* Copyright (c) 2002, Index Data.
* See the file LICENSE for details.
*
- * $Id: d1_expat.c,v 1.3 2002-07-03 10:04:04 adam Exp $
+ * $Id: d1_expat.c,v 1.7 2002-08-23 14:24:05 adam Exp $
*/
#if HAVE_EXPAT_H
#include <stdio.h>
#include <stdlib.h>
+#if HAVE_ICONV_H
+#include <errno.h>
+#include <iconv.h>
+#endif
+
#include <yaz/xmalloc.h>
#include <yaz/log.h>
#include <yaz/data1.h>
#include <expat.h>
+#define XML_CHUNK 1024
+
struct user_info {
data1_node *d1_stack[256];
int level;
data1_handle dh;
NMEM nmem;
+ int loglevel;
};
static void cb_start (void *user, const char *el, const char **attr)
ui->d1_stack[ui->level] = data1_mk_tag (ui->dh, ui->nmem, el, attr,
ui->d1_stack[ui->level-1]);
ui->level++;
- yaz_log (LOG_DEBUG, "cb_start %s", el);
+ yaz_log (ui->loglevel, "cb_start %s", el);
}
static void cb_end (void *user, const char *el)
struct user_info *ui = (struct user_info*) user;
ui->level--;
- yaz_log (LOG_DEBUG, "cb_end %s", el);
+ yaz_log (ui->loglevel, "cb_end %s", el);
}
static void cb_chardata (void *user, const char *s, int len)
{
struct user_info *ui = (struct user_info*) user;
- int i;
-
- for (i = 0; i<len; i++)
- if (!strchr ("\n\n ", s[i]))
- break;
- if (i != len)
- {
- ui->d1_stack[ui->level] = data1_mk_text_n (ui->dh, ui->nmem, s, len,
+#if 0
+ yaz_log (ui->loglevel, "cb_chardata %.*s", len, s);
+#endif
+ ui->d1_stack[ui->level] = data1_mk_text_n (ui->dh, ui->nmem, s, len,
ui->d1_stack[ui->level -1]);
- }
}
static void cb_decl (void *user, const char *version, const char*encoding,
attr_list[1] = version;
attr_list[2] = "encoding";
- attr_list[3] = encoding;
+ attr_list[3] = "UTF-8"; /* encoding */
attr_list[4] = "standalone";
attr_list[5] = standalone ? "yes" : "no";
data1_mk_preprocess (ui->dh, ui->nmem, "xml", attr_list,
ui->d1_stack[ui->level-1]);
- yaz_log (LOG_DEBUG, "decl version=%s encoding=%s",
+ yaz_log (ui->loglevel, "decl version=%s encoding=%s",
version ? version : "null",
encoding ? encoding : "null");
}
ui->d1_stack[ui->level-1]);
data1_mk_text_nf (ui->dh, ui->nmem, data, strlen(data), res);
- yaz_log (LOG_DEBUG, "decl processing target=%s data=%s",
+ yaz_log (ui->loglevel, "decl processing target=%s data=%s",
target ? target : "null",
data ? data : "null");
static void cb_comment (void *user, const char *data)
{
struct user_info *ui = (struct user_info*) user;
- yaz_log (LOG_DEBUG, "decl comment data=%s", data ? data : "null");
+ yaz_log (ui->loglevel, "decl comment data=%s", data ? data : "null");
data1_mk_comment (ui->dh, ui->nmem, data, ui->d1_stack[ui->level-1]);
}
const char *sysid, const char *pubid,
int has_internal_subset)
{
- yaz_log (LOG_DEBUG, "doctype start doctype=%s sysid=%s pubid=%s",
+ struct user_info *ui = (struct user_info*) userData;
+ yaz_log (ui->loglevel, "doctype start doctype=%s sysid=%s pubid=%s",
doctypeName, sysid, pubid);
}
static void cb_doctype_end (void *userData)
{
- yaz_log (LOG_DEBUG, "doctype end");
+ struct user_info *ui = (struct user_info*) userData;
