Merge pull request #134 from kba/page-to-alto-python

Page to ALTO Python

script/transform/page__alto

@@ -2,7 +2,6 @@
 
 SCRIPTDIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 VENDORDIR="$(cd $SCRIPTDIR/../../vendor/; pwd)"
-JAR="$VENDORDIR/JPageConverter/PageConverter.jar"
 INFILE="$1"
 OUTFILE="$2"
 ARGUMENT="$3"
@@ -16,17 +15,18 @@ if [[ "$2" = "-" ]]; then
     OUTFILE="$(mktemp)"
 fi
 
-java -jar "$JAR" -neg-coords toZero -source-xml "$INFILE" -target-xml "$OUTFILE" -convert-to ALTO 2>&1
+page-to-alto $ARGUMENT "$INFILE" > "$OUTFILE"; retval="$?"
 
 if [[ "$1" = "-" ]]; then
     rm "$INFILE"
 fi
 
+if (( retval > 0 ));then
+    rm "$OUTFILE"
+    exit $retval
+fi
+
 if [[ "$2" = "-" ]]; then
-    if [[ -z "$ARGUMENT" ]]; then
-        cat "$OUTFILE"
-    else
-        java -cp "$VENDORDIR/saxon9he.jar" net.sf.saxon.Query -s:"$OUTFILE" -qs:/ "$ARGUMENT"
-    fi
+    cat "$OUTFILE"
     rm "$OUTFILE"
 fi

script/transform/page__alto_legacy

@@ -0,0 +1,32 @@
+#!/bin/bash
+
+SCRIPTDIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+VENDORDIR="$(cd $SCRIPTDIR/../../vendor/; pwd)"
+JAR="$VENDORDIR/JPageConverter/PageConverter.jar"
+INFILE="$1"
+OUTFILE="$2"
+ARGUMENT="$3"
+
+if [[ "$1" = "-" ]]; then
+    INFILE="$(mktemp)"
+    cat >"$INFILE"
+fi
+
+if [[ "$2" = "-" ]]; then
+    OUTFILE="$(mktemp)"
+fi
+
+java -jar "$JAR" -neg-coords toZero -source-xml "$INFILE" -target-xml "$OUTFILE" -convert-to ALTO 2>&1
+
+if [[ "$1" = "-" ]]; then
+    rm "$INFILE"
+fi
+
+if [[ "$2" = "-" ]]; then
+    if [[ -z "$ARGUMENT" ]]; then
+        cat "$OUTFILE"
+    else
+        java -cp "$VENDORDIR/saxon9he.jar" net.sf.saxon.Query -s:"$OUTFILE" -qs:/ "$ARGUMENT"
+    fi
+    rm "$OUTFILE"
+fi

vendor/Makefile

@@ -4,6 +4,7 @@ RM = rm -rfv
 UNZIP = unzip -o
 WGET = wget --progress=bar:force --no-verbose
 GIT_CLONE = git clone --depth 1
+PIP = pip3
 
 SAXON_HE_VERSION_MAJOR = 9
 SAXON_HE_VERSION_MINOR = 9
@@ -45,6 +46,9 @@ ALTO2PAGE_ZIP = JPageConverter.zip
 ALTO2PAGE_URL = https://github.com/PRImA-Research-Lab/prima-page-converter/releases/download/$(ALTO2PAGE_VERSION)/JPageConverter_$(ALTO2PAGE_VERSION).zip
 ALTO2PAGE_DIR = JPageConverter
 
+PAGE_TO_ALTO_REPO = page-to-alto
+PAGE_TO_ALTO_URL = https://github.com/kba/page-to-alto
+
 # {{{
 # SAXON_BROWSER_VERSION = 1.1
 # SAXON_BROWSER_ZIP = Saxon-CE_$(SAXON_BROWSER_VERSION).zip
@@ -69,7 +73,8 @@ all:\
 	$(PAGE2HOCR_REPO)\
 	$(HOCR_SPEC_REPO)\
 	$(SAXON_HE_JAR) \
-	$(ALTO2PAGE_DIR)
+	$(ALTO2PAGE_DIR) \
+	$(PAGE_TO_ALTO_REPO)
 
 clean:
 	$(RM) $(ALTO_SCHEMA_REPO)
@@ -133,3 +138,7 @@ $(ALTO2PAGE_ZIP):
 $(ALTO2PAGE_DIR): $(ALTO2PAGE_ZIP)
 	$(UNZIP) "$<"
 	mv "JPageConverter $(ALTO2PAGE_VERSION_MAJOR_MINOR)" "$@"
+
+$(PAGE_TO_ALTO_REPO):
+	$(GIT_CLONE) "$(PAGE_TO_ALTO_URL)" "$@"
+	cd "$@"; $(PIP) install .