use GT from bags in assets instead of obsolete zip on webpage

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "repo/assets"]
+	path = repo/assets
+	url = https://github.com/OCR-D/assets

Makefile

@@ -38,8 +38,16 @@ test: test/assets deps-test
 	$(PYTHON) -m pytest test $(PYTEST_ARGS)
 
 # prepare test assets
-test/assets:
-	# TODO: instead of this, use bag repos, or add something useful to OCR-D/assets
-	test/prepare_gt.bash $@
+test/assets: repo/assets
+	mkdir -p $@
+	ocrd workspace clone $</data/kant_aufklaerung_1784/data/mets.xml -a $@/kant_aufklaerung_1784
+	bash test/prepare_gt.bash $@
 
-.PHONY: help deps deps-test install test
+repo/assets: always-update
+	git submodule sync $@
+	git submodule update --init $@
+
+clean:
+	$(RM) -r test/assets model_dta_test.h5
+
+.PHONY: help deps deps-test install test clean always-update

test/prepare_gt.bash

@@ -8,7 +8,7 @@ GT_FILES="kant_aufklaerung_1784 loeber_heuschrecken_1693"
 
 trap "rm -fr '$TMP_DIR'" ERR
 
-test ! -e "$1" # target directory must already exist
+test -d "$1" # target directory must already exist
 
 test -d "$CACHE_DIR" || mkdir -p "$CACHE_DIR"
 
@@ -18,54 +18,7 @@ for GT_FILE in $GT_FILES; do
     sed -e '//d;/^[[].*[]]$/d' < "$CACHE_DIR/$GT_FILE" > "$TMP_DIR/${GT_FILE}.txt"
 done
 
-cat <<EOF > "$TMP_DIR/page-extract-imagefilename.xsl"
-<xsl:stylesheet
-    version="1.0"
-    xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
-    xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2013-07-15">
-  <!-- unfortunately, this totally depends on the exact namespace string, i.e. version -->
-  <!-- rid of xml syntax: -->  
-  <xsl:output
-      method="text"
-      standalone="yes"
-      omit-xml-declaration="yes"/>
-  <!-- get imageFilename attribute verbatim: -->  
-  <xsl:template match="pc:PcGts/pc:Page">
-    <xsl:value-of select="@imageFilename" disable-output-escaping="yes"/>
-    <xsl:apply-templates/>
-  </xsl:template>
-  <!-- override implicit rules copying elements and attributes: -->
-  <xsl:template match="text()"/>
-</xsl:stylesheet>
-EOF
-
-for GT_FILE in $GT_FILES; do
-    test -f "$CACHE_DIR/${GT_FILE}.zip" ||
-        wget -P "$CACHE_DIR" http://www.ocr-d.de/sites/all/GTDaten/${GT_FILE}.zip
-    mkdir -p "$TMP_DIR/$GT_FILE"
-    unzip -jod "$TMP_DIR/$GT_FILE/data" "$CACHE_DIR/${GT_FILE}.zip"
-    pushd "$TMP_DIR/$GT_FILE/data"
-    ocrd workspace init .
-    mkdir -p OCR-D-IMG OCR-D-GT-PAGE
-    ZEROS=0000
-    i=0
-    for PAGE_FILE in *.xml; do
-	test "x$PAGE_FILE" = xmets.xml && continue
-        i=$((i+1))
-        ID=${ZEROS:0:$((4-${#i}))}$i
-        IMG_FILE=$(xsltproc "$TMP_DIR/page-extract-imagefilename.xsl" "$PAGE_FILE")
-        test -f "$IMG_FILE"
-        mv "$IMG_FILE" OCR-D-IMG/
-        ocrd workspace add -G OCR-D-IMG -i OCR-D-IMG_$ID -g phys_$ID -m image/tiff "OCR-D-IMG/$IMG_FILE"
-	# workaround for OCR-D/core#176 (still true for ocrd v1.0.0)
-        sed -i -e 's|imageFilename="|imageFilename="OCR-D-IMG/|' "$PAGE_FILE"
-        mv "$PAGE_FILE" OCR-D-GT-PAGE/
-        ocrd workspace add -G OCR-D-GT-PAGE -i OCR-D-GT-PAGE_$ID -g phys_$ID -m application/vnd.prima.page+xml "OCR-D-GT-PAGE/$PAGE_FILE"
-    done
-    popd
-done
-
-mv "$TMP_DIR" "$1" # atomic
+mv "$TMP_DIR"/*.txt "$1"
 
 
 

test/test_wrapper.py

@@ -21,7 +21,7 @@ def setUp(self):
 
     def runTest(self):
         resolver = Resolver()
-        workspace = resolver.workspace_from_url('test/assets/kant_aufklaerung_1784/data/mets.xml',
+        workspace = resolver.workspace_from_url('test/assets/kant_aufklaerung_1784/mets.xml',
                                                 dst_dir=WORKSPACE_DIR, download=True)
         self.assertIsNotNone(workspace)
         #