More parsing speedups

gerbonara/cam.py

@@ -141,22 +141,13 @@ class FileSettings:
 
         integer_digits, decimal_digits = self.number_format
 
-        sign = 1
-
-        if value[0] == '-':
-            sign = -1
-            value = value[1:]
-
-        elif value[0] == '+':
-            value = value[1:]
-
         if self.zeros == 'leading':
             value = self._pad + value # pad with zeros to ensure we have enough decimals
-            return sign*float(value[:-decimal_digits] + '.' + value[-decimal_digits:])
+            return float(value[:-decimal_digits] + '.' + value[-decimal_digits:])
 
         else: # no or trailing zero suppression
             value = value + self._pad
-            return sign*float(value[:integer_digits] + '.' + value[integer_digits:])
+            return float(value[:integer_digits] + '.' + value[integer_digits:])
 
     def write_gerber_value(self, value, unit=None):
         """ Convert a floating point number to a Gerber-formatted string.  """

gerbonara/excellon.py

@@ -657,19 +657,24 @@ class ExcellonParser(object):
         else:
             self.active_tool = self.tools[index]
 
-    coord = lambda name, key=None: fr'({name}(?P<{key or name}>[+-]?[0-9]*\.?[0-9]*))?'
+    coord = lambda name: fr'(?:{name}\+?(-?)([0-9]*\.?[0-9]*))?'
     xy_coord = coord('X') + coord('Y')
     xyaij_coord = xy_coord + coord('A') + coord('I') + coord('J')
 
-    @exprs.match(r'R(?P<count>[0-9]+)' + xy_coord)
+    @exprs.match(r'R([0-9]+)' + xy_coord)
     def handle_repeat_hole(self, match):
         if self.program_state == ProgramState.HEADER:
             return
 
-        dx = int(match['X'] or '0')
-        dy = int(match['Y'] or '0')
+        count, x_s, x, y_s, y = match.groups()
+        dx = self.settings.parse_gerber_value(x) or 0
+        if x_s:
+            dx = -dx
+        dy = self.settings.parse_gerber_value(y) or 0
+        if y_s:
+            dy = -dy
 
-        for i in range(int(match['count'])):
+        for i in range(int(count)):
             self.pos = (self.pos[0] + dx, self.pos[1] + dy)
             # FIXME fix API below
             if not self.ensure_active_tool():
@@ -724,18 +729,24 @@ class ExcellonParser(object):
         if match[0] == 'M00':
             self.generator_hints.append('zuken')
 
-    def do_move(self, match=None, x='X', y='Y'):
-        if self.settings.number_format == (None, None) and not '.' in match['X']:
+    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 match['X'] != '00':
+            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(match['X'])
-        y = self.settings.parse_gerber_value(match['Y'])
+        x = self.settings.parse_gerber_value(x)
+        if x_s:
+            x = -x
+        y = self.settings.parse_gerber_value(y)
+        if y_s:
+            y = -y
 
         old_pos = self.pos
 
@@ -757,7 +768,7 @@ class ExcellonParser(object):
         if self.program_state is None:
             self.warn('Routing mode command found before header.')
         self.program_state = ProgramState.ROUTING
-        self.do_move(match)
+        self.do_move(match.groups())
 
     @exprs.match('%')
     def handle_rewind_shorthand(self, match):
@@ -779,25 +790,26 @@ class ExcellonParser(object):
         self.warn('Routing command found before first tool definition.')
         return None
 
-    @exprs.match('(?P<mode>G01|G02|G03)' + xyaij_coord)
+    @exprs.match('(G01|G02|G03)' + xyaij_coord)
     def handle_linear_mode(self, match):
-        if match['mode'] == 'G01':
+        mode, *coord_groups = match.groups()
+        if mode == 'G01':
             self.interpolation_mode = InterpMode.LINEAR
         else:
-            clockwise = (match['mode'] == 'G02')
+            clockwise = (mode == 'G02')
             self.interpolation_mode = InterpMode.CIRCULAR_CW if clockwise else InterpMode.CIRCULAR_CCW
 
-        self.do_interpolation(match)
+        self.do_interpolation(coord_groups)
     
-    def do_interpolation(self, match):
-        x, y, a, i, j = match['X'], match['Y'], match['A'], match['I'], match['J']
+    def do_interpolation(self, coord_groups):
+        x_s, x, y_s, y, a_s, a, i_s, i, j_s, j = coord_groups
 
-        start, end = self.do_move(match)
+        start, end = self.do_move((x_s, x, y_s, y))
 
         if self.program_state != ProgramState.ROUTING:
             return
 
-        if not self.drill_down or not (match['X'] or match['Y']) or not self.ensure_active_tool():
+        if not self.drill_down or not (x or y) or not self.ensure_active_tool():
             return
 
         if self.interpolation_mode == InterpMode.LINEAR:
@@ -819,6 +831,8 @@ class ExcellonParser(object):
                 # Convert endpoint-radius-endpoint notation to endpoint-center-endpoint notation. We always use the
                 # smaller arc here.
                 # from https://math.stackexchange.com/a/1781546
+                if a_s:
+                    raise ValueError('Negative arc radius given')
                 r = settings.parse_gerber_value(a)
                 x1, y1 = start
                 x2, y2 = end
@@ -835,7 +849,11 @@ class ExcellonParser(object):
 
             else: # explicit center given
                 i = settings.parse_gerber_value(i)
+                if i_s:
+                    i = -i
                 j = settings.parse_gerber_value(j)
+                if j_s:
+                    j = -i
 
             self.objects.append(Arc(*start, *end, i, j, True, self.active_tool, unit=self.settings.unit))
 
