diff --git a/common/include/util.hpp b/common/include/util.hpp
index 17d4ee8add73219e0a4f4798aa89a885f110c8dd..1363383ce6708530958c5815540920ae1db195c6 100644
--- a/common/include/util.hpp
+++ b/common/include/util.hpp
@@ -40,6 +40,8 @@ utf8string splitAsUtf8(std::string_view s);
std::string int2HumanStr(int number);
+int printedLength(std::string_view s);
+
template <typename T>
bool isEmpty(const std::vector<T> & s)
{
diff --git a/common/src/util.cpp b/common/src/util.cpp
index 288b9a96fd6e14f436f2cc974236634bda701964..bee8b9078dd6740d8edd8a1f6730c438f29576d4 100644
--- a/common/src/util.cpp
+++ b/common/src/util.cpp
@@ -14,6 +14,11 @@
namespace util
{
+int printedLength(std::string_view s)
+{
+ return splitAsUtf8(s).size();
+}
+
std::string_view getFilenameFromPath(std::string_view s)
{
int indexOfSlash = s.find_last_of('/');
diff --git a/dev/src/dev.cpp b/dev/src/dev.cpp
index 3a07a91b02d98cc4244f9b5867b1f064ae894d9e..4604fbafa1d7c8ccdf4ef73664aceda6dfda507b 100644
--- a/dev/src/dev.cpp
+++ b/dev/src/dev.cpp
@@ -17,14 +17,8 @@ int main(int argc, char * argv[])
configs.emplace_back(config);
configs[0].wordIndex = 2000;
-
- configs[0].update();
- configs[0].wordIndex = 0;
-
- configs[0].update();
- configs[0].update();
configs[0].update();
- configs[0].print(stdout);
+ configs[0].printForDebug(stdout);
fmt::print(stderr, "ok\n");
std::scanf("%*c");
diff --git a/reading_machine/include/BaseConfig.hpp b/reading_machine/include/BaseConfig.hpp
index c38f926e2c5323d8e7479cf427290a5ed4b8dfc5..e7726f2348c17e7ba74f2aa38643662fb9e8a159 100644
--- a/reading_machine/include/BaseConfig.hpp
+++ b/reading_machine/include/BaseConfig.hpp
@@ -30,8 +30,7 @@ class BaseConfig : public Config
std::vector<std::string> colIndex2Name;
std::unordered_map<std::string, int> colName2Index;
- std::string rawInput;
- util::utf8string rawInputUtf8;
+ Utf8String rawInputUtf8;
private :
@@ -39,15 +38,16 @@ class BaseConfig : public Config
void readRawInput(std::string_view rawFilename);
void readTSVInput(std::string_view tsvFilename);
+ public :
+
+ BaseConfig(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename);
+
std::size_t getNbColumns() const override;
std::size_t getFirstLineIndex() const override;
std::size_t getColIndex(const std::string & colName) const override;
+ bool hasColIndex(const std::string & colName) const override;
const std::string & getColName(int colIndex) const override;
- public :
-
- BaseConfig(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename);
-
friend SubConfig;
};
diff --git a/reading_machine/include/Config.hpp b/reading_machine/include/Config.hpp
index dae3e6cf43e61020f426b3c3ac21bbc7910070d1..dee137d7aef61d1416f85cc24afe35577b0bf127 100644
--- a/reading_machine/include/Config.hpp
+++ b/reading_machine/include/Config.hpp
@@ -17,17 +17,20 @@ class Config
static constexpr const char * EOSSymbol0 = "0";
static constexpr int nbHypothesesMax = 1;
- private :
+ public :
using String = boost::flyweight<std::string>;
using Utf8String = boost::flyweight<util::utf8string>;
using ValueIterator = std::vector<String>::iterator;
using ConstValueIterator = std::vector<String>::const_iterator;
+ private :
+
std::vector<String> lines;
public :
+ const Utf8String & rawInput;
std::size_t wordIndex{0};
std::size_t characterIndex{0};
String state{"NONE"};
@@ -35,18 +38,22 @@ class Config
protected :
+ Config(const Utf8String & rawInput);
+
+ public :
+
virtual std::size_t getNbColumns() const = 0;
virtual std::size_t getColIndex(const std::string & colName) const = 0;
