};
class HttpRewrite::Within {
public:
- std::string header;
- std::string attr;
- std::string tag;
+ boost::regex header;
+ boost::regex attr;
+ boost::regex tag;
+ std::string type;
bool reqline;
RulePtr rule;
+ bool exec(std::map<std::string, std::string> &vars,
+ std::string &txt, bool anchor) const;
};
class HttpRewrite::Content {
if (it->reqline)
{
yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
- if (it->rule->test_patterns(vars, path, true))
+ if (it->exec(vars, path, true))
{
yaz_log(YLOG_LOG, "Rewritten request URL is %s", path.c_str());
hreq->path = odr_strdup(o, path.c_str());
std::list<Within>::const_iterator it = cit->within_list.begin();
for (; it != cit->within_list.end(); it++)
{
- if (it->header.length() > 0 &&
- yaz_strcasecmp(it->header.c_str(), header->name) == 0)
+ if (!it->header.empty() &&
+ regex_match(header->name, it->header))
{
+#ifdef OLDHEADERMATCH
+ // Matches and replaces the whole header line.
+ // This is good if you want to play with the header name too,
+ // but useless for patterns that want to anchor to the beginning
+ // or end of the header value, as we want to do with host-relative
+ // links. This code should probably be removed.
std::string sheader(header->name);
sheader += ": ";
sheader += header->value;
- if (it->rule->test_patterns(vars, sheader, true))
+ if (it->exec(vars, sheader, true))
{
size_t pos = sheader.find(": ");
if (pos == std::string::npos)
header->value = odr_strdup(
o, sheader.substr(pos + 2, std::string::npos).c_str());
}
+#else
+ // Match and replace only the header value
+ std::string hval(header->value);
+ if (it->exec(vars, hval, true))
+ {
+ header->value = odr_strdup(o, hval.c_str());
+ }
+
+#endif
}
}
}
int *content_len,
std::map<std::string, std::string> & vars) const
{
+ if (*content_len == 0)
+ return;
+ if (!content_type) {
+ yaz_log(YLOG_LOG, "rewrite_body: null content_type, can not rewrite");
+ return;
+ }
std::list<Content>::const_iterator cit = content_list.begin();
for (; cit != content_list.end(); cit++)
{
&& regex_match(content_type, cit->content_re))
break;
}
- if (cit == content_list.end())
+ if (cit == content_list.end()) {
+ yaz_log(YLOG_LOG,"rewrite_body: No content rule matched %s, not rewriting",
+ content_type );
return;
-
- if (*content_buf)
- {
- int i;
- for (i = 0; i < *content_len; i++)
- if ((*content_buf)[i] == 0)
- return; // binary content. skip
-
- std::string content(*content_buf, *content_len);
- cit->parse(m_verbose, content, vars);
- *content_buf = odr_strdup(o, content.c_str());
- *content_len = strlen(*content_buf);
}
+
+ int i;
+ for (i = 0; i < *content_len; i++)
+ if ((*content_buf)[i] == 0) {
+ yaz_log(YLOG_LOG,"rewrite_body: Looks like binary stuff, not rewriting");
+ return; // binary content. skip
+ }
+
+ std::string content(*content_buf, *content_len);
+ cit->parse(m_verbose, content, vars);
+ *content_buf = odr_strdup(o, content.c_str());
+ *content_len = strlen(*content_buf);
}
yf::HttpRewrite::Event::Event(const Content *p,
std::list<Within>::const_iterator it = m_content->within_list.begin();
for (; it != m_content->within_list.end(); it++)
{
- if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
- t.c_str()) == 0)
+ if (!it->tag.empty() && regex_match(t, it->tag))
