Improve allegro/orcad import

7 files changed, 167 insertions, 49 deletions

diff --git a/gerbonara/cam.py b/gerbonara/cam.py
index e6bb651..14c1e8b 100644
--- a/gerbonara/cam.py
+++ b/gerbonara/cam.py
@@ -60,6 +60,9 @@ class FileSettings:
     #: If you want to export the macros with their original formulaic expressions (which is completely fine by the
     #: Gerber standard, btw), set this parameter to ``False`` before exporting.
     calculate_out_all_aperture_macros: bool = True
+    #: Internal field used to communicate if only decimal coordinates were found inside an Excellon file, or if it
+    #: contained at least some coordinates in fixed-width notation.
+    _file_has_fixed_width_coordinates: bool = False
 
     # input validation
     def __setattr__(self, name, value):
diff --git a/gerbonara/cli.py b/gerbonara/cli.py
index 93b989e..22aa098 100644
--- a/gerbonara/cli.py
+++ b/gerbonara/cli.py
@@ -431,7 +431,7 @@ def merge(inpath, outpath, offset, rotation, input_map, command_line_units, outp
 @click.option('--version', is_flag=True, callback=_print_version, expose_value=False, is_eager=True)
 @click.option('--warnings', 'format_warnings', type=click.Choice(['default', 'ignore', 'once']), default='default',
               help='''Enable or disable file format warnings during parsing (default: on)''')
-@click.option('--units', type=Unit(), help='Output bounding box in this unit (default: millimeter)')
+@click.option('--units', type=Unit(), default='metric', help='Output bounding box in this unit (default: millimeter)')
 @click.option('--input-number-format', help='Override number format of input file (mostly useful for Excellon files)')
 @click.option('--input-units', type=Unit(), help='Override units of input file')
 @click.option('--input-zero-suppression', type=click.Choice(['off', 'leading', 'trailing']), help='Override zero suppression setting of input file')
diff --git a/gerbonara/excellon.py b/gerbonara/excellon.py
index 9b75a69..73f9592 100755
--- a/gerbonara/excellon.py
+++ b/gerbonara/excellon.py
@@ -566,6 +566,8 @@ class ExcellonParser(object):
         self.filename = None
         self.external_tools = external_tools or {}
         self.found_kicad_format_comment = False
+        self.allegro_eof_toolchange_hack = False
+        self.allegro_eof_toolchange_hack_index = 1
 
     def warn(self, msg):
         warnings.warn(f'{self.filename}:{self.lineno} "{self.line}": {msg}', SyntaxWarning)
@@ -606,18 +608,25 @@ class ExcellonParser(object):
     exprs = RegexMatcher()
 
     # NOTE: These must be kept before the generic comment handler at the end of this class so they match first.
-    @exprs.match(r';T(?P<index1>[0-9]+) Holesize (?P<index2>[0-9]+)\. = (?P<diameter>[0-9/.]+) Tolerance = \+[0-9/.]+/-[0-9/.]+ (?P<plated>PLATED|NON_PLATED|OPTIONAL) (?P<unit>MILS|MM) Quantity = [0-9]+')
+    @exprs.match(r';(?P<index1_prefix>T(?P<index1>[0-9]+))?\s+Holesize (?P<index2>[0-9]+)\. = (?P<diameter>[0-9/.]+) Tolerance = \+[0-9/.]+/-[0-9/.]+ (?P<plated>PLATED|NON_PLATED|OPTIONAL) (?P<unit>MILS|MM) Quantity = [0-9]+')
     def parse_allegro_tooldef(self, match):
         # NOTE: We ignore the given tolerances here since they are non-standard.
         self.program_state = ProgramState.HEADER # TODO is this needed? we need a test file.
         self.generator_hints.append('allegro')
 
-        if (index := int(match['index1'])) != int(match['index2']): # index1 has leading zeros, index2 not.
+        index = int(match['index2'])
+
+        if match['index1'] and index != int(match['index1']): # index1 has leading zeros, index2 not.
             raise SyntaxError('BUG: Allegro excellon tool def has mismatching tool indices. Please file a bug report on our issue tracker and provide this file!')
 
         if index in self.tools:
             self.warn('Re-definition of tool index {index}, overwriting old definition.') 
 
