From eee529601791c604a7a4b73d8d6005d44c8cf87a Mon Sep 17 00:00:00 2001
From: Alice BRENON <alice.brenon@ens-lyon.fr>
Date: Mon, 4 Sep 2023 20:06:31 +0200
Subject: [PATCH] Add source extractor for LGE using the newest syntax for
ProcessingLGE and soprano
---
scripts/LGE/extract-from-source.sh | 20 ++++++++++++++++++++
scripts/extract-from-source.sh | 4 ++--
2 files changed, 22 insertions(+), 2 deletions(-)
create mode 100755 scripts/LGE/extract-from-source.sh
diff --git a/scripts/LGE/extract-from-source.sh b/scripts/LGE/extract-from-source.sh
new file mode 100755
index 0000000..e0af410
--- /dev/null
+++ b/scripts/LGE/extract-from-source.sh
@@ -0,0 +1,20 @@
+#!/bin/sh
+
+source ${0%/*}/../lib.sh
+
+if [ "$#" != 2 ]
+then
+ die "${PROG_NAME} SOURCE_DIRECTORY TARGET_DIRECTORY"
+else
+ SOURCE="${1}"
+ TARGET="${2}"
+ [ -d "${SOURCE}" ] || die "SOURCE_DIRECTORY must be a directory (containing one ALTO/LGE/T<T> directory with one ALTO-XML file per page)"
+ [ -d "${TARGET}" ] || die "TARGET_DIRECTORY must be a directory (where output will be generated)"
+fi
+
+cd "${SOURCE}"
+for T in {1..31}
+do
+ LGEprepareVolume.sh ${T} "${TARGET}" ALTO
+ LGEencode.sh "${TARGET}/ALTO/LGE/T${T}" -k Text --metadata --text-root "${TARGET}/Text" 2> /tmp/LGE.log
+done
diff --git a/scripts/extract-from-source.sh b/scripts/extract-from-source.sh
index 0275b82..ba74ce0 100755
--- a/scripts/extract-from-source.sh
+++ b/scripts/extract-from-source.sh
@@ -10,11 +10,11 @@ then
else
SOURCE="${1}"
TARGET="${2}"
- [ -d "${SOURCE}" ] || die "SOURCE_DIRECTORY must be a directory (containing 1 .tei file per tome)"
+ [ -d "${SOURCE}" ] || die "SOURCE_DIRECTORY must be a directory (containing a folder for EDdA and one for LGE with the structure found in Source.squashfs)"
[ -d "${TARGET}" ] || die "TARGET_DIRECTORY must be a directory (where output will be generated)"
fi
FILES_TSV="${TARGET}/files.tsv"
printf "book tome rank headWord name page\n" > "${FILES_TSV}"
${BASE_DIR}/EDdA/extract-from-source.sh "${SOURCE}/EDdA/ARTFL" ${TARGET} >> "${FILES_TSV}"
-#${BASE_DIR}/LGE/extract-from-source.sh "${SOURCE}/LGE/BnF" ${TARGET} >> "${FILES_TSV}"
+${BASE_DIR}/LGE/extract-from-source.sh "${SOURCE}/LGE/BnF" ${TARGET} >> "${FILES_TSV}"
--
GitLab