X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=recctrl%2Frectext.c;h=ba618d6b30c8eeb4fe9380c2a133a2c00f7d7f46;hb=ecb3935e78cd9bcfdebafdee0834cfb1060d7b5e;hp=de72fdb5c11026e5869874cee0ba348f221456f2;hpb=e4943abc48c7750da0152b593a1e7aaf578f4b02;p=idzebra-moved-to-github.git diff --git a/recctrl/rectext.c b/recctrl/rectext.c index de72fdb..ba618d6 100644 --- a/recctrl/rectext.c +++ b/recctrl/rectext.c @@ -1,6 +1,6 @@ -/* $Id: rectext.c,v 1.18 2004-06-16 20:32:41 adam Exp $ - Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 - Index Data Aps +/* $Id: rectext.c,v 1.31 2006-05-10 08:13:30 adam Exp $ + Copyright (C) 1995-2005 + Index Data ApS This file is part of the Zebra server. @@ -25,23 +25,33 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include -#include -#include "rectext.h" +#include +#include -struct text_info { +struct filter_info { char *sep; }; -static void *text_init (RecType recType) +static void *filter_init (Res res, RecType recType) { - struct text_info *tinfo = (struct text_info *) xmalloc(sizeof(*tinfo)); + struct filter_info *tinfo = (struct filter_info *) xmalloc(sizeof(*tinfo)); tinfo->sep = 0; return tinfo; } -static void text_destroy (void *clientData) +static ZEBRA_RES filter_config(void *clientData, Res res, const char *args) { - struct text_info *tinfo = clientData; + struct filter_info *tinfo = (struct filter_info*) clientData; + xfree(tinfo->sep); + tinfo->sep = 0; + if (args && *args) + tinfo->sep = xstrdup(args); + return ZEBRA_OK; +} + +static void filter_destroy (void *clientData) +{ + struct filter_info *tinfo = clientData; xfree (tinfo->sep); xfree (tinfo); } @@ -53,7 +63,7 @@ struct buf_info { int max; }; -struct buf_info *buf_open (struct recExtractCtrl *p) +static struct buf_info *buf_open (struct recExtractCtrl *p) { struct buf_info *fi = (struct buf_info *) xmalloc (sizeof(*fi)); @@ -64,7 +74,7 @@ struct buf_info *buf_open (struct recExtractCtrl *p) return fi; } -int buf_read (struct text_info *tinfo, struct buf_info *fi, char *dst) +static int buf_getchar (struct filter_info *tinfo, struct buf_info *fi, char *dst) { if (fi->offset >= fi->max) { @@ -85,58 +95,55 @@ int buf_read (struct text_info *tinfo, struct buf_info *fi, char *dst) return 1; } -void buf_close (struct buf_info *fi) +static void buf_close (struct buf_info *fi) { xfree (fi->buf); xfree (fi); } -static int text_extract (void *clientData, struct recExtractCtrl *p) +static int filter_extract (void *clientData, struct recExtractCtrl *p) { - struct text_info *tinfo = clientData; + struct filter_info *tinfo = clientData; char w[512]; RecWord recWord; int r; struct buf_info *fi = buf_open (p); + int no_read = 0; #if 0 - yaz_log(LOG_LOG, "text_extract off=%ld", + yaz_log(YLOG_LOG, "filter_extract off=%ld", (long) (*fi->p->tellf)(fi->p->fh)); #endif - xfree(tinfo->sep); - tinfo->sep = 0; - if (p->subType) { - if (!strncmp(p->subType, "sep=", 4)) - tinfo->sep = xstrdup(p->subType+4); - } (*p->init)(p, &recWord); - recWord.reg_type = 'w'; do { int i = 0; - r = buf_read (tinfo, fi, w); + r = buf_getchar (tinfo, fi, w); while (r > 0 && i < 511 && w[i] != '\n' && w[i] != '\r') { i++; - r = buf_read (tinfo, fi, w + i); + r = buf_getchar (tinfo, fi, w + i); } if (i) { - recWord.string = w; - recWord.length = i; + no_read += i; + recWord.term_buf = w; + recWord.term_len = i; (*p->tokenAdd)(&recWord); } } while (r > 0); buf_close (fi); + if (no_read == 0) + return RECCTRL_EXTRACT_EOF; return RECCTRL_EXTRACT_OK; } -static int text_retrieve (void *clientData, struct recRetrieveCtrl *p) +static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p) { - int r, text_ptr = 0; - static char *text_buf = NULL; - static int text_size = 0; + int r, filter_ptr = 0; + static char *filter_buf = NULL; + static int filter_size = 0; int make_header = 1; int make_body = 1; const char *elementSetName = NULL; @@ -163,43 +170,44 @@ static int text_retrieve (void *clientData, struct recRetrieveCtrl *p) } while (1) { - if (text_ptr + 4096 >= text_size) + if (filter_ptr + 4096 >= filter_size) { char *nb; - text_size = 2*text_size + 8192; - nb = (char *) xmalloc (text_size); - if (text_buf) + filter_size = 2*filter_size + 8192; + nb = (char *) xmalloc (filter_size); + if (filter_buf) { - memcpy (nb, text_buf, text_ptr); - xfree (text_buf); + memcpy (nb, filter_buf, filter_ptr); + xfree (filter_buf); } - text_buf = nb; + filter_buf = nb; } - if (make_header && text_ptr == 0) + if (make_header && filter_ptr == 0) { if (p->score >= 0) { - sprintf (text_buf, "Rank: %d\n", p->score); - text_ptr = strlen(text_buf); + sprintf (filter_buf, "Rank: %d\n", p->score); + filter_ptr = strlen(filter_buf); } - sprintf (text_buf + text_ptr, "Local Number: %d\n", p->localno); - text_ptr = strlen(text_buf); + sprintf (filter_buf + filter_ptr, "Local Number: " ZINT_FORMAT "\n", + p->localno); + filter_ptr = strlen(filter_buf); if (p->fname) { - sprintf (text_buf + text_ptr, "Filename: %s\n", p->fname); - text_ptr = strlen(text_buf); + sprintf (filter_buf + filter_ptr, "Filename: %s\n", p->fname); + filter_ptr = strlen(filter_buf); } - strcpy(text_buf+text_ptr++, "\n"); + strcpy(filter_buf+filter_ptr++, "\n"); } if (!make_body) break; - r = (*p->readf)(p->fh, text_buf + text_ptr, 4096); + r = (*p->readf)(p->fh, filter_buf + filter_ptr, 4096); if (r <= 0) break; - text_ptr += r; + filter_ptr += r; } - text_buf[text_ptr] = '\0'; + filter_buf[filter_ptr] = '\0'; if (elementSetName) { if (!strcmp (elementSetName, "B")) @@ -209,7 +217,7 @@ static int text_retrieve (void *clientData, struct recRetrieveCtrl *p) } if (no_lines) { - char *p = text_buf; + char *p = filter_buf; int i = 0; while (++i <= no_lines && (p = strchr (p, '\n'))) @@ -217,21 +225,41 @@ static int text_retrieve (void *clientData, struct recRetrieveCtrl *p) if (p) { p[1] = '\0'; - text_ptr = p-text_buf; + filter_ptr = p-filter_buf; } } p->output_format = VAL_SUTRS; - p->rec_buf = text_buf; - p->rec_len = text_ptr; + p->rec_buf = filter_buf; + p->rec_len = filter_ptr; return 0; } -static struct recType text_type = { +static struct recType filter_type = { + 0, "text", - text_init, - text_destroy, - text_extract, - text_retrieve + filter_init, + filter_config, + filter_destroy, + filter_extract, + filter_retrieve +}; + +RecType +#ifdef IDZEBRA_STATIC_TEXT +idzebra_filter_text +#else +idzebra_filter +#endif + +[] = { + &filter_type, + 0, }; +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ -RecType recTypeText = &text_type;