repair regexp for attributes extraction out of dbc #330

Author: ebroecker

src/canmatrix/formats/dbc.py

@@ -472,7 +472,7 @@ def add_frame_by_id(frame):  # type: (canmatrix.Frame) -> None
         if l.__len__() == 0:
             continue
         try:
-        #if 1==1:
+#        if 1==1:
             if followUp == FollowUps.signalComment:
                 try:
                     comment += "\n" + \
@@ -792,18 +792,18 @@ def add_frame_by_id(frame):  # type: (canmatrix.Frame) -> None
                 tempba = regexp.match(decoded)
 
                 if tempba.group(1).strip().startswith("BO_ "):
-                    regexp = re.compile(r"^BA_ +\"(.*)\" +BO_ +(.+) +(.+);")
+                    regexp = re.compile(r"^BA_ +\"(.+)\" +BO_ +(\d+) +(.+);")
                     temp = regexp.match(decoded)
                     get_frame_by_id(canmatrix.ArbitrationId.from_compound_integer(int(temp.group(2)))).add_attribute(
                         temp.group(1), temp.group(3))
                 elif tempba.group(1).strip().startswith("SG_ "):
-                    regexp = re.compile(r"^BA_ +\"(.*)\" +SG_ +(.+) +(.+) +(.+);")
+                    regexp = re.compile(r"^BA_ +\"(.+)\" +SG_ +(\d+) +(\w+) +(.+);")
                     temp = regexp.match(decoded)
                     if temp != None:
                         get_frame_by_id(canmatrix.ArbitrationId.from_compound_integer(int(temp.group(2)))).signal_by_name(
                             temp.group(3)).add_attribute(temp.group(1), temp.group(4))
                 elif tempba.group(1).strip().startswith("EV_ "):
-                    regexp = re.compile(r"^BA_ +\"(.*)\" +EV_ +(.+) +(.+);")
+                    regexp = re.compile(r"^BA_ +\"(.+)\" +EV_ +(\w+) +(.*);")
                     temp = regexp.match(decoded)
                     if temp != None:
                         db.add_env_attribute(temp.group(2),temp.group(1),temp.group(3))
@@ -867,7 +867,7 @@ def add_frame_by_id(frame):  # type: (canmatrix.Frame) -> None
                         signal.mux_val_min = muxValMin
                         signal.mux_val_max = muxValMax
             elif decoded.startswith("EV_ "):
-                pattern = r"^EV_ +([\w\-]+) *\: +([0-9]+) +\[([0-9.+\-eE]+)\|([0-9.+\-eE]+)\] +\"(\w*)\" +([0-9.+\-eE]+) +([0-9.+\-eE]+) +([\w\-]+) +(.*);"
+                pattern = r"^EV_ +([\w\-\_]+) *\: +([0-9]+) +\[([0-9.+\-eE]+)\|([0-9.+\-eE]+)\] +\"(.*?)\" +([0-9.+\-eE]+) +([0-9.+\-eE]+) +([\w\-]+) +(.*);"
                 regexp = re.compile(pattern)
                 temp = regexp.match(decoded)
 

src/canmatrix/tests/test_dbc.py

@@ -233,10 +233,14 @@ def test_defines_with_spaces():
 
     BO_ 123 someFrame: 1 someOtherEcu
 
+    EV_ someEnvVar: 0 [0|0] "" 0 2 DUMMY_NODE_VECTOR0 Vector__XXX;
+
     BA_DEF_ BU_ "Node Address" INT 0 255;
     BA_DEF_ BO_ "Period [ms]" INT 0 5000;
+    BA_DEF_ EV_ "some attrib" STRING;
     BA_ "Node Address" BU_ someOtherEcu 42;
     BA_ "Period [ms]" BO_ 123 3000;
+    BA_ "some attrib" EV_ someEnvVar "some space";
      ''').encode('utf-8'))
     matrix = canmatrix.formats.dbc.load(dbc, dbcImportEncoding="utf8")
     assert matrix.ecu_defines["Node Address"].type == "INT"
@@ -245,4 +249,17 @@ def test_defines_with_spaces():
     assert matrix.frame_defines["Period [ms]"].min == 0
     assert matrix.frame_defines["Period [ms]"].max == 5000
     assert matrix.frames[0].attributes["Period [ms]"] == '3000'
+    assert matrix.env_vars["someEnvVar"]["attributes"]["some attrib"] == '"some space"'
 
+def test_defines_with_special_cars():
+    dbc = io.BytesIO(textwrap.dedent(u'''\
+    BU_: someOtherEcu
+
+    BO_ 123 someFrame: 1 someOtherEcu
+     SG_ someSignal: 1|2@0+ (1,0) [0|0] ""  CCL_TEST
+
+    BA_DEF_ SG_ "Accuracy" STRING;
+    BA_ "Accuracy" SG_ 123 someSignal "+/- 10.2 at 55.1%";
+     ''').encode('utf-8'))
+    matrix = canmatrix.formats.dbc.load(dbc, dbcImportEncoding="utf8")
+    assert matrix.frames[0].signals[0].attributes["Accuracy"] == "+/- 10.2 at 55.1%"