Make split_commands faster

4 files changed, 94 insertions, 8 deletions

diff --git a/examples/bench.py b/examples/bench.py
new file mode 100644
index 0000000..b8d8c36
--- /dev/null
+++ b/examples/bench.py
@@ -0,0 +1,60 @@
+#!/usr/bin/env python3
+
+import time
+from pathlib import Path
+
+import tqdm
+
+import gerbonara
+
+if __name__ == '__main__':
+    resources = Path(__file__).parent.parent / 'gerbonara' / 'tests' / 'resources'
+
+    TEST_FILES = [
+        'easyeda/Gerber_TopSilkLayer.GTO',
+        'allegro-2/MinnowMax_RevA1_GAF_Gerber/MinnowMax_lyr3.art',
+        'allegro-2/MinnowMax_RevA1_GAF_Gerber/MinnowMax_fab.art',
+        'eagle-newer/soldermask_bottom.gbr',
+        'eagle-newer/copper_bottom.gbr',
+# FIXME remove redundant warnings in these files
+#        'siemens/80101_0125_F200_SilkscreenBottom.gdo',
+#        'siemens/80101_0125_F200_SoldermaskBottom.gdo',
+#        'siemens/80101_0125_F200_SolderPasteBottom.gdo',
+#        'siemens/80101_0125_F200_L03.gdo',
+#        'siemens/80101_0125_F200_L01_Top.gdo',
+#        'zuken-emulated/Gerber/Symbol-A.fph',
+        'Target3001/IRNASIoTbank1.2.StopTop',
+        'Target3001/IRNASIoTbank1.2.PasteTop',
+        'Target3001/IRNASIoTbank1.2.PasteBot',
+        'Target3001/IRNASIoTbank1.2.StopBot',
+        'Target3001/IRNASIoTbank1.2.Top',
+        'pcb-rnd/power-art.gtl',
+        'pcb-rnd/power-art.gto',
+        'pcb-rnd/power-art.gtp',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G1',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G2',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G3',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G4',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G5',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G6',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G7',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G8',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G9',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G10',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G11',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.G12',
+        'altium-composite-drill/Gerber/LimeSDR-QPCIe_1v2.GBS',
+        'eagle_files/copper_top_l1.gbr',
+        'eagle_files/soldermask_top.gbr',
+        'diptrace/mainboard_Bottom.gbr',
+        'upverter/design_export.gtp',
+        'upverter/design_export.gbl',
+        ]
+
+    start = time.perf_counter()
+    for file in TEST_FILES: #tqdm.tqdm(TEST_FILES):
+        gerbonara.GerberFile.open(resources / file)
+    end = time.perf_counter()
+
+    print(f'Duration: {(end - start)*1000:.3f} ms')
+
diff --git a/examples/load_directory.py b/examples/load_directory.py
new file mode 100644
index 0000000..6e1f901
--- /dev/null
+++ b/examples/load_directory.py
@@ -0,0 +1,11 @@
+#!/usr/bin/env python3
+
+if __name__ == '__main__':
+    import argparse
+    parser = argparse.ArgumentParser()
+    parser.add_argument('input', help='Input Gerber directory')
+    args = parser.parse_args()
+
+    import gerbonara
+    print(gerbonara.LayerStack.from_directory(args.input))
+
diff --git a/gerbonara/excellon.py b/gerbonara/excellon.py
index 5f0db44..721fcca 100755
--- a/gerbonara/excellon.py
+++ b/gerbonara/excellon.py
@@ -535,6 +535,7 @@ class ExcellonParser(object):
         self.lineno = None
         self.filename = None
         self.external_tools = external_tools or {}
+        self.found_kicad_format_comment = False
 
     def warn(self, msg):
         warnings.warn(f'{self.filename}:{self.lineno} "{self.line}": {msg}', SyntaxWarning)
@@ -844,7 +845,7 @@ class ExcellonParser(object):
             integer, _, fractional = match[3][1:].partition('.')
             self.settings.number_format = len(integer), len(fractional)
 
