make it possible to specify the content type for data inputs from the…

… command line and Ant task instead of the hard-coded text/plain and thus correctly Base64-encode non-text data input and use user-given content type preferred over content type that is determined from the content or name

docs/src/anttask.xml

@@ -457,6 +457,17 @@ Calabash</citetitle> from Apache Ant using the
                                                                <td>input type, either <literal>XML</literal> or <literal>DATA</literal></td>
                                                                <td>No; default is <literal>XML</literal></td>
                                                       </tr>
+                                                      <tr>
+                                                               <td>contentType</td>
+                                                               <td>content type, if this is equal to <literal>application/xml</literal>,
+                                                                        ends in <literal>+xml</literal>, starts with <literal>text/</literal>
+                                                                        or has a charset of <literal>utf-8</literal> specified, the input
+                                                                        is treated as text, otherwise the input gets base64 encoded
+                                                                        automatically before it is given to the pipeline</td>
+                                                               <td>No; default is guessed from the resource content, as fallback from the
+                                                                        resource name and as second fallback <literal>application/octet-stream</literal>
+                                                                        is used</td>
+                                                      </tr>
                                                       <tr>
                                                                <td>if</td>
                                                                <td>This parameter behaves like standard Ant if's as documented at

docs/src/configuration.xml

@@ -545,11 +545,11 @@ left out and the input will be applied to the primary non-parameter input port.<
 <tbody>
 <row>
   <entry>Command line (long):</entry>
-  <entry><literal>--data-input</literal> <replaceable>port</replaceable>=<replaceable>uri</replaceable></entry>
+  <entry><literal>--data-input</literal> <replaceable>contentType</replaceable>@<replaceable>port</replaceable>=<replaceable>uri</replaceable></entry>
 </row>
 <row>
   <entry>Command line (short):</entry>
-  <entry><literal>-d</literal><replaceable>port</replaceable>=<replaceable>uri</replaceable></entry>
+  <entry><literal>-d</literal><replaceable>contentType</replaceable>@<replaceable>port</replaceable>=<replaceable>uri</replaceable></entry>
 </row>
 <row>
   <entry>Ant task:</entry>
@@ -559,7 +559,20 @@ left out and the input will be applied to the primary non-parameter input port.<
 </tgroup>
 </informaltable>
 
-<para>The input specified will be base64 encoded.</para>
+<para>If the port and uri do not contain an at sign, the contentType specification can be
+left out. In this case the content type will be determined from the contents of the URI.
+If the content type cannot be determined from the contents of the URI, the content type
+is guessed from the URI name. If the content type cannot be guessed from the URI name,
+it is set to "<literal>application/octet-stream</literal>". If you have to give an explicit
+content type because the port or uri contain an at sign, but you don't know the content
+type, you can set it to "<literal>content/unknown</literal>" which is the same as leaving
+it out and will try to determine the content type from content or name.</para>
+
+<para>If the detected or explicitly specified content type is equal to
+<literal>application/xml</literal>, ends in <literal>+xml</literal>, starts with
+<literal>text/</literal> or has a charset of <literal>utf-8</literal> specified,
+the input is treated as text, otherwise the input gets base64 encoded automatically
+before it is given to the pipeline.</para>
 
 <para>If the uri does not contain an equals sign, the port specification can be
 left out and the input will be applied to the primary non-parameter input port.</para>

docs/src/running.xml

@@ -311,7 +311,7 @@ will run Java with the correct class path and other arguments.</para>
       <arg choice="plain">-d</arg>
       <arg choice="plain">--data-input</arg>
     </group>
-    <arg choice="opt"><replaceable>port</replaceable>=</arg><replaceable>uri</replaceable>
+    <arg choice="opt"><replaceable>contentType</replaceable>@</arg><arg choice="opt"><replaceable>port</replaceable>=</arg><replaceable>uri</replaceable>
   </arg>
   <arg rep="repeat">
     <group choice="req">

resources/etc/usage.txt

@@ -2,35 +2,35 @@ Usage: com.xmlcalabash.drivers.Main [switches/options] [pipeline.xpl] [options]
    or: com.xmlcalabash.drivers.Main [switches/options] { [inputs/parameters] step [options] }+
 
