fixed paths in File Data Entities

README.md

@@ -36,38 +36,43 @@ This version of Calcyte targets the [R0-Crate Spec, v0.2](https://github.com/UTS
 
   npm link
 
+## Get set up for development use
+
+RO-Crate 
+
+
+
+
+
 ## Usage / instructions
 
-Calcyfy creates RO-Crates whith and HTML entry-point. This relies on 
 
-Get a web server running on local host:
 
-```
 
-```
+### Usage
+
+Calcyfy creates RO-Crates with an HTML entry-point in r
 
 Usage:
 
 ```
-> ./calcyfy
-
-  Usage: calcyfy [options] <directories...>
+>  ./calcyfy
+   Usage: calcyfy [options] <directories...>
 
-  Generates DataCrate HTML for ro-crate-metadata.jsonld files. Pass a list of directories. To create Spreadsheet files for entring metadata use -d or -r.
+  Generates [RO-Crate] HTML for ro-crate-metadata.jsonld files. Pass a list of directories. To create Spreadsheet files for entering metadata use -d or -r.
 
 
   Options:
 
-    -V, --version                   output the version number
-    -b,  --bag [bag-dir]            Create Bagit Bag(s) under [bag-dir])
-    -n,  --no                       No Citation - only applies ith --bag
-    -z,  --zip                      Create a zipped version of the bag - only applies with --bag
-    -d,  --depth [depth]            Maximum depth to recurse into directories
-    -r,  --recurse                  Recurse into directories looking for ro-crate-metadata_.xslx files
-    -s,  --save                     Save ro-crate-metadata after trimming it (and adding missing context if necessary).
-    -h,  --htmlscript [htmlscript]  HTML render script
-    -u, --url [distro]              Distribution URL
-    -h, --help                      output usage information
+    -V, --version                    output the version number
+    -b,  --bag [bag-dir]             Create Bagit Bag(s) under [bag-dir])
+    -n,  --no                        No Citation - only applies ith --bag
+    -z,  --zip                       Create a zipped version of the bag - only applies with --bag
+    -d,  --depth [depth]             Maximum depth to recurse into directories
+    -r,  --recurse                   Recurse into directories looking for CATALOG_.xslx files
+    -c,  --cratescript [cratesript]  URL of "crate.js" rendering script from the ro-crate-js library
+    -u, --url [distro]               Distribution URL
+    -h, --help                       output usage information
 ```
 
 To run Calcyte on a group of directories pass it a list of directories
@@ -83,13 +88,13 @@ One directory:
 ```
 
 calcyfy -r test_data/sample/ 
-calcyfy test_data/Glop_Pot -r -h http://localhost:8080/compiled
+calcyfy test_data/Glop_Pot -r -c /
 ```
 
 This will:
 - Traverse the entire Glop_Pot directory, and generate or update CATALOG_name.xlsx files.
 - Create or update the `test_data/Glop_Pot/`ro-crate-metadata.jsonld`` file
-- Create a *RO-Crate Website* with entry-point `test_data/Glop_Pot/CATALOG.html`
+- Create a *[RO-Crate] Website* with entry-point `test_data/Glop_Pot/ro-crate-metadata.html`
 
 All the sample directories:
 

calcyfy

@@ -52,7 +52,9 @@ program
   )
   .option("-d,  --depth [depth]", "Maximum depth to recurse into directories looking for or creating CATALOG_.xlsx file")
   .option("-r,  --recurse", "Recurse into directories looking for or creating CATALOG_.xslx files")
-  .option("-c,  --cratescript [cratesript]", "URL of Crate-script directtory")
+  .option("-c,  --cratescript [cratesript]", "URL of Crate-script directory")
+  .option("-m,  --maxfiles [maxfiles]", `Maximum number of files to itemise per directory (default is ${defaults.max_files_in_dir})`)
+
   .option(
     "-u, --url [distro]",
     "Distribution URL"
@@ -65,6 +67,9 @@ if (program.recurse) {
 if (program.depth) {
   depth = program.depth;
 }
+if (program.maxfiles) {
+  defaults.max_files_in_dir = program.maxfiles;
+}
 if (!program.rawArgs.length || !dirs) program.help();
 
 // TODO: Refactor this by if statement - there is a lot of repetition

lib/collection.js

@@ -369,6 +369,9 @@ module.exports = function() {
               this.file_info_by_filename[f.filename.replace(/.*\//, "")] = f;
             }
           }
+          else  {
+            console.log(`WARNING: Max ${defaults.max_files_in_dir} files exceeded (${items.length}) in ${dir}\n HINT: Re-run \n     ${process.argv.join(" ")} -m ${items.length + 1}`)
+          }
           //console.log(dir, metadatas[0]);
           metadata_path = path.join(dir, metadatas[0]);
           //console.log("Opening spreadsheet", metadata_path)
@@ -433,11 +436,9 @@ module.exports = function() {
               );
               for (var item of extra_context) {
                 if (item["Key"] && item["Value"]) {
-                  
                   context[item["Key"]] = item["Value"]
                   if (item["SameAs"]) {
                     this.same_as.push({"@id": item["Value"], "sameAs": item["SameAs"]})
-
                   }
                 }
               }
@@ -466,7 +467,7 @@ module.exports = function() {
               //console.log("making new child:", this.depth, this.max_depth, subdirs[i]);
               child.read(
                 path.join(dir, subdirs[i]),
-                path.join(this.rel_path, subdirs[i]),
+                path.join(this.rel_path, subdirs[i] + "/"),
                 this,
                 this.max_depth
               );

