url
get_from_web (url name) {
if (!is_rooted_web (name)) return url_none ();
url res= get_cache (name);
if (!is_none (res)) return res;
#ifdef OS_WIN32
string urlString = as_string (name);
url tmp = url_temp();
if (starts (urlString, "www."))
urlString = "http://" * urlString;
else if (starts (urlString, "ftp."))
urlString = "ftp://" * urlString;
else if (starts (urlString, "ftp://"))
urlPath = NULL;
else if (starts (urlString, "http://"))
urlPath = NULL;
else
urlString = "http://" * urlString;
urlString= web_encode (urlString);
c_string urlPath (urlString);
c_string tempFilePath (as_string (tmp));
if(!URL_Get(urlPath, tempFilePath))
return url_none();
else return set_cache (name, tmp);
#else
string test= var_eval_system ("which wget");
if (!ends (test, "wget")) return url_none ();
url tmp= url_temp ();
string tmp_s= escape_sh (concretize (tmp));
string cmd= "wget --header='User-Agent: TeXmacs-" TEXMACS_VERSION "' -q";
cmd << " -O " << tmp_s << " " << escape_sh (web_encode (as_string (name)));
// cout << cmd << "\n";
system (cmd);
// cout << "got " << name << " as " << tmp << "\n";
if (var_eval_system ("cat " * tmp_s * " 2> /dev/null") == "") {
remove (tmp);
return url_none ();
}
else return set_cache (name, tmp);
#endif
}
std::string encode_field(const std::string& v, const std::string& local_domain_url)
{
std::string r;
r.reserve(v.length() << 1);
for (size_t i = 0; i < v.length(); )
{
if (boost::istarts_with(v.c_str() + i, "ftp://")
|| boost::istarts_with(v.c_str() + i, "http://")
|| boost::istarts_with(v.c_str() + i, "https://")
|| boost::istarts_with(v.c_str() + i, "mailto:"))
{
size_t p = i;
while (p < v.length()
&& !isspace(v[p] & 0xff)
&& v[p] != '\"'
&& v[p] != '<'
&& v[p] != '>')
{
p++;
}
if (v[p - 1] == '!' || v[p - 1] == ',' || v[p - 1] == '.' || v[p - 1] == '?')
p--;
if (v[p - 1] == ')')
p--;
std::string url = web_encode(v.substr(i, p - i));
if (boost::istarts_with(v.c_str() + i, "ftp."))
r += web_link(url, "ftp://" + url, false);
else if (boost::istarts_with(v.c_str() + i, "www."))
r += web_link(url, "http://" + url, false);
else
r += web_link(boost::istarts_with(v.c_str() + i, "mailto:") ? url.substr(7) : encode_local_url(url, local_domain_url), url, false);
i = p;
}
else
{
char c = v[i++];
switch (c)
{
case '<':
r += "<";
break;
case '&':
r += "&";
break;
default:
r += c;
}
}
}
return r;
}
static std::string web_link(str_ref title, str_ref link)
{
return "<a href=\"" + web_encode(link) + "\">" + web_encode(title.empty() ? link : title) + "</a>";
}
static std::string web_link(const std::string& link_title, const std::string& link, bool encode)
{
return encode
? web_link(web_encode(link_title), web_encode(link), false)
: (boost::format("<a href=\"%s\">%s</a>") % link % (link_title.empty() ? link : link_title)).str();
}
void do_url1(FILE *fp, const struct mngmsg *m_msp, const struct pstruct *ppst,
char *l_name, int n1,
const struct a_struct *aln_p, const char *annot_var_s,
const struct annot_str *q_annot_p,
const struct annot_str *l_annot_p )
{
char my_q_name[200], my_l_name[200], json_l_name[200];
char *db, *bp;
char pgm[10], o_pgm[10], lib[MAX_LSTR];
char *tmp_annot_s, *q_domain_s, *l_domain_s, *tmp_domain_s, *etmp_domain_s;
int n_tmp_annot_s, n_tmp_domain;