-        elif self.settings.number_format == (None, None) and not metric:
+        elif self.settings.number_format == (None, None) and not metric and not self.found_kicad_format_comment:
             self.warn('Using implicit number format from bare "INCH" statement. This is normal for Fritzing, Diptrace, Geda and pcb-rnd.')
             self.settings.number_format = (2,4)
     
@@ -981,6 +982,7 @@ class ExcellonParser(object):
         x = None if x == '-' else int(x)
         y = None if y == '-' else int(y)
         self.settings.number_format = x, y
+        self.found_kicad_format_comment = True
         self.settings.notation = match[2]
         self.settings.unit = Inch if match[3] == 'inch' else MM
 
diff --git a/gerbonara/rs274x.py b/gerbonara/rs274x.py
index 82ee047..e29ecd8 100644
--- a/gerbonara/rs274x.py
+++ b/gerbonara/rs274x.py
@@ -520,11 +520,11 @@ class GerberParser:
     NAME = r"[a-zA-Z_$\.][a-zA-Z_$\.0-9+\-]+"
 
     STATEMENT_REGEXES = {
-        'region_start': r'G36$',
-        'region_end': r'G37$',
         'coord': fr"(?P<interpolation>G0?[123]|G74|G75|G54|G55)?(X(?P<x>{NUMBER}))?(Y(?P<y>{NUMBER}))?" \
             fr"(I(?P<i>{NUMBER}))?(J(?P<j>{NUMBER}))?" \
             fr"(?P<operation>D0?[123])?$",
+        'region_start': r'G36$',
+        'region_end': r'G37$',
         'aperture': r"(G54|G55)?D(?P<number>\d+)",
         # Allegro combines format spec and unit into one long illegal extended command.
         'allegro_format_spec': r"FS(?P<zero>(L|T|D))?(?P<notation>(A|I))[NG0-9]*X(?P<x>[0-7][0-7])Y(?P<y>[0-7][0-7])[DM0-9]*\*MO(?P<unit>IN|MM)",
@@ -581,12 +581,25 @@ class GerberParser:
         self.lineno = None
         self.line = None
 
+    def _shorten_line(self):
+        line_joined = self.line.replace('\r', '').replace('\n', '\\n')
+        if len(line_joined) > 80:
+            return f'{line_joined[:20]}[...]{line_joined[-20:]}'
+        else:
+            return line_joined
+
     def warn(self, msg, kls=SyntaxWarning):
-        line_joined = self.line.replace('\n', '\\n')
-        warnings.warn(f'{self.filename}:{self.lineno} "{line_joined}": {msg}', kls)
+        warnings.warn(f'{self.filename}:{self.lineno} "{self._shorten_line()}": {msg}', kls)
 
     @classmethod
     def _split_commands(kls, data):
+        for match in re.finditer(r'G04.*?\*|%.*?%|[^*%]*\*', data, re.DOTALL):
+            cmd = match[0].strip().strip('%').rstrip('*').replace('\r', '').replace('\n', '')
+            if cmd:
+                yield 1, cmd
+        return
+
+        #######
         start = 0
         extended_command = False
         lineno = 1
@@ -637,13 +650,13 @@ class GerberParser:
                     try:
                         getattr(self, f'_parse_{name}')(match)
                     except Exception as e:
-                        raise SyntaxError(f'{filename}:{lineno} "{line}": {e}') from e
+                        raise SyntaxError(f'{filename}:{lineno} "{self._shorten_line()}": {e}') from e
                     line = line[match.end(0):]
                     break
 
             else:
-                self.warn(f'Unknown statement found: "{line}", ignoring.', UnknownStatementWarning)
-                self.target.comments.append(f'Unknown statement found: "{line}", ignoring.')
+                self.warn(f'Unknown statement found: "{self._shorten_line()}", ignoring.', UnknownStatementWarning)
+                self.target.comments.append(f'Unknown statement found: "{self._shorten_line()}", ignoring.')
         
         self.target.apertures = list(self.aperture_map.values())
         self.target.import_settings = self.file_settings