From ffca1378172da2061d56101f8b0a3461d64a13e1 Mon Sep 17 00:00:00 2001
From: Franck Dary <franck.dary@lis-lab.fr>
Date: Mon, 8 Feb 2021 14:02:51 +0100
Subject: [PATCH] Fixed filing of columns rawRange
---
reading_machine/src/Action.cpp | 15 +++++++++------
reading_machine/src/Transition.cpp | 4 ++++
2 files changed, 13 insertions(+), 6 deletions(-)
diff --git a/reading_machine/src/Action.cpp b/reading_machine/src/Action.cpp
index c158808..75ca4fa 100644
--- a/reading_machine/src/Action.cpp
+++ b/reading_machine/src/Action.cpp
@@ -80,9 +80,6 @@ Action Action::consumeCharacterIndex(const util::utf8string & consumed)
{
auto apply = [consumed](Config & config, Action &)
{
- config.getLastNotEmptyHyp(Config::rawRangeStartColName, config.getWordIndex()) = fmt::format("{}", config.getCharacterIndex());
- config.getLastNotEmptyHyp(Config::rawRangeEndColName, config.getWordIndex()) = fmt::format("{}", config.getCharacterIndex() + consumed.size());
-
config.moveCharacterIndex(consumed.size());
};
@@ -586,8 +583,7 @@ Action Action::addCharsToCol(const std::string & col, int n, Config::Object obje
{
if (util::isEmpty(config.getAsFeature(Config::rawRangeStartColName, index)))
config.getLastNotEmptyHyp(Config::rawRangeStartColName, index) = fmt::format("{}", config.getCharacterIndex());
- if (util::isEmpty(config.getAsFeature(Config::rawRangeEndColName, index)))
- config.getLastNotEmptyHyp(Config::rawRangeEndColName, index) = fmt::format("{}", config.getCharacterIndex());
+ config.getLastNotEmptyHyp(Config::rawRangeEndColName, index) = fmt::format("{}", config.getCharacterIndex());
int curEndValue = std::stoi(config.getAsFeature(Config::rawRangeEndColName, index));
config.getLastNotEmptyHyp(Config::rawRangeEndColName, index) = fmt::format("{}", curEndValue+n);
}
@@ -753,8 +749,15 @@ Action Action::updateIds(int bufferIndex)
if (config.has(0,firstIndexOfSentence,0))
{
std::string textMetadata = "# text = ";
- for (auto i = std::stoi(config.getAsFeature(Config::rawRangeStartColName, firstIndexOfSentence)); i < std::stoi(config.getAsFeature(Config::rawRangeEndColName, lineIndex)); i++)
+ int firstIndex = 0;
+ int lastIndex = 0;
+ try {firstIndex = std::stoi(config.getAsFeature(Config::rawRangeStartColName, firstIndexOfSentence));}
+ catch (std::exception & e) {util::myThrow(fmt::format("{} : '{}'", e.what(), config.getAsFeature(Config::rawRangeStartColName, firstIndexOfSentence)));}
+ try {lastIndex = std::stoi(config.getAsFeature(Config::rawRangeEndColName, lineIndex));}
+ catch (std::exception & e) {util::myThrow(fmt::format("{} : '{}'", e.what(), config.getAsFeature(Config::rawRangeEndColName, lineIndex)));}
+ for (auto i = firstIndex; i < lastIndex; i++)
textMetadata = fmt::format("{}{}", textMetadata, config.getLetter(i));
+
config.getLastNotEmptyHyp(Config::commentsColName, firstIndexOfSentence) = fmt::format("{}\n# sent_id = {}", textMetadata, config.getAsFeature(Config::sentIdColName, firstIndexOfSentence));
}
};
diff --git a/reading_machine/src/Transition.cpp b/reading_machine/src/Transition.cpp
index 1b8d309..db34591 100644
--- a/reading_machine/src/Transition.cpp
+++ b/reading_machine/src/Transition.cpp
@@ -317,7 +317,11 @@ void Transition::initSplitWord(std::vector<std::string> words)
sequence.emplace_back(Action::addCharsToCol("FORM", consumedWord.size(), Config::Object::Buffer, 0));
sequence.emplace_back(Action::consumeCharacterIndex(consumedWord));
for (unsigned int i = 1; i < words.size(); i++)
+ {
sequence.emplace_back(Action::addHypothesisRelativeRelaxed("FORM", Config::Object::Buffer, i, words[i]));
+ sequence.emplace_back(Action::transformSuffix(Config::rawRangeStartColName, Config::Object::Buffer, 0, Config::rawRangeStartColName, Config::Object::Buffer, i, util::utf8string(), util::utf8string()));
+ sequence.emplace_back(Action::transformSuffix(Config::rawRangeEndColName, Config::Object::Buffer, 0, Config::rawRangeEndColName, Config::Object::Buffer, i, util::utf8string(), util::utf8string()));
+ }
sequence.emplace_back(Action::setMultiwordIds(words.size()-1));
costDynamic = [words](const Config & config)
--
GitLab