long q_offset, l_offset;
char *ref_url, *lbp=NULL;
char *srch_url, *srch_url1, *dom_url;
/* set the database */
if (m_msp->ldb_info.ldnaseq==SEQT_DNA) db="nucleotide";
else db="Protein";
/* set the program type */
if (strncmp(m_msp->f_id0,"rss",3)==0) {
strncpy(pgm,"fa",sizeof(pgm));
}
else if (strncmp(m_msp->f_id0,"rfx",3)==0) {
strncpy(pgm,"fx",sizeof(pgm));
}
else { strncpy(pgm,m_msp->f_id0,sizeof(pgm)); }
SAFE_STRNCPY(o_pgm, pgm, sizeof(o_pgm));
/* get a library name (probably does not work for %, + abbreviations */
if (m_msp->lname[0]!='%') {
SAFE_STRNCPY(lib,m_msp->lname,sizeof(lib));
}
else {
SAFE_STRNCPY(lib,"%25",sizeof(lib));
SAFE_STRNCAT(lib,&m_msp->lname[1],sizeof(lib));
}
lib[sizeof(lib)-1]='\0';
if ((lbp = strchr(l_name,'|'))==NULL) {
lbp = l_name;
}
else {
lbp++;
}
SAFE_STRNCPY(my_q_name,m_msp->qtitle,sizeof(my_q_name));
if ((bp=strchr(my_q_name,' '))!=NULL) *bp='\0';
SAFE_STRNCPY(my_l_name,lbp,sizeof(my_l_name));
if (pgm[0]=='t' || !strcmp(pgm,"fx") || !strcmp(pgm,"fy")==0 ) {
if ((lbp=strchr(my_l_name,':'))!=NULL) *lbp='\0';
lbp = &my_l_name[strlen(my_l_name)-2];
if ( *lbp == '_' ) *lbp = '\0';
}
/* change the program name for fastx, tfastx, tfasta */
/* fastx returns proteins */
if (strcmp(pgm,"fx")==0 || strcmp(pgm,"fy")==0) {SAFE_STRNCPY(pgm,"fa",sizeof(pgm));}
else if (strcmp(pgm,"ff")==0) {SAFE_STRNCPY(pgm,"fa",sizeof(pgm));}
else if (pgm[0]=='t') {
SAFE_STRNCPY(pgm,"fx",sizeof(pgm));
SAFE_STRNCPY(lib,DEF_PROT_LIB,sizeof(lib));
}
fflush(fp);
q_offset = aln_p->q_offset;
l_offset = aln_p->l_offset;
/* set up ref_url, srch_url, srch_url1, dom_url */
fflush(fp);
ref_url = getenv("REF_URL");
srch_url = getenv("SRCH_URL");
srch_url1 = getenv("SRCH_URL1");
dom_url = NULL;
dom_url = getenv("DOMAIN_PLOT_URL");
if (ref_url || srch_url || srch_url1 || dom_url) {
fprintf(fp,"<!-- LINK_START %s -->",l_name);
/* REF_URL should provide */
/* "<A HREF=\"http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=%s&fcmd=Search&doptcmd1=DocSum&term=%s\">Entrez lookup</A> " */
if (ref_url != NULL) {fprintf(fp,ref_url,db,my_l_name);}
/* SRCH_URL should provide */
/* "<A HREF=\"http://localhost/fasta_www2/searchfa.cgi?query=%s&db=fasta_www.cgi&lib=%s&pgm=%s&start=%ld&stop=%ld&n1=%d&o_pgm=%s\">Re-search database</A> " */
if (srch_url != NULL) {
fprintf(fp,srch_url,my_l_name,db,lib,pgm,
l_offset+aln_p->amin1+1,l_offset+aln_p->amax1,n1,m_msp->f_id0);
}
/* SRCH_URL1 should provide: */
/* "<A HREF=\"http://localhost/fasta_www2/searchxf.cgi?query=%s&db=%s&lib=%s&pgm=%s&start=%ld&stop=%ld&n1=%d&o_pgm=%s\">General re-search</A>\n" */
if (srch_url1 != NULL) {
fprintf(fp,srch_url1,my_l_name,db,lib,pgm,
l_offset+aln_p->amin1+1,l_offset+aln_p->amax1,n1,m_msp->f_id0);
}
if (dom_url!=NULL) {
if (annot_var_s && annot_var_s[0]) {
tmp_annot_s = web_encode(annot_var_s);
}
else tmp_annot_s = "";
q_domain_s = l_domain_s = NULL;
if (q_annot_p && q_annot_p->n_domains > 0 &&
(q_domain_s = display_domains('q',q_annot_p->s_annot_arr_p, q_annot_p->n_annot))!=NULL) {
}
if (l_annot_p && l_annot_p->n_domains > 0 &&
(l_domain_s = display_domains('l',l_annot_p->s_annot_arr_p, l_annot_p->n_annot))!=NULL) {
}
/* combine domain strings */
n_tmp_domain = 0;
if (q_domain_s) n_tmp_domain += strlen(q_domain_s)+1;
