X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=recctrl%2Fxslt.c;h=b5d5ffea65967b4b4cb7aeaebe55ddfca41a948d;hb=e77e69cb9cf741189af0cd5ffb165c3d67cd7fa7;hp=9c7de499f8e4994949cd4f67d776619692c0a9ba;hpb=824db602e8767538b3eb075a604f557ebc9c3085;p=idzebra-moved-to-github.git diff --git a/recctrl/xslt.c b/recctrl/xslt.c index 9c7de49..b5d5ffe 100644 --- a/recctrl/xslt.c +++ b/recctrl/xslt.c @@ -1,4 +1,4 @@ -/* $Id: xslt.c,v 1.6 2005-05-31 17:36:16 adam Exp $ +/* $Id: xslt.c,v 1.9 2005-06-07 13:10:52 adam Exp $ Copyright (C) 1995-2005 Index Data ApS @@ -28,6 +28,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include #include +#include #include #include @@ -40,21 +41,33 @@ struct filter_schema { const char *stylesheet; struct filter_schema *next; const char *default_schema; + const char *include_snippet; xsltStylesheetPtr stylesheet_xsp; }; struct filter_info { xmlDocPtr doc; char *fname; - int split_depth; + const char *split_level; + const char *split_path; ODR odr; struct filter_schema *schemas; xmlTextReaderPtr reader; }; -#define ZEBRA_INDEX_NS "http://indexdata.dk/zebra/indexing/1" -#define ZEBRA_SCHEMA_IDENTITY_NS "http://indexdata.dk/zebra/identity/1" -static const char *zebra_index_ns = ZEBRA_INDEX_NS; +#define ZEBRA_SCHEMA_XSLT_NS "http://indexdata.dk/zebra/xslt/1" + +static const char *zebra_xslt_ns = ZEBRA_SCHEMA_XSLT_NS; + +static void set_param_xml(const char **params, const char *name, + const char *value, ODR odr) +{ + while (*params) + params++; + params[0] = name; + params[1] = value; + params[2] = 0; +} static void set_param_str(const char **params, const char *name, const char *value, ODR odr) @@ -81,15 +94,50 @@ static void set_param_int(const char **params, const char *name, } +int zebra_xmlInputMatchCallback (char const *filename) +{ + yaz_log(YLOG_LOG, "match %s", filename); + return 0; +} + + +void * zebra_xmlInputOpenCallback (char const *filename) +{ + return 0; +} + +int zebra_xmlInputReadCallback (void * context, char * buffer, int len) +{ + return 0; +} + +int zebra_xmlInputCloseCallback (void * context) +{ + return 0; +} + + + + + static void *filter_init_xslt(Res res, RecType recType) { struct filter_info *tinfo = (struct filter_info *) xmalloc(sizeof(*tinfo)); tinfo->reader = 0; tinfo->fname = 0; - tinfo->split_depth = 0; + tinfo->split_level = 0; + tinfo->split_path = 0; tinfo->odr = odr_createmem(ODR_ENCODE); tinfo->doc = 0; tinfo->schemas = 0; + +#if 0 + xmlRegisterDefaultInputCallbacks(); + xmlRegisterInputCallbacks(zebra_xmlInputMatchCallback, + zebra_xmlInputOpenCallback, + zebra_xmlInputReadCallback, + zebra_xmlInputCloseCallback); +#endif return tinfo; } @@ -97,7 +145,7 @@ static void *filter_init_xslt1(Res res, RecType recType) { struct filter_info *tinfo = (struct filter_info *) filter_init_xslt(res, recType); - tinfo->split_depth = 1; + tinfo->split_level = "1"; return tinfo; } @@ -144,8 +192,9 @@ static ZEBRA_RES create_schemas(struct filter_info *tinfo, const char *fname) return ZEBRA_FAIL; for (ptr = ptr->children; ptr; ptr = ptr->next) { - if (ptr->type == XML_ELEMENT_NODE && - !strcmp(ptr->name, "schema")) + if (ptr->type != XML_ELEMENT_NODE) + continue; + if (!strcmp(ptr->name, "schema")) { struct _xmlAttr *attr; struct filter_schema *schema = xmalloc(sizeof(*schema)); @@ -155,6 +204,7 @@ static ZEBRA_RES create_schemas(struct filter_info *tinfo, const char *fname) schema->default_schema = 0; schema->next = tinfo->schemas; schema->stylesheet_xsp = 0; + schema->include_snippet = 0; tinfo->schemas = schema; for (attr = ptr->properties; attr; attr = attr->next) { @@ -162,12 +212,27 @@ static ZEBRA_RES create_schemas(struct filter_info *tinfo, const char *fname) attr_content(attr, "name", &schema->name); attr_content(attr, "stylesheet", &schema->stylesheet); attr_content(attr, "default", &schema->default_schema); + attr_content(attr, "snippet", &schema->include_snippet); } if (schema->stylesheet) schema->stylesheet_xsp = xsltParseStylesheetFile( (const xmlChar*) schema->stylesheet); } + else if (!strcmp(ptr->name, "split")) + { + struct _xmlAttr *attr; + for (attr = ptr->properties; attr; attr = attr->next) + { + attr_content(attr, "level", &tinfo->split_level); + attr_content(attr, "path", &tinfo->split_path); + } + } + else + { + yaz_log(YLOG_WARN, "Bad element %s in %s", ptr->name, fname); + return ZEBRA_FAIL; + } } return ZEBRA_OK; } @@ -244,7 +309,7 @@ static void index_node(struct filter_info *tinfo, struct recExtractCtrl *ctrl, { index_node(tinfo, ctrl, ptr->children, recWord); if (ptr->type != XML_ELEMENT_NODE || !ptr->ns || - strcmp(ptr->ns->href, zebra_index_ns)) + strcmp(ptr->ns->href, zebra_xslt_ns)) continue; if (!strcmp(ptr->name, "index")) { @@ -277,10 +342,10 @@ static int extract_doc(struct filter_info *tinfo, struct recExtractCtrl *p, xmlChar *buf_out; int len_out; - struct filter_schema *schema = lookup_schema(tinfo, ZEBRA_INDEX_NS); + struct filter_schema *schema = lookup_schema(tinfo, zebra_xslt_ns); params[0] = 0; - set_param_str(params, "schema", ZEBRA_INDEX_NS, tinfo->odr); + set_param_str(params, "schema", zebra_xslt_ns, tinfo->odr); (*p->init)(p, &recWord); recWord.reg_type = 'w'; @@ -312,6 +377,7 @@ static int extract_doc(struct filter_info *tinfo, struct recExtractCtrl *p, static int extract_split(struct filter_info *tinfo, struct recExtractCtrl *p) { int ret; + int split_depth = 0; if (p->first_record) { if (tinfo->reader) @@ -325,12 +391,15 @@ static int extract_split(struct filter_info *tinfo, struct recExtractCtrl *p) if (!tinfo->reader) return RECCTRL_EXTRACT_ERROR_GENERIC; + if (tinfo->split_level) + split_depth = atoi(tinfo->split_level); ret = xmlTextReaderRead(tinfo->reader); while (ret == 1) { int type = xmlTextReaderNodeType(tinfo->reader); int depth = xmlTextReaderDepth(tinfo->reader); - if (tinfo->split_depth == 0 || - (type == XML_READER_TYPE_ELEMENT && tinfo->split_depth == depth)) + if (split_depth == 0 || + (split_depth > 0 && + type == XML_READER_TYPE_ELEMENT && split_depth == depth)) { xmlNodePtr ptr = xmlTextReaderExpand(tinfo->reader); xmlNodePtr ptr2 = xmlCopyNode(ptr, 1); @@ -371,7 +440,7 @@ static int filter_extract(void *clientData, struct recExtractCtrl *p) odr_reset(tinfo->odr); - if (tinfo->split_depth == 0) + if (tinfo->split_level == 0 && tinfo->split_path == 0) return extract_full(tinfo, p); else { @@ -391,14 +460,62 @@ static int ioclose_ret(void *context) } +static const char *snippet_doc(struct recRetrieveCtrl *p, int text_mode, + int window_size) +{ + const char *xml_doc_str; + int ord = 0; + WRBUF wrbuf = wrbuf_alloc(); + zebra_snippets *res = + zebra_snippets_window(p->doc_snippet, p->hit_snippet, window_size); + zebra_snippet_word *w = zebra_snippets_list(res); + + if (text_mode) + wrbuf_printf(wrbuf, "\'"); + else + wrbuf_printf(wrbuf, "\n", zebra_xslt_ns); + for (; w; w = w->next) + { + if (ord == 0) + ord = w->ord; + else if (ord != w->ord) + + break; + if (text_mode) + wrbuf_printf(wrbuf, "%s%s%s ", + w->match ? "*" : "", + w->term, + w->match ? "*" : ""); + else + { + wrbuf_printf(wrbuf, " ", + w->ord, w->seqno, + (w->match ? "match='1'" : "")); + wrbuf_xmlputs(wrbuf, w->term); + wrbuf_printf(wrbuf, "\n"); + } + } + if (text_mode) + wrbuf_printf(wrbuf, "\'"); + else + wrbuf_printf(wrbuf, "\n"); + + xml_doc_str = odr_strdup(p->odr, wrbuf_buf(wrbuf)); + + zebra_snippets_destroy(res); + wrbuf_free(wrbuf, 1); + return xml_doc_str; +} + static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p) { - const char *esn = ZEBRA_SCHEMA_IDENTITY_NS; + const char *esn = zebra_xslt_ns; const char *params[10]; struct filter_info *tinfo = clientData; xmlDocPtr resDoc; xmlDocPtr doc; struct filter_schema *schema; + int window_size = -1; if (p->comp) { @@ -418,6 +535,9 @@ static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p) return 0; } + if (schema->include_snippet) + window_size = atoi(schema->include_snippet); + params[0] = 0; set_param_str(params, "schema", esn, p->odr); if (p->fname) @@ -426,6 +546,9 @@ static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p) set_param_int(params, "score", p->score, p->odr); set_param_int(params, "size", p->recordSize, p->odr); + if (window_size >= 0) + set_param_xml(params, "snippet", snippet_doc(p, 1, window_size), + p->odr); doc = xmlReadIO(ioread_ret, ioclose_ret, p /* I/O handler */, 0 /* URL */, 0 /* encoding */, @@ -436,6 +559,13 @@ static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p) return 0; } + if (window_size >= 0) + { + xmlNodePtr node = xmlDocGetRootElement(doc); + const char *snippet_str = snippet_doc(p, 0, window_size); + xmlDocPtr snippet_doc = xmlParseMemory(snippet_str, strlen(snippet_str)); + xmlAddChild(node, xmlDocGetRootElement(snippet_doc)); + } if (!schema->stylesheet_xsp) resDoc = doc; else