Merge 281e869 into 963ccc2

nielstron · Oct 24, 2020 · 99278b4 · 99278b4
2 parents 963ccc2 + 281e869
commit 99278b4
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 15 deletions.
diff --git a/quantulum3/_lang/en_US/parser.py b/quantulum3/_lang/en_US/parser.py
@@ -62,26 +62,27 @@ def extract_spellout_values(text):
     """
 
     values = []
-    try:
-        for item in reg.text_pattern_reg(lang).finditer(text):
+    for item in reg.text_pattern_reg(lang).finditer(text):
+        try:
             surface, span = clean_surface(item.group(0), item.span())
             if not surface or surface.lower() in reg.scales(lang):
                 continue
             curr = result = 0.0
-            for word in surface.split():
+            for word in surface.lower().split():
                 try:
                     scale, increment = (
                         1,
                         float(
                             re.sub(
                                 r"(-$|[%s])" % reg.grouping_operators_regex(lang),
                                 "",
-                                word.lower(),
+                                word,
                             )
                         ),
                     )
                 except ValueError:
-                    scale, increment = reg.numberwords(lang)[word.lower()]
+                    match = re.search(reg.numberwords_regex(), word)
+                    scale, increment = reg.numberwords(lang)[match.group(0)]
                 curr = curr * scale + increment
                 if scale > 100:
                     result += curr
@@ -93,9 +94,9 @@ def extract_spellout_values(text):
                     "new_surface": str(result + curr),
                 }
             )
-    except KeyError:
-        # just ignore the match if an error occurred
-        values = []
+        except (KeyError, AttributeError):
+            # just ignore the match if an error occurred
+            values = []
 
     return sorted(values, key=lambda x: x["old_span"][0])
 

diff --git a/quantulum3/_lang/en_US/regex.py b/quantulum3/_lang/en_US/regex.py
@@ -54,7 +54,7 @@
     "ninth": 1 / 9,
 }
 
-MISCNUM = {"and": (1, 0), "a": (1, 1), "an": (1, 1)}
+MISCNUM = {"&": (1, 0), "and": (1, 0), "a": (1, 1), "an": (1, 1)}
 
 ###############################################################################
 
@@ -73,11 +73,8 @@
         (?<![a-zA-Z0-9+.-])    # lookbehind, avoid "Area51"
         {number_pattern_no_groups}
     )?
-    [ -]?(?:{numberwords_regex})
-    [ -]?(?:{numberwords_regex})?
-    [ -]?(?:{numberwords_regex})?[ -]?(?:{numberwords_regex})?
-    [ -]?(?:{numberwords_regex})?[ -]?(?:{numberwords_regex})?
-    [ -]?(?:{numberwords_regex})?
+    [, ]?(?:{numberwords_regex})
+    (?:[, -]*(?:{numberwords_regex}))*
     (?!\s?{number_pattern_no_groups}) # Disallow being followed by only a
                                       # number
 """

diff --git a/quantulum3/_lang/en_US/tests/quantities.json b/quantulum3/_lang/en_US/tests/quantities.json
@@ -1351,5 +1351,15 @@
   {
     "req": "I sent some money using a PayPal account",
     "res": []
+  },
+  {
+    "req": "three million, two hundred & forty, you say?",
+    "res": [
+        {
+          "value": 3000240,
+          "unit": "dimensionless",
+          "surface": "three million, two hundred & forty"
+        }
+      ]
   }
 ]
diff --git a/quantulum3/regex.py b/quantulum3/regex.py
@@ -90,7 +90,9 @@ def numberwords(lang="en_US"):
 
 @cached
 def numberwords_regex(lang="en_US"):
-    all_numbers = r"|".join(r"\b%s\b" % i for i in list(numberwords(lang).keys()) if i)
+    all_numbers = r"|".join(
+        r"((?<=\W)|^)%s((?=\W)|$)" % i for i in list(numberwords(lang).keys()) if i
+    )
     return all_numbers