if (l_domain_s) n_tmp_domain += strlen(l_domain_s)+1;
etmp_domain_s = "";
if (n_tmp_domain > 0) {
if ((tmp_domain_s=(char *)calloc(n_tmp_domain,sizeof(char)))==NULL) {
fprintf(stderr,"*** error [%s:%d] *** cannot allocate tmp_domain_s[%d]\n",
__FILE__, __LINE__,n_tmp_domain);
}
else {
tmp_domain_s[0] = '\0';
if (q_domain_s) SAFE_STRNCAT(tmp_domain_s, q_domain_s, n_tmp_domain);
if (l_domain_s) SAFE_STRNCAT(tmp_domain_s, l_domain_s, n_tmp_domain);
etmp_domain_s = web_encode(tmp_domain_s);
}
}
/* appropriate format string: */
/*
pgm=%s -- program abbrev that created alignment
q_name=%s -- query info
q_cstart=%ld
q_cstop=%ld
q_astart=%ld
q_astop=%ld
l_name=%s -- library info
l_cstart=%ld
l_cstop=%ld
l_astart=%ld
l_astop=%ld
region=%s -- aligned domain and variant information
doms=%s
DOMAIN_PLOT_URL = "pgm=%s;q_name=%s;q_cstart=%ld;q_cstop=%ld&q_astart=%ld&q_astop=%ld&l_name=%s&l_cstart=%ld&l_cstop=%ld&l_astart=%ld&l_astop=%ld®ions=%s&doms=%s"
*/
/* think about the alternative of running a script
rather than embedding it */
fprintf(fp,dom_url,o_pgm,
my_q_name, q_offset+seq_pos(1,aln_p->qlrev,2),q_offset+seq_pos(m_msp->n0,aln_p->qlrev,2),
q_offset+seq_pos(aln_p->amin0+1,aln_p->qlrev,1), q_offset+seq_pos(aln_p->amax0, aln_p->qlrev,2),
my_l_name, l_offset+seq_pos(1,aln_p->llrev,2), l_offset+seq_pos(n1,aln_p->llrev,2),
l_offset+seq_pos(aln_p->amin1+1,aln_p->llrev,1),l_offset+seq_pos(aln_p->amax1,aln_p->llrev,2),
tmp_annot_s, etmp_domain_s);
if (n_tmp_domain>0 && tmp_domain_s) {
free(tmp_domain_s);
free(etmp_domain_s);
}
if (l_annot_p && l_annot_p->n_domains && l_domain_s) {
free(l_domain_s);
}
if (q_annot_p && q_annot_p->n_domains && q_domain_s) {
free(q_domain_s);
}
if (annot_var_s && annot_var_s[0] && tmp_annot_s) free(tmp_annot_s);
}
fprintf(fp,"\n<!-- LINK_STOP -->");
fflush(fp);
}
/*
if ((srch_url2 = getenv("SRCH_URL2"))==NULL)
fprintf(fp,"<A HREF=\"http://fasta.bioch.virginia.edu/fasta/cgi/lalignx.cgi?seq1=\"%s\"&in_seq1=\"FASTA\"&seq2=\"%s\"&in_seq2=\"Accession\"&ssr2=%ld:%ld\">lalign</A>\n<p>\n",my_l_name,db,lib,pgm,l_offset+aln_p->amin1+1,l_offset+aln_p->amax1,n1);
else
fprintf(fp,srch_url1,my_l_name,db,lib,pgm,
l_offset+aln_p->amin1+1,l_offset+aln_p->amax1,n1);
*/
if (getenv("JSON_HTML")) {
/* replace '|' with '_' */
SAFE_STRNCPY(json_l_name, l_name, sizeof(json_l_name));
for (bp=strchr(json_l_name,'|'); bp; bp=strchr(bp+1,'|')) { *bp = '_'; }
/* replace '.' with '_' */
for (bp=strchr(json_l_name,'.'); bp; bp=strchr(bp+1,'.')) { *bp = '_'; }
fprintf(fp,"\n<script type=\"text/javascript\">\n//<![CDATA[\n var json_%s = {\n",json_l_name);
encode_json_str(fp, "db", db, 1);
encode_json_str(fp, "l_acc", l_name, 0);
encode_json_str(fp, "acc", my_l_name, 0);
encode_json_str(fp, "lib", lib, 0);
encode_json_str(fp, "pgm", pgm, 0);
encode_json_str(fp, "o_pgm", m_msp->f_id0, 0);
encode_json_aln(fp, aln_p, q_offset, l_offset, 0);
if (annot_var_s && annot_var_s[0]) { encode_json_lines(fp, "annot", annot_var_s, 0); }
if (q_annot_p && q_annot_p->n_domains > 0) { encode_json_domains(fp, "q_domains", q_annot_p, 0); }
if (l_annot_p && l_annot_p->n_domains > 0) { encode_json_domains(fp, "l_domains", l_annot_p, 0); }
fprintf(fp, "\n}\n//]]>\n</script>");
fflush(fp);
}
}