+ virtual bool hasColIndex(const std::string & colName) const = 0;
virtual std::size_t getFirstLineIndex() const = 0;
virtual const std::string & getColName(int colIndex) const = 0;
- protected :
-
std::size_t getIndexOfLine(int lineIndex) const;
std::size_t getIndexOfCol(int colIndex) const;
std::size_t getNbLines() const;
void addLines(unsigned int nbLines);
void resizeLines(unsigned int nbLines);
+ bool has(int colIndex, int lineIndex, int hypothesisIndex) const;
String & get(int colIndex, int lineIndex, int hypothesisIndex);
const String & getConst(int colIndex, int lineIndex, int hypothesisIndex) const;
String & getLastNotEmpty(int colIndex, int lineIndex);
@@ -58,6 +65,8 @@ class Config
virtual ~Config() {}
void print(FILE * dest) const;
+ void printForDebug(FILE * dest) const;
+ bool has(const std::string & colName, int lineIndex, int hypothesisIndex) const;
String & get(const std::string & colName, int lineIndex, int hypothesisIndex);
const String & getConst(const std::string & colName, int lineIndex, int hypothesisIndex) const;
String & getLastNotEmpty(const std::string & colName, int lineIndex);
diff --git a/reading_machine/include/SubConfig.hpp b/reading_machine/include/SubConfig.hpp
index 60faa69a2e296df860a9b545a25998af7e677bd9..e8c8efe89118eeaaee9341ed6794accba4b3b71a 100644
--- a/reading_machine/include/SubConfig.hpp
+++ b/reading_machine/include/SubConfig.hpp
@@ -15,17 +15,15 @@ class SubConfig : public Config
const BaseConfig & model;
std::size_t firstLineIndex{0};
- private :
+ public :
+ SubConfig(BaseConfig & model);
+ bool update();
std::size_t getNbColumns() const override;
std::size_t getColIndex(const std::string & colName) const override;
+ bool hasColIndex(const std::string & colName) const override;
const std::string & getColName(int colIndex) const override;
std::size_t getFirstLineIndex() const override;
-
- public :
-
- SubConfig(BaseConfig & model);
- bool update();
};
#endif
diff --git a/reading_machine/src/BaseConfig.cpp b/reading_machine/src/BaseConfig.cpp
index 0a33c514f34f79faf0e983dabfed8499edca6854..6853e4eea2deae0df10dde3595138ae41873e4c6 100644
--- a/reading_machine/src/BaseConfig.cpp
+++ b/reading_machine/src/BaseConfig.cpp
@@ -41,12 +41,14 @@ void BaseConfig::readRawInput(std::string_view rawFilename)
if (not file)
util::myThrow(fmt::format("Cannot open file '{}'", rawFilename));
+ std::string rawInputTemp;
+
while (not std::feof(file))
- rawInput.push_back(std::fgetc(file));
+ rawInputTemp.push_back(std::fgetc(file));
std::fclose(file);
- rawInputUtf8 = util::splitAsUtf8(rawInput);
+ rawInputUtf8 = util::splitAsUtf8(rawInputTemp);
}
void BaseConfig::readTSVInput(std::string_view tsvFilename)
@@ -102,7 +104,7 @@ void BaseConfig::readTSVInput(std::string_view tsvFilename)
std::fclose(file);
}
-BaseConfig::BaseConfig(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename)
+BaseConfig::BaseConfig(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename) : Config(rawInputUtf8)
{
if (tsvFilename.empty() and rawFilename.empty())
util::myThrow("tsvFilename and rawFilenames can't be both empty");
@@ -128,6 +130,11 @@ std::size_t BaseConfig::getColIndex(const std::string & colName) const
return colName2Index.at(colName);
}
+bool BaseConfig::hasColIndex(const std::string & colName) const
+{
+ return colName2Index.count(colName);
+}
+
const std::string & BaseConfig::getColName(int colIndex) const
{
return colIndex2Name[colIndex];
diff --git a/reading_machine/src/Config.cpp b/reading_machine/src/Config.cpp
