#include <boost/regex.hpp>
#include <boost/lexical_cast.hpp>
-#include <list>
#include <map>
-#if HAVE_SYS_TYPES_H
-#include <sys/types.h>
-#endif
-
namespace mp = metaproxy_1;
namespace yf = mp::filter;
-yf::HttpRewrite::HttpRewrite()
+namespace metaproxy_1 {
+ namespace filter {
+ class HttpRewrite::Replace {
+ public:
+ std::string regex;
+ std::string recipe;
+ std::map<int, std::string> group_index;
+ const std::string search_replace(
+ std::map<std::string, std::string> & vars,
+ const std::string & txt) const;
+ std::string sub_vars (
+ const std::map<std::string, std::string> & vars) const;
+ void parse_groups();
+ };
+
+ class HttpRewrite::Rule {
+ public:
+ std::list<Replace> replace_list;
+ const std::string test_patterns(
+ std::map<std::string, std::string> & vars,
+ const std::string & txt) const;
+ };
+ class HttpRewrite::Within {
+ public:
+ std::string header;
+ std::string attr;
+ std::string tag;
+ RulePtr rule;
+ };
+
+ class HttpRewrite::Phase {
+ public:
+ std::list<Within> within_list;
+ void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
+ std::map<std::string, std::string> & vars) const;
+ void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
+ std::map<std::string, std::string> & vars) const;
+ void rewrite_body(mp::odr & o,
+ char **content_buf, int *content_len,
+ std::map<std::string, std::string> & vars) const;
+ };
+ }
+}
+
+yf::HttpRewrite::HttpRewrite() :
+ req_phase(new Phase), res_phase(new Phase)
{
}
{
}
-void yf::HttpRewrite::process(mp::Package & package) const
+void yf::HttpRewrite::process(mp::Package & package) const
{
- yaz_log(YLOG_DEBUG, "HttpRewrite begins....");
+ yaz_log(YLOG_LOG, "HttpRewrite begins....");
Z_GDU *gdu = package.request().get();
//map of request/response vars
std::map<std::string, std::string> vars;
{
Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
mp::odr o;
- rewrite_reqline(o, hreq, vars);
- yaz_log(YLOG_DEBUG, ">> Request headers");
- rewrite_headers(o, hreq->headers, vars);
- rewrite_body(o, &hreq->content_buf, &hreq->content_len, vars);
+ req_phase->rewrite_reqline(o, hreq, vars);
+ yaz_log(YLOG_LOG, ">> Request headers");
+ req_phase->rewrite_headers(o, hreq->headers, vars);
+ req_phase->rewrite_body(o,
+ &hreq->content_buf, &hreq->content_len, vars);
package.request() = gdu;
}
package.move();
if (gdu && gdu->which == Z_GDU_HTTP_Response)
{
Z_HTTP_Response *hres = gdu->u.HTTP_Response;
- yaz_log(YLOG_DEBUG, "Response code %d", hres->code);
+ yaz_log(YLOG_LOG, "Response code %d", hres->code);
mp::odr o;
- yaz_log(YLOG_DEBUG, "<< Respose headers");
- rewrite_headers(o, hres->headers, vars);
- rewrite_body(o, &hres->content_buf, &hres->content_len, vars);
+ yaz_log(YLOG_LOG, "<< Respose headers");
+ res_phase->rewrite_headers(o, hres->headers, vars);
+ res_phase->rewrite_body(o, &hres->content_buf,
+ &hres->content_len, vars);
package.response() = gdu;
}
}
-void yf::HttpRewrite::rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq,
- std::map<std::string, std::string> & vars) const
+void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
+ Z_HTTP_Request *hreq,
+ std::map<std::string, std::string> & vars) const
{
//rewrite the request line
std::string path;
if (strstr(hreq->path, "http://") == hreq->path)
{
- yaz_log(YLOG_DEBUG, "Path in the method line is absolute, "
+ yaz_log(YLOG_LOG, "Path in the method line is absolute, "
"possibly a proxy request");
path += hreq->path;
}
else
{
//TODO what about proto
+ path += "http://";
path += z_HTTP_header_lookup(hreq->headers, "Host");
- path += hreq->path;
+ path += hreq->path;
}
- yaz_log(YLOG_DEBUG, "Proxy request URL is %s", path.c_str());
- std::string npath =
- test_patterns(vars, path, req_uri_pats, req_groups_bynum);
- if (!npath.empty())
+
+ std::list<Within>::const_iterator it = within_list.begin();
+ if (it != within_list.end())
{
- yaz_log(YLOG_DEBUG, "Rewritten request URL is %s", npath.c_str());
- hreq->path = odr_strdup(o, npath.c_str());
+ RulePtr rule = it->rule;
+
+ yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
+ std::string npath = rule->test_patterns(vars, path);
+ if (!npath.empty())