 Where switches are:
-  -a, --schema-aware                   Turn on schema-aware processing,
-                                       this also sets saxon-processor to 'ee' implicitly
-  -b, --binding prefix=uri             Specify namespace binding
-  -c, --config configfile              Specify a particular configuration file
-  -d, --data-input [port=]uri          Bind the specified input port to data, if no port is specified,
-                                       the first unbound input port is used
-  -D, --debug                          Turn on debugging
-  -E, --entity-resolver className      Specify a resolver class for entity resolution
-  -G, --log-style logstyle             Specify the default style for p:log output;
-                                       Must be 'off', 'plain', 'wrapped' (default), or 'directory'
-  -i, --input [port=]uri               Bind the specified input port, if no port is specified,
-                                       the first unbound input port is used
-  -l, --library library.xpl            Load the specified library
-  -o, --output [port=]uri              Bind the specified output port, if no port is specified,
-                                       the first unbound output port is used
-  -p, --with-param [port@]param=value  Specify a parameter
-      --profile file                   Specify a file, or '-' for console output,
-                                       where to write profiling information of the
-                                       pipeline that was run
-  -P, --saxon-processor edition        Request a specific edition of Saxon;
-                                       Must be 'he' (default), 'pe' or 'ee'
-  -s, --step stepname                  Run the step named 'stepname' instead of a pipeline
-  -S, --safe-mode                      Request 'safe' execution
-  -U, --uri-resolver className         Specify a resolver class for URI resolution
-  -v, --version                        Show XML Calabash version
-  -X, --extension extname              Enable the 'extname' extension;
-                                       valid extname values are 'general-values', 'xpointer-on-text',
-                                       'transparent-json', 'json-flavor=<flavor>' and 'use-xslt-10'
-      --saxon-configuration file       Load the specified Saxon configuration
+  -a, --schema-aware                         Turn on schema-aware processing,
+                                             this also sets saxon-processor to 'ee' implicitly
+  -b, --binding prefix=uri                   Specify namespace binding
+  -c, --config configfile                    Specify a particular configuration file
+  -d, --data-input [contentType@][port=]uri  Bind the specified input port to data, if no port is specified,
+                                             the first unbound input port is used
+  -D, --debug                                Turn on debugging
+  -E, --entity-resolver className            Specify a resolver class for entity resolution
+  -G, --log-style logstyle                   Specify the default style for p:log output;
+                                             Must be 'off', 'plain', 'wrapped' (default), or 'directory'
+  -i, --input [port=]uri                     Bind the specified input port, if no port is specified,
+                                             the first unbound input port is used
+  -l, --library library.xpl                  Load the specified library
+  -o, --output [port=]uri                    Bind the specified output port, if no port is specified,
+                                             the first unbound output port is used
+  -p, --with-param [port@]param=value        Specify a parameter
+      --profile file                         Specify a file, or '-' for console output,
+                                             where to write profiling information of the
+                                             pipeline that was run
+  -P, --saxon-processor edition              Request a specific edition of Saxon;
+                                             Must be 'he' (default), 'pe' or 'ee'
+  -s, --step stepname                        Run the step named 'stepname' instead of a pipeline
+  -S, --safe-mode                            Request 'safe' execution
+  -U, --uri-resolver className               Specify a resolver class for URI resolution
+  -v, --version                              Show XML Calabash version
+  -X, --extension extname                    Enable the 'extname' extension;
+                                             valid extname values are 'general-values', 'xpointer-on-text',
+                                             'transparent-json', 'json-flavor=<flavor>' and 'use-xslt-10'
+      --saxon-configuration file             Load the specified Saxon configuration
 
 For more information, consult http://xmlcalabash.com/docs/reference/ or
 the documentation that came with the distribution.

src/com/xmlcalabash/drivers/CalabashTask.java

@@ -48,6 +48,7 @@
 import org.apache.tools.ant.types.resources.Union;
 import org.apache.tools.ant.util.FileNameMapper;
 
+import static com.xmlcalabash.util.Input.Type.DATA;
 import static com.xmlcalabash.util.Input.Type.XML;
 import static java.lang.Long.MAX_VALUE;
 import static java.util.Arrays.asList;
@@ -282,7 +283,7 @@ public void addConfiguredInput(Input input) {
                 return;
             }
 
