/* query.c -- apply a boolean query to a keyword file */ /* Bruce Tanner -- Cerritos College */ /* Version history: Modifications by Richard D. Piccard -- Ohio University 2000/07/07 Add home_page function immediately before sorting, to change the ranking of the hits based on several criteria. The numerical values can be customized readily, and two environment variables can be used to suppress particular hosts or paths. 2000/07/10 Use include files custom_head.c and custom_foot.c for cosmetic and site-specific changes to the HTML at the top and bottom of the generated page. 2000/08/01 Add capability for REQUIRED_REALM to home_page. 2000/08/24 Removed \n prior to > closing tags, and abolished in order to work around bug in Mac Internet Explorer V5. ------------------------------------- 0.0 1993/06/18 Start the program 0.1 1993/07/03 Finish coding 0.8 1993/07/06 Squashed most of the bugs 1.0 1993/07/13 Released 1.1 1993/07/29 Changed grammar to allow Query inside Factor 1.2 1993/07/29 Invoke interactive mode if argc < 4 1.3 1993/08/04 Change name to query, calling it search is confusing 1.4 1993/08/06 Move wildcard processing from shell to inside program 1.5 1993/11/17 Include punctuation in query string to match build_index 1.6 1993/12/01 Handle multiple topic field sizes 1.7 1994/03/01 Fix bad reference to freed pointer 1.7a 1994/06/29 Added some include files and related stuff for DECC. - FM 1.8 1994/11/04 Handle host/port included in selector file 1.9 1994/12/31 Trim off host/port from file_name in select_result(), and report any failure to access file_name. - FM 2.0 1995/06/14 Handle additional index file field 2.1 1995/11/03 Make near() use last result with word position 2.2 1995/11/06 Merge with Foteos Macrides' WWWquery.c 2.3 1995/12/30 Added check for a WWW_GATEWAY_INTERFACE environment variable which acts as a /www switch. - FM 1995/12/30 Added sort qualifier, as default, so that /nosort will block sorting based on the score. - FM 1995/12/30 Added check for a WWW_NOSORT environment variable which acts as a /nosort switch . - FM 1995/12/30 Added check for a WWW_SHOW_SCORE enviroment variable which acts as a /score switch. - FM 1995/12/31 Changed default to nosort, check for WWW_SORT. - BT 2.4 1996/01/04 Ignore words smaller than WWW_MINIMUM_WORD - BT 1996/01/05 Added WWW_ITEMS_FOUND environment variable for changing the string which indicates the total number of hits (must have exactly one %d in it). - FM 1996/01/05 Added WWW_ITEMS_LISTED environment variable for changing the string which indicates the first and last hit in the list, if all of them are not listed (must have exactly two %d's in it). - FM 1996/01/05 Added WWW_STARTING_HIT environment variable for specifying the value of the first hit to be listed. This can be used together with WWW_MAXIMUM_HITS to specify a range of hits within the total, when using a FORM interface to the search engine. The starting number is passed as an OL attribute for clients which support SEQNUM or START for OLs. - FM 1996/01/05 Added code in www_emit() to not act of WWW_SHOW_DATE or WWW_SHOW_SIZE if they depend on a stat() and it fails. This deals with the situation in which the indexing was done on temporary text files, e.g., from a Lynx traversal, which have been deleted after indexing. - FM 1996/01/07 Added WWW_ADD_CONTEXT environment variable for specifying whether to add First or Previous and Next or Last submit buttons, with NAMEs "ph" and "nh", respectively, as LH content following the OL start tag, and hidden INPUTs following the OL end tag, with NAMEs "os", "om", "ot", and "oq", indicating the current (which on resubmission will be the "old") starting, maximum and total hits, and query. When using a FORM interface, their VALUEs can be used by the calling command file to do the same search (i.e., with the previous query) but with the starting hit shifted in relation to the previous starting and maximum hit values. - FM 1996/01/07 Added WWW_END_FORM environment variable for specifying whether to add a FORM end tag following the OL end tag, or following the hidden INPUTs if included. - FM 1996/01/09 Made use of WWW_END_FORM and WWW_ADD_CONTENT environment variables depended on WWW_VIA_FORM having been set, and the insertion of hidden inputs dependent on named submit buttons having been add, (if all the hits are listed, no named submit buttons are added, and so the hidden inputs aren't needed). This makes it easier for the calling command file to regulate its behavior based on the WWW_KEY_COUNT. - FM 1996/01/10 Added WWW_SEND_CONTENTS environment variable for specifying that unique search results get displayed immediately. - BT 1996/01/11 Changed _tolower() Arne Vajhøj's international version. - BT 1996/01/30 Output the hidden INPUTs before the hit list, to be sure they are included if the transmission is stopped during the listing. - FM If there are 10 or more items in the hit list, not all are being listed, and WWW_ADD_CONTEXT is set, append another pseudo toolbar, following the list, for getting a previous and/or next set of hits. - FM 1996/02/01 Added WWW_HITS_TARGET environment variable, which, if non-NULL and a non-zero-length string, will cause the TARGET attribute to be included in anchors for the hit list with its value equated to the string. - FM 1996/02/21 Fixed WWW_SEND_CONTENT handling so that it returns a complete URL in the Location: header with the symbolic path substitutions, so that the client will replace its base URL for the document and any partial references within it will be resolved properly, and so that it will work properly with /~username databases. - FM 1996/02/23 Added TITLE attribute for anchors with gopher URLs in www_emit() hit lists. - FM 1996/03/12 Fixed near() processing of /noposition index - BT 1996/07/08 Fixed ww_emit not writing host:port when necessary - BT ** Foteos Macrides -- Worcester Foundation for Experimental Biology ** WWWquery.c Version History: 0.0 1994/04/08 Made mods for use with CERN v2.16betavms httpd. Uses a query constructed from WWW_KEY fields by an htbin script. Returns an HTML menu of hits with "escaped" URL's or the requested document/section. - FM 1.0 1994/04/24 Seems to working OK, so posted the Initial Release. - FM 1.1 1994/04/28 Added WWW_SEND_RAW_FILE and WWW_OMIT_PRE_TAG symbols, and fixed up partial URL handling. - FM 1.2 1994/04/29 Added code for invoking the native CERN httpd and VMS gopherd security measures / authorization checks for raw file requests. - FM Added requirement that the path to sectional documents begin with the string "www_root" or "gopher_root" for sections from them to be sent (versus a 403 error message). This meets all the security measures for the VMS gopherd. For sectional databases in the httpd data tree(s), you can implement additional authorization checks only for access to the htbin script. The only subsequent security check is the requirement that the database reside in a secure path, based on it beginning with one of the above two strings. For raw file requests, if a lead "www_root" string is in a device field, a lead "000000." is inserted into the directory field (if it was not already present), and all paths passed back to the httpd for checking are first converted to a www symbolic hierarchy - FM 1.3 1994/04/30 Allow both "TEXT= R" and "TEXT=R" as flag for a file or database section fetch. - FM Fix up misuse of argv[1], and return error messages if maximum argument lengths are exceeded. - FM 1.4 1994/05/06 Various fixes to path and filename handling. - FM 1.4a 1994/06/29 Added some include files and related stuff for DECC. - FM 1994/08/22 Modified to return direct URL's if WWW_SEND_RAW_FILE was defined, avoiding the need for redirection via Location: on requests for the file. - FM 1994/10/03 Added WWW_MAXIMUM_HITS symbol for setting the maximum number of hits to return to the client, if made non-NULL in the calling script. - FM 1994/10/13 Added code for handling both text and binary files in searches, based on Bruce Tanner's modification of the indexer (build_index.c) for indexing of binary files. - FM Added WWW_SHOW_FILETYPE symbol for inserting icons (or ALT text) into hit lists to indicate the file types. Use this if your indexes include binary file types. - FM 1994/10/14 Fixed select_result() to handle name and path fields of of any size in the .SEL files. - FM 1994/10/16 Added code for WWW_SHOW_SIZE, WWW_SHOW_DATE, WWW_SHOW_TIME, WWW_USE_MDATE symbols - FM. 1994/10/17 Efficiency tweeks for latest build_index.c. - FM 1.9 1994/12/31 Added code for handling host/path in selectors. - FM 1995/03/14 Omit tag in hit list if WWW_VIA_FORM is set. - FM Include "TEXT R" in the tests for a range request. - FM Check for WWW_DATADIRECTORY and WWW_FORBIDDEN in addition to "WWW_Root" and "Gopher_Root" as lead strings in the path fields for fetches via RANGE selectors. Also check for WWW_FORCE_000000 when formatting such selectors. - FM 1995/04/13 Indicate MaxHits restriction if hit list is truncated. - FM Added WWW_OMIT_CONTENT_TYPE, WWW_OMIT_HEAD, WWW_OMIT_FOOT, and WWW_VMSINDEX_TITLE symbols. - FM 1995/04/26 Added WWW_IPATH_INFO, WWW_IPATH_TRANSLATED and WWW_OMIT_IPATH_CONVERSION symbols. - FM 1995/04/30 Added support for searches of sectional databases in /~userdir/ directories. - FM ** WWWquery.c -- Apply a boolean query to (a) keyword file(s), and return a ** hit list, a requested section with HTML "packaging", or a ** "raw" HTML, text or binary file from an indexed file set. ** The symbol WWW_SHOW_FILETYPE can be made non-NULL by the ** calling script to have icons (or ALT text) indicating the ** file types to be inserted into the hit lists returned from ** searches. ** The symbol WWW_SEND_RAW_FILE can be made non-NULL by the ** calling script for requested files to be sent raw (i.e., ** without additional headers and PRE tags). The hit lists ** for such files have direct URL's. If a ?TEXT=R... URL is ** used to fetch them from gerry-rigged URL's, it will be ** redirected via "Location:", so that the server will check ** access authorization via it's native procedures. Otherwise, ** the path to the document must begin with "www_root" or ** "gopher_root" (case-insensitive) or the client will be ** sent a 