Multi document support

ARCHITECTURE.md

@@ -0,0 +1,121 @@
+# Inline XBRL Viewer architecture 
+
+This document provides developer documentation for various aspects of the
+Inline XBRL Viewer.
+
+## JSON data
+
+The viewer relies on a block of JSON data that is embedded within the primary
+XHTML file of the viewer.  This avoids having to do full XBRL processing within
+the viewer, and avoids the need to access the report's taxonomy from the
+viewer.
+
+The format of the JSON data is described below.  Newer viewers are expected to
+work correctly with data generated by an older plugin version within the same
+major version, degrading new features gracefully if required.
+
+
+```
+{
+    "features": [],
+    "prefixes": {},
+    "roles": {},
+    "languages": {},
+    "sourceReports": [
+        {
+            "docSetFiles": [],
+            "targetReports": [
+                {
+                    "localDocs": {},
+                    "roleDefs": {},
+                    "concepts": {},
+                    "facts": {},
+                    "rels": {},
+                }
+            ]
+        }
+    ],
+    "validation": [],
+    "filingDocuments": "foo.zip"
+}
+```
+
+A viewer can contain one or more "source reports", each producing one or more
+target report.
+
+A source report is an Inline XBRL Document Set, comprising one or more Inline
+XBRL Documents.  A target report is the XBRL data obtained from extracting a
+given target document from the source report.
+
+Target documents are not yet supported, so at present, there will always be one
+target report per source report.
+
+### prefixes
+
+`prefixes` is a map of namespace prefixes to namespace URIs.  It is shared by
+all Target Reports.
+
+### languages
+
+
+```
+"languages": {
+    "en-us": "English (US)",
+    "en": "English",
+    "fr": "French",
+},
+
+```
+
+A map of language codes to language labels.  There should be an entry here for
+every language code used by a label in a report.
+
+### sourceReports
+
+`sourceReports` is an array of objects of report data.  Prior to multi-document
+support, this property was not present, and the `docSetFiles` property, and the
+properties of the objects within `targetReports` appeared at the top-level.
+
+There is an object in this array for each separate Inline XBRL Document Set in
+the viewer.
+
+### docSetFiles
+
+`docSetFiles` is a list of files in the document set.  It is used to load other
+documents in the document set into iframes in tabs in the viewer.  For single
+file, non-stub viewers, this property is not present.
+
+### targetReports
+
+An array of objects, one for each target document obtained from the enclosing
+source report.
+
+### localDocs
+
+```
+{
+    "filename.xsd": ["schema"],
+    "inline.xhtml": ["inline"],
+}
+```
+
+`localDocs` is the set of XBRL files local to the filing (i.e. the Inline XBRL
+files, and any extension taxonomies).
+
+The property is used to populate a summary of the files in the filing in the
+document summary panel.
+
+## IDs and Viewer Unique IDs
+
+The viewer requires all iXBRL facts to have `id` attributes.  The Python plugin
+will add `id` attributes to facts if not present.  IDs are required to be unique
+within an Inline XBRL Document Set, but in order to avoid collisions between IDs
+in different Inline XBRL Document Sets when multiple source reports are
+present, the JavaScript viewer internally uses "viewer unique IDs", which are
+the `id` attribute values prefixed by the source report index (e.g `0-f1` is
+`f1` in the first report).  The original `id` attributes are only used to
+resolve footnotes and continuation chains during the pre-process phase.  Facts
+are selected via the `ivids` data attribute on wrapper nodes, which uses viewer
+unique IDs.
+
+

iXBRLViewerPlugin/__init__.py

@@ -81,6 +81,11 @@ def iXBRLViewerCommandLineOptionExtender(parser, *args, **kwargs):
                       default=False,
                       dest="zipViewerOutput",
                       help="Converts the viewer output into a self contained zip")
+
+    # Force "keepOpen" to true, so that all models are retained.  Needed for
+    # multi-instance viewers.
+    parser.set_defaults(keepOpen = True)
+
     featureGroup = OptionGroup(parser, "Viewer Features",
                             "See viewer README for information on enabling/disabling features.")
     for featureConfig in FEATURE_CONFIGS:
@@ -110,7 +115,9 @@ def generateViewer(
     :param features: List of feature names to enable via generated JSON data.
     """
     # extend XBRL-loaded run processing for this option
-    if cntlr.modelManager is None or cntlr.modelManager.modelXbrl is None or not cntlr.modelManager.modelXbrl.modelDocument:
+    if (cntlr.modelManager is None 
+        or len(cntlr.modelManager.loadedModelXbrls) == 0 
+        or any(not mx.modelDocument for mx in cntlr.modelManager.loadedModelXbrls)):
         cntlr.addToLog("No taxonomy loaded.")
         return
     modelXbrl = cntlr.modelManager.modelXbrl
@@ -137,7 +144,7 @@ def generateViewer(
     try:
         out = saveViewerDest
         if out:
-            viewerBuilder = IXBRLViewerBuilder(modelXbrl)
+            viewerBuilder = IXBRLViewerBuilder(cntlr.modelManager.loadedModelXbrls)
             if features:
                 for feature in features:
                     viewerBuilder.enableFeature(feature)
@@ -306,7 +313,7 @@ def load_plugin_url():
     'author': 'Paul Warren',
     'copyright': 'Copyright :: Workiva Inc. :: 2019',
     'CntlrCmdLine.Options': commandLineOptionExtender,
-    'CntlrCmdLine.Xbrl.Run': commandLineRun,
+    'CntlrCmdLine.Filing.End': commandLineRun,
     'CntlrWinMain.Menu.Tools': toolsMenuExtender,
     'CntlrWinMain.Xbrl.Loaded': guiRun,
 }