-            inputMappers.put(port, new TypedFileNameMapper(inputMapper, input.getType()));
+            inputMappers.put(port, new TypedFileNameMapper(inputMapper, input.getType(), input.getContentType()));
         } else {
             if (inputMappers.containsKey(port)) {
                 handleError("Resources used on input port that already has a mapper: " + port);
@@ -294,7 +295,7 @@ public void addConfiguredInput(Input input) {
             }
 
             for (Resource resource : resources.listResources()) {
-                inputResources.get(port).add(new TypedResource(resource, input.getType()));
+                inputResources.get(port).add(new TypedResource(resource, input.getType(), input.getContentType()));
             }
         }
     }
@@ -1048,7 +1049,7 @@ public void execute() {
                             for (String fileName : inputFileNames) {
                                 FileResource mappedResource = new FileResource(baseDir, fileName);
                                 if (mappedResource.isExists()) {
-                                    mappedResources.add(new TypedResource(mappedResource, inputMapper.getType()));
+                                    mappedResources.add(new TypedResource(mappedResource, inputMapper.getType(), inputMapper.getContentType()));
                                 } else {
                                     log("Skipping non-exstent mapped resource: " + mappedResource.toString(), Project.MSG_DEBUG);
                                 }
@@ -1141,7 +1142,7 @@ public void execute() {
                             List<TypedResource> mappedResources = new ArrayList<TypedResource>();
                             for (String fileName : inputFileNames) {
                                 FileResource mappedResource = new FileResource(baseDir, fileName);
-                                mappedResources.add(new TypedResource(mappedResource, inputMapper.getType()));
+                                mappedResources.add(new TypedResource(mappedResource, inputMapper.getType(), inputMapper.getContentType()));
                             }
                             useInputResources.put(port, mappedResources);
                         }
@@ -1269,13 +1270,13 @@ private void process(Map<String, List<TypedResource>> inputResources, Map<String
             for (String port : inputResources.keySet()) {
                 for (TypedResource typedResource : inputResources.get(port)) {
                     Resource resource = typedResource.getResource();
-                    userArgs.addInput(port, resource.getInputStream(), resource.toString(), typedResource.getType());
+                    userArgs.addInput(port, resource.getInputStream(), resource.toString(), typedResource.getType(), typedResource.getContentType());
                 }
             }
             for (Step step : steps) {
                 for (Input input : step.getInputs()) {
                     for (Resource resource : input.getResources().listResources()) {
-                        userArgs.addInput(input.getPort(), resource.getInputStream(), resource.toString(), input.getType());
+                        userArgs.addInput(input.getPort(), resource.getInputStream(), resource.toString(), input.getType(), input.getContentType());
                     }
                 }
                 for (Parameter parameter : step.getParameters()) {
@@ -1499,6 +1500,11 @@ public static class Input extends Port {
          */
         private Type type = XML;
 
+        /**
+         * The content type
+         */
+        private String contentType;
+
         /**
          * Get the input type
          *
@@ -1516,6 +1522,27 @@ public Type getType() {
         public void setType(Type type) {
             this.type = type;
         }
+
+        /**
+         * Get the content type
+         *
+         * @return the content type
+         */
+        public String getContentType() {
+            if ((contentType != null) && (type != DATA)) {
+                throw new IllegalStateException("contentType of input can only be set if type is DATA");
+            }
+            return contentType;
+        }
+
+        /**
+         * Set the content type
+         *
+         * @param contentType the content type
+         */
+        public void setContentType(String contentType) {
+            this.contentType = contentType;
+        }
     } // Input
 
     /**
@@ -1757,8 +1784,13 @@ public boolean shouldUse() {
     private static class TypedResource {
         private Resource resource = null;
         private Type type = null;
+        private String contentType = null;
 
         private TypedResource(Resource resource, Type type) {
+            this(resource, type, null);
+        }
+
+        private TypedResource(Resource resource, Type type, String contentType) {
             if (resource == null) {
                 throw new IllegalArgumentException("resource must not be null");
             }
@@ -1767,6 +1799,7 @@ private TypedResource(Resource resource, Type type) {
                 throw new IllegalArgumentException("type must not be null");
             }
             this.type = type;
+            this.contentType = contentType;
         }
 