@@ -865,7 +883,8 @@ class ExcellonParser(object):
 
     @exprs.match('G93' + xy_coord)
     def handle_absolute_mode(self, match):
-        if int(match['X'] or 0) != 0 or int(match['Y'] or 0) != 0:
+        _x_s, x, _y_s, y = match.groups()
+        if int(x or 0) != 0 or int(y or 0) != 0:
             # Siemens tooling likes to include a meaningless G93X0Y0 after its header.
             raise NotImplementedError('G93 zero set command is not supported.')
         self.generator_hints.append('siemens')
@@ -890,23 +909,11 @@ class ExcellonParser(object):
     def handle_unhandled(self, match):
         self.warn(f'{match[0]} excellon command intended for CAM tools found in EDA file.')
 
-    @exprs.match(coord('X', 'x1') + coord('Y', 'y1') + 'G85' + coord('X', 'x2') + coord('Y', 'y2'))
-    def handle_slot_dotted(self, match):
-        self.warn('Weird G85 excellon slot command used. Please raise an issue on our issue tracker and provide this file for testing.')
-        self.do_move(match, 'X1', 'Y1')
-        start, end = self.do_move(match, 'X2', 'Y2')
-        
-        if self.program_state in (ProgramState.DRILLING, ProgramState.HEADER): # FIXME should we realy handle this in header?
-            if self.ensure_active_tool():
-                # We ignore whether a slot is a "routed" G00/G01 slot or a "drilled" G85 slot and export both as routed
-                # slots.
-                self.objects.append(Line(*start, *end, self.active_tool, unit=self.settings.unit))
-
     @exprs.match(xyaij_coord)
     def handle_bare_coordinate(self, match):
         # Yes, drills in the header doesn't follow the specification, but it there are many files like this.
         if self.program_state in (ProgramState.DRILLING, ProgramState.HEADER):
-            _start, end = self.do_move(match)
+            _start, end = self.do_move(match.groups()[:4])
 
             if not self.ensure_active_tool():
                 return
@@ -916,7 +923,7 @@ class ExcellonParser(object):
         elif self.program_state == ProgramState.ROUTING:
             # Bare coordinates for routing also seem illegal, but Siemens actually uses these.
             # Example file: siemens/80101_0125_F200_ContourPlated.ncd
-            self.do_interpolation(match)
+            self.do_interpolation(match.groups())
 
         else:
             self.warn('Bare coordinate after end of file')

gerbonara/rs274x.py

@@ -292,7 +292,6 @@ class GraphicsState:
         self.polarity_dark = True
         self.point = None
         self.aperture = None
-        self.file_settings = None
         self.interpolation_mode = InterpMode.LINEAR
         self.multi_quadrant_mode = None # used only for syntax checking
         self.aperture_mirroring = (False, False) # LM mirroring (x, y)
@@ -305,6 +304,7 @@ class GraphicsState:
         self.image_scale = (1.0, 1.0) # SF image scaling (x, y); deprecated
         self._mat = None
         self.file_settings = file_settings
+        self.unit = file_settings.unit if file_settings else None
         self.aperture_map = aperture_map or {}
         self.warn = warn
         self.unit_warning = False
@@ -362,27 +362,23 @@ class GraphicsState:
             return rx, ry
 
     def flash(self, x, y):
-        if self.file_settings.unit is None and not self.unit_warning:
-            self.warn('Gerber file does not contain a unit definition.')
-            self.unit_warning = True
+        if self.unit is None:
+            raise SyntaxError('Gerber file does not contain a unit definition.')
         self.update_point_native(x, y)
         obj = go.Flash(*self.map_coord(*self.point), self.aperture,
                 polarity_dark=self._polarity_dark,
-                unit=self.file_settings.unit,
+                unit=self.unit,
                 attrs=self.object_attrs)
         return obj
 
     def interpolate(self, x, y, i=None, j=None, aperture=True, multi_quadrant=False):
         old_point = self.map_coord(*self.update_point_native(x, y))
-        unit = self.file_settings.unit
 
-        if not self.unit_warning and unit is None:
-            self.warn('Gerber file does not contain a unit definition.')
-        self.unit_warning = True
+        if (unit := self.unit) is None:
+            raise SyntaxError('Gerber file does not contain a unit definition.')
 
