improve tool

Signed-off-by: Maximilian Huber <[email protected]>

entr_puml.sh

@@ -5,16 +5,26 @@ set -euo pipefail
 cd "$( dirname "${BASH_SOURCE[0]}" )"
 
 spdx="$1"
+shift
 puml="$spdx.puml"
 png="$spdx.png"
 
 export spdx
 export puml
 entr_task() (
   set -ex
+
   nix develop --command poetry run python -m spdx3_to_graph -- -v "$spdx"
-  plantuml "$puml"
+  PLANTUML_LIMIT_SIZE=$(( 8192 * 10 )) plantuml "$puml"
 )
+
+if [[ "$#" -ge 1 ]]; then
+  if [[ "$1" == "--once" ]]; then
+    entr_task
+    exit 0
+  fi
+fi
+
 export -f entr_task
 
 if [[ ! -f "$puml" || ! -f "$png" ]]; then

spdx3_to_graph/spdx30.py

@@ -767,7 +767,7 @@ def _make_object(cls, typ):
     def decode(cls, decoder, *, objectset=None):
         typ, obj_d = decoder.read_object()
         if typ is None:
-            raise TypeError("Unable to determine type for object")
+            raise TypeError(f"Unable to determine type for object {decoder.data}")
 
         obj = cls._make_object(typ)
         for key in (obj.ID_ALIAS, obj._OBJ_IRIS["_id"]):

spdx3_to_graph/spdx3_to_graph.py

@@ -13,9 +13,14 @@
 import argparse
 from pathlib import Path
 import collections.abc
+import re
 
 from . import spdx30 as spdx
 
+def escape_string(s):
+    # escapes newlines and quotes
+    return s.replace("\n", "\\n").replace("\"", "\\\"")
+
 class SPDXPumlGraph():
 
     def _get_idx(self, _id):
@@ -32,7 +37,6 @@ def _get_idx(self, _id):
 
     def _get_idx_of_shaclobject(self, o):
         if o is None:
-            logging.error(f"None object")
             raise ValueError("None object")
         _id = o._id
         if o._id is None:
@@ -41,14 +45,26 @@ def _get_idx_of_shaclobject(self, o):
                 logging.warn(f"None id for {type(o)}, use {_id}")
         return self._get_idx(_id)
 
-    def _create_node(self, o):
-        idx = self._get_idx_of_shaclobject(o)
-        if idx in self.inserted:
+    def _get_idx_of_str(self, o):
+        if o is None:
+            raise ValueError("None str")
+        if o not in self.id_to_idx:
+            idx = hashlib.md5(o.encode()).hexdigest()
+            self.id_to_idx[o] = idx
+            self.idx_to_id[idx] = o
             return idx
-        self.inserted.add(idx)
+        else:
+            return self.id_to_idx[o]
 
-        self.logging.debug(f"Create node: {o._id} as {idx}: {o}")
+
+    def _create_node(self, o):
         if isinstance(o, spdx.SHACLObject):
+            idx = self._get_idx_of_shaclobject(o)
+            if idx in self.inserted:
+                return idx
+            self.inserted.add(idx)
+            self.logging.debug(f"Create node: {o._id} as {idx}: {o}")
+
             if isinstance(o, spdx.Element) and o.name is not None:
                 self.lines_defs.append(f"object \"<b>{o.name}</b>\\n{o.__class__.__name__}\" as {idx}")
             elif o._id is None:
@@ -83,9 +99,16 @@ def _create_node(self, o):
                         continue
 
                 if isinstance(value, str):
-                    self.lines.append(f"{idx} : {compact} = \"{value}\"")
+                    self.lines.append(f"{idx} : {compact} = \"{escape_string(value)}\"")
                 else:
                     self.lines.append(f"{idx} : {compact} = {value}")
+        elif isinstance(o, str):
+            idx = self._get_idx_of_str(o)
+            self.logging.debug(f"Create str node: {idx}")
+            self.lines_defs.append(f"object \"{escape_string(o)}\" as {idx}")
+        else:
+            raise ValueError(f"Unknown type {type(o)}")
+
         return idx