+ {
+ yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
+ hreq->path = odr_strdup(o, npath.c_str());
+ }
}
}
-void yf::HttpRewrite::rewrite_headers (mp::odr & o, Z_HTTP_Header *headers,
- std::map<std::string, std::string> & vars) const
+void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
+ Z_HTTP_Header *headers,
+ std::map<std::string, std::string> & vars) const
{
for (Z_HTTP_Header *header = headers;
- header != 0;
- header = header->next)
+ header != 0;
+ header = header->next)
{
std::string sheader(header->name);
sheader += ": ";
sheader += header->value;
- yaz_log(YLOG_DEBUG, "%s: %s", header->name, header->value);
- std::string out = test_patterns(vars,
- sheader,
- req_uri_pats, req_groups_bynum);
- if (!out.empty())
+ yaz_log(YLOG_LOG, "%s: %s", header->name, header->value);
+
+ std::list<Within>::const_iterator it = within_list.begin();
+ if (it == within_list.end())
+ continue;
+ RulePtr rule = it->rule;
+
+ std::string out = rule->test_patterns(vars, sheader);
+ if (!out.empty())
{
size_t pos = out.find(": ");
if (pos == std::string::npos)
{
- yaz_log(YLOG_DEBUG, "Header malformed during rewrite, ignoring");
+ yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
continue;
}
header->name = odr_strdup(o, out.substr(0, pos).c_str());
- header->value = odr_strdup(o, out.substr(pos+2,
- std::string::npos).c_str());
+ header->value = odr_strdup(o, out.substr(pos+2,
+ std::string::npos).c_str());
}
}
}
-void yf::HttpRewrite::rewrite_body (mp::odr & o, char **content_buf, int *content_len,
- std::map<std::string, std::string> & vars) const
+void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
+ char **content_buf,
+ int *content_len,
+ std::map<std::string, std::string> & vars) const
{
if (*content_buf)
{
- std::string body(*content_buf);
- std::string nbody =
- test_patterns(vars, body, req_uri_pats, req_groups_bynum);
- if (!nbody.empty())
+
+ std::list<Within>::const_iterator it = within_list.begin();
+ if (it != within_list.end())
{
- *content_buf = odr_strdup(o, nbody.c_str());
- *content_len = nbody.size();
+ RulePtr rule = it->rule;
+
+ std::string body(*content_buf);
+ std::string nbody = rule->test_patterns(vars, body);
+ if (!nbody.empty())
+ {
+ *content_buf = odr_strdup(o, nbody.c_str());
+ *content_len = nbody.size();
+ }
}
}
}
* Tests pattern from the vector in order and executes recipe on
the first match.
*/
-const std::string yf::HttpRewrite::test_patterns(
+const std::string yf::HttpRewrite::Rule::test_patterns(
std::map<std::string, std::string> & vars,
- const std::string & txt,
- const spair_vec & uri_pats,
- const std::vector<std::map<int, std::string> > & groups_bynum_vec)
- const
+ const std::string & txt) const
{
- for (unsigned i = 0; i < uri_pats.size(); i++)
+ std::list<Replace>::const_iterator it = replace_list.begin();
+
+ for (; it != replace_list.end(); it++)
{
- std::string out = search_replace(vars, txt,
- uri_pats[i].first, uri_pats[i].second,
- groups_bynum_vec[i]);
+ std::string out = it->search_replace(vars, txt);
if (!out.empty()) return out;
}
return "";
}
-
-const std::string yf::HttpRewrite::search_replace(
+const std::string yf::HttpRewrite::Replace::search_replace(
std::map<std::string, std::string> & vars,
- const std::string & txt,
- const std::string & uri_re,
- const std::string & uri_pat,
- const std::map<int, std::string> & groups_bynum) const
+ const std::string & txt) const
{
//exec regex against value
- boost::regex re(uri_re);
+ boost::regex re(regex);
boost::smatch what;
std::string::const_iterator start, end;
start = txt.begin();
std::string out;
while (regex_search(start, end, what, re)) //find next full match
{
- unsigned i;
+ size_t i;
for (i = 1; i < what.size(); ++i)
{
//check if the group is named
std::map<int, std::string>::const_iterator it
- = groups_bynum.find(i);
- if (it != groups_bynum.end())
+ = group_index.find(i);
+ if (it != group_index.end())
{ //it is
- std::string name = it->second;
if (!what[i].str().empty())
- vars[name] = what[i];
+ vars[it->second] = what[i];
}
}
//prepare replacement string
- std::string rvalue = sub_vars(uri_pat, vars);