         public Resource getResource() {
@@ -1776,13 +1809,22 @@ public Resource getResource() {
         public Type getType() {
             return type;
         }
+
+        public String getContentType() {
+            return contentType;
+        }
     } // TypedResource
 
     private static class TypedFileNameMapper implements FileNameMapper {
         private FileNameMapper fileNameMapper;
         private Type type;
+        private String contentType;
 
         private TypedFileNameMapper(FileNameMapper fileNameMapper, Type type) {
+            this(fileNameMapper, type, null);
+        }
+
+        private TypedFileNameMapper(FileNameMapper fileNameMapper, Type type, String contentType) {
             if (fileNameMapper == null) {
                 throw new IllegalArgumentException("fileNameMapper must not be null");
             }
@@ -1791,12 +1833,17 @@ private TypedFileNameMapper(FileNameMapper fileNameMapper, Type type) {
                 throw new IllegalArgumentException("type must not be null");
             }
             this.type = type;
+            this.contentType = contentType;
         }
 
         public Type getType() {
             return type;
         }
 
+        public String getContentType() {
+            return contentType;
+        }
+
         @Override
         public void setFrom(String s) {
             fileNameMapper.setFrom(s);

src/com/xmlcalabash/drivers/Main.java

@@ -242,13 +242,13 @@ boolean run(UserArgs userArgs, XProcConfiguration config) throws SaxonApiExcepti
                             ReadableData rd;
                             switch (input.getKind()) {
                                 case URI:
-                                    rd = new ReadableData(runtime, c_data, input.getUri(), "text/plain");
+                                    rd = new ReadableData(runtime, c_data, input.getUri(), input.getContentType());
                                     doc = rd.read();
                                     break;
 
                                 case INPUT_STREAM:
                                     InputStream inputStream = input.getInputStream();
-                                    rd = new ReadableData(runtime, c_data, inputStream, "text/plain");
+                                    rd = new ReadableData(runtime, c_data, inputStream, input.getContentType());
                                     doc = rd.read();
                                     inputStream.close();
                                     break;

src/com/xmlcalabash/io/ReadableData.java

@@ -101,7 +101,7 @@ private DocumentSequence ensureDocuments() {
             stream = (uri == null) ? inputStream : ("-".equals(uri) ? System.in : getStream(dataURI));
             String serverContentType = getContentType();
 
-            if ("content/unknown".equals(serverContentType) && contentType != null) {
+            if ((contentType != null) && !"content/unknown".equals(contentType)) {
                 // pretend...
                 serverContentType = contentType;
             }

src/com/xmlcalabash/util/Input.java

@@ -5,21 +5,28 @@
 import static com.xmlcalabash.util.Input.Kind.INPUT_STREAM;
 import static com.xmlcalabash.util.Input.Kind.NONE;
 import static com.xmlcalabash.util.Input.Kind.URI;
+import static com.xmlcalabash.util.Input.Type.DATA;
 import static com.xmlcalabash.util.Input.Type.XML;
 
 public class Input {
     private String uri;
     private InputStream inputStream;
     private Type type;
+    private String contentType;
     private Kind kind = NONE;
 
     public Input(String uri) {
         this(uri, XML);
     }
 
     public Input(String uri, Type type) {
+        this(uri, type, null);
+    }
+
+    public Input(String uri, Type type, String contentType) {
         this.uri = uri;
         this.type = type;
+        this.contentType = contentType;
         kind = URI;
     }
 
@@ -28,9 +35,12 @@ public Input(InputStream inputStream, String uri) {
     }
 
     public Input(InputStream inputStream, String uri, Type type) {
+        this(inputStream, uri, type, null);
+    }
+
+    public Input(InputStream inputStream, String uri, Type type, String contentType) {
+        this(uri, type, contentType);
         this.inputStream = inputStream;
-        this.uri = uri;
-        this.type = type;
         kind = INPUT_STREAM;
     }
 
@@ -59,6 +69,13 @@ public Type getType() {
         return type;
     }
 
+    public String getContentType() {
+        if ((contentType != null) && (type != DATA)) {
+            throw new IllegalStateException("contentType of input can only be set if type is DATA");
+        }
+        return contentType;
+    }
+
     public Kind getKind() {
         return kind;
     }