index a1b88982fefce7a2d1899d2b879fc3a58e3170d7..297b2864cbc9484fb9acc7f66a3951a3bbfa671e 100644
--- a/reading_machine/src/Config.cpp
+++ b/reading_machine/src/Config.cpp
@@ -2,6 +2,10 @@
#include "Config.hpp"
#include "util.hpp"
+Config::Config(const Utf8String & rawInput) : rawInput(rawInput)
+{
+}
+
std::size_t Config::getIndexOfLine(int lineIndex) const
{
return lineIndex * getNbColumns() * (nbHypothesesMax+1);
@@ -22,6 +26,16 @@ void Config::resizeLines(unsigned int nbLines)
lines.resize(nbLines*getNbColumns()*(nbHypothesesMax+1));
}
+bool Config::has(int colIndex, int lineIndex, int hypothesisIndex) const
+{
+ return colIndex >= 0 && colIndex < (int)getNbColumns() && lineIndex >= (int)getFirstLineIndex() && lineIndex < (int)getFirstLineIndex() + (int)getNbLines() && hypothesisIndex >= 0 && hypothesisIndex < nbHypothesesMax+1;
+}
+
+bool Config::has(const std::string & colName, int lineIndex, int hypothesisIndex) const
+{
+ return hasColIndex(colName) && has(getColIndex(colName), lineIndex, hypothesisIndex);
+}
+
Config::String & Config::get(const std::string & colName, int lineIndex, int hypothesisIndex)
{
return get(getColIndex(colName), lineIndex, hypothesisIndex);
@@ -58,6 +72,43 @@ void Config::print(FILE * dest) const
}
}
+void Config::printForDebug(FILE * dest) const
+{
+ static constexpr int windowSize = 5;
+ int firstLineToPrint = wordIndex;
+ int lastLineToPrint = wordIndex;
+ while (wordIndex-firstLineToPrint < windowSize and has(0, firstLineToPrint, 0))
+ --firstLineToPrint;
+ while (lastLineToPrint - wordIndex < windowSize and has(0, lastLineToPrint, 0))
+ ++lastLineToPrint;
+
+ std::vector<std::vector<std::string>> toPrint;
+
+ for (int line = firstLineToPrint; line <= lastLineToPrint; line++)
+ {
+ toPrint.emplace_back();
+ toPrint.back().emplace_back(line == (int)wordIndex ? "=>" : "");
+ for (unsigned int i = 0; i < getNbColumns(); i++)
+ toPrint.back().emplace_back(getLastNotEmptyConst(i, line));
+ }
+
+ std::vector<std::size_t> colLength(toPrint[0].size(), 0);
+ for (auto & line : toPrint)
+ for (unsigned int col = 0; col < line.size()-1; col++)
+ colLength[col] = std::max((int)colLength[col], util::printedLength(line[col]));
+
+ for (auto & line : toPrint)
+ {
+ for (unsigned int col = 0; col < line.size()-1; col++)
+ if (col == 0)
+ fmt::print(dest, "{:>{}}", line[col], colLength[col]);
+ else
+ fmt::print(dest, "{:<{}}{}", line[col], colLength[col], col == line.size()-2 ? "\n" : "\t");
+ if (line.back() == EOSSymbol1)
+ fmt::print(dest, "\n");
+ }
+}
+
Config::String & Config::getLastNotEmpty(int colIndex, int lineIndex)
{
int baseIndex = getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex);
diff --git a/reading_machine/src/SubConfig.cpp b/reading_machine/src/SubConfig.cpp
index eeb284aec8cd390f681d000dbc2428e45625fc2a..118e62bb737f08a4fd9da2d61245add7fdcfe0fc 100644
--- a/reading_machine/src/SubConfig.cpp
+++ b/reading_machine/src/SubConfig.cpp
@@ -1,6 +1,6 @@
#include "SubConfig.hpp"
-SubConfig::SubConfig(BaseConfig & model) : model(model)
+SubConfig::SubConfig(BaseConfig & model) : Config(model.rawInput), model(model)
{
wordIndex = model.wordIndex;
characterIndex = model.characterIndex;
@@ -71,6 +71,11 @@ std::size_t SubConfig::getColIndex(const std::string & colName) const
return model.getColIndex(colName);
}
+bool SubConfig::hasColIndex(const std::string & colName) const
+{
+ return model.getColIndex(colName);
+}
+
const std::string & SubConfig::getColName(int colIndex) const
{
return model.getColName(colIndex);