lib/defaults.js

@@ -23,7 +23,7 @@ const datacite_file_name = "datacite.xml";
 const metadata_json_file_name = `${ro_crate_name}.jsonld`;
 const html_file_name = "ro-crate-preview.html";
 const max_depth = "10"; // Number of dirs to recurse into
-const max_files_in_dir = "30"; // Don't list files in a directory if there are more than this
+const max_files_in_dir = "20"; // Don't list files in a directory if there are more than this
 const ignore_file_regex = new RegExp(
   `(^~)|(^\\.)|(${datacite_file_name})|^${html_file_name}$|^${metadata_json_file_name}$`
 );
@@ -38,7 +38,7 @@ const ROCrate_profile_file = "defaults/profile-datacrate-v" + ROCrate_version +
 const path = require("path")
 const DEFAULTS = path.join(__dirname, '../defaults');
 const defaults_dir = DEFAULTS;
-const context = "https://raw.githubusercontent.com/ResearchObject/ro-crate/master/docs/0.2-DRAFT/context.json";
+const context = "https://raw.githubusercontent.com/ResearchObject/ro-crate/master/docs/0.3-DRAFT/context.json";
 const default_context = context;
 const metadata_template = path.join(DEFAULTS, 'metadata_template.html');
 const render_script = "https://code.research.uts.edu.au/eresearch/CalcyteJS/raw/feature/or-crate/lib/render.js";

lib/item.js

@@ -48,21 +48,6 @@ module.exports = function() {
     return links_to;
   }
 
-  function make_link(string, collection, el) {
-    //console.log("Linking", string);
-    var link_id = links_to_id(string, collection);
-
-    if (link_id && link_id.match(/^https?:\/\//i)) {
-      el.ele("a", string).att("href", link_id);
-      string = "<a href='" + link_id + "'>" + string + "</a>";
-    } else if (link_id) {
-      el.ele("a", string).att("href", "#" + link_id);
-      string = "<a href='#" + link_id + "'>" + string + "</a>";
-    } else {
-      el.ele("span", string);
-    }
-    return string;
-  }
 
   return {
     metadata: this.metadata,
@@ -121,12 +106,7 @@ module.exports = function() {
       }
       frag["@type"] = this.types;
       
-      var ident =  helper.value_as_array(this.id)
-      for (let id of ident){
-        if (id && id.match(/(https?:\/\/)(dx\.)?(doi.org\/10\.)/i)) {
-          helper.push_value(frag, "identifier", id.replace(/https?:\/\//i, ""))
-        }
-      }
+      
       this.json_ld_fragment = frag;
       //console.log(frag);
       return frag;
@@ -145,27 +125,31 @@ module.exports = function() {
         property.parse(key, value);
         //this.colum_names.push(k);
         //console.log("Fresh property", property.name, value);
-
         if (property.is_file) {
           //console.log("FILE ", this.collection.path, value);
           this.id = path.join(this.collection.rel_path, value);
           this.types.push("File");
-          //var pr = new metadata_property_name();
-          //pr.parse("path", this.id);
-          //this.properties[pr.name] = pr;
           this.is_file = value;
         } else if (property.is_id) {
           //console.log("Got an ID", value);
-          this.id = value;
+          if (Array.isArray(value)) {
+            value = value[0];
+          }
+          if (value.endsWith("/" ) || value.startsWith(".") || value.match(/^(https?:\/\/|mailto:.*@|_b:)/i)) {
+            this.id = value;
+          }
+          else {
+            this.id = "#" + value;
+          }
         } else if (property.is_name) {
           this.name = value;
         } else if (property.is_type) {
-          this.types = property.data;
+          this.types = this.types.concat(property.data);
         }
  
         // Add to lookup table
         if (!this.id) {
-          this.id = uuidv4();
+          this.id = "#" + uuidv4();
         } 
         //console.log(this.name)
         if (this.name) {

lib/property.js

@@ -98,7 +98,6 @@ module.exports = function() {
       //Assume what's left of header is a property, and hence starts with lowecase letter
       this.name = lowercase_first(this.name);
       var nest;
-
       if (this.name.endsWith("*")) {
         this.name = this.name.slice(0, -1);
         this.is_repeating = true;
@@ -127,7 +126,6 @@ module.exports = function() {
 
       if (this.property_URI == "http://schema.org/identifier") {
         this.is_id = true;
-       
       }
 
       if (this.property_URI == "http://schema.org/name") {

package.json

@@ -31,7 +31,7 @@
     "json": "^9.0.6",
     "jsonld": "^1.6.2",
     "node-json2html": "^1.1.1",
-    "ro-crate": "^1.2.11",
+    "ro-crate": "^1.2.12",
     "sha1": "^1.1.1",
     "shelljs": "^0.8.1",
     "tmp": "0.0.33",