403 error message. ** The symbol WWW_OMIT_PRE_TAG can be made non-NULL by the ** calling script to have document sections returned without ** PRE tags bounding them, but with the other HTML "packaging" ** retained. ** The symbol WWW_MAXIMUM_HITS can be made non-NULL by the ** calling script to set a maximum number of hits to be ** returned to the client. ** The symbol WWW_SHOW_SIZE can be made non-NULL by the calling ** script to have the size of each file or database section ** indicated in the hit lists. ** The symbol WWW_SHOW_DATE can be made non-NULL by the calling ** script to have the dates of files (or date of the database) ** indicated in the hit lists (DD-MMM-YY). If WWW_SHOW_TIME ** also is made non-NULL, for the current year the hour and ** minute will be indicated instead of the year (DD-MMM HH:MM). ** If WWW_USE_MDATE is made non-NULL, the date of the last ** modification (e.g., from an APPEND) will be used. The ** default is the file creation date (i.e., of the highest ** version). ** The symbol WWW_OMIT_CONTENT_TYPE can be made non-NULL by the ** calling script if it sends the Content-type header itself. ** Otherwise "Content-type: text/html\n\n" will be sent by ** WWWquery. ** The symbol WWW_OMIT_HEAD can be made non-NULL by the calling ** script if it sends the , ... and ** tags itself. ** The symbol WWW_VMSINDEX_TITLE can be set to a string by the ** calling script for use in the tag. Otherwise, ** WWWquery uses "VMS Indexed Database Search" (not used if ** WWW_OMIT_HEAD is set). ** The symbol WWW_VIA_FORM can be made non-NULL by the calling ** script if the query was submitted via a FORM and the ** <ISINDEX> tag should be omitted from the hit lists (not ** used if WWW_OMIT_HEAD is set). ** The symbol WWW_OMIT_FOOT can be made non-NULL by the calling ** script if it sends the </BODY> and </HTML> end tags itself. ** The symbol WWW_FORCE_000000 can be made non-NULL by the ** calling script to force insertion of "000000." into ** RANGE selectors whose paths begin with "WWW_Rootxxx:[" ** before processing them for a fetch (for backward ** compatibility; you don't need this if you've upgraded ** all of your VMSIndex software to the latest versions). ** The symbols WWW_IPATH_INFO and WWW_IPATH_TRANSLATION can ** be set to strings by the calling script for conversion ** of device:[directory]file paths from the .SEL files which ** have been cast to /device/directory/file further back to ** the URL paths in the server's configuation file. Defaults ** for conversion of /WWW_Root/000000/ or /WWW_Root/documents/ ** to /www/ are set in WWWsearch.com. In the example htbin ** calling script, VMSIndex.com, the strings are modified on ** the basis of the standard CGI variables WWW_PATH_INFO and ** and WWW_PATH_TRANSLATED, so you can append a path to the ** URL, e.g., /htbin/vmsindex/foo/ to have WWW_IPATH_INFO ** set to /foo/ and WWW_IPATH_TRANSLATION set to the ** translation for it in the server's map/pass rules. If ** the path is /~username/ then its translated path from ** the .SEL file will be converted in the hit list back to ** a form which is acceptible to the server. See the example ** UserSearch.com htbin script for more information. These ** features also are implemented for searches of sectional ** databases in server data trees and /~username/ directories. ** VMSGopherServer.com and HTTP_IndexUserDoc.com for more ** information. The calling script can make the symbol ** WWW_OMIT_IPATH_CONVERSION non-NULL to suppress these ** conversions. ** The symbol WWW_DATADIRECTORY can be set by the calling script ** to a string which will be checked (case insensitive) in ** addition to the standard "WWW_Root" and "Gopher_Root" ** strings for access to paths in RANGE selectors (all other ** fetches are handled by the server and submitted to its own ** access checks). These checks are bypassed if IPATH ** conversions are implemented. ** The symbol WWW_FORBIDDEN can be set by the calling script to ** a string which will be checked like WWW_DATADIRECTORY, but ** to block access. This check is performed whether or not ** IPATH conversions are implemented. ** The symbol WWW_SEND_CONTENTS can be made non-NULL to specify ** that if the search results in a single item found, and that ** the script has NOT set WWW_OMIT_CONTENT_TYPE, that item ** will automaticlly be selected (actually, a redirect to the ** URL of the item is done). ** */ /* ** Usage: ** query database query ** query/www database query ** query/gopher database out-file query host port */ /* Query: expr {expr} implicit 'near' between expressions */ /* Expr: term {or term} */ /* Term: factor {and|not factor} */ /* Factor: (query) | token */ /* Token: a-z{a-z}[*] */ #include <ssdef.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <ctype.h> #include <rms.h> #include <descrip.h> #include <fscndef.h> #include <unixlib.h> #include <climsgdef.h> #include <lib$routines.h> #include <starlet.h> #include <stat.h> #include <time.h> #include "intctype.h" #define KEY_NAME 32 /* maximum size of key name */ #define MAX_QUERY 500 /* maximum size of query */ #define SPEC_SIZE 256 /* file specification size */ #define DEBUG 0 /* perform debugging printfs */ #define MOD_REALLOC 1 /* handle VAXC initial realloc() problem */ typedef struct { int index; /* selector index */ int file; /* file index */ int word; /* word index */ int score; /* result of near/and/or operator */ int pos; /* position in source document */ } Select; typedef struct { int count; /* number of members in result set */ Select *select; /* set of selectors */ } Result; typedef struct { char *str; /* token string */ } Token; extern query_commands(); extern int cli$dcl_parse(); extern int cli$get_value(); extern int cli$present(); static struct FAB idxfab, selfab, posfab; static struct RAB idxrab, selrab, posrab; static struct XABSUM xabsum; static struct XABKEY xabkey; static FILE *outfile; static int index_offset, last_field = 0; static char *index_field; static char *index_type = ".IDX"; static char *position_type = ".POS"; static char *selector_type = ".SEL"; static char orig_qstr[MAX_QUERY]; static char orig_idx[SPEC_SIZE]; /* buffer for index filespec from argv[1] */ static char **selector_name; /* array of selector file names */ static int selector_index = 0; /* selector file index */ static int *index_size; /* array of selector file index field sizes */ static Result last_result; static char Host[SPEC_SIZE]; /* server_name, and port if not 80 */ static char *BAD_HOST_LENGTH = "Deprecated host too long or too short."; static char *BAD_PATH_LENGTH = "Deprecated path too long or too short."; static char *BAD_REALM_LENGTH = "Required realm too long or too short."; static char *MISSING_ARGUMENTS = "Script invoked with missing arguments."; static char *QUERY_TOO_BIG = "Query exceeds maximum length."; static char *SPEC_TOO_BIG = "Path exceeds maximum length."; static char *IDX_NOT_INDEXED = "Idx file must be indexed."; static char *SEL_NOT_INDEXED = "Selector file must be indexed."; static char *POS_NOT_INDEXED = "Position file must be indexed."; static char *CLI_ERROR = "CLI processing error."; static char *TITLE = NULL; static short omit_content_type = FALSE; static short omit_head = FALSE; static short omit_foot = FALSE; static short show_score = FALSE; static char *hex = "0123456789ABCDEF"; /* For escaping URL's */ static unsigned char isAcceptable[96] = /* For escaping URL's */ /* 0 1 2 3 4 5 6 7 8 9 A B C D E F */ { 0,0,0,0,0,0,0,0,0,0,7,6,0,7,7,4, /* 2x !"#$%&'()*+,-./ */ 7,7,7,7,7,7,7,7,7,7,0,0,0,0,0,0, /* 3x 0123456789:;<=>? */ 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, /* 4x @ABCDEFGHIJKLMNO */ 7,7,7,7,7,7,7,7,7,7,7,0,0,0,0,7, /* 5X PQRSTUVWXYZ[\]^_ */ 0,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, /* 6x `abcdefghijklmno */ 7,7,7,7,7,7,7,7,7,7,7,0,0,0,0,0 }; /* 7X pqrstuvwxyz{\}~ DEL */ /* Search functions */ static int query(char *qstr, Result *result, Token *op); static void get_token(char *qstr, Token *token); static void *my_realloc(void *mem, int size); static void find(Token token, Result *rx); static int select_lt(Select s1, Select s2); static int select_eq(Select s1, Select s2); static void and(Result r1, Result r2, Result *r3); static void or(Result r1, Result r2, Result *r3); static void not(Result r1, Result r2, Result *r3); static int factor(char *qstr, Result *result, Token *op); static int term(char *qstr, Result *result, Token *op); static int expr(char *qstr, Result *result, Token *op); static int query(char *qstr, Result *result, Token *op); static void find_selector(Select sel); static void gopher_emit(Select, char*, char*); static int compare(const void *, const void *); static int switch_present(char *sw); static char *switch_value(char *sw); /* Format and output a hit list. */ static void select_result(Result result); static void www_emit(Result result); static void home_page(Result result); void parse_selector(char *str, int *new_style, char **title, char *gtype, char *ptype, char **file, char **host, char **port, char *method, char **path); /* Redirect or output a client's request for a RANGE fetch. */ static void display_result(char *file_name); /* Utilities */ static char *URLescape(char *str); static char *HTVMS_wwwName(char *vmsname); static char *FileType(char gtype); static void too_bad(char *reason); static int strcasecomp(char *a, char *b); static int strncasecomp(char *a, char *b, int n); int main(int argc, char *argv[]) { int status, ind, size, RunQuery, context = 0; short leng; static char input_spec[SPEC_SIZE], idx_spec[SPEC_SIZE]; static char file_name[SPEC_SIZE], cli_input[SPEC_SIZE]; static char p1[SPEC_SIZE], p2[MAX_QUERY], p3[MAX_QUERY]; static char p4[100], p5[100]; char *cp, qstr[MAX_QUERY], *port; Result file_result, result; Token token; struct fscndef scan_list[] = {{(short) 0, (short) FSCN$_NODE, (long) 0}, {(short) 0, (short) FSCN$_DEVICE, (long) 0}, {(short) 0, (short) FSCN$_DIRECTORY, (long) 0}, {(short) 0, (short) FSCN$_NAME, (long) 0}, {(short) 0, (short) FSCN$_VERSION, (long) 0}, {(short) 0, (short) 0, (long) 0}}; $DESCRIPTOR(cli_dsc, cli_input); $DESCRIPTOR(input_dsc, input_spec); $DESCRIPTOR(idx_dsc, idx_spec); $DESCRIPTOR(file_name_dsc, file_name); $DESCRIPTOR(p1_dsc, "p1"); $DESCRIPTOR(p1_value, p1); $DESCRIPTOR(p2_dsc, "p2"); $DESCRIPTOR(p2_value, p2); $DESCRIPTOR(p3_dsc, "p3"); $DESCRIPTOR(p3_value, p3); $DESCRIPTOR(p4_dsc, "p4"); $DESCRIPTOR(p4_value, p4); $DESCRIPTOR(p5_dsc, "p5"); $DESCRIPTOR(p5_value, p5); /* ** Get the host if present, and port if present and not 80. */ strcpy(Host, (cp=getenv("WWW_SERVER_NAME")) ? cp : ""); for (cp=Host; *cp; cp++) *cp = tolower(*cp); if ((port=getenv("WWW_SERVER_PORT")) != NULL && strcmp(port, "80")) { strcat(Host, ":"); strcat(Host, port); } /* ** Should we send the Content-type header? */ if (getenv("WWW_OMIT_CONTENT_TYPE") != NULL) omit_content_type = TRUE; /* ** Should we send the <HTML>, <HEAD>, <TITLE>, ** optional <ISINDEX>, </HEAD> and <BODY> tags? */ if (getenv("WWW_OMIT_HEAD") != NULL) omit_head = TRUE; /* ** Should we send the </BODY> and </HTML> tags? */ if (getenv("WWW_OMIT_FOOT") != NULL) omit_foot = TRUE; /* ** Get the TITLE if present. */ TITLE = getenv("WWW_VMSINDEX_TITLE"); /* ** Set up the CLI. */ status = lib$get_foreign(&cli_dsc, 0, &leng, 0); for (ind = leng; ind >= 0; ind--) cli_input[ind+6] = cli_input[ind]; strncpy(cli_input, "query ", 6); cli_dsc.dsc$w_length = leng+6; status = cli$dcl_parse(&cli_dsc, query_commands); if (status != CLI$_NORMAL) { /* error in parse, exit */ too_bad(CLI_ERROR); exit(1); } /* ** Should we include scores in titles? */ if (switch_present("score") || getenv("WWW_SHOW_SCORE") != NULL) show_score = TRUE; RunQuery = (!switch_present("gopher") && !