hdx-python-utilities 4.2.0__tar.gz → 4.2.1__tar.gz

{hdx_python_utilities-4.2.0 → hdx_python_utilities-4.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: hdx-python-utilities
-Version: 4.2.0
+Version: 4.2.1
 Summary: HDX Python Utilities for streaming tabular data, date and time handling and other helpful functions
 Project-URL: Homepage, https://github.com/OCHA-DAP/hdx-python-utilities
 Author-email: Michael Rans <rans@email.com>

{hdx_python_utilities-4.2.0 → hdx_python_utilities-4.2.1}/documentation/index.md

@@ -894,6 +894,12 @@ Examples:
     matcher = RapidFuzzMatcher(RapidFuzzMatcher.place_name_scorer)
     assert matcher.match(kalimantan, "central kalimantan") is None
 
+    # place_name_scorer also expands a leading initial to a word starting
+    # with it and requires a trailing single letter in the possible name
+    shewa = ["west shewa", "east shewa"]
+    assert matcher.match(shewa, "e shewa") == 1
+    assert matcher.match(["region iv a calabarzon"], "region iv b") is None
+
     # Phonetics is an alternative matcher using Refined Soundex
     phonetics = Phonetics()
     assert phonetics.match(possible_names, "al dali") == 1

{hdx_python_utilities-4.2.0 → hdx_python_utilities-4.2.1}/src/hdx/utilities/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '4.2.0'
-__version_tuple__ = version_tuple = (4, 2, 0)
+__version__ = version = '4.2.1'
+__version_tuple__ = version_tuple = (4, 2, 1)
 
 __commit_id__ = commit_id = None

{hdx_python_utilities-4.2.0 → hdx_python_utilities-4.2.1}/src/hdx/utilities/matching.py

@@ -13,6 +13,50 @@ LIST_SEPARATORS = re.compile(r"\s*(?:[,;/:&\n]|\s(?:and|et)\s)\s*", re.IGNORECAS
 HYPHEN = re.compile(r"\s*-\s*")
 
 
+def expand_initials(words: list[str], other_words: list[str]) -> list[str] | None:
+    """Check single letters at the start or end of words against
+    other_words. A leading letter is an initial and is replaced by an unused
+    word in other_words that is not in words and starts with it ("e shewa"
+    -> "east shewa", "b gumuz" -> "benishangul gumuz"). A trailing letter
+    distinguishes places ("region iv b") so must be in other_words. Single
+    letters between other words are left alone as they are often conjunctions
+    (eg. Spanish "y").
+
+    Args:
+        words: Words that may contain single letters
+        other_words: Words to expand single letters to
+
+    Returns:
+        Expanded words or None if a single letter has no counterpart
+    """
+    leading = []
+    for word in words:
+        if len(word) != 1:
+            break
+        leading.append(word)
+    trailing = []
+    for word in reversed(words[len(leading) :]):
+        if len(word) != 1:
+            break
+        trailing.append(word)
+    for letter in trailing:
+        if letter.isalpha() and letter not in other_words:
+            return None
+    unmatched = [word for word in other_words if word not in words and len(word) > 1]
+    expanded = list(words)
+    for i, letter in enumerate(leading):
+        if not letter.isalpha() or letter in other_words:
+            continue
+        for other in unmatched:
+            if other[0] == letter:
+                unmatched.remove(other)
+                expanded[i] = other
+                break
+        else:
+            return None
+    return expanded
+
+
 class Phonetics(RefinedSoundex):
     def match(
         self,
@@ -119,6 +163,11 @@ class RapidFuzzMatcher:
         distinguishes places, but not names like sectors, where extra words
         are often descriptive (eg. "food security and livelihood").
 
+        Single letters at the start or end of name are checked first with
+        expand_initials ("e shewa" scores 0 against "west shewa" but matches
+        "east shewa"), unless the names differ only in spaces and punctuation
+        (eg. "sana a" vs "sanaa").
+
         Args:
             name: Name to match
             possible_name: Possible name
@@ -126,6 +175,15 @@ class RapidFuzzMatcher:
         Returns:
             Similarity score from 0-100
         """
+        joined_score = rapidfuzz_fuzz.ratio(
+            NON_WORD_CHARACTERS.sub("", name),
+            NON_WORD_CHARACTERS.sub("", possible_name),
+        )
+        if joined_score < 90:
+            name_words = expand_initials(name.split(), possible_name.split())
+            if name_words is None:
+                return 0.0
+            name = " ".join(name_words)
         words = set(name.split())
         possible_words = set(possible_name.split())