- //rewrite value
- std::string rhvalue = what.prefix().str()
- + rvalue + what.suffix().str();
- yaz_log(YLOG_DEBUG, "! Rewritten '%s' to '%s'",
+ std::string rvalue = sub_vars(vars);
+ yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
what.str(0).c_str(), rvalue.c_str());
- out += rhvalue;
+ out.append(start, what[0].first);
+ out.append(rvalue);
start = what[0].second; //move search forward
}
+ //if we had a match cat the last part
+ if (start != txt.begin())
+ out.append(start, end);
return out;
}
-void yf::HttpRewrite::parse_groups(
- const spair_vec & uri_pats,
- std::vector<std::map<int, std::string> > & groups_bynum_vec)
+void yf::HttpRewrite::Replace::parse_groups()
{
- for (unsigned h = 0; h < uri_pats.size(); h++)
+ int gnum = 0;
+ bool esc = false;
+ const std::string & str = regex;
+ std::string res;
+ yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
+ for (size_t i = 0; i < str.size(); ++i)
{
- int gnum = 0;
- bool esc = false;
- //regex is first, subpat is second
- std::string str = uri_pats[h].first;
- //for each pair we have an indexing map
- std::map<int, std::string> groups_bynum;
- for (unsigned i = 0; i < str.size(); ++i)
+ res += str[i];
+ if (!esc && str[i] == '\\')
{
- if (!esc && str[i] == '\\')
- {
- esc = true;
- continue;
- }
- if (!esc && str[i] == '(') //group starts
+ esc = true;
+ continue;
+ }
+ if (!esc && str[i] == '(') //group starts
+ {
+ gnum++;
+ if (i+1 < str.size() && str[i+1] == '?') //group with attrs
{
- gnum++;
- if (i+1 < str.size() && str[i+1] == '?') //group with attrs
+ i++;
+ if (i+1 < str.size() && str[i+1] == ':') //non-capturing
{
+ if (gnum > 0) gnum--;
+ res += str[i];
i++;
- if (i+1 < str.size() && str[i+1] == ':') //non-capturing
- {
- if (gnum > 0) gnum--;
- i++;
- continue;
- }
- if (i+1 < str.size() && str[i+1] == 'P') //optional, python
- i++;
- if (i+1 < str.size() && str[i+1] == '<') //named
+ res += str[i];
+ continue;
+ }
+ if (i+1 < str.size() && str[i+1] == 'P') //optional, python
+ i++;
+ if (i+1 < str.size() && str[i+1] == '<') //named
+ {
+ i++;
+ std::string gname;
+ bool term = false;
+ while (++i < str.size())
{
- i++;
- std::string gname;
- bool term = false;
- while (++i < str.size())
- {
- if (str[i] == '>') { term = true; break; }
- if (!isalnum(str[i]))
- throw mp::filter::FilterException
- ("Only alphanumeric chars allowed, found "
- " in '"
- + str
- + "' at "
- + boost::lexical_cast<std::string>(i));
- gname += str[i];
- }
- if (!term)
+ if (str[i] == '>') { term = true; break; }
+ if (!isalnum(str[i]))
throw mp::filter::FilterException
- ("Unterminated group name '" + gname
- + " in '" + str +"'");
- groups_bynum[gnum] = gname;
- yaz_log(YLOG_DEBUG, "Found named group '%s' at $%d",
- gname.c_str(), gnum);
+ ("Only alphanumeric chars allowed, found "
+ " in '"
+ + str
+ + "' at "
+ + boost::lexical_cast<std::string>(i));
+ gname += str[i];
}
+ if (!term)
+ throw mp::filter::FilterException
+ ("Unterminated group name '" + gname
+ + " in '" + str +"'");
+ group_index[gnum] = gname;
+ yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
+ gname.c_str(), gnum);
}
}
- esc = false;
}
- groups_bynum_vec.push_back(groups_bynum);
+ esc = false;
}
+ regex = res;
}
-std::string yf::HttpRewrite::sub_vars (const std::string & in,
- const std::map<std::string, std::string> & vars)
+std::string yf::HttpRewrite::Replace::sub_vars (
+ const std::map<std::string, std::string> & vars) const
{
std::string out;
bool esc = false;
- for (unsigned i = 0; i < in.size(); ++i)
+ const std::string & in = recipe;
+ for (size_t i = 0; i < in.size(); ++i)
{
if (!esc && in[i] == '\\')
{
++i;
std::string name;
bool term = false;
- while (++i < in.size())
+ while (++i < in.size())
{
if (in[i] == '}') { term = true; break; }
name += in[i];
{
throw mp::filter::FilterException
("Malformed or trimmed var ref in '"
- +in+"' at "+boost::lexical_cast<std::string>(i));
+ +in+"' at "+boost::lexical_cast<std::string>(i));
}
continue;
}
return out;
}
-void yf::HttpRewrite::configure(
- const spair_vec req_uri_pats,
- const spair_vec res_uri_pats)
-{
- //TODO should we really copy them out?