(switch_present("www") || getenv("WWW_GATEWAY_INTERFACE") != NULL)); /* assume that the wrapper command file handles validation */ if (switch_present("gopher") && !switch_present("p5")) { too_bad(MISSING_ARGUMENTS); exit(1); } if ((switch_present("www") || getenv("WWW_GATEWAY_INTERFACE") != NULL) && !switch_present("p2")) { too_bad(MISSING_ARGUMENTS); exit(1); } strcpy(p1, switch_value("p1")); strcpy(p2, switch_value("p2")); strcpy(p3, switch_value("p3")); strcpy(p4, switch_value("p4")); strcpy(p5, switch_value("p5")); if (RunQuery && !switch_present("p1")) { printf("Index file to search: "); fgets(orig_idx, SPEC_SIZE, stdin); orig_idx[strlen(orig_idx)-1] = '\0'; /* remove \n */ } else strcpy(orig_idx, p1); /* ** Get the index specification. */ if (strlen(orig_idx) >= SPEC_SIZE) { too_bad(SPEC_TOO_BIG); exit(1); } if (RunQuery && !switch_present("p2")) { printf("Enter query: "); fgets(orig_qstr, MAX_QUERY, stdin); orig_qstr[strlen(orig_qstr)-1] = '\0'; /* remove \n */ } /* get the www and interactive query string */ if ((switch_present("www") || getenv("WWW_GATEWAY_INTERFACE") != NULL) || (RunQuery && switch_present("p2"))) { /* ** Get the query-string. */ if (!strlen(p2)) { /* passed an empty query? */ too_bad(MISSING_ARGUMENTS); exit(1); } if (strlen(p2) >= MAX_QUERY) { too_bad(QUERY_TOO_BIG); exit(1); } strcpy(orig_qstr, p2); /* make our copy of the query */ } /* get the gopher query */ if (switch_present("gopher")) { if (!strlen(p3)) { too_bad(MISSING_ARGUMENTS); exit(1); } if (strlen(p3) >= MAX_QUERY) { too_bad(QUERY_TOO_BIG); exit(1); } strcpy(orig_qstr, p3); /* make our copy of the query */ } /* If it's a WWW RANGE fetch, display_result() */ if ((switch_present("www") || getenv("WWW_GATEWAY_INTERFACE") != NULL) && ((strncasecomp(orig_qstr, "TEXT= R", 7) == 0) || (strncasecomp(orig_qstr, "TEXT=R", 6) == 0) || (strncasecomp(orig_qstr, "TEXT R", 6) == 0))) { display_result(orig_idx); exit(1); } else { for (cp = orig_qstr; *cp; cp++) *cp = tolower(*cp); } strcat(orig_qstr, " "); /* query ends with a space */ result.count = 0; /* init result */ result.select = NULL; strcpy(input_spec, orig_idx); /* set up descriptor to wildcard input spec */ input_dsc.dsc$w_length = (short) strlen(input_spec); if (((status = sys$filescan(&input_dsc, scan_list, 0)) & 1) != SS$_NORMAL) lib$stop(status); cp = NULL; size = 0; for (ind = 0; ind < 4; ind++) { if (cp == NULL) cp = (char *) scan_list[ind].fscn$l_addr; size += scan_list[ind].fscn$w_length; } strncpy(idx_spec, cp, size); /* copy node, dev, dir, name */ idx_spec[size] = '\0'; strcat(idx_spec, index_type); /* add .idx */ idx_dsc.dsc$w_length = (short) strlen(idx_spec); while (((status = lib$find_file(&idx_dsc, &file_name_dsc, &context, 0, 0, 0, 0)) & 1) == SS$_NORMAL) { /* while lib$find_file finds file names */ #if (DEBUG) printf("Find_file returned %s\n", file_name); #endif cp = strchr(file_name, ' '); if (cp) *cp = '\0'; /* chop off trailing spaces */ /* save the file names for when we need to get the selectors */ selector_name = (char **) my_realloc((char **) selector_name, (++selector_index+1) * sizeof(char *)); selector_name[selector_index] = (char *) calloc(strlen(file_name)+1, sizeof(char)); cp = strrchr(file_name, '.'); if (cp) *cp = '\0'; /* once again throw out file type */ strcpy(selector_name[selector_index], file_name); #if (DEBUG) printf("Saving selector index %d = %s\n", selector_index, selector_name[selector_index]); #endif /* initialize index fab and rab */ idxfab = cc$rms_fab; idxrab = cc$rms_rab; xabsum = cc$rms_xabsum; xabkey = cc$rms_xabkey; idxfab.fab$l_fna = file_name; idxfab.fab$b_fns = strlen(file_name); idxfab.fab$l_dna = index_type; idxfab.fab$b_dns = strlen(index_type); idxfab.fab$b_shr = FAB$M_SHRGET; /* don't allow writers */ idxfab.fab$l_xab = (char *) &xabsum; idxrab.rab$l_fab = (struct FAB *) &idxfab; idxrab.rab$b_rac = RAB$C_KEY; /* open index file */ if (((status = sys$open(&idxfab)) & 1) != SS$_NORMAL) lib$stop(status); if (idxfab.fab$b_org != FAB$C_IDX) { too_bad(IDX_NOT_INDEXED); exit(1); } idxfab.fab$l_xab = (char *) &xabkey; /* fill in the key XAB */ if (((status = sys$display(&idxfab)) & 1) != SS$_NORMAL) lib$stop(status); /* add another index_size element for new idx file */ index_size = (int *) my_realloc((int *) index_size, (selector_index+1) * sizeof(int)); /* index field size is (MRS - Key 0 size)/2 if 3 fields */ /* MRS = Keyword+topic+word+count (last 3 fields are equal size) */ /* Key 0 = Keyword+topic */ if ((idxfab.fab$w_mrs - xabkey.xab$b_siz0) < 10) /* 2 fields? */ index_size[selector_index] = (idxfab.fab$w_mrs - xabkey.xab$b_siz0); else index_size[selector_index] = (idxfab.fab$w_mrs - xabkey.xab$b_siz0)/2; index_offset = xabkey.xab$w_pos0 + xabkey.xab$b_siz0 - index_size[selector_index]; index_field = (char *) my_realloc((char *) index_field, (index_size[selector_index]+1) * sizeof(char)); strncpy(index_field, "", index_size[selector_index]+1); #if (DEBUG) printf("Selector %d index size = %d key_size = %d\n", selector_index, index_size[selector_index], xabkey.xab$b_siz0 - index_size[selector_index]); #endif /* make the index file record */ idxrab.rab$w_usz = idxfab.fab$w_mrs; idxrab.rab$l_ubf = (char *) my_realloc((char *) idxrab.rab$l_ubf, (idxrab.rab$w_usz+1) * sizeof(char)); strncpy(idxrab.rab$l_ubf, "", idxrab.rab$w_usz+1); /* connect record streams */ if (((status = sys$connect(&idxrab)) & 1) != SS$_NORMAL) lib$stop(status); strcpy(qstr, orig_qstr); /* query and friends mangle qstr */ query(qstr, &file_result, &token); /* evaluate the query on this file */ or(result, file_result, &result); /* accumulate results */ status = sys$close(&idxfab); } /* while finding input files */ if (switch_present("sort") || getenv("WWW_SORT") != NULL) { /* * Before sorting, elevate the ranking of any hit that is a "home" * page or otherwise preferred. - RDP */ if (result.count >= 2) home_page(result); for (ind = 0; ind < result.count; ind++) result.select[ind].pos = ind; /* remember position */ qsort(result.select, result.count, sizeof(Select), compare); /* sort the results */ /* * discard any results with zero score -- RDP */ for (ind = result.count - 1; ind >= 0; ind--) { if (result.select[ind].score == 0) result.count = ind; } } if (RunQuery) { select_result(result); } else if (switch_present("www") || getenv("WWW_GATEWAY_INTERFACE") != NULL) { orig_qstr[strlen(orig_qstr)-1] = '\0'; /* Trim trailing space */ www_emit(result); } else if (switch_present("gopher")) { /* open the output file and write the resulting selector set */ outfile = fopen(p2, "a"); if (outfile == NULL) { perror("Output file could not be opened"); exit(1); } for (ind = 0; ind < result.count; ind++) gopher_emit(result.select[ind], p4, p5); fclose(outfile); } free(selrab.rab$l_ubf); status = sys$close(&selfab); } /* qsort compare function */ int compare(const void *s1, const void *s2) { int result; result = ((*(Select *) s2).score - (*(Select *) s1).score); /* if the score is the same, sort on source file position */ if (result == 0) result = ((*(Select *) s1).pos - (*(Select *) s2).pos); return result; } void open_selector(char *file_name) { int status; /* initialize selector fab and rab */ selfab = cc$rms_fab; selrab = cc$rms_rab; selfab.fab$l_fna = file_name; selfab.fab$b_fns = strlen(file_name); selfab.fab$l_dna = selector_type; selfab.fab$b_dns = strlen(selector_type); selfab.fab$b_shr = FAB$M_SHRGET; selrab.rab$l_fab = (struct FAB *) &selfab; selrab.rab$b_rac = RAB$C_KEY; /* open selector file */ if (((status = sys$open(&selfab)) & 1) != SS$_NORMAL) lib$stop(status); if (selfab.fab$b_org != FAB$C_IDX) { too_bad(SEL_NOT_INDEXED); exit(1); } /* make the selector file record */ selrab.rab$w_usz = selfab.fab$w_mrs; selrab.rab$l_ubf = (char *) my_realloc((char *) selrab.rab$l_ubf, (selrab.rab$w_usz+1) * sizeof(char)); strncpy(selrab.rab$l_ubf, "", selrab.rab$w_usz+1); if (((status = sys$connect(&selrab)) & 1) != SS$_NORMAL) lib$stop(status); } int open_position(char *file_name) { int status; /* initialize position fab and rab */ posfab = cc$rms_fab; posrab = cc$rms_rab; posfab.fab$l_fna = file_name; posfab.fab$b_fns = strlen(file_name); posfab.fab$l_dna = position_type; posfab.fab$b_dns = strlen(position_type); posfab.fab$b_shr = FAB$M_SHRGET; posrab.rab$l_fab = (struct FAB *) &posfab; posrab.rab$b_rac = RAB$C_KEY; /* open position file */ if (((status = sys$open(&posfab)) & 1) != SS$_NORMAL) return 1; if (posfab.fab$b_org != FAB$C_IDX) { too_bad(POS_NOT_INDEXED); exit(1); } /* make the position file record */ posrab.rab$w_usz = posfab.fab$w_mrs; posrab.rab$l_ubf = (char *) my_realloc((char *) posrab.rab$l_ubf, (posrab.rab$w_usz+1) * sizeof(char)); strncpy(posrab.rab$l_ubf, "", posrab.rab$w_usz+1); if (((status = sys$connect(&posrab)) & 1) != SS$_NORMAL) return 1; return 0; /* 0 = success */ } /* return the next token */ void get_token(char *qstr, Token *token) { char *str = NULL, *cp; #if (DEBUG) printf("Token: '%s' ", qstr); #endif while (*qstr && (*qstr <= ' ')) /* remove leading spaces and junk */ strcpy(qstr, qstr+1); /* if at a paren, strip and return it as the token */ if ((*qstr == '(') || (*qstr == ')')) { str = (char *) calloc(2, sizeof(char)); strncpy(str, qstr, 1); strcpy(qstr, qstr+1); } else { /* point cp to the end of the token */ if (strchr(qstr, ' ') && strchr(qstr, ')')) cp = strchr(qstr, ' ') < strchr(qstr, ')') ? strchr(qstr, ' ') : strchr(qstr, ')'); else if (strchr(qstr, ')') == NULL) cp = strchr(qstr, ' '); else if (strchr(qstr, ' ') == NULL) cp = strchr(qstr, ')'); else cp = NULL; if (cp) { str = (char *) calloc(xabkey.xab$b_siz0+1, sizeof(char)); strncpy(str, qstr, cp - qstr); if (*cp == ')') strcpy(qstr, cp); /* save ending ')' */ else strcpy(qstr, cp+1); /* strip space */ } else str = (char *) calloc(1, sizeof(char)); } #if (DEBUG) printf("Returns: '%s' Op: '%s'\n", qstr, str); #endif token->str = str; } /* Lance was right, realloc sometimes blows when initially allocating memory MOD_REALLOC indicates whether to use malloc() on initial allocation */ void *my_realloc(void *mem, int size) { void *mem_ptr; if ((mem == (void *) 0) && (MOD_REALLOC)) return ((void *) malloc(size)); else { mem_ptr = (void *) realloc(mem, size); return mem_ptr; } } /* create a set of selectors that are associated with the token */ void find(Token token, Result *rx) { int status, ind, value, word_id, count; rx->count = 0; /* assume no match */ rx->select = NULL; idxrab.rab$b_rac = RAB$C_KEY; idxrab.rab$b_krf = 0; idxrab.rab$l_kbf = token.str; idxrab.rab$l_rop = 0; /* set up exact match */ idxrab.rab$b_ksz = xabkey.xab$b_siz0 - index_size[selector_index]; if (token.str[strlen(token.str) - 1] == '*') { idxrab.rab$b_ksz = strlen(token.str) - 1; idxrab.rab$l_rop = RAB$M_KGE; /* set up approximate generic match */ } /* key can't be shorter than field size */ while (strlen(token.str) < idxrab.rab$b_ksz) strcat(token.str, " "); /* find the start record */ if (((status = sys$find(&idxrab)) & 1) != SS$_NORMAL) { #if (DEBUG) printf("Find: %s no match\n", token.str); #endif return; /* no match */ } idxrab.rab$b_rac = RAB$C_SEQ; while (((status = sys$get(&idxrab)) & 1) == SS$_NORMAL) { if (strncmp((char *) (idxrab.rab$l_ubf + xabkey.xab$w_pos0), token.str, idxrab.rab$b_ksz) != 0) break; /* no match */ /* get first field (selector ID) */ strncpy(index_field, (char *) (idxrab.rab$l_ubf + index_offset), index_size[selector_index]); value = atoi(index_field); /* get second field */ strncpy(index_field, (char *) (idxrab.rab$l_ubf + index_offset + index_size[selector_index]), index_size[selector_index]); word_id = atoi(index_field); /* assume that it's the word ID */ if (word_id == 0) /* word_id = 0 means /noposition */ word_id = -1; /* /* is there a third field? */ if (idxfab.fab$w_mrs <= (index_offset + index_size[selector_index] * 2)) { count = word_id; /* no, second field = count */ word_id = -1; /* there is no word id */ } else { /* get third field (count) */ strncpy(index_field, (char *) (idxrab.rab$l_ubf + index_offset + index_size[selector_index] * 2), index_size[selector_index]); count = atoi(index_field); } for (ind = 0; ind < rx->count; ind++) if ((rx->select[ind].index == value) && (rx->select[ind].file == selector_index)) /* if the value already there */ break; /* don't add it */ /* unfortunately, you can't put a 'continue' in the previous line */ if ((ind < rx->count) && (rx->select[ind].index == value) && (rx->select[ind].file == selector_index)) continue; rx->select = (Select *) my_realloc((Select *) rx->select, (rx->count+1) * sizeof(Select)); /* keep the values in ascending order */ for (ind = rx->count; ind >= 0; ind--) if ((ind == 0) || (rx->select[ind - 1].file < selector_index) || ((rx->select[ind - 1].file == selector_index) && (rx->select[ind - 1].index < value))) { rx->select[ind].file = selector_index; rx->select[ind].index = value; rx->select[ind].word = word_id; rx->select[ind].score = count * 20; /* fudge to scale up mult single word hits */ break; } else rx->select[ind] = rx->select[ind - 1]; rx->count++; } #if (DEBUG) printf("Find: %s -> ", token.str); for (ind = 0; ind < rx->count; ind++) printf("%d-%d:%d:%d ", rx->select[ind].file, rx->select[ind].index, rx->select[ind].word, rx->select[ind].score); printf("\n"); #endif } /* selector booleans */ int select_lt(Select s1, Select s2) { return ((s1.file < s2.file) || ((s1.file == s2.file) && (s1.index < s2.index))); } int select_eq(Select s1, Select s2) { return ((s1.file == s2.file) && (s1.index == s2.index)); } /* return the proximity of word1 to word2 */ /* closer words get higher scores */ int proximity(Select s1, Select s2) { int siz1 = 0, siz2 = 0, ind1, ind2, *array1 = (int *)0, *array2 = (int *)0; int status, score, prox = 999999; #if (DEBUG) printf("Enter proximity()\n"); printf("S1: %d-%d %d %d\n", s1.file, s1.index, s1.word, s1.score); printf("S2: %d-%d %d %d\n", s2.file, s2.index, s2.word, s2.score); #endif if ((s1.word <= 0) || (s2.word <= 0)) /* s1 or s2 aren't words, but booleans */ return (s1.score > s2.score ? s1.score : s2.score); /* quit now */ if (open_position(selector_name[s1.file])) return 10; /* position file not there, return a constant proximity */ sprintf(index_field, "%0*d", index_size[s1.file], s1.word); posrab.rab$b_rac = RAB$C_KEY; posrab.rab$b_krf = 0; posrab.rab$l_kbf = index_field; posrab.rab$l_rop = 0; /* set up exact match */ posrab.rab$b_ksz = index_size[s1.file]; for (;;) { /* find the word */ if (((status = sys$get(&posrab)) & 1) != SS$_NORMAL) break; * (char *) (posrab.rab$l_ubf + posrab.rab$w_rsz) = '\0'; /* terminate string */ posrab.rab$b_rac = RAB$C_SEQ; if (strncmp(posrab.rab$l_ubf, index_field, index_size[s1.file])) break; array1 = (int *) my_realloc((int *) array1, (siz1+1) * sizeof(int)); array1[siz1++] = atoi(posrab.rab$l_ubf + index_size[s1.file]); } sprintf(index_field, "%0*d", index_size[s2.file], s2.word); posrab.rab$b_rac = RAB$C_KEY; for (;;) { /* find the word */ if (((status = sys$get(&posrab)) & 1) != SS$_NORMAL) break; * (char *) (posrab.rab$l_ubf + posrab.rab$w_rsz) = '\0'; /* terminate string */ posrab.rab$b_rac = RAB$C_SEQ; if (strncmp(posrab.rab$l_ubf, index_field, index_size[s2.file])) break; array2 = (int *) my_realloc((int *) array2, (siz2+1) * sizeof(int)); array2[siz2++] = atoi(posrab.rab$l_ubf + index_size[s2.file]); } free(posrab.rab$l_ubf); status = sys$close(&posfab); for (ind1 = 0; ind1 < siz1; ind1++) { for (ind2 = 0; ind2 < siz2; ind2++) { score = array2[ind2] - array1[ind1]; if (score < 0) score = -(score*2); /* penalize them if they are in the wrong order */ if (score < prox) prox = score; /* record best score */ } } free(array1); free(array2); /* here, the best prox = 1; lets scale it to 100 */ if (prox > 100) prox = 1; /* very bad scores become 1 */ else prox = 101 - prox; #if (DEBUG) printf("Exit proximity with score %d\n", prox); #endif return prox; } /* return selector where word1 is near word2 */ void near(Result r1, Result r2, Result *r3) { int ind1, ind2; Result rx, temp_result; rx.count = 0; rx.select = NULL; temp_result.count = 0; temp_result.select = NULL; #if (DEBUG) printf("Enter near()\nR1: "); for (ind1 = 0; ind1 < r1.count; ind1++) printf("%d-%d ", r1.select[ind1].file, r1.select[ind1].index); printf("\n"); printf("R2: "); for (ind2 = 0; ind2 < r2.count; ind2++) printf("%d-%d ", r2.select[ind2].file, r2.select[ind2].index); printf("\n"); #endif ind1 = ind2 = 0; if (r1.count && (r1.select[0].word == 0)) { /* r1 is a composite */ near(last_result, r2, &temp_result); /* compare r2 with last token */ and(r1, temp_result, &rx); /* intersect results */ } else if (r1.count && (r1.select[0].word == -1)) { /* r1 has no positions */ and(r1, r2, &rx); /* just intersect results */ } else { for (;;) { if ((ind1 == r1.count) || (ind2 == r2.count)) break; else if (select_lt(r1.select[ind1], r2.select[ind2])) ind1++; else if (select_lt(r2.select[ind2], r1.select[ind1])) ind2++; else if (select_eq(r1.select[ind1], r2.select[ind2])) { rx.select = (Select *) my_realloc((Select *) rx.select, (rx.count+1) * sizeof(Select)); rx.select[rx.count].file = r1.select[ind1].file; rx.select[rx.count].index = r1.select[ind1].index; rx.select[rx.count].word = 0; rx.select[rx.count].score = proximity(r1.select[ind1], r2.select[ind2]); rx.count++; ind1++; ind2++; } } /* make a copy of the result in last_result for use by near() */ last_result.select = (Select *) my_realloc((Select *) last_result.select, (r2.count) * sizeof(Select)); last_result.count = r2.count; for (ind2 = 0; ind2 < r2.count; ind2++) { last_result.select[ind2].file = r2.select[ind2].file; last_result.select[ind2].index = r2.select[ind2].index; last_result.select[ind2].word = r2.select[ind2].word; last_result.select[ind2].score = r2.select[ind2].score; } } if (r3->select) free(r3->select); *r3 = rx; #if (DEBUG) printf("NEAR: "); for (ind1 = 0; ind1 < rx.count; ind1++) printf("%d-%d:%d ", rx.select[ind1].file, rx.select[ind1].index, rx.select[ind1].score); printf("\n"); #endif } /* perform set intersection */ void and(Result r1, Result r2, Result *r3) { int ind1, ind2; Result rx; rx.count = 0; rx.select = NULL; #if (DEBUG) printf("Enter and()\nR1: "); for (ind1 = 0; ind1 < r1.count; ind1++) printf("%d-%d:%d ", r1.select[ind1].file, r1.select[ind1].index, r1.select[ind1].score); printf("\n"); printf("R2: "); for (ind2 = 0; ind2 < r2.count; ind2++) printf("%d-%d:%d ", r2.select[ind2].file, r2.select[ind2].index, r2.select[ind2].score); printf("\n"); #endif ind1 = ind2 = 0; for (;;) { if ((ind1 == r1.count) || (ind2 == r2.count)) break; else if (select_lt(r1.select[ind1], r2.select[ind2])) ind1++; else if (select_lt(r2.select[ind2], r1.select[ind1])) ind2++; else if (select_eq(r1.select[ind1], r2.select[ind2])) { rx.select = (Select *) my_realloc((Select *) rx.select, (rx.count+1) * sizeof(Select)); rx.select[rx.count].file = r1.select[ind1].file; rx.select[rx.count].index = r1.select[ind1].index; rx.select[rx.count].score = r1.select[ind1].score < r2.select[ind2].score ? r1.select[ind1].score : r2.select[ind2].score; rx.count++; ind1++; ind2++; } } if (r3->select) free(r3->select); *r3 = rx; #if (DEBUG) printf("AND: "); for (ind1 = 0; ind1 < rx.count; ind1++) printf("%d-%d:%d ", rx.select[ind1].file, rx.select[ind1].index, rx.select[ind1].score); printf("\n"); #endif } /* perform set inclusion */ void or(Result r1, Result r2, Result *r3) { int ind1, ind2; Result rx; rx.count = 0; rx.select = NULL; #if (DEBUG) printf("Enter or()\nR1: "); for (ind1 = 0; ind1 < r1.count; ind1++) printf("%d-%d:%d ", r1.select[ind1].file, r1.select[ind1].index, r1.select[ind1].score); printf("\n"); printf("R2: "); for (ind2 = 0; ind2 < r2.count; ind2++) printf("%d-%d:%d ", r2.select[ind2].file, r2.select[ind2].index, r2.select[ind2].score); printf("\n"); #endif ind1 = ind2 = 0; for (;;) { if ((ind1 == r1.count) && (ind2 == r2.count)) break; else if ((ind2 == r2.count) || ((ind1 < r1.count) && select_lt(r1.select[ind1], r2.select[ind2]))) { rx.select = (Select *) my_realloc((Select *) rx.select, (rx.count+1) * sizeof(Select)); rx.select[rx.count].file = r1.select[ind1].file; rx.select[rx.count].index = r1.select[ind1].index; rx.select[rx.count].score = r1.select[ind1].score; rx.count++; ind1++; } else if ((ind1 == r1.count) || ((ind2 < r2.count) && select_lt(r2.select[ind2], r1.select[ind1]))) { rx.select = (Select *) my_realloc((Select *) rx.select, (rx.count+1) * sizeof(Select)); rx.select[rx.count].file = r2.select[ind2].file; rx.select[rx.count].index = r2.select[ind2].index; rx.select[rx.count].score = r2.select[ind2].score; rx.count++; ind2++; } else if (select_eq(r1.select[ind1], r2.select[ind2])) { rx.select = (Select *) my_realloc((Select *) rx.select, (rx.count+1) * sizeof(Select)); rx.select[rx.count].file = r1.select[ind1].file; rx.select[rx.count].index = r1.select[ind1].index; rx.select[rx.count].score = r1.select[ind1].score > r2.select[ind2].score ? r1.select[ind1].score : r2.select[ind2].score; rx.count++; ind1++; ind2++; } } if (r3->select) free(r3->select); *r3 = rx; #if (DEBUG) printf("OR: "); for (ind1 = 0; ind1 < rx.count; ind1++) printf("%d-%d:%d ", rx.select[ind1].file, rx.select[ind1].index, rx.select[ind1].score); printf("\n"); #endif } /* perform set exclusion */ void not(Result r1, Result r2, Result *r3) { int ind1, ind2; Result rx; rx.count = 0; rx.select = NULL; #if (DEBUG) printf("Enter not()\nR1: "); for (ind1 = 0; ind1 < r1.count; ind1++) printf("%d-%d:%d ", r1.select[ind1].file, r1.select[ind1].index, r1.select[ind1].score); printf("\n"); printf("R2: "); for (ind2 = 0; ind2 < r2.count; ind2++) printf("%d-%d:%d ", r2.select[ind2].file, r2.select[ind2].index, r2.select[ind2].score); printf("\n"); #endif ind1 = ind2 = 0; for (;;) { if (ind1 == r1.count) break; else if ((ind2 == r2.count) || select_lt(r1.select[ind1], r2.select[ind2])) { rx.select = (Select *) my_realloc((Select *) rx.select, (rx.count+1) * sizeof(Select)); rx.select[rx.count].file = r1.select[ind1].file; rx.select[rx.count].index = r1.select[ind1].index; rx.select[rx.count].score = r1.select[ind1].score; rx.count++; ind1++; } else if (select_lt(r2.select[ind2], r1.select[ind1])) ind2++; else if (select_eq(r1.select[ind1], r2.select[ind2])) { ind1++; ind2++; } } if (r3->select) free(r3->select); *r3 = rx; #if (DEBUG) printf("NOT: "); for (ind1 = 0; ind1 < rx.count; ind1++) printf("%d-%d:%d ", rx.select[ind1].file, rx.select[ind1].index, rx.select[ind1].score); printf("\n"); #endif } /* parse factor */ int factor(char *qstr, Result *result, Token *op) { Token token; #if (DEBUG) printf("Enter factor()\n"); #endif get_token(qstr, &token); if (strcmp(token.str, "(") == 0) { if (!query(qstr, result, op)) { #if (DEBUG) printf("Query failed; exit Factor with failure\n"); #endif return 0; } if (strcmp(op->str, ")") != 0) { #if (DEBUG) printf("Query success, no rparen; exit Factor with failure\n"); #endif return 0; } } else find(token, result); free(token.str); get_token(qstr, op); #if (DEBUG) printf("Exit factor()\n"); #endif return 1; } /* parse term */ int term(char *qstr, Result *result, Token *op) { Result temp; Token token; #if (DEBUG) printf("Enter term()\n"); #endif if (factor(qstr, result, op)) { while ((strcmp(op->str, "and") == 0) || (strcmp(op->str, "not") == 0)) { if (!factor(qstr, &temp, &token)) return 0; if (strcmp(op->str, "and") == 0) and(*result, temp, result); else if (strcmp(op->str, "not") == 0) not(*result, temp, result); free(op->str); *op = token; free(temp.select); } #if (DEBUG) printf("Exit term()\n"); #endif return 1; } #if (DEBUG) printf("Exit Term with failure\n"); #endif return 0; } /* parse expression */ int expr(char *qstr, Result *result, Token *op) { Result temp; Token token; #if (DEBUG) printf("Enter expr()\n"); #endif if (term(qstr, result, op)) { while (strcmp(op->str, "or") == 0) { if (!term(qstr, &temp, &token)) return 0; or(*result, temp, result); free(op->str); *op = token; free(temp.select); } #if (DEBUG) printf("Exit expr()\n"); #endif return 1; } #if (DEBUG) printf("Exit Expr with failure\n"); #endif return 0; } /* parse query */ int query(char *qstr, Result *result, Token *op) { Result result2; char temp_qstr[MAX_QUERY], *cp; int status, too_small, min_word = 0; #if (DEBUG) printf("Enter query()\n"); #endif status = expr(qstr, result, op); /* evaluate the expr */ while ((strlen(op->str) > 0) && /* non-boolean at end of expression */ (strcmp(op->str, ")") != 0)) { /* and it isn't a ")" */ #if (DEBUG) printf("Found \"%s\" at end of expression; 'near' assumed\n", op->str); #endif /* Check whether we've set a minimum word length. */ if ((cp=getenv("WWW_MINIMUM_WORD")) != NULL) { min_word = atoi(cp); } too_small = (strlen(op->str) < min_word); strcpy(temp_qstr, op->str); /* put token */ strcat(temp_qstr, " "); /* back on the */ strcat(temp_qstr, qstr); /* front of the query */ strcpy(qstr, temp_qstr); free(op->str); status = expr(qstr, &result2, op); if (!too_small) near(*result, result2, result); /* how close are the expressions? */ free(result2.select); } free(op->str); #if (DEBUG) printf("Exit query()\n"); #endif return status; } /* read selector record */ void find_selector(Select sel) { int status; static int current_file = 0; sprintf(index_field, "%0*d", index_size[sel.file], sel.index); if (sel.file != current_file) { current_file = sel.file; status = sys$close(&selfab); open_selector(selector_name[current_file]); } selrab.rab$b_rac = RAB$C_KEY; selrab.rab$b_krf = 0; selrab.rab$l_kbf = index_field; selrab.rab$l_rop = 0; /* set up exact match */ selrab.rab$b_ksz = index_size[sel.file]; /* find the selector record */ if (((status = sys$get(&selrab)) & 1) != SS$_NORMAL) lib$stop(status); *(char *) (selrab.rab$l_ubf + selrab.rab$w_rsz) = '\0'; /* terminate string */ } /* parse the selector record into it's parts ** this reads two different formats and returns which type in new_style ** old style: <gtype>title<tab><ptype>file<tab>host<tab>port ** new style: title<tab>|gtype|ptype|file|host:port|method|path ** Note: gtype is a pointer to a single character ** ptype must already be allocated at least 22 bytes ** method must already be allocated at least 5 bytes ** The contents of '*str' will be riddled with nulls */ void parse_selector(char *str, int *new_style, char **title, char *gtype, char *ptype, char **file, char **host, char **port, char *method, char **path) { char *ptr1, *ptr2; ptr1 = strchr(str, '\t'); /* find end of topic title */ *ptr1++ = '\0'; /* mark end of title */ *new_style = (*ptr1 == '|'); /* determine record type */ if (*new_style == 0) { /* handle the old style record */ *gtype = *str; *title = str+1; if (*ptr1 == 'R') { /* old style range */ ptr2 = strchr(ptr1, '-'); /* skip to second dash */ ptr2 = strchr(ptr2+1, '-'); *ptr2 = '\0'; /* mark end of ptype */ strcpy(ptype, ptr1); /* copy ptype Rnnnn-nnnn */ } else { *ptype = *ptr1; /* copy single character ptype */ *(ptype+1) = '\0'; ptr2 = ptr1; /* ptr2 = last character of ptype */ } *file = ptr2+1; /* start of file name */ ptr1 = strchr(*file, '\t'); /* find end of old style filename */ if (ptr1) *ptr1++ = '\0'; /* mark end of file name */ else { /* record ends after file name */ method[0] = '\0'; *host = *port = *path = ""; return; /* other fields are empty; done */ } *host = ptr1; /* start of host name */ ptr1 = strchr(*host, '\t'); /* find end of host name */ if (ptr1) *ptr1++ = '\0'; /* mark end of host name */ else { /* record ends after file name */ method[0] = '\0'; *port = *path = ""; return; /* other fields are empty; done */ } *port = ptr1; /* start of port number */ method[0] = '\0'; *path = ""; return; /* other fields are empty; done */ } /* new style selector */ *title = str; /* record title location */ ptr1++; /* skip over first | */ ptr2 = strchr(ptr1, '|'); /* find end of gtype field */ if (ptr2) *ptr2++ = '\0'; /* mark end of gtype field */ else ptr2 = ""; /* else point to empty string */ *gtype = *(ptr1); /* copy gtype */ ptr1 = ptr2; /* ptr1 = start of next field */ ptr2 = strchr(ptr1, '|'); /* find end of ptype field */ if (ptr2) *ptr2++ = '\0'; /* mark end of ptype field */ else ptr2 = ""; /* else point to empty string */ strcpy(ptype, ptr1); /* copy ptype */ ptr1 = ptr2; /* ptr1 = start of next field */ ptr2 = strchr(ptr1, '|'); /* find end of filename field */ if (ptr2) *ptr2++ = '\0'; /* mark end of ptype field */ else ptr2 = ""; /* else point to empty string */ *file = ptr1; /* record filename location */ ptr1 = ptr2; /* ptr1 = start of next field */ ptr2 = strchr(ptr1, '|'); /* find end of hostname field */ if (ptr2) *ptr2++ = '\0'; /* mark end of hostname field */ else ptr2 = ""; /* else point to empty string */ *host = ptr1; /* record host location */ ptr1 = ptr2; /* ptr1 = start of next field */ ptr2 = strchr(ptr1, '|'); /* find end of method field */ if (ptr2) *ptr2++ = '\0'; /* mark end of method field */ else ptr2 = ""; /* else point to empty string */ strcpy(method, ptr1); /* record method */ ptr1 = ptr2; /* ptr1 = start of next field */ ptr2 = strchr(ptr1, '|'); /* find end of path field */ if (ptr2) *ptr2++ = '\0'; /* mark end of path field */ else ptr2 = ""; /* else point to empty string */ *path = ptr1; /* record path location */ ptr1 = ptr2; /* ptr1 = start of next field */ ptr1 = strchr(*host, ':'); /* find