{
- std::vector<std::string> attr;
- boost::split(attr, it->attr, boost::is_any_of(","));
- size_t i;
- for (i = 0; i < attr.size(); i++)
+ if (!it->attr.empty() && regex_match("#text", it->attr))
{
- if (attr[i].compare("#text") == 0)
- {
- s_within.push(it);
- return;
- }
+ s_within.push(it);
+ return;
}
}
}
{
std::list<Within>::const_iterator it = s_within.top();
std::string t(tag, tag_len);
- if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
+ if (regex_match(t, it->tag))
s_within.pop();
}
}
for (; it != m_content->within_list.end(); it++)
{
std::string t(tag, tag_len);
- if (it->tag.length() == 0 ||
- yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
+ if (it->tag.empty() || regex_match(t, it->tag))
{
std::string a(attr, attr_len);
- std::vector<std::string> attr;
- boost::split(attr, it->attr, boost::is_any_of(","));
- size_t i;
- for (i = 0; i < attr.size(); i++)
- {
- if (attr[i].compare("#text") &&
- yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
- subst = true;
- }
+ if (!it->attr.empty() && regex_match(a, it->attr))
+ subst = true;
}
if (subst)
break;
if (subst)
{
std::string s(value, val_len);
- it->rule->test_patterns(m_vars, s, true);
+ it->exec(m_vars, s, true);
wrbuf_puts(m_w, s.c_str());
}
else
{
std::list<Within>::const_iterator it = s_within.top();
std::string t(tag, tag_len);
- if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
+ if (regex_match(t, it->tag))
s_within.pop();
}
wrbuf_puts(m_w, "</");
if (it != m_content->within_list.end())
{
std::string s(value, len);
- it->rule->test_patterns(m_vars, s, false);
+ it->exec(m_vars, s, false);
wrbuf_puts(m_w, s.c_str());
}
else
wrbuf_write(m_w, value, len);
}
+static bool embed_quoted_literal(
+ std::string &content,
+ std::map<std::string, std::string> &vars,
+ mp::filter::HttpRewrite::RulePtr ruleptr,
+ bool html_context)
+{
+ bool replace = false;
+ std::string res;
+ const char *cp = content.c_str();
+ const char *cp0 = cp;
+ while (*cp)
+ {
+ if (html_context && !strncmp(cp, """, 6))
+ {
+ cp += 6;
+ res.append(cp0, cp - cp0);
+ cp0 = cp;
+ while (*cp)
+ {
+ if (!strncmp(cp, """, 6))
+ break;
+ if (*cp == '\n')
+ break;
+ cp++;
+ }
+ if (!*cp)
+ break;
+ std::string s(cp0, cp - cp0);
+ if (ruleptr->test_patterns(vars, s, true))
+ replace = true;
+ cp0 = cp;
+ res.append(s);
+ }
+ else if (*cp == '"' || *cp == '\'')
+ {
+ int m = *cp;
+ cp++;
+ res.append(cp0, cp - cp0);
+ cp0 = cp;
+ while (*cp)
+ {
+ if (cp[-1] != '\\' && *cp == m)
+ break;
+ if (*cp == '\n')
+ break;
+ cp++;
+ }
+ if (!*cp)
+ break;
+ std::string s(cp0, cp - cp0);
+ if (ruleptr->test_patterns(vars, s, true))
+ replace = true;
+ cp0 = cp;
+ res.append(s);
+ }
+ else if (*cp == '/' && cp[1] == '/')
+ {
+ while (cp[1] && cp[1] != '\n')
+ cp++;
+ }
+ cp++;
+ }
+ res.append(cp0, cp - cp0);
+ content = res;
+ return replace;
+}
+
+bool yf::HttpRewrite::Within::exec(
+ std::map<std::string, std::string> & vars,
+ std::string & txt, bool anchor) const
+{
+ if (type == "quoted-literal")
+ {
+ return embed_quoted_literal(txt, vars, rule, true);
+ }
+ else
+ {
+ return rule->test_patterns(vars, txt, anchor);
+ }
+}
+
bool yf::HttpRewrite::Rule::test_patterns(
std::map<std::string, std::string> & vars,