+ yaz_log (ui->loglevel, "doctype end");
}
const char *base, const char *systemId,
const char *publicId, const char *notationName)
{
- yaz_log (LOG_DEBUG,
- "entity %s is_para_entry=%d value=%.*s base=%s systemId=%s"
+ struct user_info *ui = (struct user_info*) userData;
+ yaz_log (ui->loglevel,
+ "entity decl %s is_para_entry=%d value=%.*s base=%s systemId=%s"
" publicId=%s notationName=%s",
entityName, is_parameter_entity, value_length, value,
base, systemId, publicId, notationName);
}
-#define XML_CHUNK 1024
+static int cb_external_entity (XML_Parser pparser,
+ const char *context,
+ const char *base,
+ const char *systemId,
+ const char *publicId)
+{
+ struct user_info *ui = (struct user_info*) XML_GetUserData(pparser);
+ FILE *inf;
+ int done = 0;
+ XML_Parser parser;
+
+ yaz_log (ui->loglevel,
+ "external entity context=%s base=%s systemid=%s publicid=%s",
+ context, base, systemId, publicId);
+ if (!systemId)
+ return 1;
+
+ if (!(inf = fopen (systemId, "rb")))
+ {
+ yaz_log (LOG_WARN|LOG_ERRNO, "fopen %s", systemId);
+ return 0;
+ }
+
+ parser = XML_ExternalEntityParserCreate (pparser, "", 0);
+ while (!done)
+ {
+ int r;
+ void *buf = XML_GetBuffer (parser, XML_CHUNK);
+ if (!buf)
+ {
+ yaz_log (LOG_WARN, "XML_GetBuffer fail");
+ break;
+ }
+ r = fread (buf, 1, XML_CHUNK, inf);
+ if (r == 0)
+ {
+ if (ferror(inf))
+ {
+ yaz_log (LOG_WARN|LOG_ERRNO, "fread %s", systemId);
+ break;
+ }
+ done = 1;
+ }
+ if (!XML_ParseBuffer (parser, r, done))
+ {
+ yaz_log (LOG_WARN, "XML_ParseBuffer failed %s",
+ XML_ErrorString(XML_GetErrorCode(parser)));
+ }
+ }
+ fclose (inf);
+ XML_ParserFree (parser);
+ return done;
+}
+
+
+#if HAVE_ICONV_H
+static int cb_encoding_convert (void *data, const char *s)
+{
+ iconv_t t = (iconv_t) data;
+ size_t ret;
+ size_t outleft = 2;
+ char outbuf_[2], *outbuf = outbuf_;
+ size_t inleft = 4;
+ char *inbuf = (char *) s;
+ unsigned short code;
+
+ ret = iconv (t, &inbuf, &inleft, &outbuf, &outleft);
+ if (ret == (size_t) (-1) && errno != E2BIG)
+ {
+ iconv (t, 0, 0, 0, 0);
+ return -1;
+ }
+ if (outleft != 0)
+ return -1;
+ memcpy (&code, outbuf_, sizeof(short));
+ return code;
+}
+
+static void cb_encoding_release (void *data)
+{
+ iconv_t t = (iconv_t) data;
+ iconv_close (t);
+}
+
+static int cb_encoding_handler (void *userData, const char *name,
+ XML_Encoding *info)
+{
+ int i = 0;
+ int no_ok = 0;
+ struct user_info *ui = (struct user_info*) userData;
+
+ iconv_t t = iconv_open ("UNICODE", name);
+ if (t == (iconv_t) (-1))
+ return 0;
+
+ info->data = 0; /* signal that multibyte is not in use */
+ yaz_log (ui->loglevel, "Encoding handler of %s", name);
+ for (i = 0; i<256; i++)
+ {
+ size_t ret;
+ char outbuf_[5];
+ char inbuf_[5];
+ char *inbuf = inbuf_;
+ char *outbuf = outbuf_;
+ size_t inleft = 1;
+ size_t outleft = 2;
+ inbuf_[0] = i;
+
+ iconv (t, 0, 0, 0, 0); /* reset iconv */
+
+ ret = iconv(t, &inbuf, &inleft, &outbuf, &outleft);
+ if (ret == (size_t) (-1))
+ {
+ if (errno == EILSEQ)
