; } BUF; BUF strengh[10], model[10], maker[10]; int n_strengh = 0, n_model = 0, n_maker = 0; char buffer[100]; p = pCard->Field(2); while (p) { p = get_words(p, buffer); if (strcmp(buffer, "")) { strcpy(model[n_model].buf, buffer); n_model++; } } p = pCard->Field(3); while (p) { p = get_words(p, buffer); if (strcmp(buffer, "")) { strcpy(maker[n_maker].buf, buffer); n_maker++; } } p = pCard->Field(4); while (p) { p = get_words(p, buffer); if (strcmp(buffer, "")) { strcpy(strengh[n_strengh].buf, buffer); n_strengh++; } } int max_number = n_model; if (n_maker > max_number) max_number = n_maker; if (n_strengh > max_number) max_number = n_strengh; for (i = 0; i < n_strengh; i++) { p = strstr(strengh[i].buf, "MHZ"); if (p) *p = '\0'; } while (pCard) pCard = _CardInfo[NMRSPM_TOKEN-1].delete_a_card(pCard); for (i = 0; i < max_number; i++) { pCard = _CardInfo[NMRSPM_TOKEN-1]. add_a_new_card(NdbToken[NMRSPM_TOKEN-1], 0); pCard->AddElementToField(i+1, 1); if (n_model) { j = i; if (j >= n_model) j = n_model - 1; pCard->AddElementToField(model[j].buf, 2); } if (n_maker) { j = i; if (j >= n_maker) j = n_maker - 1; pCard->AddElementToField(maker[j].buf, 3); } if (n_strengh) { j = i; if (j >= n_strengh) j = n_strengh - 1; pCard->AddElementToField(strengh[j].buf, 4); } } } pCard = _CardInfo[NMRSMP_TOKEN-1].Head(); if (pCard) { pCard->AddElementToField(1, 1); pCard->AddElementToField(1, 2); } int NMRSDT_TOKEN = find_card_id("NMRSDT"); pCard = _CardInfo[NMRSDT_TOKEN].Head(); if (pCard) pCard->AddElementToField(1, 1); } void _MAXIT::delete_atom_count() { int card_id = find_card_id("ATNUMS"); _CardInfo[card_id].clear_Cards(); } static int find_a_match(const char *context, REMARKS *remark) { char temp_string[80], temp_string1[80], string[80]; int label = 0, i, strip_blanks, not_empty, len, pos; strcpy(temp_string1, context); strip_blanks = 1; for (i = 0; i < remark->NumField; i++) { if (!strcmp(remark->FieldList[i].TokenName, "") && strcmp(remark->FieldList[i].Text, "")) { strcpy(temp_string, remark->FieldList[i].Text); ndb_set_string_to_upper(temp_string); ndb_strip_leading_blanks(temp_string); len = strlen(temp_string); if (!len) strip_blanks = 0; break; } } if (strip_blanks) ndb_strip_leading_blanks(temp_string1); for (i = 0; i < remark->NumField; i++) { if (!strcmp(remark->FieldList[i].TokenName, "") && strcmp(remark->FieldList[i].Text, "")) { label = 1; strcpy(temp_string, remark->FieldList[i].Text); ndb_set_string_to_upper(temp_string); ndb_strip_leading_blanks(temp_string); len = strlen(temp_string); not_empty = 1; if (!len) { strcpy(temp_string, remark->FieldList[i].Text); not_empty = 0; len = strlen(temp_string); } if (strip_blanks) pos = remark->FieldList[i].FieldStCol - remark->FieldList[0].FieldStCol; else pos = remark->FieldList[i].FieldStCol; if (strncmp(&temp_string1[pos], temp_string, len)) { if (!not_empty) { label = 0; break; } ndb_clean_string(temp_string); strcpy(string, temp_string1); ndb_clean_string(string); len = strlen(temp_string); if (strncmp(string, temp_string, len)) { label = 0; break; } } } } return label; } static void clean_middle_space(char *string) { int i, j = 0, k, length; length = strlen(string); for (i = 0; i < length; i++) if (string[i] != ' ') j = i; for (i = 0; i < j; i++) { if (string[i] == ' ') { for (k = i; k < length - 1; k++) string[k] = string[k+1]; length--; j--; } } string[length] = '\0'; } static int string_compare(const char *remark, const char *string, const int pos, const int len) { char tmp_remark[80], tmp_tpt[80]; int i = 0, j = 0; while (i < pos && remark[j - 1] != ' ') { i++; j--; } strncpy(tmp_remark, &remark[j], len); strncpy(tmp_tpt, string, len); tmp_remark[len] = '\0'; tmp_tpt[len] = '\0'; ndb_strip_leading_blanks(tmp_remark); ndb_strip_leading_blanks(tmp_tpt); clean_middle_space(tmp_remark); clean_middle_space(tmp_tpt); return (strcmp(tmp_remark, tmp_tpt)); } static void get_remark_answer(char *answer, One_Card_Info_Format *remarkCard, REMARKS *remark, int field_no) { int i, k, tmp, end, width, strip_blanks; char temp_string[80]; strcpy(answer, ""); if (remarkCard == NULL) return; strip_blanks = 1; strcpy(temp_string, remark->FieldList[0].Text); ndb_set_string_to_upper(temp_string); ndb_strip_leading_blanks(temp_string); k = strlen(temp_string); if (!k) strip_blanks = 0; strcpy(temp_string, remarkCard->Field(2)); if (strip_blanks) ndb_strip_leading_blanks(temp_string); if (field_no == 0) { if (strcmp(temp_string,"NULL") && strcmp(temp_string,"NONE") && strcmp(temp_string,"N/A")) strncpy(answer, temp_string, remark->FieldList[field_no].FieldWidth); else strcpy(answer, "NULL"); } else { if (strip_blanks) { tmp = remark->FieldList[field_no].FieldStCol - remark->FieldList[0].FieldStCol; end = remark->FieldList[field_no].FieldStCol - remark->FieldList[0].FieldStCol + remark->FieldList[field_no].FieldWidth; } else { tmp = remark->FieldList[field_no].FieldStCol; end = remark->FieldList[field_no].FieldStCol + remark->FieldList[field_no].FieldWidth; } if (field_no == remark->NumField - 1) { if (strip_blanks) k = 67 - remark->FieldList[0].FieldStCol; else k = 67; if (k > end) end = k; } if (end > (int) strlen(temp_string)) end = strlen(temp_string); for (i = tmp; i < end; i++) if (temp_string[i] != ' ' && temp_string[i] != ':' && temp_string[i] != ';' && temp_string[i] != '(' && temp_string[i] != ')') break; while (i && temp_string[i-1] != ' ' && temp_string[i-1] != ':' && temp_string[i-1] != ';' && temp_string[i-1] != '(' && temp_string[i] != ')') i--; width = 0; for (k = i; k < end; k++) { if (temp_string[k] == ';') break; if ((remark->FieldList[field_no].FieldType == 1 || remark->FieldList[field_no].FieldType == 2) && (temp_string[k] == ' ' || temp_string[k] == '(' || temp_string[k] == ')')) break; answer[width] = temp_string[k]; width++; } answer[width] = '\0'; } } static void get_name_and_version(char *name, char *version, char *program) { unsigned int i; name[0] = '\0'; version[0] = '\0'; char *p = get_word(program, name); while (p) { p = get_word(p, version); } if (strcmp(version, "")) { for (i = 0; i < strlen(version); i++) if (!isdigit(version[i]) && version[i] != '.') break; if (i == strlen(version)) { p = strstr(program, version); *p = '\0'; } } strcpy(name, program); ndb_clean_string(name); } static void clean_author_name(char *authors, char *name) { char *buff = new char[strlen(name)+3]; sprintf(buff, "(%s)", name); char *p = strstr(authors, buff); delete buff; if (p) { *p = '\0'; ndb_clean_string(authors); return; } p = strstr(authors, name); if (p) { *p = '\0'; ndb_clean_string(authors); } }