o Better code structure for normalization.

1 files changed, 54 insertions, 46 deletions

diff --git a/src/ee/digikey/normalize_facts.py b/src/ee/digikey/normalize_facts.py
index 1ddf78c..62aef0d 100644
--- a/src/ee/digikey/normalize_facts.py
+++ b/src/ee/digikey/normalize_facts.py
@@ -1,6 +1,6 @@
 import re
 from pathlib import Path
-from typing import List, Tuple, Union
+from typing import List, Tuple, Union, Mapping, Callable, Any
 
 from ee import EeVal, EeException
 from ee.part import PartDb, load_db, save_db
@@ -65,10 +65,52 @@ def handle_tolerance(tolerance: bomFile.Fact, capacitance: bomFile.Fact) -> List
     return [bomFile.Fact(key=uris.make_fact_key(key), label=label, value=str(value)) for key, label, value in facts]
 
 
+def re_parser(pattern, kis: List[Tuple[str, int]]):
+    r = re.compile(pattern)
+
+    def parse(value):
+        m = r.match(value)
+
+        if not m:
+            return
+
+        return [(k, m.group(i)) for k, i in kis]
+
+    return parse
+
+
+def no_parser(key):
+    def parse(value):
+        return key, value
+
+    return parse
+
+
+def ee_val_parser(key):
+    def parse(value):
+        return key, EeVal.parse(value)
+
+    return parse
+
+
 _operating_temperature_re = re.compile("-([0-9]+)°C ~ ([0-9])+°C")
 _height_seated_re = re.compile("([0-9]+\\.[0-9]+)\" \\(([0-9]+\\.[0-9]+)mm\\)")
 _land_size_re = re.compile("([0-9]+\\.[0-9]+)\" L x ([0-9]+\\.[0-9]+)\" W "
                            "\\(([0-9]+\\.[0-9]+)mm x ([0-9]+\\.[0-9]+)mm\\)")  # 0.260" L x 0.260" W (6.60mm x 6.60mm)
+parsers: Mapping[int, Callable[[str], Any]] = {
+    3: no_parser("tolerance"),
+    1471: ee_val_parser("voltage-input-min"),  # Voltage - Input (Min)
+    573: ee_val_parser("voltage-input-max"),  # Voltage - Input (Max)
+    1429: ee_val_parser("voltage-output-max"),  # Voltage - Output (Max)
+    252: re_parser(_operating_temperature_re, [("temperature-min", 1), ("temperature-max", 2)]),
+    1686: re_parser(_operating_temperature_re, [("temperature-junction-min", 1), ("temperature-junction-max", 2)]),
+
+    1500: re_parser(_height_seated_re, [("part-height", 2)]),
+    884: re_parser(_land_size_re, [("land-size-x", 3), ("land-size-y", 4)]),
+
+    2049: ee_val_parser("capacitance"),
+    2079: ee_val_parser("voltage-rating"),
+}
 
 
 def normalize_facts(in_dir: Path, out_dir: Path):
@@ -88,57 +130,23 @@ def normalize_facts(in_dir: Path, out_dir: Path):
         for f in in_facts:
             key = uris.parse_digikey_fact_key(f.keyProp)
 
-            out_fact = None
-
             value = f.valueProp
 
             if value == "-":
                 continue
 
-            try:
-                if key == 3:
-                    out_fact = bomFile.Fact(key=uris.make_fact_key("tolerance"), label="Tolerance", value=value)
-                elif key == 252:
-                    m = _operating_temperature_re.match(value)
-                    if m:
-                        mn = m.group(1)
-                        mx = m.group(2)
-
-                        out_facts.extend([
-                            bomFile.Fact(key=uris.make_fact_key("temperature-min"), label="Min temperature", value=mn),
-                            bomFile.Fact(key=uris.make_fact_key("temperature-max"), label="Max temperature", value=mx)])
-                elif key == 1500:
-                    m = _height_seated_re.match(value)
-
-                    if m:
-                        out_fact = bomFile.Fact(key=uris.make_fact_key("part-height"), label="Height", value=m.group(2))
+            parser = parsers.get(key)
+
+            if parser is not None:
+                res = parser(value)
+
+                if res is not None:
+                    if isinstance(res, list):
+                        results = res
                     else:
-                        print("{}: bad str: '{}'".format(key, value))
-                elif key == 884:
-                    m = _land_size_re.match(value)
-
-                    if m:
-                        out_facts.extend([
-                            bomFile.Fact(key=uris.make_fact_key("land-size-x"),
-                                         label="Land size, X",
-                                         value=m.group(3)),
-                            bomFile.Fact(key=uris.make_fact_key("land-size-y"),
-                                         label="Land size, Y",
-                                         value=m.group(4))])
-                elif key == 2049:
-                    val = EeVal.parse(value)
-                    out_fact = bomFile.Fact(key=uris.make_fact_key("capacitance"), label="Capacitance", value=str(val))
-                elif key == 2079:
-                    val = EeVal.parse(value)
-                    out_fact = bomFile.Fact(key=uris.make_fact_key("voltage-rating"), label="Voltage rating",
-                                            value=str(val))
-                else:
-                    pass  # fact ignored
-            except EeException:
-                pass
-
-            if out_fact:
-                out_facts.append(out_fact)
+                        results = [res]
+                    facts = [bomFile.Fact(key=uris.make_fact_key(key), value=value) for key, value in results]
+                    out_facts.extend(facts)
 
         if len(out_facts) == 0:
             continue