blob: b17a163a25f4d9b6d9c14813d5d0178ed50fd4b3 [file] [log] [blame]
/* -*- Mode:C++; c-file-style:"gnu"; indent-tabs-mode:nil -*- */
/**
* Copyright (C) 2013 Regents of the University of California.
* @author: Jeff Thompson <jefft0@remap.ucla.edu>
* @author: Alexander Afanasyev <alexander.afanasyev@ucla.edu>
* See COPYING for copyright and distribution information.
*/
#include <stdexcept>
#include <algorithm>
#include <string.h>
#include <ndn-cpp/name.hpp>
#include "c/name.h"
#include "c/util/ndn_memory.h"
using namespace std;
namespace ndn {
static const char *WHITESPACE_CHARS = " \n\r\t";
/**
* Modify str in place to erase whitespace on the left.
* @param str
*/
static inline void
trimLeft(string& str)
{
size_t found = str.find_first_not_of(WHITESPACE_CHARS);
if (found != string::npos) {
if (found > 0)
str.erase(0, found);
}
else
// All whitespace
str.clear();
}
/**
* Modify str in place to erase whitespace on the right.
* @param str
*/
static inline void
trimRight(string& str)
{
size_t found = str.find_last_not_of(WHITESPACE_CHARS);
if (found != string::npos) {
if (found + 1 < str.size())
str.erase(found + 1);
}
else
// All whitespace
str.clear();
}
/**
* Modify str in place to erase whitespace on the left and right.
* @param str
*/
static void
trim(string& str)
{
trimLeft(str);
trimRight(str);
}
/**
* Convert the hex character to an integer from 0 to 15, or -1 if not a hex character.
* @param c
* @return
*/
static int
fromHexChar(uint8_t c)
{
if (c >= '0' && c <= '9')
return (int)c - (int)'0';
else if (c >= 'A' && c <= 'F')
return (int)c - (int)'A' + 10;
else if (c >= 'a' && c <= 'f')
return (int)c - (int)'a' + 10;
else
return -1;
}
/**
* Return a copy of str, converting each escaped "%XX" to the char value.
* @param str
*/
static string
unescape(const string& str)
{
ostringstream result;
for (size_t i = 0; i < str.size(); ++i) {
if (str[i] == '%' && i + 2 < str.size()) {
int hi = fromHexChar(str[i + 1]);
int lo = fromHexChar(str[i + 2]);
if (hi < 0 || lo < 0)
// Invalid hex characters, so just keep the escaped string.
result << str[i] << str[i + 1] << str[i + 2];
else
result << (uint8_t)(16 * hi + lo);
// Skip ahead past the escaped value.
i += 2;
}
else
// Just copy through.
result << str[i];
}
return result.str();
}
uint64_t Name::Component::toNumberWithMarker(uint8_t marker) const
{
struct ndn_NameComponent componentStruct;
get(componentStruct);
uint64_t result;
ndn_Error error;
if ((error = ndn_NameComponent_toNumberWithMarker(&componentStruct, marker, &result)))
throw runtime_error(ndn_getErrorString(error));
return result;
}
Name::Component
Name::Component::fromNumber(uint64_t number)
{
ptr_lib::shared_ptr<vector<uint8_t> > value(new vector<uint8_t>());
// First encode in little endian.
while (number != 0) {
value->push_back(number & 0xff);
number >>= 8;
}
// Make it big endian.
reverse(value->begin(), value->end());
return Blob(value);
}
Name::Component
Name::Component::fromNumberWithMarker(uint64_t number, uint8_t marker)
{
ptr_lib::shared_ptr<vector<uint8_t> > value(new vector<uint8_t>());
// Add the leading marker.
value->push_back(marker);
// First encode in little endian.
while (number != 0) {
value->push_back(number & 0xff);
number >>= 8;
}
// Make it big endian.
reverse(value->begin() + 1, value->end());
return Blob(value);
}
void
Name::Component::get(struct ndn_NameComponent& componentStruct) const
{
value_.get(componentStruct.value);
}
uint64_t
Name::Component::toNumber() const
{
struct ndn_NameComponent componentStruct;
get(componentStruct);
return ndn_NameComponent_toNumber(&componentStruct);
}
int
Name::Component::compare(const Name::Component& other) const
{
// Imitate ndn_Exclude_compareComponents.
if (value_.size() < other.value_.size())
return -1;
if (value_.size() > other.value_.size())
return 1;
// The components are equal length. Just do a byte compare.
return ndn_memcmp((uint8_t*)value_.buf(), (uint8_t*)other.value_.buf(), value_.size());
}
void
Name::set(const char *uri_cstr)
{
components_.clear();
string uri = uri_cstr;
trim(uri);
if (uri.size() == 0)
return;
size_t iColon = uri.find(':');
if (iColon != string::npos) {
// Make sure the colon came before a '/'.
size_t iFirstSlash = uri.find('/');
if (iFirstSlash == string::npos || iColon < iFirstSlash) {
// Omit the leading protocol such as ndn:
uri.erase(0, iColon + 1);
trim(uri);
}
}
// Trim the leading slash and possibly the authority.
if (uri[0] == '/') {
if (uri.size() >= 2 && uri[1] == '/') {
// Strip the authority following "//".
size_t iAfterAuthority = uri.find('/', 2);
if (iAfterAuthority == string::npos)
// Unusual case: there was only an authority.
