License text in json

rules/gather_licenses_info.bzl

@@ -71,15 +71,12 @@ def _write_licenses_info_impl(target, ctx):
         return [OutputGroupInfo(licenses = depset())]
 
     # Write the output file for the target
-    name = "%s_licenses_info.json" % ctx.label.name
-    lic_info, _ = licenses_info_to_json(info)
-    content = "[\n%s\n]\n" % ",\n".join(lic_info)
-    out = ctx.actions.declare_file(name)
-    ctx.actions.write(
-        output = out,
-        content = content,
-    )
-    outs.append(out)
+    json_out = ctx.actions.declare_file("%s_licenses_info.json" % ctx.label.name)
+    licenses_files = write_licenses_info(ctx, target, json_out)
+    outs.append(json_out)
+    json_out_full = ctx.actions.declare_file("%s_licenses_info_full.json" % ctx.label.name)
+    embed_licenses_text(ctx, json_out, licenses_files, json_out_full)
+    outs.append(json_out_full)
 
     if ctx.attr._trace[TraceInfo].trace:
         trace = ctx.actions.declare_file("%s_trace_info.json" % ctx.label.name)
@@ -100,6 +97,12 @@ gather_licenses_info_and_write = aspect(
     attr_aspects = ["*"],
     attrs = {
         "_trace": attr.label(default = "@rules_license//rules:trace_target"),
+        "_embed_licenses_text_tool": attr.label(
+            default = Label("@rules_license//tools:embed_licenses_text"),
+            executable = True,
+            allow_files = True,
+            cfg = "exec",
+        ),
     },
     provides = [OutputGroupInfo],
     requires = [gather_licenses_info],
@@ -146,6 +149,10 @@ def write_licenses_info(ctx, deps, json_out):
     Returns:
       A list of License File objects for each of the license paths referenced in the json.
     """
+
+    if type(deps) != type([]):
+        deps = [deps]
+
     licenses_json = []
     licenses_files = []
     for dep in deps:
@@ -163,6 +170,30 @@ def write_licenses_info(ctx, deps, json_out):
     )
     return licenses_files
 
+def embed_licenses_text(ctx, json_in, licenses_files, json_out):
+    """Reads the license text from the referenced file and attaches it to the json
+
+    Args:
+      ctx: context of the caller
+      json_in: input handle to read the JSON info
+      licenses_files: input handles for all LICENSE files
+      json_out: output handle to write the JSON info
+    """
+
+    inputs = [json_in]+licenses_files
+    outputs = [json_out]
+    args = ctx.actions.args()
+    args.add("--licenses_in", json_in.path)
+    args.add("--licenses_out", json_out.path)
+    ctx.actions.run(
+        mnemonic = "EmbedLICENSETexts",
+        progress_message = "Embedding license texts...",
+        inputs = inputs,
+        outputs = outputs,
+        executable = ctx.executable._embed_licenses_text_tool,
+        arguments = [args],
+    )
+
 def licenses_info_to_json(licenses_info):
     """Render a single LicenseInfo provider to JSON
 

rules_gathering/gather_metadata.bzl

@@ -85,14 +85,9 @@ def _write_metadata_info_impl(target, ctx):
         return [OutputGroupInfo(licenses = depset())]
 
     # Write the output file for the target
-    name = "%s_metadata_info.json" % ctx.label.name
-    content = "[\n%s\n]\n" % ",\n".join(metadata_info_to_json(info))
-    out = ctx.actions.declare_file(name)
-    ctx.actions.write(
-        output = out,
-        content = content,
-    )
-    outs.append(out)
+    json_out = ctx.actions.declare_file("%s_metadata_info.json" % ctx.label.name)
+    write_metadata_info(ctx, target, json_out)
+    outs.append(json_out)
 
     if ctx.attr._trace[TraceInfo].trace:
         trace = ctx.actions.declare_file("%s_trace_info.json" % ctx.label.name)
@@ -140,8 +135,8 @@ def write_metadata_info(ctx, deps, json_out):
 
       def _foo_impl(ctx):
         ...
-        out = ctx.actions.declare_file("%s_licenses.json" % ctx.label.name)
-        write_metadata_info(ctx, ctx.attr.deps, metadata_file)
+        json_file = ctx.actions.declare_file("%s_licenses.json" % ctx.label.name)
+        write_metadata_info(ctx, ctx.attr.deps, json_file)
 
     Args:
       ctx: context of the caller
@@ -150,10 +145,15 @@ def write_metadata_info(ctx, deps, json_out):
             aspect over them
       json_out: output handle to write the JSON info
     """
+
+    if type(deps) != type([]):
+        deps = [deps]
+
     licenses = []
     for dep in deps:
         if TransitiveMetadataInfo in dep:
-            licenses.extend(metadata_info_to_json(dep[TransitiveMetadataInfo]))
+            metadata_info = dep[TransitiveMetadataInfo]
+            licenses.extend(metadata_info_to_json(metadata_info))
     ctx.actions.write(
         output = json_out,
         content = "[\n%s\n]\n" % ",\n".join(licenses),

tools/BUILD

@@ -44,3 +44,10 @@ py_binary(
     python_version = "PY3",
     visibility = ["//visibility:public"],
 )
+
+py_binary(
+    name = "embed_licenses_text",
+    srcs = ["embed_licenses_text.py"],
+    python_version = "PY3",
+    visibility = ["//visibility:public"],
+)

tools/embed_licenses_text.py

@@ -0,0 +1,30 @@
+#!/usr/bin/env python3
+import argparse
+import codecs
+import json
+
+
+def main():
+  parser = argparse.ArgumentParser(
+      description='Demonstraton license compliance checker')
+
+  parser.add_argument('--licenses_in', help='path to JSON file containing all license info')
+  parser.add_argument('--licenses_out', help='path to JSON file where augmented license info should be stored')
+  args = parser.parse_args()
+
+  with codecs.open(args.licenses_in, encoding='utf-8') as inp:
+    license_data = json.loads(inp.read())
+
+  for target in license_data:
+    for license in target['licenses']:
+      license_path = license.get('license_text')
+      if license_path:
+        with codecs.open(license_path, encoding='utf-8') as license_file:
+          license['license_text'] = license_file.read()
+
+  with codecs.open(args.licenses_out, mode='w', encoding='utf-8') as out:
+    json.dump(license_data, out, indent=4)
+
+
+if __name__ == '__main__':
+  main()