Merge pull request #4 from someth2say/ruby

Added perl, ruby and docker variants

Author: someth2say

.pre-commit-hooks.yaml

@@ -1,11 +1,40 @@
 ---
-
-- id: xmlformat
-  name: XML Formatter
+- id: xmlformat_docker_perl
+  name: XML Formatter (perl bash docker)
   description: >
     Formats XML files according GLS formating rules
-  entry: quay.io/someth2say/xmlformat:0.1 
+  entry: quay.io/someth2say/xmlformat:0.2 
   types: [xml]
   language: docker_image
   require_serial: true
-  
+
+- id: xmlformat_ruby
+  name: XML Formatter (ruby)
+  description: >
+    Formats XML files according GLS formating rules
+  entry: xmlformat.rb
+  #args: ["-i","--config-file","xmlformat.conf"]
+  args: ["-i"]
+  language: ruby
+  require_serial: true
+
+- id: xmlformat_bash_ruby
+  name: XML Formatter (ruby bash)
+  description: >
+    Formats XML files according GLS formating rules
+  entry: bin/xmlformat.sh
+  args: ["-l","rb"]
+  types: [xml]
+  language: script
+  require_serial: true
+
+- id: xmlformat_bash_perl
+  name: XML Formatter (perl bash)
+  description: >
+    Formats XML files according GLS formating rules
+  entry: bin/xmlformat.sh
+  args: ["-l","pl"]
+  types: [xml]
+  language: script
+  require_serial: true
+

Dockerfile

@@ -1,40 +1,45 @@
-FROM registry.access.redhat.com/ubi8/ubi-minimal as perl
+FROM registry.access.redhat.com/ubi8/ubi-minimal as ubi8-perl-ruby
 MAINTAINER Jordi Sola <[email protected]>
 
 LABEL maintainer="Jordi Sola <[email protected]>" \
-      name="XMLFormat" \
-      vcs-ref="TBD" \
+      name="ubi8-perl-ruby" \
       version="0.1-SNAPSHOT" \
-      build-date="TBD" \
+      description="UBI8 based image containing perl and ruby interpreters" \
+      summary="UBI8 based image containing perl and ruby interpreters" \
+      io.k8s.description="UBI8 based image containing perl and ruby interpreters" \
+      io.k8s.display-name="ubi8-perl-ruby" \
+      io.openshift.tags="perl ruby" 
+
+# Install perl and ruby interpreter
+RUN microdnf install -y perl ruby
+###
+
+FROM ubi8-perl-ruby
+LABEL name="XMLFormat" \
+      version="0.2-SNAPSHOT" \
       description="Container image wrapping XML formater by Kitebird (http://www.kitebird.com/software/xmlformat/)" \
       summary="Container image wrapping XML formater by Kitebird (http://www.kitebird.com/software/xmlformat/)" \
       io.k8s.description="Container image wrapping XML formater by Kitebird (http://www.kitebird.com/software/xmlformat/)" \
       io.k8s.display-name="XMLFormat" \
       io.openshift.tags="xml format perl" 
 
-# Install perl interpreter
-RUN microdnf install -y perl
-
-###
-
-FROM perl
-
 # Docker arguments to facilitate image extension
-ARG SCRIPTS_DIR="/xmlformat" 
-ARG EXTRA_ARGS="-i"
-ARG CFG_FILE="${SCRIPTS_DIR}/xmlformat.cfg" 
+ARG SCRIPTS_DIR="/bin" 
+ARG EXTRA_ARGS=""
+ARG CFG_FILE="${SCRIPTS_DIR}/xmlformat.conf" 
+ARG SCRIPT_LANG="pl"
 
 # Environment to be used by the xmlformat.sh script
 ENV PATH=${SCRIPTS_DIR}:$PATH \
-    XMLFORMAT_CFG_FILE=${CFG_FILE} \
-    XMLFORMAT_SCRIPTS_DIR=${SCRIPTS_DIR} \
+    XMLFORMAT_LANG=${SCRIPT_LANG} \
+    XMLFORMAT_CONF=${CFG_FILE} \
     XMLFORMAT_EXTRA_ARGS=${EXTRA_ARGS}
 
 # Copy the root scripts to the scripts folder
-ADD scripts ${SCRIPTS_DIR}
+ADD bin ${SCRIPTS_DIR}
 
 # Make scripts executable by root group, just in case
 RUN chgrp -R 0 ${SCRIPTS_DIR} && \
     chmod -R g=u ${SCRIPTS_DIR}
 
-ENTRYPOINT [ "xmlformat.sh" ]
+ENTRYPOINT [ "xmlformat.sh" ]

scripts/xmlformat.cfg renamed to bin/xmlformat.conf

@@ -1095,8 +1095,12 @@ sub tree_canonize2
           if (!defined ($next_child) && $par_opts->{format} eq "block")
             || $self->non_normalized_node ($next_child);
 
+        # After whitespace normalization, split the text into sentences
+        # This is done by a simple regexp replacement, 
+        # looking for period and interrogation symbols and adding a break after them.
+        # Note also that the split should also be indented correctly after the break.
         my $indent_str = (" " x ( $indent+$par_opts->{"subindent"}));
-        $child->{content} =~ s/([.?]) \s*/$1 \n$indent_str/g;
+        $child->{content} =~ s/([.?]) \s*/$1\n$indent_str/g;
 
         # If resulting text is empty, discard the node.
         next if $child->{content} =~ /^$/;
@@ -1655,7 +1659,7 @@ package main;
 # are used. (These are set up in new().)
 
 my $env_conf_file = $ENV{XMLFORMAT_CONF};
-my $def_conf_file = "./xmlformat.conf";
+my $def_conf_file = "./xmlformat.conf"; 
 
 # If no config file was named, but XMLFORMAT_CONF is set, use its value
 # as the config file name.