+ {
+ yaz_log (ui->loglevel, "Encoding %d: invalid sequence", i);
+ info->map[i] = -1; /* invalid sequence */
+ }
+ if (errno == EINVAL)
+ { /* multi byte input */
+ int len = 2;
+ int j = 0;
+ info->map[i] = -1;
+
+ while (len <= 4)
+ {
+ char sbuf[80];
+ int k;
+ inbuf = inbuf_;
+ inleft = len;
+ outbuf = outbuf_;
+ outleft = 2;
+
+ inbuf_[len-1] = j;
+ iconv (t, 0,0,0,0);
+
+ assert (i >= 0 && i<255);
+
+ *sbuf = 0;
+ for (k = 0; k<len; k++)
+ {
+ sprintf (sbuf+strlen(sbuf), "%d ", inbuf_[k]&255);
+ }
+ ret = iconv (t, &inbuf, &inleft, &outbuf, &outleft);
+ if (ret == (size_t) (-1))
+ {
+ if (errno == EILSEQ || errno == E2BIG)
+ {
+ j++;
+ if (j > 255)
+ break;
+ }
+ else if (errno == EINVAL)
+ {
+ len++;
+ j = 7;
+ }
+ }
+ else if (outleft == 0)
+ {
+ info->map[i] = -len;
+ info->data = t; /* signal that multibyte is in use */
+ break;
+ }
+ else
+ {
+ break;
+ }
+ }
+ if (info->map[i] < -1)
+ yaz_log (ui->loglevel, "Encoding %d: multibyte input %d",
+ i, -info->map[i]);
+ else
+ yaz_log (ui->loglevel, "Encoding %d: multibyte input failed",
+ i);
+ }
+ if (errno == E2BIG)
+ {
+ info->map[i] = -1; /* no room for output */
+ yaz_log (LOG_WARN, "Encoding %d: no room for output",
+ i);
+ }
+ }
+ else if (outleft == 0)
+ {
+ unsigned short code;
+ memcpy (&code, outbuf_, sizeof(short));
+ info->map[i] = code;
+ no_ok++;
+ }
+ else
+ { /* should never happen */
+ info->map[i] = -1;
+ yaz_log (LOG_DEBUG, "Encoding %d: bad state", i);
+ }
+ }
+ if (info->data)
+ { /* at least one multi byte */
+ info->convert = cb_encoding_convert;
+ info->release = cb_encoding_release;
+ }
+ else
+ {
+ /* no multi byte - we no longer need iconv handler */
+ iconv_close(t);
+ info->convert = 0;
+ info->release = 0;
+ }
+ if (!no_ok)
+ return 0;
+ return 1;
+}
+/* HAVE_ICONV_H */
+#endif
+
data1_node *data1_read_xml (data1_handle dh,
int (*rf)(void *, char *, size_t), void *fh,
struct user_info uinfo;
int done = 0;
+ uinfo.loglevel = LOG_LOG;
uinfo.level = 1;
uinfo.dh = dh;
uinfo.nmem = m;
XML_SetCommentHandler (parser, cb_comment);
XML_SetDoctypeDeclHandler (parser, cb_doctype_start, cb_doctype_end);
XML_SetEntityDeclHandler (parser, cb_entity_decl);
-
+ XML_SetExternalEntityRefHandler (parser, cb_external_entity);
+#if HAVE_ICONV_H
+ XML_SetUnknownEncodingHandler (parser, cb_encoding_handler, &uinfo);
+#endif
while (!done)
{
int r;
if (!buf)
{
/* error */
- return 0;
+ yaz_log (LOG_WARN, "XML_GetBuffer fail");
+ break;
}
r = (*rf)(fh, buf, XML_CHUNK);
if (r < 0)
{
/* error */
- return 0;
+ yaz_log (LOG_WARN, "XML read fail");
+ break;
}
else if (r == 0)
done = 1;
- XML_ParseBuffer (parser, r, done);
+ if (!XML_ParseBuffer (parser, r, done))
+ {
+ yaz_log (LOG_WARN, "XML_ParseBuffer (1) failed %s",
+ XML_ErrorString(XML_GetErrorCode(parser)));
+ }
}
XML_ParserFree (parser);
- if (!uinfo.d1_stack[1])
+ if (!uinfo.d1_stack[1] || !done)
return 0;
return uinfo.d1_stack[0];
}
+/* HAVE_EXPAT_H */
#endif
projects / yaz-moved-to-github.git / blobdiff