+        if not match['index1_prefix']:
+            # This is a really nasty orcad file without tool change commands, that instead just puts all holes in order
+            # of the hole size definitions with M00's in between.
+            self.allegro_eof_toolchange_hack = True
+
         # NOTE: We map "optionally" plated holes to plated holes for API simplicity. If you hit a case where that's a
         # problem, please raise an issue on our issue tracker, explain why you need this and provide an example file.
         is_plated = None if match['plated'] is None else (match['plated'] in ('PLATED', 'OPTIONAL'))
@@ -630,13 +639,19 @@ class ExcellonParser(object):
         else:
             unit = MM
 
-        if unit != self.settings.unit:
+        if self.settings.unit is None:
+            self.settings.unit = unit
+
+        elif unit != self.settings.unit:
             self.warn('Allegro Excellon drill file tool definitions in {unit.name}, but file parameters say the '
                     'file should be in {settings.unit.name}. Please double-check that this is correct, and if it is, '
                     'please raise an issue on our issue tracker.')
 
         self.tools[index] = ExcellonTool(diameter=diameter, plated=is_plated, unit=unit)
 
+        if self.allegro_eof_toolchange_hack and self.active_tool is None:
+            self.active_tool = self.tools[index]
+
     # Searching Github I found that EasyEDA has two different variants of the unit specification here.
     @exprs.match(';Holesize (?P<index>[0-9]+) = (?P<diameter>[.0-9]+) (?P<unit>INCH|inch|METRIC|mm)')
     def parse_easyeda_tooldef(self, match):
@@ -753,6 +768,12 @@ class ExcellonParser(object):
     def handle_end_of_program(self, match):
         if self.program_state in (None, ProgramState.HEADER):
             self.warn('M30 statement found before end of header.')
+
+        if self.allegro_eof_toolchange_hack:
+            self.allegro_eof_toolchange_hack_index = min(max(self.tools), self.allegro_eof_toolchange_hack_index + 1)
+            self.active_tool = self.tools[self.allegro_eof_toolchange_hack_index]
+            return
+
         self.program_state = ProgramState.FINISHED
         # TODO: maybe add warning if this is followed by other commands.
 
@@ -762,14 +783,17 @@ class ExcellonParser(object):
     def do_move(self, coord_groups):
         x_s, x, y_s, y = coord_groups
 
-        if self.settings.number_format == (None, None) and '.' not in x:
-            # TARGET3001! exports zeros as "00" even when it uses an explicit decimal point everywhere else.
-            if x != '00':
-                raise SyntaxError('No number format set and value does not contain a decimal point. If this is an Allegro '
-                    'Excellon drill file make sure either nc_param.txt or ncdrill.log ends up in the same folder as '
-                    'it, because Allegro does not include this critical information in their Excellon output. If you '
-                    'call this through ExcellonFile.from_string, you must manually supply from_string with a '
-                    'FileSettings object from excellon.parse_allegro_ncparam.')
+        if '.' not in x:
+            self.settings._file_has_fixed_width_coordinates = True
+
+            if self.settings.number_format == (None, None):
+                # TARGET3001! exports zeros as "00" even when it uses an explicit decimal point everywhere else.
+                if x != '00':
+                    raise SyntaxError('No number format set and value does not contain a decimal point. If this is an Allegro '
+                        'Excellon drill file make sure either nc_param.txt or ncdrill.log ends up in the same folder as '
+                        'it, because Allegro does not include this critical information in their Excellon output. If you '
+                        'call this through ExcellonFile.from_string, you must manually supply from_string with a '
+                        'FileSettings object from excellon.parse_allegro_ncparam.')
 
         x = self.settings.parse_gerber_value(x)
         if x_s:
diff --git a/gerbonara/layer_rules.py b/gerbonara/layer_rules.py
index aae52ae..eb3e0e5 100644
--- a/gerbonara/layer_rules.py
+++ b/gerbonara/layer_rules.py
@@ -82,6 +82,7 @@ MATCH_RULES = {
     'bottom paste':     r'.*_boardoutline\.\w+', # FIXME verify this
     'drill plated':     r'.*\.(drl)', # diptrace has unplated drills on the outline layer
     'other netlist':    r'.*\.ipc', # default rule due to lack of tool-specific examples
+    'header regex':     [['sufficient', r'top .*|bottom .*', r'G04 DipTrace [.-0-9a-z]*\*']],
     },
 
