Push v0.03

Author: mrueda

Changes

@@ -1,10 +1,12 @@
 Revision history for Perl distribution OMOP-CSV-Validator
 
-0.03 2025-04-02T00:00:00Z (Manuel Rueda <mrueda@cpan.org>)
+0.03 2025-04-03T00:00:00Z (Manuel Rueda <mrueda@cpan.org>)
 
  - Added blank_is_undef => 1 to TEXT::CSV_XS
  - Fixed issue with non-required properties with NULL values
  - Added a custom "_coerce" property on numeric fields to enable coercion into number
+ - Added Term::ANSIColor + emojis to stdout
+ - Added optional <--no-colors> parameter to turn off Term::ANSIColor
 
 0.02 2025-04-02T00:00:00Z (Manuel Rueda <mrueda@cpan.org>)
 

README.md

@@ -5,6 +5,7 @@ The OMOP CSV Validator is a **CLI tool** (and module) that **validates OMOP CDM
 ## Features
 
 - **DDL Parsing:** Automatically converts PostgreSQL OMOP CDM DDL into JSON schemas.
+- **Version Independent** Works with any DDL (e.g., 5.3, 5.4).
 - **CSV Validation:** Validates CSV files using JSON::Validator.
 - **Numeric Coercion:** Coerces CSV numeric fields (including handling of null values).
 - **Modular Design:** Separate CLI and module for easy testing and integration.

bin/omop-csv-validator

@@ -16,9 +16,11 @@ use Path::Tiny;
 use OMOP::CSV::Validator;
 use JSON::XS;
 use Pod::Usage;
-use Data::Dumper;
+#use Data::Dumper;
 use Term::ANSIColor;    # For colored output
+use File::Basename;
 binmode STDOUT, ':encoding(UTF-8)';
+binmode STDERR, ':encoding(UTF-8)';
 
 # Module version
 my $VERSION = $OMOP::CSV::Validator::VERSION;
@@ -29,13 +31,15 @@ my $csv_file;
 my $sep = ',';
 my $table_name;         # Optional parameter to override table name
 my $schemas_file;       # Optional: file to save schemas
+my $nocolor = 0;
 
 GetOptions(
     'ddl=s'          => \$ddl_file,        # Path to the DDL file (PostgreSQL)
     'input=s'        => \$csv_file,        # Path to the input CSV file
     'sep=s'          => \$sep,             # Field separator (default: comma)
     'table|t=s'      => \$table_name,      # Optional: override table name
     'save-schemas=s' => \$schemas_file,    # Optional: file to save schemas
+    'no-color|nc'    => \$nocolor,         # Optional: Turn off STDOUT color
     'help|h'         => \my $help,         # Show help message
     'version|V'      => sub {
         say color("cyan"), "$0 Version $VERSION", color("reset");
@@ -45,12 +49,24 @@ GetOptions(
 
 pod2usage(1) if $help;
 
+# Turning color off if argument <--no-color>
+$ENV{'ANSI_COLORS_DISABLED'} = 1 if $nocolor;
+
 unless ( $ddl_file && $csv_file ) {
     warn color("red"), "[ERROR] --ddl and --input are required parameters.\n",
       color("reset");
     pod2usage(1);
 }
 
+# Define the hash with emoji values
+my %msg_emoji = (
+    error   => '❌',
+    warning => '⚠️ ',
+    dot     => '✖',
+    success => '✅',
+    save    => '💾'
+);
+
 # Read the DDL file (UTF-8)
 my $ddl_text = path($ddl_file)->slurp_utf8;
 
@@ -64,40 +80,42 @@ my $schemas = $validator->load_schemas_from_ddl($ddl_text);
 if ($schemas_file) {
     my $json = JSON::XS->new->utf8->pretty->encode($schemas);
     path($schemas_file)->spew_utf8($json);
-    say color("green"), "💾 Schemas saved to '$schemas_file'", color("reset");
+    say color("green"), "$msg_emoji{save} Schemas saved to '$schemas_file'",
+      color("reset");
 }
 