start of port number field */ if (ptr1) *ptr1++ = '\0'; /* mark end of hostname field */ else ptr1 = ""; /* else point to empty string */ *port = ptr1; /* record port location */ } /* output gopher-style selector */ /* lookup selector with key 'index'; print selector, host, port */ void gopher_emit(Select sel, char *host, char *port) { char *ptr, score[10], ptype[25], method[10]; char *title, gtype, *file, *host2, *port2, *path; int new_style; #if (DEBUG) printf("Selector %d\n", sel.index); #endif find_selector(sel); /* this assumes that the first 'index_size' characters of the selector record is the selector index field and the rest is the selector itself */ ptr = (char *) (selrab.rab$l_ubf + index_size[sel.file]); parse_selector(ptr, &new_style, &title, >ype, ptype, &file, &host2, &port2, method, &path); if (show_score) /* include scores in title? */ sprintf(score, "[%d] ", sel.score); else strcpy(score, ""); if (ptype[0] == 'R') /* parse_selector() trims trailing dash */ strcat(ptype,"-"); fprintf(outfile, "%c%s%s\t%s%s\t%s\t%s\n", gtype, title, score, ptype, file, strlen(host2) ? host2 : host, strlen(port2) ? port2 : port); } void select_result(Result result) { int ind, start, end, new_style; char *ptr1, *title, gtype, *file_name, *host, *port, *path; char ptype[25], inputline[SPEC_SIZE], answer[20], score[10], method[10]; FILE *fp = NULL; printf("There are %d topic(s) found:\n", result.count); /* list all the selectors */ for (ind = 0; ind < result.count; ind++) { find_selector(result.select[ind]); ptr1 = (char *) (selrab.rab$l_ubf + index_size[result.select[ind].file]); parse_selector(ptr1, &new_style, &title, >ype, ptype, &file_name, &host, &port, method, &path); if (show_score) /* include scores in title? */ sprintf(score, "[%d]", result.select[ind].score); else strcpy(score, ""); printf("%d. %s (%c) %s\n", ind+1, score, ptype[0], title); } /* display the topics selected */ for (;;) { do { ind = 0; /* if non-numeric, ind will stay 0 */ printf("\nSelect topic to view [0 to quit]: "); fgets(answer, 10, stdin); sscanf(answer, "%d", &ind); } while ((ind < 0) || (ind > result.count)); if (ind == 0) return; find_selector(result.select[ind-1]); ptr1 = (char *) (selrab.rab$l_ubf + index_size[result.select[ind-1].file]); parse_selector(ptr1, &new_style, &title, >ype, ptype, &file_name, &host, &port, method, &path); if ((ptype[0] == 'R')) { /* Section from a text database */ sscanf(ptype+1, "%d-%d", &start, &end); if ((fp = fopen(file_name, "r", "shr=get", "mbc=32")) != NULL) { fseek(fp, start, SEEK_SET); printf("\n\nThis is from the document %s\n\n", file_name); while (fgets(inputline, sizeof(inputline), fp) != NULL) { printf("%s", inputline); if (ftell(fp) >= end) break; } } else { printf("\n\nCould not access the document %s\n", file_name); } } else if (ptype[0] == '0') { /* Whole text file */ if ((fp = fopen(file_name, "r", "shr=get", "mbc=32")) != NULL) { while (fgets(inputline, sizeof(inputline), fp) != NULL) printf("%s", inputline); } else { printf("\n\nCould not access the document %s\n", file_name); } } else printf("\n\nThis is not a text document\n"); if (fp) fclose(fp); } } /* ** Format and output a hit list. */ static void www_emit(Result result) { int ind, start, end, bytes, status; struct stat buf; char *ThisYear; time_t elapsed; struct tm *local; char cdate[26]; #define MMM cdate+4 #define DD cdate+8 #define HHMM cdate+11 #define YY cdate+22 unsigned short send_raw = FALSE; unsigned short show_date = FALSE, show_time = FALSE, use_mdate = FALSE; unsigned short show_size = FALSE; unsigned short max_hits = FALSE; unsigned short use_ipath_info = FALSE; int MaxHits = 0; int StartingHit, EndingHit, TotalHits; unsigned short via_form = FALSE; unsigned short add_context = FALSE; unsigned short HavePrevious = FALSE, HaveNext = FALSE; char *cp, *cp1, *ptr1, *name, *path; char score[10], ptype[25], method[10]; char *title, *file; int new_style; char *file_name, size[128], DateAndOrSize[256]; char items_found[256], items_listed[256]; char ipath_info[SPEC_SIZE],required_realm[250]; char ipath_translated[SPEC_SIZE], alt_ipath_translated[SPEC_SIZE]; char *alt_path=NULL; char *hits_target=NULL; char gtype; char *host, *port; char *gopher_title=NULL; FILE *fp; /* * Check for required realm */ if ((cp=getenv("WWW_REQUIRED_REALM")) == NULL) { required_realm[0] = '\0'; } else { if (strlen(cp) <= 2) { required_realm[0] ='\0'; } else { if (strlen(cp) <= 249) { strcpy(required_realm, cp); } else { too_bad(BAD_REALM_LENGTH); exit(1); } } } /* Check whether we're using whole files. */ if (getenv("WWW_SEND_RAW_FILE") != NULL) send_raw = TRUE; /* Check whether we've set an items found string with a %d. */ if ((cp=getenv("WWW_ITEMS_FOUND")) != NULL && strstr(cp, "%d") != NULL) { strcpy(items_found, cp); } else { strcpy(items_found, "There are %d item(s) found"); } /* Check whether we've set an items listed string with two %d's. */ if ((cp=getenv("WWW_ITEMS_LISTED")) != NULL && (ptr1=strstr(cp, "%d")) != NULL && strstr((ptr1+2), "%d") != NULL) { sprintf(items_listed, ".\n%s", cp); } else { strcpy(items_listed, ".\nListing %d to %d"); } /* Check whether we've set a maximum number of hits. */ if ((cp=getenv("WWW_MAXIMUM_HITS")) != NULL) { MaxHits = atoi(cp); max_hits = TRUE; } /* Check whether we've set a starting number for hits. */ if ((cp=getenv("WWW_STARTING_HIT")) != NULL) { StartingHit = atoi(cp); } else { StartingHit = 1; } /* Check whether we're showing the size. */ if ((cp=getenv("WWW_SHOW_SIZE")) != NULL) show_size = TRUE; /* Check whether we're showing the date. */ if ((cp=getenv("WWW_SHOW_DATE")) != NULL) { show_date = TRUE; /* Do we prefer the modification date? */ if ((cp=getenv("WWW_USE_MDATE")) != NULL) use_mdate = TRUE; /* Should we show the time in place of the current year? */ if ((cp=getenv("WWW_SHOW_TIME")) != NULL) show_time = TRUE; } /* Check whether the query is from a form. */ if (getenv("WWW_VIA_FORM") != NULL) via_form = TRUE; /* Check whether the hit list should include TARGET attributes. */ if ((cp=getenv("WWW_HITS_TARGET")) != NULL && *cp) { hits_target = (char *)malloc(strlen(cp) + 12); sprintf(hits_target, " TARGET=\"%s\"", cp); } /* * If via_form is set, check whether to add context via * named submit buttons and hidden inputs. If add_context * is set, the hidden inputs will not be added if all the * hits were reported and the named submit buttons were * not added. */ if (via_form && getenv("WWW_ADD_CONTEXT") != NULL) add_context = TRUE; /* * Check whether to substitute ipath_info for ipath_translated * in hit lists, and if so, set up the translation. */ if (((cp=getenv("WWW_OMIT_IPATH_CONVERSION")) == NULL) && ( cp=getenv("WWW_IPATH_INFO") ) != NULL) { if (strlen(cp) > SPEC_SIZE) cp[SPEC_SIZE] = '\0'; strcpy(ipath_info, cp); if ((cp=getenv("WWW_IPATH_TRANSLATED")) != NULL) { if (strlen(cp) > SPEC_SIZE) cp[SPEC_SIZE] = '\0'; strcpy(ipath_translated, cp); /* * Make sure we handle both root (device/000000/) * and subdirectory (device/foo/) paths. */ if ((cp=strstr(ipath_translated, "000000/")) != NULL) { *cp = '\0'; strcpy(alt_ipath_translated, ipath_translated); *cp = '0'; strcat(alt_ipath_translated, (cp+7)); } else { alt_ipath_translated[0] = '\0'; } use_ipath_info = TRUE; } } /* should we redirect to the only selector? */ if (!omit_content_type && (result.count == 1) && (getenv("WWW_SEND_CONTENTS") != NULL)) { find_selector(result.select[0]); ptr1 = (char *) (selrab.rab$l_ubf + index_size[result.select[0].file]); parse_selector(ptr1, &new_style, &name, >ype, ptype, &file_name, &host, &port, method, &path); /* Get the path field for the URL. */ if (ptype[0] != 'R') { /* No range is indicated. Use a direct URL. */ if (strlen(path) == 0) /* no path given in selector */ path = HTVMS_wwwName(file_name); /* use file name */ } else { /* Get file_name and number of bytes from the Range selector. */ sscanf(ptype+1, "%d-%d", &start, &end); bytes = (end - start) + 1; if (send_raw == TRUE) { /* Ignore the Range and use a direct URL. */ ptype[0] = '\0'; if (strlen(path) == 0) /* no path given in selector */ path = HTVMS_wwwName(file_name); /* use file name */ } else if (use_ipath_info == TRUE) { /* Include the Range but convert the pathspec. */ path = HTVMS_wwwName(file_name); } else { /* We're including a Range and VMS pathspecs so hex escape. */ alt_path = URLescape(file_name); path = alt_path; } } /* * Check for ipath_translated or alt_ipath_translated * and substitute ipath_info (if set). */ if (use_ipath_info == TRUE) { alt_path = (char *)malloc(strlen(path) + strlen(ipath_info) + 1); if (strncasecomp(path, ipath_translated, strlen(ipath_translated)) == 0) { strcpy(alt_path, ipath_info); strcat(alt_path, (char*)&path[strlen(ipath_translated)]); path = alt_path; } else if(alt_ipath_translated[0] != '\0' && strncasecomp(path, alt_ipath_translated, strlen(alt_ipath_translated)) == 0) { strcpy(alt_path, ipath_info); strcat(alt_path, (char*)&path[strlen(alt_ipath_translated)]); path = alt_path; } } /* Output the hit. */ if (ptype[0] != 'R' || send_raw == TRUE) { /* Direct URL. */ printf ("Location: %s://%s%s%s%s\n\n", strlen(method) ? method : "http", strlen(host) ? host : Host, !strlen(host) && strlen(port) ? ":" : "", !strlen(host) && strlen(port) ? port : "", strlen(path) ? path : HTVMS_wwwName(file_name)); } else { /* Hex escaped URL with range. */ printf ("Location: %s://%s%s%s%s%s?TEXT=%s-%s\n\n", strlen(method) ? method : "http", strlen(host) ? host : Host, !strlen(host) && strlen(port) ? ":" : "", !strlen(host) && strlen(port) ? port : "", (cp=getenv("WWW_SCRIPT_NAME")) ? cp : "/htbin/query", (use_ipath_info) ? ipath_info : "", ptype, strlen(path) ? path : HTVMS_wwwName(file_name)); } if (alt_path != NULL) { free(alt_path); alt_path = NULL; } return; } /* Set up the HTML rendition. */ if (!omit_content_type) printf("Content-Type: text/html\n\n"); if (!omit_head) { printf("<HTML>\n<HEAD>\n"); if (TITLE != NULL) printf("<TITLE>%s\n", TITLE); else printf("VMS Indexed Database Search\n"); /* * You can leave the part active, if you don't care about * functioning with Mac Internet Explorer V5 -- none of the hit links * will be clickable if you do, though! -- RDP 8/21/00 * * if (!via_form) * printf("\n"); */ #include "custom_head.c" } printf("