 'target': {
@@ -151,22 +152,25 @@ MATCH_RULES = {
 
 'allegro': {
     # Allegro doesn't have any widespread convention, so we rely heavily on the layer name auto-guesser here.
-    'drill mech': r'.*\.(drl|rou)',
-    'generic gerber': r'.*\.art',
+    'drill plated':     r'.*\.(drl)',
+    'drill nonplated':  r'.*\.(rou)',
+    'other unknown':    r'.*(place|assembly|keep.?in|keep.?out).*\.art',
+    'autoguess':        r'.*\.art',
     'excellon params':  r'nc_param\.txt|ncdrill\.log|ncroute\.log',
     'other netlist':    r'.*\.ipc', # default rule due to lack of tool-specific examples
+    'header regex':     [['required,sufficient', r'.*\.art', r'G04 File Origin:\s+Cadence Allegro [0-9]+\.[0-9]+[-a-zA-Z0-9]*']],
     },
 
 'pads': {
     # Pads also does not seem to have a factory-default naming schema. Or it has one but everyone ignores it.
-    'generic gerber':   r'.*\.pho',
-    'drill mech':       r'.*\.drl',
+    'autoguess':        r'.*\.pho',
+    'drill plated':     r'.*\.drl',
     },
 
 'zuken': {
-    'generic gerber': r'.*\.fph',
+    'autoguess': r'.*\.fph',
     'gerber params': r'.*\.fpl',
-    'drill mech': r'.*\.fdr',
+    'drill unknown': r'.*\.fdr',
     'excellon params': r'.*\.fdl',
     'other netlist': r'.*\.ipc',
     'ipc-2581': r'.*\.xml',
diff --git a/gerbonara/layers.py b/gerbonara/layers.py
index 4b9d360..e200e62 100644
--- a/gerbonara/layers.py
+++ b/gerbonara/layers.py
@@ -112,31 +112,61 @@ class NamingScheme:
     }

 

 

+def apply_rules(filenames, rules):

+    certain = False

+    gen = {}

+    already_matched = set()

+    header_regex = rules.pop('header regex', [])

+    header_regex_matched = [False] * len(header_regex)

+

+    file_headers = {}

+    def get_header(path):

+        if path not in file_headers:

+            with open(path) as f:

+                file_headers[path] = f.read(16384)

+        return file_headers[path]

+

+    for layer, regex in rules.items():

+        for fn in filenames:

+            if fn in already_matched:

+                continue

 

-def _match_files(filenames):

-    matches = {}

-    for generator, rules in MATCH_RULES.items():

-        already_matched = set()

-        gen = {}

-        matches[generator] = gen

-        for layer, regex in rules.items():

-            for fn in filenames:

-                if fn in already_matched:

-                    continue

+            target = None

+            if (m := re.fullmatch(regex, fn.name, re.IGNORECASE)):

+                if layer == 'inner copper':

+                    target = 'inner_' + ''.join(e or '' for e in m.groups()) + ' copper'

+                else:

+                    target = layer

 

-                if (m := re.fullmatch(regex, fn.name, re.IGNORECASE)):

-                    if layer == 'inner copper':

-                        target = 'inner_' + ''.join(e or '' for e in m.groups()) + ' copper'

-                    else:

-                        target = layer

+                gen[target] = gen.get(target, []) + [fn]

+                already_matched.add(fn)

 

-                    gen[target] = gen.get(target, []) + [fn]

-                    already_matched.add(fn)

-    return matches

+            for i, (match_type, layer_match, header_match) in enumerate(header_regex):

+                if re.fullmatch(layer_match, fn.name, re.IGNORECASE) or (

+                        target is not None and re.fullmatch(layer_match, target, re.IGNORECASE)):

+                    if re.search(header_match, get_header(fn)):

 

+                        if 'sufficient' in match_type:

+                            certain = True

+

+                        header_regex_matched[i] = True

+

+    if any('required' in match_type and not match

+           for match, (match_type, *_) in zip(header_regex_matched, header_regex)):

+        return False, {}

+

+    return certain, gen

 

 def _best_match(filenames):

-    matches = _match_files(filenames)

+    matches = {}

+    for generator, rules in MATCH_RULES.items():

+        certain, candidate = apply_rules(filenames, rules)

+

+        if certain:

+            return generator, candidate

+

+        matches[generator] = candidate

+

     matches = sorted(matches.items(), key=lambda pair: len(pair[1]))

     generator, files = matches[-1]

     return generator, files

@@ -243,7 +273,7 @@ def _layername_autoguesser(fn):
     elif re.search('film', fn):

         use = 'copper'

 

-    elif re.search('out(line)?', fn):