 # Determine the schema to use (either from --table or derived from CSV filename)
-my $schema;
+my ($schema, $schema_name);
 if ($table_name) {
-    $schema = $schemas->{ lc $table_name }
-      or die color("red"), "❌ No schema found for table '$table_name'\n",
-      color("reset");
+   $schema = $schemas->{ lc $table_name } or die color("red"), "$msg_emoji{error} No schema found for table '$table_name'\n", color("reset");
+   $schema_name = $table_name;
 }
 else {
-    $schema = $validator->get_schema_from_csv_filename( $csv_file, $schemas )
-      or die color("red"),
-      "❌ No schema found for table derived from '$csv_file'\n", color("reset");
+    $schema = $validator->get_schema_from_csv_filename( $csv_file, $schemas ) or die color("red"), "$msg_emoji{error} No schema found for table derived from '$csv_file'\n", color("reset");
+     $schema_name = basename($csv_file);
+     $schema_name =~ s/\.csv$//i;
 }
 
 # Validate the CSV file
 my $errors = $validator->validate_csv_file( $csv_file, $schema, $sep );
 
 if (@$errors) {
-    say color("bold white on_red"), "❌ Validation errors found:",
+    say color("bold white on_red"),
+      "$msg_emoji{error} Validation errors found:",
       color("reset");
     foreach my $err (@$errors) {
-        say color("red"), "⚠️  Row $err->{row} validation failed:",
+        say color("red"),
+          "$msg_emoji{warning} Row $err->{row} validation failed:",
           color("reset");
         foreach my $msg ( @{ $err->{errors} } ) {
-            say "   ✖ $msg";
+            say "   $msg_emoji{dot} $msg";
         }
     }
     exit 1;
 }
 else {
     say color("bold white on_green"),
-      "✅ CSV file 'person.csv' is valid against the 'person' schema.",
+"$msg_emoji{success} CSV file '$csv_file' is valid against the '$schema_name' schema.",
       color("reset");
     exit 0;
 }
@@ -137,6 +155,10 @@ to derive the table name from the CSV filename.
 
 (optional) Path to a file where the DDL-derived schemas should be saved (in JSON format).
 
+=item B<--no-color>, B<-nc>
+
+(Optional) Turn off STDOUT color
+
 =item B<--help>, B<-h>
 
 Display this help message.

lib/OMOP/CSV/Validator.pm

@@ -10,7 +10,7 @@ use Scalar::Util qw(looks_like_number);
 use Path::Tiny;
 use Data::Dumper;
 
-our $VERSION = '0.02_1';
+our $VERSION = '0.03';
 
 =head1 NAME
 
@@ -26,7 +26,7 @@ OMOP::CSV::Validator - Validates OMOP CDM CSV files against their expected data
     my $schemas = $validator->load_schemas_from_ddl($ddl_text);
 
     # Retrieve specific table schema for a CSV file
-    my $schema  = $validator->get_schema_from_csv_filename($csv_file, $schemas);
+    my $schema = $validator->get_schema_from_csv_filename($csv_file, $schemas);
 
     # Validate CSV file
     my $errors  = $validator->validate_csv_file($csv_file, $schema);
@@ -121,12 +121,12 @@ sub _build_schema {
             my $prop = {};
 
             if ( $type =~ /int/ ) {
-                $prop->{type}  = 'integer';
-                $prop->{_type} = 'integer';
+                $prop->{type}    = 'integer';
+                $prop->{_coerce} = 1;
             }
             elsif ( $type =~ /numeric|real|double/ ) {
-                $prop->{type}  = 'number';
-                $prop->{_type} = 'number';
+                $prop->{type}    = 'number';
+                $prop->{_coerce} = 1;
 
             }
             elsif ( $type eq 'date' ) {
@@ -210,12 +210,7 @@ sub validate_csv_file {
         for my $col ( keys %{ $schema->{properties} } ) {
             if ( exists $record->{$col} ) {
                 my $prop = $schema->{properties}->{$col};
-                if (
-                    defined $prop->{_type}
-                    && (   $prop->{_type} eq 'integer'
-                        or $prop->{_type} eq 'number' )
-                  )
-                {
+                if ( defined $prop->{_coerce} && $prop->{_coerce} ) {
                     $record->{$col} =
                       $self->dotify_and_coerce_number( $record->{$col} );
                 }