Searching for:

%s

\n", orig_qstr); if (required_realm[0] != '\0') { printf("

in pages whose URLs contain:

%s

\n", required_realm); } EndingHit = TotalHits = result.count; fprintf(stdout, items_found, TotalHits); if (StartingHit > 1 && StartingHit > TotalHits) { if (max_hits) { StartingHit = TotalHits - MaxHits + 1; if (StartingHit < 1) StartingHit = 1; } else { StartingHit = 1; } } if (max_hits && EndingHit > (StartingHit + MaxHits - 1)) { EndingHit = (StartingHit + MaxHits - 1); fprintf(stdout, items_listed, StartingHit, EndingHit); } else if (StartingHit > 1) { fprintf(stdout, items_listed, StartingHit, EndingHit); } if (!max_hits) MaxHits = TotalHits; if (StartingHit > 1) HavePrevious = TRUE; if (EndingHit < TotalHits) HaveNext = TRUE; if (!(HavePrevious || HaveNext)) add_context = FALSE; if (add_context) { printf(":\n\n", StartingHit); printf("\n", MaxHits); printf("\n", TotalHits); printf("\n", orig_qstr); printf("

[ ", MaxHits); } else { printf( "[ ", MaxHits); } } if (HaveNext) { int next = (((TotalHits - MaxHits) < EndingHit) ? (TotalHits - EndingHit) : MaxHits); if (!HavePrevious) { printf("[ "); } else { printf(" | "); } if ((EndingHit + next) >= TotalHits) { printf( "", next); } else { printf( "", next); } } if (HavePrevious || HaveNext) { printf(" ]\n"); } } /* List the selectors. */ for (ind = (StartingHit - 1); ind < EndingHit; ind++) { find_selector(result.select[ind]); ptr1 = (char *) (selrab.rab$l_ubf + index_size[result.select[ind].file]); parse_selector(ptr1, &new_style, &name, >ype, ptype, &file_name, &host, &port, method, &path); status = -1; bytes = 0; DateAndOrSize[0] = '\0'; /* Get the path field for the URL. */ if (ptype[0] != 'R') { /* No range is indicated. Use a direct URL. */ if (strlen(path) == 0) /* no path given in selector */ path = HTVMS_wwwName(file_name); /* use file name */ } else { /* Get file_name and number of bytes from the Range selector. */ sscanf(ptype+1, "%d-%d", &start, &end); bytes = (end - start) + 1; if (send_raw == TRUE) { /* Ignore the Range and use a direct URL. */ ptype[0] = '\0'; if (strlen(path) == 0) /* no path given in selector */ path = HTVMS_wwwName(file_name); /* use file name */ } else if (use_ipath_info == TRUE) { /* Include the Range but convert the pathspec. */ path = HTVMS_wwwName(file_name); } else { /* We're including a Range and VMS pathspecs so hex escape. */ alt_path = URLescape(file_name); path = alt_path; } } /* Check whether to load the DateAndOrSize string. */ if (show_date || show_size) { /* Check whether we need to call stat(). */ if (show_date || ptype[0] != 'R') status = stat(file_name, &buf); /* Check whether we need to load the size string. */ if (show_size && (bytes || !status)) { if (!bytes) bytes = (int)buf.st_size; sprintf(size, "%d%s", (bytes >= 1000) ? (bytes+1023)/1024 : bytes, (bytes >= 1000) ? "KB" : " Bytes"); } /* Check whether we need to load the date fields. */ if (show_date && !status) { if (use_mdate) /* Use the last modification date. */ strcpy(cdate,ctime((unsigned long *) &buf.st_mtime)); else /* Default to the creation date. */ strcpy(cdate,ctime((unsigned long *) &buf.st_ctime)); /* Terminate each date field. */ cdate[3] = cdate[7] = cdate[10] = cdate[16] = cdate[24]= '\0'; if (cdate[8]==' ') cdate[8] = '0'; /* Check whether we need to know ThisYear. */ if (show_time) { elapsed = time(&elapsed); local = localtime(&elapsed); ThisYear = asctime(local) + 22; } } /* Load the DateAndOrSize string. */ if (show_date && !status) { if (show_time && show_size) { if (strncmp(YY,ThisYear,2)==0) sprintf(DateAndOrSize, " [%s-%s %s, %s]", DD, MMM, HHMM, size); else sprintf(DateAndOrSize, " [%s-%s-%s, %s]", DD, MMM, YY, size); } else if (show_time) { if (strncmp(YY,ThisYear,2)==0) sprintf(DateAndOrSize, " [%s-%s %s]", DD, MMM, HHMM); else sprintf(DateAndOrSize, " [%s-%s-%s]", DD, MMM, YY); } else if (show_size) { sprintf(DateAndOrSize, " [%s-%s-%s, %s]", DD, MMM, YY, size); } else { sprintf(DateAndOrSize, " [%s-%s-%s]", DD, MMM, YY); } } else if (show_size && bytes) { sprintf(DateAndOrSize, " [%s]", size); } } /* * Create the score string or dummy. */ if (show_score) sprintf(score, " (%d)", result.select[ind].score); else strcpy(score, ""); /* * Check for ipath_translated or alt_ipath_translated * and substitute ipath_info (if set). */ if (use_ipath_info == TRUE) { alt_path = (char *)malloc(strlen(path) + strlen(ipath_info) + 1); if (strncasecomp(path, ipath_translated, strlen(ipath_translated)) == 0) { strcpy(alt_path, ipath_info); strcat(alt_path, (char*)&path[strlen(ipath_translated)]); path = alt_path; } else if(alt_ipath_translated[0] != '\0' && strncasecomp(path, alt_ipath_translated, strlen(alt_ipath_translated)) == 0) { strcpy(alt_path, ipath_info); strcat(alt_path, (char*)&path[strlen(alt_ipath_translated)]); path = alt_path; } } /* provide http default if a host is given with no method */ if ((strlen(method) == 0) && strlen(host)) strcpy(method, "http"); if (strlen(method)) strcat(method, "://"); /* load gopher_title if it's a gopher URL */ if (strlen(method) && !strcmp(method, "gopher://")) { gopher_title = (char *)malloc(strlen(name) + 12); sprintf(gopher_title, " TITLE=\"%s\"", name); } /* Output the hit. */ if (ptype[0] != 'R' || send_raw == TRUE) { /* Direct URL. */ printf ("

%s%s%s%s\n", FileType(gtype), strlen(method) ? method : "", strlen(host) ? host : "", strlen(port) ? ":" : "", strlen(port) ? port : "", path, (gopher_title != NULL) ? gopher_title : "", (hits_target != NULL) ? hits_target : "", name, DateAndOrSize, score); } else { /* Hex escaped URL with range. */ printf ("

%s%s%s%s\n", FileType(gtype), strlen(method) ? method : "", strlen(host) ? host : "", strlen(port) ? ":" : "", strlen(port) ? port : "", (cp=getenv("WWW_SCRIPT_NAME")) ? cp : "/htbin/query", (use_ipath_info) ? ipath_info : "", ptype, path, (gopher_title != NULL) ? gopher_title : "", (hits_target != NULL) ? hits_target : "", name, DateAndOrSize, score); } if (alt_path != NULL) { free(alt_path); alt_path = NULL; } if (gopher_title != NULL) { free(gopher_title); gopher_title = NULL; } } if (hits_target != NULL) free(hits_target); /* Complete the HTML rendition. */ if (add_context && (EndingHit - StartingHit + 1) >= 10 && (HavePrevious || HaveNext)) { if (HavePrevious) { if ((StartingHit - MaxHits) > 1) { printf( "
[ ", MaxHits); } else { printf( "
[ ", MaxHits); } } if (HaveNext) { int next = (((TotalHits - MaxHits) < EndingHit) ? (TotalHits - EndingHit) : MaxHits); if (!HavePrevious) { printf("
[ "); } else { printf(" | "); } if ((EndingHit + next) >= TotalHits) { printf( "", next); } else { printf( "", next); } } printf(" ]\n"); } printf("