+    elif re.search('out(line)?|board.?geometry', fn):

         use = 'outline'

         side = 'mechanical'

 

@@ -385,7 +415,7 @@ class LayerStack:
         with ZipFile(file) as f:

             f.extractall(path=tmp_indir)

 

-        inst = kls.open_dir(tmp_indir, board_name=board_name, lazy=lazy)

+        inst = kls.open_dir(tmp_indir, board_name=board_name, lazy=lazy, overrides=overrides, autoguess=autoguess)

         inst.tmpdir = tmpdir

         inst.original_path = Path(original_path or file)

         inst.was_zipped = True

@@ -445,6 +475,11 @@ class LayerStack:
                                 filemap[layer].remove(fn)

                             filemap[layer] = filemap.get(layer, []) + [fn]

 

+        if 'autoguess' in filemap:

+            warnings.warn(f'This generator ({generator}) often exports ambiguous filenames. Falling back to autoguesser for some files. Use at your own peril.')

+            for key, values in _do_autoguess(filemap.pop('autoguess')).items():

+                filemap[key] = filemap.get(key, []) + values

+

         if sum(len(files) for files in filemap.values()) < 6 and autoguess:

             warnings.warn('Ambiguous gerber filenames. Trying last-resort autoguesser.')

             generator = None

@@ -453,6 +488,8 @@ class LayerStack:
                 raise ValueError('Cannot figure out gerber file mapping. Partial map is: ', filemap)

 

         excellon_settings, external_tools = None, None

+        automatch_drill_scale = False

+

         if generator == 'geda':

             # geda is written by geniuses who waste no bytes of unnecessary output so it doesn't actually include the

             # number format in files that use imperial units. Unfortunately it also doesn't include any hints that the

@@ -470,16 +507,18 @@ class LayerStack:
                     if (external_tools := parse_allegro_logfile(file.read_text())):

                         break

                 del filemap['excellon params']

-            # Ignore if we can't find the param file -- maybe the user has convinced Allegro to actually put this

-            # information into a comment, or maybe they have made Allegro just use decimal points like XNC does.

+            else:

+                # Ignore if we can't find the param file -- maybe the user has convinced Allegro to actually put this

+                # information into a comment, or maybe they have made Allegro just use decimal points like XNC does.

+                # We'll run an automatic scale matching later.

+                excellon_settings = FileSettings(number_format=(2, 4))

+                automatch_drill_scale = True

 

-            filemap = _do_autoguess([ f for files in filemap.values() for f in files ])

             if len(filemap) < 6:

                 raise SystemError('Cannot figure out gerber file mapping')

             # FIXME use layer metadata from comments and ipc file if available

 

         elif generator == 'zuken':

-            filemap = _do_autoguess([ f for files in filemap.values() for f in files ])

             if len(filemap) < 6:

                 raise SystemError('Cannot figure out gerber file mapping')

             # FIXME use layer metadata from comments and ipc file if available

@@ -503,7 +542,9 @@ class LayerStack:
         else:

             excellon_settings = None

 

-        ambiguous = [ f'{key} ({", ".join(x.name for x in value)})' for key, value in filemap.items() if len(value) > 1 and not 'drill' in key ]

+        ambiguous = [ f'{key} ({", ".join(x.name for x in value)})'

+                     for key, value in filemap.items()

+                     if len(value) > 1 and not 'drill' in key and not key == 'other unknown']

         if ambiguous:

             raise SystemError(f'Ambiguous layer names: {", ".join(ambiguous)}')

 

@@ -512,8 +553,8 @@ class LayerStack:
         netlist = None

         layers = {} # { tuple(key.split()): None for key in STANDARD_LAYERS }

         for key, paths in filemap.items():

-            if len(paths) > 1 and not 'drill' in key:

-                raise ValueError(f'Multiple matching files found for {key} layer: {", ".join(value)}')

+            if len(paths) > 1 and not 'drill' in key and not key == 'other unknown':

+                raise ValueError(f'Multiple matching files found for {key} layer: {", ".join(map(str, value))}')

 

             for path in paths:

                 id_result = identify_file(path.read_text())

@@ -574,6 +615,35 @@ class LayerStack:
             board_name = re.sub(r'^\W+', '', board_name)

             board_name = re.sub(r'\W+$', '', board_name)

 

+        if automatch_drill_scale:

+            top_copper = layers[('top', 'copper')].to_excellon(errors='ignore', holes_only=True)

