separate out token replacer as a standalone function in preparation for #300
This commit is contained in:
@@ -10,6 +10,12 @@ typedef util::variant<bool, int64_t, uint64_t, double, std::string> Value;
|
||||
|
||||
std::string toString(const Value &value);
|
||||
|
||||
inline std::string& operator+=(std::string& out, const Value& value) {
|
||||
out += toString(value);
|
||||
return out;
|
||||
}
|
||||
|
||||
|
||||
Value parseValue(pbf data);
|
||||
|
||||
namespace util {
|
||||
|
||||
49
include/llmr/util/token.hpp
Normal file
49
include/llmr/util/token.hpp
Normal file
@@ -0,0 +1,49 @@
|
||||
#ifndef LLMR_UTIL_TOKEN
|
||||
#define LLMR_UTIL_TOKEN
|
||||
|
||||
#ifdef __linux__
|
||||
#include <boost/regex.hpp>
|
||||
namespace regex_impl = boost;
|
||||
#else
|
||||
#include <regex>
|
||||
namespace regex_impl = std;
|
||||
#endif
|
||||
|
||||
namespace llmr {
|
||||
namespace util {
|
||||
|
||||
namespace detail {
|
||||
const regex_impl::regex tokenRegex("\\{\\{(\\w+)\\}\\}");
|
||||
const regex_impl::sregex_token_iterator tokensEnd = regex_impl::sregex_token_iterator();
|
||||
}
|
||||
|
||||
template <typename Tokens>
|
||||
std::string replaceTokens(const std::string &source, const Tokens &tokens) {
|
||||
std::string result;
|
||||
result.reserve(source.size());
|
||||
|
||||
bool token = false;
|
||||
for (auto token_it = regex_impl::sregex_token_iterator(source.begin(), source.end(),
|
||||
detail::tokenRegex, {-1, 1});
|
||||
token_it != detail::tokensEnd; ++token_it, token = !token) {
|
||||
if (!token_it->matched) {
|
||||
continue;
|
||||
}
|
||||
|
||||
if (token) {
|
||||
const auto it_prop = tokens.find(token_it->str());
|
||||
if (it_prop != tokens.end()) {
|
||||
result += it_prop->second;
|
||||
}
|
||||
} else {
|
||||
result += token_it->str();
|
||||
}
|
||||
}
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
} // end namespace util
|
||||
} // end namespace llmr
|
||||
|
||||
#endif
|
||||
@@ -9,6 +9,7 @@
|
||||
#include <llmr/renderer/raster_bucket.hpp>
|
||||
#include <llmr/util/raster.hpp>
|
||||
#include <llmr/util/constants.hpp>
|
||||
#include <llmr/util/token.hpp>
|
||||
#include <llmr/geometry/glyph_atlas.hpp>
|
||||
#include <llmr/text/glyph_store.hpp>
|
||||
#include <llmr/text/glyph.hpp>
|
||||
@@ -195,9 +196,6 @@ std::unique_ptr<Bucket> TileParser::createTextBucket(const VectorTileLayer &laye
|
||||
tile.textVertexBuffer, tile.triangleElementsBuffer, bucket_desc, placement);
|
||||
|
||||
util::utf8_to_utf32 ucs4conv;
|
||||
regex_impl::regex token_regex("\\{\\{(\\w+)\\}\\}");
|
||||
const auto tokens_end = regex_impl::sregex_token_iterator();
|
||||
|
||||
|
||||
std::vector<std::pair<std::u32string, pbf>> labels;
|
||||
|
||||
@@ -211,34 +209,8 @@ std::unique_ptr<Bucket> TileParser::createTextBucket(const VectorTileLayer &laye
|
||||
return nullptr;
|
||||
VectorTileFeature feature{feature_pbf, layer};
|
||||
|
||||
const std::string &field = bucket_desc.geometry.field;
|
||||
std::string source_string;
|
||||
source_string.reserve(field.size());
|
||||
|
||||
bool token = false;
|
||||
for (auto token_it = regex_impl::sregex_token_iterator(field.begin(), field.end(), token_regex, {-1, 1}); token_it != tokens_end; ++token_it, token = !token) {
|
||||
if (!token_it->matched) {
|
||||
continue;
|
||||
}
|
||||
|
||||
if (token) {
|
||||
auto it_prop = feature.properties.find(token_it->str());
|
||||
if (it_prop == feature.properties.end()) {
|
||||
// feature does not have the correct property
|
||||
if (debug::labelTextMissingWarning) {
|
||||
fprintf(stderr,
|
||||
"[WARNING] feature doesn't have property '%s' required for labelling\n",
|
||||
token_it->str().c_str());
|
||||
}
|
||||
continue;
|
||||
}
|
||||
source_string += toString(it_prop->second);
|
||||
} else {
|
||||
source_string += token_it->str();
|
||||
}
|
||||
}
|
||||
|
||||
const std::u32string string = ucs4conv.convert(source_string);
|
||||
const std::u32string string = ucs4conv.convert(
|
||||
util::replaceTokens(bucket_desc.geometry.field, feature.properties));
|
||||
|
||||
// Loop through all characters of this text and collect unique codepoints.
|
||||
for (uint32_t chr : string) {
|
||||
|
||||
Reference in New Issue
Block a user