return;
else {
uri.erase(0, iAfterAuthority + 1);
trim(uri);
}
}
else {
uri.erase(0, 1);
trim(uri);
}
}
size_t iComponentStart = 0;
// Unescape the components.
while (iComponentStart < uri.size()) {
size_t iComponentEnd = uri.find("/", iComponentStart);
if (iComponentEnd == string::npos)
iComponentEnd = uri.size();
Component component(fromEscapedString(&uri[0], iComponentStart, iComponentEnd));
// Ignore illegal components. This also gets rid of a trailing '/'.
if (component.getValue())
components_.push_back(Component(component));
iComponentStart = iComponentEnd + 1;
}
}
void
Name::get(struct ndn_Name& nameStruct) const
{
if (nameStruct.maxComponents < components_.size())
throw runtime_error("nameStruct.maxComponents must be >= this name getNComponents()");
nameStruct.nComponents = components_.size();
for (size_t i = 0; i < nameStruct.nComponents; ++i)
components_[i].get(nameStruct.components[i]);
}
void
Name::set(const struct ndn_Name& nameStruct)
{
clear();
for (size_t i = 0; i < nameStruct.nComponents; ++i)
append(nameStruct.components[i].value.value, nameStruct.components[i].value.length);
}
Name&
Name::append(const Name& name)
{
if (&name == this)
// Copying from this name, so need to make a copy first.
return append(Name(name));
for (size_t i = 0; i < name.components_.size(); ++i)
components_.push_back(name.components_[i]);
return *this;
}
string
Name::toUri() const
{
if (components_.size() == 0)
return "/";
ostringstream result;
for (size_t i = 0; i < components_.size(); ++i) {
result << "/";
toEscapedString(*components_[i].getValue(), result);
}
return result.str();
}
Name
Name::getSubName(size_t iStartComponent, size_t nComponents) const
{
Name result;
size_t iEnd = iStartComponent + nComponents;
for (size_t i = iStartComponent; i < iEnd && i < components_.size(); ++i)
result.components_.push_back(components_[i]);
return result;
}
Name
Name::getSubName(size_t iStartComponent) const
{
Name result;
for (size_t i = iStartComponent; i < components_.size(); ++i)
result.components_.push_back(components_[i]);
return result;
}
bool
Name::equals(const Name& name) const
{
if (components_.size() != name.components_.size())
return false;
for (size_t i = 0; i < components_.size(); ++i) {
if (*components_[i].getValue() != *name.components_[i].getValue())
return false;
}
return true;
}
bool
Name::match(const Name& name) const
{
// Imitate ndn_Name_match.
// This name is longer than the name we are checking it against.
if (components_.size() > name.components_.size())
return false;
// Check if at least one of given components doesn't match.
for (size_t i = 0; i < components_.size(); ++i) {
if (*components_[i].getValue() != *name.components_[i].getValue())
return false;
}
return true;
}
Blob
Name::fromEscapedString(const char *escapedString, size_t beginOffset, size_t endOffset)
{
string trimmedString(escapedString + beginOffset, escapedString + endOffset);
trim(trimmedString);
string value = unescape(trimmedString);
if (value.find_first_not_of(".") == string::npos) {
// Special case for component of only periods.
if (value.size() <= 2)
// Zero, one or two periods is illegal. Ignore this component.
return Blob();
else
// Remove 3 periods.
return Blob((const uint8_t *)&value[3], value.size() - 3);
}
else
return Blob((const uint8_t *)&value[0], value.size());
}
Blob
Name::fromEscapedString(const char *escapedString)
{
return fromEscapedString(escapedString, 0, ::strlen(escapedString));
}
void
Name::toEscapedString(const vector<uint8_t>& value, ostringstream& result)
{
bool gotNonDot = false;
for (unsigned i = 0; i < value.size(); ++i) {
if (value[i] != 0x2e) {
gotNonDot = true;
break;
}
}
if (!gotNonDot) {
// Special case for component of zero or more periods. Add 3 periods.
result << "...";
for (size_t i = 0; i < value.size(); ++i)
result << '.';
}
else {
// In case we need to escape, set to upper case hex and save the previous flags.
ios::fmtflags saveFlags = result.flags(ios::hex | ios::uppercase);
for (size_t i = 0; i < value.size(); ++i) {
uint8_t x = value[i];
// Check for 0-9, A-Z, a-z, (+), (-), (.), (_)
if ((x >= 0x30 && x <= 0x39) || (x >= 0x41 && x <= 0x5a) ||
(x >= 0x61 && x <= 0x7a) || x == 0x2b || x == 0x2d ||
x == 0x2e || x == 0x5f)
result << x;
else {
result << '%';
if (x < 16)
result << '0';
result << (unsigned int)x;
}
}
// Restore.
result.flags(saveFlags);
}
}
string
Name::toEscapedString(const vector<uint8_t>& value)
{
ostringstream result;
toEscapedString(value, result);
return result.str();
}
bool
Name::breadthFirstLess(const Name& name1, const Name& name2)
{
for (size_t i = 0; i < name1.size() && i < name2.size(); ++i) {
if (name1[i] == name2[i])
// The components at this index are equal, so check the next components.
continue;
// Otherwise, the result is based on the components at this index.
return name1[i] < name2[i];
}
// The components up to min(name1.size(), name2.size()) are equal, so sort on the shorter name.
return name1.size() < name2.size();
}
}