+

+            # precision is matching precision in mm

+            def map_coords(obj, precision=0.01, scale=1):

+                obj = obj.converted(MM)

+                return round(obj.x*scale/precision), round(obj.y*scale/precision)

+

+            aper_coords = {map_coords(obj) for obj in top_copper.drills()}

+

+            for drill_file in [drill_pth, drill_npth, *drill_layers]:

+                if not drill_file or not drill_pth.import_settings._file_has_fixed_width_coordinates:

+                    continue

+

+                scale_matches = {}

+                for exp in range(-6, 6):

+                    scale = 10**exp

+                    hole_coords = {map_coords(obj, scale=scale) for obj in drill_file.drills()}

+

+                    scale_matches[scale] = len(aper_coords - hole_coords), len(hole_coords - aper_coords)

+                scales_out = [(max(a, b), scale) for scale, (a, b) in scale_matches.items()]

+                _matches, scale = sorted(scales_out)[0]

+                warnings.warn(f'Performing automatic alignment of poorly exported drill layer. Scale matching results: {scale_matches}. Chosen scale: {scale}')

+

+                # Note: This is only used with allegro files, which use decimal points and explicit units in their tool

+                # definitions. Thus, we only scale object coordinates, and not apertures.

+                for obj in drill_file.objects:

+                    obj.scale(scale)

+

         return kls(layers, drill_pth, drill_npth, drill_layers, board_name=board_name,

                 original_path=original_path, was_zipped=was_zipped, generator=[*all_generator_hints, None][0])

 

diff --git a/gerbonara/rs274x.py b/gerbonara/rs274x.py
index fd83c5b..108ccd6 100644
--- a/gerbonara/rs274x.py
+++ b/gerbonara/rs274x.py
@@ -152,7 +152,7 @@ class GerberFile(CamFile):
 
         self.map_apertures(lookup)
 
-    def to_excellon(self, plated=None, errors='raise'):
+    def to_excellon(self, plated=None, errors='raise', holes_only=False):
         """ Convert this excellon file into a :py:class:`~.excellon.ExcellonFile`. This will convert interpolated lines
         into slots, and circular aperture flashes into holes. Other features such as ``G36`` polygons or flashes with
         non-circular apertures will result in a :py:obj:`ValueError`. You can, of course, programmatically remove such
@@ -160,7 +160,10 @@ class GerberFile(CamFile):
         new_objs = []
         new_tools = {}
         for obj in self.objects:
-            if not (isinstance(obj, go.Line) or isinstance(obj, go.Arc) or isinstance(obj, go.Flash)) or \
+            if holes_only and not isinstance(obj, go.Flash):
+                continue
+
+            if not isinstance(obj, (go.Line, go.Arc, go.Flash)) or \
                 not isinstance(getattr(obj, 'aperture', None), apertures.CircleAperture):
                     if errors == 'raise':
                         raise ValueError(f'Cannot convert {obj} to excellon.')
diff --git a/gerbonara/tests/test_layers.py b/gerbonara/tests/test_layers.py
index a5103de..b473f91 100644
--- a/gerbonara/tests/test_layers.py
+++ b/gerbonara/tests/test_layers.py
@@ -302,7 +302,21 @@ REFERENCE_DIRS = {
         'Drill/8seg_Driver__routed_Drill_thru_plt.fdr/8seg_Driver__routed_Drill_thru_plt.fdl': None,
         'Drill/8seg_Driver__routed_Drill_thru_plt.fdr/8seg_Driver__routed_Drill_thru_plt.fdr': 'drill plated',
         'Drill/8seg_Driver__routed_Drill_thru_nplt.fdr': 'drill nonplated',
-        }
+        },
+    'orcad': {
+        'Assembly.art': None,
+        'BOTTOM.art': 'bottom copper',
+        'GND2.art': 'inner_3 copper',
+        'LAYER_1.art': 'inner_2 copper',
+        'LAYER_2.art': 'inner_4 copper',
+        'PWR.art': 'inner_2 copper',
+        'Solder_Mask_Bottom.art': 'bottom mask',
+        'Solder_Mask_Top.art': 'top mask',
+        'TOP.art': 'top copper',
+        'arena_12-12_v6_L1-L6.drl': 'drill plated',
+        'silk_screen_bottom.art': 'bottom silk',
+        'silk_screen_top.art': 'top silk',
+        },
     }
 
 @filter_syntax_warnings