Merge pull request #82 from bertsky/use-bashlib-input-files

simplify file iterator by delegating to bashlib input-files
OCR-D · Dec 8, 2021 · 9117389 · 9117389
2 parents 9487dfd + 85a9819
commit 9117389
Showing 1 changed file with 9 additions and 54 deletions.
diff --git a/ocrd-olena-binarize b/ocrd-olena-binarize
@@ -356,7 +356,7 @@ function main {
     # shellcheck source=../core/ocrd/bashlib/lib.bash
     source $(ocrd bashlib filename)
     ocrd__wrap "$SHAREDIR/ocrd-tool.json" "ocrd-olena-binarize" "$@"
-    ocrd__minversion 2.13.0
+    ocrd__minversion 2.29.0
 
     scribo_options=(--enable-negate-output)
     case ${params[impl]} in
@@ -386,66 +386,21 @@ function main {
     esac
 
     cd "${ocrd__argv[working_dir]}"
-    page_id=${ocrd__argv[page_id]:-}
-    in_file_grp=${ocrd__argv[input_file_grp]}
     out_file_grp=${ocrd__argv[output_file_grp]}
-    mkdir -p $out_file_grp
 
-    local IFS=$'\n'
-    files=($(ocrd workspace find     \
-                  ${page_id:+-g} ${page_id:-} \
-                  -G $in_file_grp    \
-                  -k url             \
-                  -k ID              \
-                  -k mimetype        \
-                  -k pageId          \
-                  --download))
-    page_pages=($(ocrd workspace find \
-                       -G $in_file_grp \
-                       -m $MIMETYPE_PAGE \
-                       -k pageId))
-    multi_pages=($(ocrd workspace find \
-                        -G $in_file_grp \
-                        -m //image/.* \
-                        -k pageId | sort | uniq -d))
-    declare -A is_page is_multi
-    for page in "${page_pages[@]}"; do
-        is_page[$page]=1
-    done
-    for page in "${multi_pages[@]}"; do
-        is_multi[$page]=1
-    done
-    local IFS=$' \t\n'
-    local n=0 zeros=0000
-    for csv in "${files[@]}"; do
-        let n+=1
-        # Parse comma separated fields
-        local IFS=$'\t'
-        local fields=($csv)
-        local IFS=$' \t\n'
-
-        local in_fpath="${fields[0]}"
-        local in_id="${fields[1]}"
-        local in_mimetype="${fields[2]}"
-        local in_pageId="${fields[3]:-}"
+    for ((n=0; n<${#ocrd__files[*]}; n++)); do
+        local in_fpath="$(ocrd__input_file $n url)"
+        local in_id="$(ocrd__input_file $n ID)"
+        local in_mimetype="$(ocrd__input_file $n mimetype)"
+        local in_pageId="$(ocrd__input_file $n pageId)"
+        local out_id="$(ocrd__input_file $n outputFileId)"
+        local out_fpath="$out_file_grp/${out_id}.xml"
 
         if ! test -f "${in_fpath#file://}"; then
            ocrd log error "input file ID=${in_id} (pageId=${in_pageId} MIME=${in_mimetype}) is not on disk"
            continue
         fi
-        # fileGrps may contain PAGE and (derived) images
-        # so if this pageId has a PAGE file, ignore all others
-        # and otherwise if it has multiple images, raise an error
-        if ((${is_page[${in_pageId:--}]:-0})); then
-            test x$in_mimetype != x$MIMETYPE_PAGE && continue
-        elif ((${is_multi[${in_pageId:--}]:-0})); then
-            ocrd__raise "No PAGE-XML for page '$in_pageId' in fileGrp '$in_file_grp' but multiple images."
-        fi
-        local out_id="${in_id//$in_file_grp/$out_file_grp}"
-        if [ "x$out_id" = "x$in_id" ]; then
-            out_id=${out_file_grp}_${zeros:0:$((4-${#n}))}$n
-        fi
-        local out_fpath="$out_file_grp/${out_id}.xml"
+        mkdir -p $out_file_grp
 
         if [ "x${in_mimetype}" = x${MIMETYPE_PAGE} ]; then
             ocrd log info "processing PAGE-XML input file $in_id ($in_pageId)"