Omit sort key by default in yaz-icu's output. It may be enabled with
option -s. The sort key was used in previous tests. But since it is
encoded differently between ICU version we omit it in our tests.
<arg choice="opt" rep="repeat">commands</arg>
<arg>-c <replaceable>config</replaceable></arg>
<arg>-p <replaceable>opt</replaceable></arg>
<arg choice="opt" rep="repeat">commands</arg>
<arg>-c <replaceable>config</replaceable></arg>
<arg>-p <replaceable>opt</replaceable></arg>
<arg>-x</arg>
</cmdsynopsis>
</refsynopsisdiv>
<arg>-x</arg>
</cmdsynopsis>
</refsynopsisdiv>
</varlistentry>
<varlistentry>
</varlistentry>
<varlistentry>
- <term>-x <replaceable>config</replaceable></term>
+ <term>-s</term>
+ <listitem><para>
+ Specifies that output should include sort key as well. Note that
+ sort key differs between ICU versions.
+ </para></listitem>
+ </varlistentry>
+
+ <varlistentry>
+ <term>-x</term>
<listitem><para>
Specifies that output should be XML based rather than
"text" based.
<listitem><para>
Specifies that output should be XML based rather than
"text" based.
-1 1 'børn' 'Børn' '+EKC\x01\x85\xA5\x06\x01\x09'
-2 2 'le' 'Le' '?1\x01\x06\x01\x06'
-3 2 'carré' 'Carré' '-)KK1\x01\x82\x8D\x01\x0A'
-4 3 'le' 'Le' '?1\x01\x06\x01\x06'
-5 3 'carre' 'Carre' '-)KK1\x01\x09\x01\x09'
-6 4 'le' 'Le' '?1\x01\x06\x01\x06'
-7 4 'carré' 'Carré' '-)KK1\x01\x82\x8D\x01\x0A'
+1 1 'børn' 'Børn'
+2 2 'le' 'Le'
+3 2 'carré' 'Carré'
+4 3 'le' 'Le'
+5 3 'carre' 'Carre'
+6 4 'le' 'Le'
+7 4 'carré' 'Carré'
-1 1 'børn' 'Børn' '+EKC\x01\x85\xA5\x06\x01\x09'
-2 2 'le' 'Le' '?1\x01\x06\x01\x06'
-3 2 'carre' 'Carre' '-)KK1\x01\x09\x01\x09'
-4 3 'le' 'Le' '?1\x01\x06\x01\x06'
-5 3 'carre' 'Carre' '-)KK1\x01\x09\x01\x09'
-6 4 'le' 'Le' '?1\x01\x06\x01\x06'
-7 4 'carre' 'Carre' '-)KK1\x01\x09\x01\x09'
+1 1 'børn' 'Børn'
+2 2 'le' 'Le'
+3 2 'carre' 'Carre'
+4 3 'le' 'Le'
+5 3 'carre' 'Carre'
+6 4 'le' 'Le'
+7 4 'carre' 'Carre'
char conffile[1024];
char print[1024];
int xmloutput;
char conffile[1024];
char print[1024];
int xmloutput;
yaz_icu_chain_t chain;
FILE * infile;
FILE * outfile;
yaz_icu_chain_t chain;
FILE * infile;
FILE * outfile;
fprintf(stderr, "yaz-icu\n"
" [-c (path/to/config/file.xml)]\n"
" [-p (a|c|l|t)] print ICU info \n"
fprintf(stderr, "yaz-icu\n"
" [-c (path/to/config/file.xml)]\n"
" [-p (a|c|l|t)] print ICU info \n"
+ " [-s] Show sort normalization key\n"
" [-x] XML output\n"
"\n"
"Examples:\n"
" [-x] XML output\n"
"\n"
"Examples:\n"
p_config->conffile[0] = 0;
p_config->print[0] = 0;
p_config->xmloutput = 0;
p_config->conffile[0] = 0;
p_config->print[0] = 0;
p_config->xmloutput = 0;
+ p_config->sortoutput = 0;
p_config->chain = 0;
p_config->infile = stdin;
p_config->outfile = stdout;
/* set up command line parameters */
p_config->chain = 0;
p_config->infile = stdin;
p_config->outfile = stdout;
/* set up command line parameters */
- while ((ret = options("c:p:x", argv, argc, &arg)) != -2)
+ while ((ret = options("c:p:xs", argv, argc, &arg)) != -2)
case 'p':
strcpy(p_config->print, arg);
break;
case 'p':
strcpy(p_config->print, arg);
break;
+ case 's':
+ p_config->sortoutput = 1;
+ break;
case 'x':
p_config->xmloutput = 1;
break;
default:
case 'x':
p_config->xmloutput = 1;
break;
default:
+ printf("Got %d\n", ret);
print_option_error(p_config);
}
}
print_option_error(p_config);
}
}
/* should XML encode this. Bug #1902 */
fprintf(config.outfile,
"<token id=\"%lu\" line=\"%lu\""
/* should XML encode this. Bug #1902 */
fprintf(config.outfile,
"<token id=\"%lu\" line=\"%lu\""
- " norm=\"%s\" display=\"%s\" sortkey=\"%s\"/>\n",
+ " norm=\"%s\" display=\"%s\"",
token_count,
line_count,
icu_chain_token_norm(config.chain),
token_count,
line_count,
icu_chain_token_norm(config.chain),
- icu_chain_token_display(config.chain),
- wrbuf_cstr(sw));
+ icu_chain_token_display(config.chain));
+ if (p_config->sortoutput)
+ {
+ fprintf(config.outfile, " sortkey=\"%s\"",
+ wrbuf_cstr(sw));
+ }
+ fprintf(config.outfile, "/>\n");
- fprintf(config.outfile, "%lu %lu '%s' '%s' '%s'\n",
+ {
+ fprintf(config.outfile, "%lu %lu '%s' '%s'",
token_count,
line_count,
icu_chain_token_norm(config.chain),
token_count,
line_count,
icu_chain_token_norm(config.chain),
- icu_chain_token_display(config.chain),
- wrbuf_cstr(sw));
+ icu_chain_token_display(config.chain));
+ if (p_config->sortoutput)
+ {
+ fprintf(config.outfile, " '%s'", wrbuf_cstr(sw));
+ }
+ fprintf(config.outfile, "\n");
+ }
/*
* Local variables:
* c-basic-offset: 4
/*
* Local variables:
* c-basic-offset: 4
+ * c-file-style: "Stroustrup"
* indent-tabs-mode: nil
* End:
* vim: shiftwidth=4 tabstop=8 expandtab
* indent-tabs-mode: nil
* End:
* vim: shiftwidth=4 tabstop=8 expandtab