std::string & txt, bool anchor)
std::string &content,
std::map<std::string, std::string> &vars) const
{
- std::string res;
- const char *cp = content.c_str();
- const char *cp0 = cp;
- while (*cp)
- {
- if (*cp == '"' || *cp == '\'')
- {
- int m = *cp;
- cp++;
- res.append(cp0, cp - cp0);
- cp0 = cp;
- while (*cp)
- {
- if (cp[-1] != '\\' && *cp == m)
- break;
- if (*cp == '\n')
- break;
- cp++;
- }
- if (!*cp)
- break;
- std::list<Within>::const_iterator it = within_list.begin();
- std::string s(cp0, cp - cp0);
- if (it != within_list.end())
- it->rule->test_patterns(vars, s, true);
- cp0 = cp;
- res.append(s);
- }
- else if (*cp == '/' && cp[1] == '/')
- {
- while (cp[1] && cp[1] != '\n')
- cp++;
- }
- cp++;
- }
- res.append(cp0, cp - cp0);
- content = res;
+ std::list<Within>::const_iterator it = within_list.begin();
+ if (it != within_list.end())
+ embed_quoted_literal(content, vars, it->rule, false);
}
void yf::HttpRewrite::Content::configure(
continue;
if (!strcmp((const char *) ptr->name, "within"))
{
- static const char *names[6] =
- { "header", "attr", "tag", "rule", "reqline", 0 };
- std::string values[5];
+ static const char *names[7] =
+ { "header", "attr", "tag", "rule", "reqline", "type", 0 };
+ std::string values[6];
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())
+ if (values[0].length() > 0)
+ w.header.assign(values[0], boost::regex_constants::icase);
+ if (values[1].length() > 0)
+ w.attr.assign(values[1], boost::regex_constants::icase);
+ if (values[2].length() > 0)
+ w.tag.assign(values[2], boost::regex_constants::icase);
+
+ std::vector<std::string> rulenames;
+ boost::split(rulenames, values[3], boost::is_any_of(","));
+ if (rulenames.size() == 0)
+ {
throw mp::filter::FilterException
- ("Reference to non-existing rule '" + values[3] +
+ ("Empty rule in '" + values[3] +
"' in http_rewrite filter");
- w.rule = it->second;
+ }
+ else if (rulenames.size() == 1)
+ {
+ std::map<std::string,RulePtr>::const_iterator it =
+ rules.find(rulenames[0]);
+ if (it == rules.end())
+ throw mp::filter::FilterException
+ ("Reference to non-existing rule '" + rulenames[0] +
+ "' in http_rewrite filter");
+ w.rule = it->second;
+
+ }
+ else
+ {
+ RulePtr rule(new Rule);
+ size_t i;
+ for (i = 0; i < rulenames.size(); i++)
+ {
+ std::map<std::string,RulePtr>::const_iterator it =
+ rules.find(rulenames[i]);
+ if (it == rules.end())
+ throw mp::filter::FilterException
+ ("Reference to non-existing rule '" + rulenames[i] +
+ "' in http_rewrite filter");
+ RulePtr subRule = it->second;
+ std::list<Replace>::iterator rit =
+ subRule->replace_list.begin();
+ for (; rit != subRule->replace_list.end(); rit++)
+ rule->replace_list.push_back(*rit);
+ }
+ w.rule = rule;
+ }
w.reqline = values[4] == "1";
+ w.type = values[5];
+ if (w.type.empty() || w.type == "quoted-literal")
+ ;
+ else
+ throw mp::filter::FilterException
+ ("within type must be quoted-literal or none in "
+ " in http_rewrite filter");
within_list.push_back(w);
}
}
Content c;
c.type = values[0];
- // if (!values[1].empty())
- c.content_re = values[1];
+ if (!values[1].empty())
+ c.content_re.assign(values[1], boost::regex::icase);
c.configure(ptr->children, rules);
phase.content_list.push_back(c);
}