- this->req_uri_pats = req_uri_pats;
- this->res_uri_pats = res_uri_pats;
- //pick up names
- parse_groups(req_uri_pats, req_groups_bynum);
- parse_groups(res_uri_pats, res_groups_bynum);
-}
-
-static void configure_rules(const xmlNode *ptr, yf::HttpRewrite::spair_vec & dest)
+void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
{
+ std::map<std::string, RulePtr > rules;
for (ptr = ptr->children; ptr; ptr = ptr->next)
{
if (ptr->type != XML_ELEMENT_NODE)
continue;
- else if (!strcmp((const char *) ptr->name, "rewrite"))
+ else if (!strcmp((const char *) ptr->name, "rule"))
{
- std::string from, to;
- const struct _xmlAttr *attr;
- for (attr = ptr->properties; attr; attr = attr->next)
+ static const char *names[2] = { "name", 0 };
+ std::string values[1];
+ values[0] = "default";
+ mp::xml::parse_attr(ptr, names, values);
+
+ RulePtr rule(new Rule);
+ for (xmlNode *p = ptr->children; p; p = p->next)
{
- if (!strcmp((const char *) attr->name, "from"))
- from = mp::xml::get_text(attr->children);
- else if (!strcmp((const char *) attr->name, "to"))
- to = mp::xml::get_text(attr->children);
+ if (p->type != XML_ELEMENT_NODE)
+ continue;
+ if (!strcmp((const char *) p->name, "rewrite"))
+ {
+ Replace replace;
+ const struct _xmlAttr *attr;
+ for (attr = p->properties; attr; attr = attr->next)
+ {
+ if (!strcmp((const char *) attr->name, "from"))
+ replace.regex = mp::xml::get_text(attr->children);
+ else if (!strcmp((const char *) attr->name, "to"))
+ replace.recipe = mp::xml::get_text(attr->children);
+ else
+ throw mp::filter::FilterException
+ ("Bad attribute "
+ + std::string((const char *) attr->name)
+ + " in rewrite section of http_rewrite");
+ }
+ yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
+ replace.regex.c_str(), replace.recipe.c_str());
+ replace.parse_groups();
+ if (!replace.regex.empty())
+ rule->replace_list.push_back(replace);
+ }
else
throw mp::filter::FilterException
- ("Bad attribute "
- + std::string((const char *) attr->name)
- + " in rewrite section of http_rewrite");
+ ("Bad element "
+ + std::string((const char *) p->name)
+ + " in http_rewrite filter");
}
- yaz_log(YLOG_DEBUG, "Found rewrite rule from '%s' to '%s'",
- from.c_str(), to.c_str());
- if (!from.empty())
- dest.push_back(std::make_pair(from, to));
+ if (!rule->replace_list.empty())
+ rules[values[0]] = rule;
+ }
+ else if (!strcmp((const char *) ptr->name, "within"))
+ {
+ static const char *names[5] =
+ { "header", "attr", "tag", "rule", 0 };
+ std::string values[4];
+ mp::xml::parse_attr(ptr, names, values);
+ Within w;
+ w.header = values[0];
+ w.attr = values[1];
+ w.tag = values[2];
+ std::map<std::string,RulePtr>::const_iterator it =
+ rules.find(values[3]);
+ if (it == rules.end())
+ throw mp::filter::FilterException
+ ("Reference to non-existing rule '" + values[3] +
+ "' in http_rewrite filter");
+ w.rule = it->second;
+ phase.within_list.push_back(w);
}
else
{
throw mp::filter::FilterException
- ("Bad element o"
+ ("Bad element "
+ std::string((const char *) ptr->name)
- + " in http_rewrite1 filter");
+ + " in http_rewrite filter");
}
}
}
void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
const char *path)
{
- spair_vec req_uri_pats;
- spair_vec res_uri_pats;
for (ptr = ptr->children; ptr; ptr = ptr->next)
{
if (ptr->type != XML_ELEMENT_NODE)
continue;
else if (!strcmp((const char *) ptr->name, "request"))
{
- configure_rules(ptr, req_uri_pats);
+ configure_phase(ptr, *req_phase);
}
else if (!strcmp((const char *) ptr->name, "response"))
{
- configure_rules(ptr, res_uri_pats);
+ configure_phase(ptr, *res_phase);
}
else
{
+ " in http_rewrite1 filter");
}
}
- configure(req_uri_pats, res_uri_pats);
}
static mp::filter::Base* filter_creator()