         if aperture:
-            aperture = self.aperture
-            if not aperture:
+            if (aperture := self.aperture) is None:
                 raise SyntaxError('Interpolation attempted without selecting aperture first')
 
             if math.isclose(aperture.equivalent_width(), 0):
@@ -401,7 +397,6 @@ class GraphicsState:
                     polarity_dark=self._polarity_dark, unit=unit, attrs=self.object_attrs)
 
         else:
-
             if i is None and j is None:
                 self.warn('Linear segment implied during arc interpolation mode through D01 w/o I, J values')
                 return go.Line(*old_point, *self.map_coord(*self.point), aperture,
@@ -508,8 +503,8 @@ class GerberParser:
     NAME = r"[a-zA-Z_$\.][a-zA-Z_$\.0-9+\-]+"
 
     STATEMENT_REGEXES = {
-        'coord': fr"(G0?[123]|G74|G75|G54|G55)?\s*(?:X({NUMBER}))?(?:Y({NUMBER}))?" \
-            fr"(?:I({NUMBER}))?(?:J({NUMBER}))?\s*" \
+        'coord': fr"(G0?[123]|G74|G75|G54|G55)?\s*(?:X\+?(-?)({NUMBER}))?(?:Y\+?(-?)({NUMBER}))?" \
+            fr"(?:I\+?(-?)({NUMBER}))?(?:J\+?(-?)({NUMBER}))?\s*" \
             fr"(?:D0?([123]))?$",
         'region_start': r'G36$',
         'region_end': r'G37$',
@@ -578,14 +573,16 @@ class GerberParser:
     def _split_commands(self, data):
         # Ignore '%' signs within G04 commments because eagle likes to put completely broken file attributes inside G04
         # comments, and those contain % signs. Best of all, they're not even balanced.
-        self.lineno = 0
-        for match in re.finditer(r'G04.*?\*|%.*?%|[^*%]*\*', data, re.DOTALL):
-            cmd = match[0].strip().strip('%').rstrip('*')
+        self.lineno = 1
+        for match in re.finditer(r'G04.*?\*\s*|%.*?%\s*|[^*%]*\*\s*', data, re.DOTALL):
+            cmd = match[0]
+            newlines = cmd.count('\n')
+            cmd = cmd.strip().strip('%').rstrip('*')
             if cmd:
                 # Expensive, but only used in case something goes wrong.
                 self.line = cmd
                 yield cmd
-            self.lineno += cmd.count('\n')
+            self.lineno += newlines
         self.lineno = 0
         self.line = ''
 
@@ -622,7 +619,7 @@ class GerberParser:
                     self.warn('File is missing mandatory M02 EOF marker. File may be truncated.')
 
     def _parse_coord(self, match):
-        interp, x, y, i, j, op = match.groups() # faster than name-based group access
+        interp, x_s, x, y_s, y, i_s, i, j_s, j, op = match.groups() # faster than name-based group access
         has_coord = x or y or i or j
 
         if not interp:
@@ -647,7 +644,11 @@ class GerberParser:
             self.generator_hints.append('zuken')
 
         x = self.file_settings.parse_gerber_value(x)
+        if x_s:
+            x = -x
         y = self.file_settings.parse_gerber_value(y)
+        if y_s:
+            y = -y
 
         if not op and has_coord:
             if self.last_operation == '1':
@@ -680,17 +681,19 @@ class GerberParser:
                     self.warn('Deprecated G74 multi-quadant mode arc found. G74 is bad and you should feel bad.')
 
             i = self.file_settings.parse_gerber_value(i)
+            if i_s:
+                i = -i
             j = self.file_settings.parse_gerber_value(j)
+            if j_s:
+                j = -j
 
             if self.current_region is None:
                 # in multi-quadrant mode this may return None if start and end point of the arc are the same.
-                obj = self.graphics_state.interpolate(x, y, i, j,
-                        multi_quadrant=bool(self.multi_quadrant_mode))
+                obj = self.graphics_state.interpolate(x, y, i, j, multi_quadrant=self.multi_quadrant_mode)
                 if obj is not None:
                     self.target.objects.append(obj)
             else:
-                obj = self.graphics_state.interpolate(x, y, i, j, aperture=False,
-                        multi_quadrant=bool(self.multi_quadrant_mode))
+                obj = self.graphics_state.interpolate(x, y, i, j, aperture=False, multi_quadrant=self.multi_quadrant_mode)
                 if obj is not None:
                     self.current_region.append(obj)
 
@@ -774,9 +777,9 @@ class GerberParser:
 
     def _parse_unit_mode(self, match):
         if match['unit'] == 'MM':
-            self.file_settings.unit = MM
+            self.graphics_state.unit = self.file_settings.unit = MM
         else:
-            self.file_settings.unit = Inch
+            self.graphics_state.unit = self.file_settings.unit = Inch
 
     def _parse_allegro_format_spec(self, match):
         self._parse_format_spec(match)
@@ -921,7 +924,7 @@ class GerberParser:
         self.current_region = None
 
     def _parse_old_unit(self, match):
-        self.file_settings.unit = Inch if match['mode'] == 'G70' else MM
+        self.graphics_state.unit = self.file_settings.unit = Inch if match['mode'] == 'G70' else MM
         self.warn(f'Deprecated {match["mode"]} unit mode statement found. This deprecated since 2012.', DeprecationWarning)
         self.target.comments.append('Replaced deprecated {match["mode"]} unit mode statement with MO statement')