parse train attributes

3 files changed, 105 insertions, 6 deletions

diff --git a/README.md b/README.md
index d7e142f..0376efe 100644
--- a/README.md
+++ b/README.md
@@ -83,13 +83,15 @@ as noted in the PDF file.
 	"10": {
 		"rawType": "ICE-W",
 		"type": "ICE 3",
-		"shortType": "3" /* optional */,
-		"name": "ICE International", /* optional */
+		"shortType": "3",
+		"name": "ICE International",
 		"route": { /* scheduled route */ },
-		"schedules": [ /* scheduled service days and route deviations */ ] /* optional */,
-		"cycle": { /* cycle ("Umlauf") data */ } /* optional */,
-		"hasWagon": { /* wagon type map */ } /* optional */,
-		"wagons": [ /* wagon list */ ] /* optional */,
+		"commonAttr": { /* train/powercar attributes */ },
+		"attrVariants": [ { /* train/powercar attributes */ } /* ... */ ],
+		"schedules": [ /* scheduled service days and route deviations */ ],
+		"cycle": { /* cycle ("Umlauf") data */ },
+		"hasWagon": { /* wagon type map */ },
+		"wagons": [ /* wagon list */ ],
 	}
 }
 ```
@@ -99,6 +101,9 @@ long-distance trains, but may be used by other european operators as well. For
 instance, the IC services Amsterdam – Berlin and Koebenhavns – Aarhus often use
 identical three-letter numbers.
 
+rawType, type, and route are always present. The remaining properties are
+optional.
+
 #### rawType
 
 The train type as specified in the PDF file, e.g. **ICE-A**, **IC**, or
@@ -190,6 +195,21 @@ names used in iris-tts or HAFAS. A missing preStart / postEnd entry does not
 imply that the train is prepared / parked at the first / last station with
 passenger service. Station names may be surrounded by brackets.
 
+#### train/powercar attributes
+
+```js
+{
+	"brakingPercentage": 177,
+	"length": 402,
+	"series": "406",
+	"station": "FF",
+	"vmax": 300
+}
+```
+
+Taken as-is from the PDF file. commonAttr contains only properties which are
+identical in each variant. *station* is a DS100 identifier with unknown meaning.
+
 #### schedules
 
 *work in progress*
diff --git a/bin/db-zugbildung-to-json b/bin/db-zugbildung-to-json
index 849d685..09bd3c5 100755
--- a/bin/db-zugbildung-to-json
+++ b/bin/db-zugbildung-to-json
@@ -549,6 +549,53 @@ for my $train ( values %map ) {
 	}
 }
 
+for my $train ( values %map ) {
+
+	if ( not $train->{details} ) {
+		next;
+	}
+
+	my @details = @{ $train->{details} // [] };
+	my %common_setup;
+	my @setups;
+
+	for my $line (@details) {
+		my %setup;
+		if ( $line =~ m{ Tfz1:(\d+) }x ) {
+			$setup{series} = $1;
+		}
+		if ( $line =~ m{ Tfz2:(\d+) }x ) {
+			$setup{series2} = $1;
+		}
+		if ( $line =~ m{ Hg(\d+) }x ) {
+			$setup{vmax} = 0 + $1;
+		}
+		if ( $line =~ m{ BrH(\d+) }x ) {
+			$setup{brakingPercentage} = 0 + $1;
+		}
+		if ( $line =~ m{ (\d+)m }x ) {
+			$setup{length} = 0 + $1;
+		}
+		if ( %setup and $line =~ m{ ^ \s* ([+]? [A-Z ]{1,7}) }x ) {
+			$setup{station} = $1;
+			$setup{station} =~ s{\s+$}{};
+		}
+		push( @setups, {%setup} );
+	}
+
+	for my $key (qw(brakingPercentage length series series2 vmax)) {
+		if ( ( my @uniq = uniq grep { $_ } map { $_->{$key} } @setups ) == 1 ) {
+			$common_setup{$key} = $key eq 'series' ? $uniq[0] : 0 + $uniq[0];
+		}
+	}
+
+	delete $train->{details};
+	if (%common_setup) {
+		$train->{commonAttr} = {%common_setup};
+	}
+	$train->{attrVariants} = [@setups];
+}
+
 for my $train_number ( keys %map ) {
 	my $wagon_numbers_ok = 1;
 	my $wagon_types_ok   = 1;
diff --git a/schema.yml b/schema.yml
index 87746aa..3fc21ec 100644
--- a/schema.yml
+++ b/schema.yml
@@ -66,6 +66,12 @@ components:
           description: true if this is an empty train without passenger service ("Leerfahrt")
         route:
           $ref: '#/components/schemas/route'
+        commonAttr:
+          $ref: '#/components/schemas/trainAttr'
+        attrVariants:
+          type: array
+          items:
+            $ref: '#/components/schemas/trainAttr'
         cycle:
           $ref: '#/components/schemas/cycle'
         hasWagon:
@@ -94,6 +100,32 @@ components:
         postEnd:
           type: string
           description: station where the train is parked
+    trainAttr:
+      type: object
+      properties:
+        brakingPercentage:
+          type: integer
+          example: 177
+          description: Bremshundertstel
+        length:
+          type: integer
+          example: 402
+          description: train length (including powercar)
+        series:
+          type: string
+          example: "406"
+          description: powercar series
+        series2:
+          type: string
+          description: series of second powercar (if present)
+        station:
+          type: string
+          example: "FF"
+          description: DS100 station code, details unknown
+        vmax:
+          type: integer
+          example: 300
+          description: rated maximum speed in km/h
     cycle:
       type: object
       properties: