From 3916915e222e458a0f9e72956c07f7821fd3f950 Mon Sep 17 00:00:00 2001
From: Bertrand Benjamin <benjamin.bertrand@opytex.org>
Date: Sat, 2 Mar 2024 09:45:38 +0100
Subject: [PATCH] Feat: add import from Nextcloud

---
 Makefile | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

diff --git a/Makefile b/Makefile
index e20935b..9e86597 100644
--- a/Makefile
+++ b/Makefile
@@ -1,3 +1,40 @@
+DATA_BASE=./datas
+
+PDF_BASE=$(DATA_BASE)/pdfs
+PDF_YEARS=$(wildcard $(PDF_BASE)/*)
+
+RAW_BASE=$(DATA_BASE)/raw
+RAW_CRG=$(RAW_BASE)/CRG
+RAW_CRG_YEARS=$(subst $(PDF_BASE), $(RAW_CRG), $(PDF_YEARS))
+
+
+$(RAW_CRG)/%/: $(wildcard $(PDF_BASE)/%/*)
+	echo $(wildcard $(PDF_BASE)/$*/*)
+	@echo ----
+	ls $(PDF_BASE)/$*/
+	@echo ----
+	echo $*
+	@echo ----
+	echo $^
+	@echo ----
+	echo $?
+
+#./datas/raw/CRG/%: 
+#pdf-oralia extract all --src $$year --dest $$(subst $$PDF_BASE, $$RAW_CRG, $$year)
+# $(RAW_CRG_YEARS): $(PDF_PATHS)
+# 	for year in $(PDF_PATHS); do \
+# 		echo $$year; \
+# 		echo $$(subst $$PDF_BASE, $$RAW_CRG, $$year); \
+# 		echo "----"; \
+# 	done;
+
+extract_pdfs:
+	for year in 2021 2022 2023 2024; do \
+		mkdir -p $(RAW_CRG)/$$year/extracted;\
+		pdf-oralia extract all --src $(PDF_BASE)/$$year/ --dest $(RAW_CRG)/$$year/extracted; \
+		pdf-oralia join --src $(RAW_CRG)/$$year/extracted/ --dest $(RAW_CRG)/$$year/; \
+	done
+
 clean_raw:
 	rm -rf ./PLESNA Compta SYSTEM/raw/**/*.csv
 
@@ -8,3 +45,5 @@ clean_built:
 
 clean_all: clean_built clean_raw
 
+import_nextcloud:
+	rsync -a ~/Nextcloud/PLESNA\ Compta\ SYSTEM/Histoire/ ./datas/Histoire