\n
\n"); #include "custom_research.c" #include "custom_foot.c" } return; } /* * Modify the scores to elevate the ranking of home pages (using the path) * and to elevate the ranking of pages where the hit was early enough on the * page that it might have been from meta-tagged keywords or description. * Based on www_emit (there may be some relic lines of code that are not * needed). - RDP */ static void home_page (Result result) { int ind, rawplace, rawscore, start, end, bytes, status; int slashcheck, slashcount; struct stat buf; time_t elapsed; struct tm *local; char cdate[26]; #define MMM cdate+4 #define DD cdate+8 #define HHMM cdate+11 #define YY cdate+22 unsigned short send_raw = FALSE; unsigned short use_ipath_info = FALSE; char *cp, *cp1, *ptr1, *name, *path; char score[10], ptype[25], method[10], bad_host[250], bad_path[250]; char required_realm[250], this_url[250]; char *title, *file; int new_style; char *file_name, size[128]; char items_found[256], items_listed[256]; char ipath_info[SPEC_SIZE]; char ipath_translated[SPEC_SIZE], alt_ipath_translated[SPEC_SIZE]; char *alt_path=NULL; char gtype; char *host, *port; FILE *fp; int bad_path_cut, bad_host_cut, slash_max, slash_add; int place_top, top_boost, place_early, place_scale, home_boost; /* * Initialize cut and boost parameters */ if ((cp=getenv("WWW_BAD_PATH_CUT")) != NULL) { bad_path_cut = atoi(cp); } else { bad_path_cut = 2; } if ((cp=getenv("WWW_BAD_HOST_CUT")) != NULL) { bad_host_cut = atoi(cp); } else { bad_host_cut = 2; } if ((cp=getenv("WWW_SLASH_MAX")) != NULL) { slash_max = atoi(cp); } else { slash_max = 4; } if ((cp=getenv("WWW_SLASH_ADD")) != NULL) { slash_add = atoi(cp); } else { slash_add = 50; } if ((cp=getenv("WWW_PLACE_TOP")) != NULL) { place_top = atoi(cp); } else { place_top = 50; } if ((cp=getenv("WWW_TOP_BOOST")) != NULL) { top_boost = atoi(cp); } else { top_boost = 50; } if ((cp=getenv("WWW_PLACE_EARLY")) != NULL) { place_early = atoi(cp); } else { place_early = 150; } if ((cp=getenv("WWW_PLACE_SCALE")) != NULL) { place_scale = atoi(cp); } else { place_scale = 2; } if ((cp=getenv("WWW_HOME_BOOST")) != NULL) { home_boost = atoi(cp); } else { home_boost = 400; } /* * Check for deprecated path */ if ((cp=getenv("WWW_DEPRECATED_PATH")) != NULL) { if ((strlen(cp) <= 249) && (strlen(cp) >= 4)) { strcpy(bad_path, cp); } else { too_bad(BAD_PATH_LENGTH); exit(1); } } else { bad_path[0] = '\0'; } /* * Check for deprecated host */ if ((cp=getenv("WWW_DEPRECATED_HOST")) != NULL) { if ((strlen(cp) <= 249) && (strlen(cp) >= 4)) { strcpy(bad_host, cp); } else { too_bad(BAD_HOST_LENGTH); exit(1); } } else { bad_host[0] = '\0'; } /* * Check for required realm */ if ((cp=getenv("WWW_REQUIRED_REALM")) == NULL) { required_realm[0] = '\0'; } else { if (strlen(cp) <= 2) { required_realm[0] ='\0'; } else { if (strlen(cp) <= 249) { strcpy(required_realm, cp); } else { too_bad(BAD_REALM_LENGTH); exit(1); } } } /* * Check whether to substitute ipath_info for ipath_translated * in hit lists, and if so, set up the translation. */ if (((cp=getenv("WWW_OMIT_IPATH_CONVERSION")) == NULL) && ( cp=getenv("WWW_IPATH_INFO") ) != NULL) { if (strlen(cp) > SPEC_SIZE) cp[SPEC_SIZE] = '\0'; strcpy(ipath_info, cp); if ((cp=getenv("WWW_IPATH_TRANSLATED")) != NULL) { if (strlen(cp) > SPEC_SIZE) cp[SPEC_SIZE] = '\0'; strcpy(ipath_translated, cp); /* * Make sure we handle both root (device/000000/) * and subdirectory (device/foo/) paths. */ if ((cp=strstr(ipath_translated, "000000/")) != NULL) { *cp = '\0'; strcpy(alt_ipath_translated, ipath_translated); *cp = '0'; strcat(alt_ipath_translated, (cp+7)); } else { alt_ipath_translated[0] = '\0'; } use_ipath_info = TRUE; } } /* Go through the selectors one at a time. */ for (ind = 0; ind < result.count; ind++) { find_selector(result.select[ind]); ptr1 = (char *) (selrab.rab$l_ubf + index_size[result.select[ind].file]); parse_selector(ptr1, &new_style, &name, >ype, ptype, &file_name, &host, &port, method, &path); status = -1; bytes = 0; /* Get the path field for the URL. */ if (ptype[0] != 'R') { /* No range is indicated. Use a direct URL. */ if (strlen(path) == 0) /* no path given in selector */ path = HTVMS_wwwName(file_name); /* use file name */ } else { /* Get file_name and number of bytes from the Range selector. */ sscanf(ptype+1, "%d-%d", &start, &end); bytes = (end - start) + 1; if (send_raw == TRUE) { /* Ignore the Range and use a direct URL. */ ptype[0] = '\0'; if (strlen(path) == 0) /* no path given in selector */ path = HTVMS_wwwName(file_name); /* use file name */ } else if (use_ipath_info == TRUE) { /* Include the Range but convert the pathspec. */ path = HTVMS_wwwName(file_name); } else { /* We're including a Range and VMS pathspecs so hex escape. */ alt_path = URLescape(file_name); path = alt_path; } } /* * Check for ipath_translated or alt_ipath_translated * and substitute ipath_info (if set). */ if (use_ipath_info == TRUE) { alt_path = (char *)malloc(strlen(path) + strlen(ipath_info) + 1); if (strncasecomp(path, ipath_translated, strlen(ipath_translated)) == 0) { strcpy(alt_path, ipath_info); strcat(alt_path, (char*)&path[strlen(ipath_translated)]); path = alt_path; } else if(alt_ipath_translated[0] != '\0' && strncasecomp(path, alt_ipath_translated, strlen(alt_ipath_translated)) == 0) { strcpy(alt_path, ipath_info); strcat(alt_path, (char*)&path[strlen(alt_ipath_translated)]); path = alt_path; } } /* provide http default if a host is given with no method */ if ((strlen(method) == 0) && strlen(host)) strcpy(method, "http"); if (strlen(method)) strcat(method, "://"); /* * At this point, method, host, port, and path are the obvious parts * of the HREF that will be created for the link. - RDP */ rawscore = result.select[ind].score; rawplace = result.select[ind].pos; if ((strlen(path) + strlen(host)) <= (250 - 7 - 6)) { sprintf(this_url, "%s%s%s%s", strlen(method) ? method : "", strlen(host) ? host : "", strlen(port) ? port : "", path); } else { /* deal with pathologically long URL */ sprintf(this_url, "%s", strlen(method) ? method : ""); strncat(this_url, host, (250 - strlen(this_url) - 3)); strncat(this_url, port, (250 - strlen(this_url) - 2)); strncat(this_url, path, (250 - strlen(this_url) - 1)); } if (required_realm[0] == '\0' || (required_realm[0] != '\0' && (strstr(this_url,required_realm) != NULL)) ) { /* * either no required realm or it matches, so go ahead * with the boost and cut calculations * */ if (strstr(path,"index.") != NULL || strstr(path,"welcome.") != NULL || strstr(path,"default.") != NULL || path[strlen(path)-1] == '/') { /* * we have a home page, folks! */ rawscore = rawscore + home_boost; /* home page boost */ } /* * Apply boost for early location in page -- primary target * is keywords and description text from META tags. */ if (rawplace <= place_top) rawscore = rawscore + top_boost; if (rawplace <= place_early) rawscore = rawscore + (place_early - rawplace)/place_scale; /* * Apply boost for high-level path (small number of slashes). */ slashcount = 0; for (slashcheck=0; slashcheck < strlen(path); slashcheck++) { if (path[slashcheck] == '/') slashcount = slashcount + 1; } if (slashcount <= slash_max) rawscore = rawscore + slash_add*(slash_max - slashcount); /* * Apply cut for pages on deprecated host. */ if ((strlen(bad_host) >= 2) && (strstr(host, bad_host) != NULL)) { rawscore = rawscore/bad_host_cut; } /* * Apply cut for pages on deprecated path. */ if ((strlen(bad_path) >= 2) && (strstr(path, bad_path) != NULL)) { rawscore = rawscore/bad_path_cut; } } else { /* * * required realm and it did not match * */ rawscore = 0; } result.select[ind].score = rawscore; if (alt_path != NULL) { free(alt_path); alt_path = NULL; } } return; } /* ** Redirect or output a client's request for a RANGE fetch. */ static void display_result(char *file_name) { int start, end, i; unsigned short omit_pre = FALSE; char *cp, inputline[SPEC_SIZE]; FILE *fp; /* Get the range and VMS filename. */ if (orig_qstr[5] == ' ') /** "TEXT= R..." **/ i = 7; else /** "TEXT=R..." or "TEXT R..." **/ i = 6; sscanf(orig_qstr+i, "%d-%d-%s", &start, &end, file_name); /* Force any paths whose device begins with WWW_Root to begin at root? */ if (!strncasecomp(file_name, "WWW_Root", 8) && (cp=strstr(file_name, ":[")) != NULL && getenv("WWW_FORCE_000000") != NULL) { cp += 2; if (*cp && strncmp(cp, "000000", 6)) { strncpy(inputline, file_name, cp-file_name); inputline[cp-file_name] = '\0'; strcat(inputline, "000000."); strcat(inputline, cp); strcpy(file_name, inputline); } } /* Use Location: for raw files to implement httpd authorization checks. */ /* (still need this for old or gerry-rigged URL's) */ if (getenv("WWW_SEND_RAW_FILE") != NULL) { printf("Location: %s%s%s\n\n", strlen(Host) ? "http://" : "", strlen(Host) ? Host : "", HTVMS_wwwName(file_name)); return; } /* Set up the HTML rendition for document sections. */ if (!omit_content_type) printf("Content-Type: text/html\n\n"); if (!omit_head) { printf("\n\n"); if (TITLE != NULL) printf("%s\n", TITLE); else printf("VMS Indexed Database Search\n"); #include "custom_head.c" } /* * If we are omitting the IPATH conversions, then * require that the path to the document begin with * WWW_Root, Gopher_Root, or a string optionally set by * the calling script as the WWW_DATADIRECTORY symbol. * * Always require that it not begin with the WWW_FORBIDDEN * symbol if the calling script has set a string for it. * * These checks are done for files or database sections * fetched with a RANGE specification because the path * to them is not included in the server's access checks * (the server only checks access for the script which * invoked us). We want to block spoofs like: * R0-10000-SYS$COMMON:[SYSMGR]FOO.BLAH */ if ( getenv("WWW_OMIT_CONVERSION") != NULL && ( strncasecomp(file_name, "WWW_Root", 8) && strncasecomp(file_name, "Gopher_Root", 11) && !(((cp=getenv("WWW_DATADIRECTORY")) != NULL) && strncasecomp(file_name, cp, strlen(cp))) ) || ( ((cp=getenv("WWW_FORBIDDEN")) != NULL) && 0==strncasecomp(file_name, cp, strlen(cp)) ) ) { printf("

ERROR 403

\n");

    /* Get and send the section with appropriate further packaging. */ 
    if ((cp=getenv("WWW_IPATH_TRANSLATED")) != NULL &&
        getenv("WWW_OMIT_IPATH_CONVERSION") == NULL)
        fp = fopen(cp, "r", "shr=get", "mbc=32");
    else
        fp = fopen(file_name, "r", "shr=get", "mbc=32");
    if (fp == NULL) {
	printf("ERROR 403\nUnable to fopen() database.");
#include "custom_foot.c"
	return;
    }
    fseek(fp, start, SEEK_SET);
    if(omit_pre)
        printf("This is from the document %s\n\n", file_name);
    else
	printf("\nThis is from the document %s\n\n", file_name);
    while (fgets(inputline, sizeof(inputline), fp) != NULL) {
        printf("%s", inputline);
	if (ftell(fp) >= end)
            break;
    }
    fclose(fp);
    if (omit_pre)
        printf("\n");
    else
        printf("\n

$\"\"$

$\"[DIR]\"$

$\"[HEX]\"$

$\"[BIN]\"$

$\"[UUE]\"$

$\"[IDX]\"$

$\"[IMG]\"$

$\"[SND]\"$

$\"[MOV]\"$

$\"[UNK]\"$

VMS Indexed Database Search