Merge pull request #2 from ZeroGachis/feature/regex-extract

✨ Add 'regex-extract' pre-processor

README.md

@@ -77,6 +77,7 @@ assert not errors
 
 - left-pad-zeroes
 - map
+- regex-extract
 - replace
 - strip-whitespaces
 

magicparse/pre_processors.py

@@ -1,3 +1,4 @@
+import re
 from .transform import Transform
 
 
@@ -68,4 +69,27 @@ def key() -> str:
         return "strip-whitespaces"
 
 
-builtins = [LeftPadZeroes, Map, Replace, StripWhitespaces]
+class RegexExtract(PreProcessor):
+    def __init__(self, pattern: str) -> None:
+        pattern = re.compile(pattern)
+        if "value" not in pattern.groupindex:
+            raise ValueError(
+                "regex-extract's pattern must contain a group named 'value'"
+            )
+
+        self.pattern = pattern
+
+    def apply(self, value: str) -> str:
+        match = re.match(self.pattern, value)
+        if not match:
+            raise ValueError(
+                f"cannot extract value from pattern '{self.pattern.pattern}'"
+            )
+
+        return match.group("value")
+
+    def key() -> str:
+        return "regex-extract"
+
+
+builtins = [LeftPadZeroes, Map, RegexExtract, Replace, StripWhitespaces]

tests/test_pre_processors.py

@@ -1,7 +1,9 @@
+import re
 from magicparse.pre_processors import (
     LeftPadZeroes,
     Map,
     PreProcessor,
+    RegexExtract,
     Replace,
     StripWhitespaces,
 )
@@ -36,6 +38,17 @@ def test_strip_whitespaces(self):
         pre_processor = PreProcessor.build({"name": "strip-whitespaces"})
         assert isinstance(pre_processor, StripWhitespaces)
 
+    def test_regex_extract(self):
+        pre_processor = PreProcessor.build(
+            {
+                "name": "regex-extract",
+                "parameters": {"pattern": "^xxx(?P<value>\\d{13})xxx$"},
+            }
+        )
+        assert isinstance(pre_processor, RegexExtract)
+        assert isinstance(pre_processor.pattern, re.Pattern)
+        assert pre_processor.pattern.pattern == "^xxx(?P<value>\\d{13})xxx$"
+
     def test_unknown(self):
         with pytest.raises(ValueError, match="invalid pre-processor 'anything'"):
             PreProcessor.build({"name": "anything"})
@@ -101,6 +114,41 @@ def test_success(self):
         assert pre_processor.apply("    an input     ") == "an input"
 
 
+class TestRegexExtract(TestCase):
+    def test_build_without_value_group(self):
+        with pytest.raises(
+            ValueError,
+            match=r"regex-extract's pattern must contain a group named 'value'",
+        ):
+            PreProcessor.build(
+                {"name": "regex-extract", "parameters": {"pattern": "xxx"}}
+            )
+
+    def test_pattern_not_found(self):
+        pre_processor = PreProcessor.build(
+            {
+                "name": "regex-extract",
+                "parameters": {"pattern": "^xxx(?P<value>\\d{13})xxx$"},
+            }
+        )
+        with pytest.raises(ValueError) as error:
+            pre_processor.apply("an input")
+
+        assert (
+            error.value.args[0]
+            == "cannot extract value from pattern '^xxx(?P<value>\\d{13})xxx$'"
+        )
+
+    def test_pattern_found(self):
+        pre_processor = PreProcessor.build(
+            {
+                "name": "regex-extract",
+                "parameters": {"pattern": "^xxx(?P<value>\\d{13})xxx$"},
+            }
+        )
+        pre_processor.apply("xxx9780201379624xxx") == "9780201379624"
+
+
 class TestRegister(TestCase):
     class YesPreProcessor(PreProcessor):
         def key() -> str: