Improve date filter regexes to extract structured, natural, partial dates

- Much faster than using dateparser
  - It took 2x-4x for improved regex to extracts 1-15% more dates
  - Whereas It took 33x to 100x for dateparser to extract 65% - 400% more dates
  - Improve date extractor tests to test deduping dates, natural,
    structured date extraction from content

- Extract some natural, partial dates and more structured dates
  Using regex is much faster than using dateparser. It's a little
  crude but should pay off in performance.

  Supports dates of form:
  - (Day-of-Month) Month|AbbreviatedMonth Year|2DigitYear
  - Month|AbbreviatedMonth (Day-of-Month) Year|2DigitYear

src/khoj/search_filter/date_filter.py

@@ -1,9 +1,10 @@
+import calendar
 import logging
 import re
 from collections import defaultdict
 from datetime import datetime, timedelta
 from math import inf
-from typing import List
+from typing import List, Tuple
 
 import dateparser as dtparse
 from dateparser.search import search_dates
@@ -23,48 +24,82 @@ class DateFilter(BaseFilter):
     # - dt>="last week"
     # - dt:"2 years ago"
     date_regex = r"dt([:><=]{1,2})[\"'](.*?)[\"']"
-    raw_date_regex = r"\d{4}[-/]\d{2}[-/]\d{2}"
 
     def __init__(self, entry_key="compiled"):
         self.entry_key = entry_key
         self.date_to_entry_ids = defaultdict(set)
         self.cache = LRU()
+        self.dtparser_regexes = self.compile_date_regexes()
+        self.dtparser_ordinal_suffixes = re.compile(r"(st|nd|rd|th)")
         self.dtparser_settings = {
             "PREFER_DAY_OF_MONTH": "first",
             "DATE_ORDER": "YMD",  # Prefer YMD and DMY over MDY when parsing ambiguous dates
         }
 
+    def compile_date_regexes(self):
+        months = calendar.month_name[1:]
+        abbr_months = calendar.month_abbr[1:]
+        # Extract natural dates from content like 1st April 1984, 31 April 84, Apr 4th 1984, 13 Apr 84
+        dBY_regex = re.compile(r"\b\d{1,2}(?:st|nd|rd|th)? (?:" + "|".join(months) + r") \d{4}\b", re.IGNORECASE)
+        dBy_regex = re.compile(r"\b\d{1,2}(?:st|nd|rd|th)? (?:" + "|".join(months) + r") \d{2}\b", re.IGNORECASE)
+        BdY_regex = re.compile(r"\b(?:" + "|".join(months) + r") \d{1,2}(?:st|nd|rd|th)? \d{4}\b", re.IGNORECASE)
+        Bdy_regex = re.compile(r"\b(?:" + "|".join(months) + r") \d{1,2}(?:st|nd|rd|th)? \d{2}\b", re.IGNORECASE)
+        dbY_regex = re.compile(r"\b\d{1,2}(?:st|nd|rd|th)? (?:" + "|".join(abbr_months) + r") \d{4}\b", re.IGNORECASE)
+        dby_regex = re.compile(r"\b\d{1,2}(?:st|nd|rd|th)? (?:" + "|".join(abbr_months) + r") \d{2}\b", re.IGNORECASE)
+        bdY_regex = re.compile(r"\b(?:" + "|".join(abbr_months) + r") \d{1,2}(?:st|nd|rd|th)? \d{4}\b", re.IGNORECASE)
+        bdy_regex = re.compile(r"\b(?:" + "|".join(abbr_months) + r") \d{1,2}(?:st|nd|rd|th)? \d{2}\b", re.IGNORECASE)
+        # Extract natural of form Month, Year like January 2021, Jan 2021, Jan 21
+        BY_regex = re.compile(r"\b(?:" + "|".join(months) + r") \d{4}\b", re.IGNORECASE)
+        By_regex = re.compile(r"\b(?:" + "|".join(months) + r") \d{2}\b", re.IGNORECASE)
+        bY_regex = re.compile(r"\b(?:" + "|".join(abbr_months) + r") \d{4}\b", re.IGNORECASE)
+        by_regex = re.compile(r"\b(?:" + "|".join(abbr_months) + r") \d{2}\b", re.IGNORECASE)
+        # Extract structured dates from content like 1984-04-01, 1984/04/01, 01-04-1984, 01/04/1984, 01.04.1984, 01-04-84, 01/04/84
+        Ymd_date_regex = re.compile(r"\b\d{4}[-\/]\d{2}[-\/]\d{2}\b", re.IGNORECASE)
+        dmY_date_regex = re.compile(r"\b\d{2}[-\/]\d{2}[-\/]\d{4}\b", re.IGNORECASE)
+        dmy_date_regex = re.compile(r"\b\d{2}[-\/]\d{2}[-\/]\d{2}\b", re.IGNORECASE)
+        dmY_dot_date_regex = re.compile(r"\b\d{2}[\.]\d{2}[\.]\d{4}\b", re.IGNORECASE)
+
+        # Combine date formatter and date identifier regex pairs
+        dtparser_regexes: List[Tuple[str, re.Pattern[str]]] = [
+            # Structured dates
+            ("%Y-%m-%d", Ymd_date_regex),
+            ("%Y/%m/%d", Ymd_date_regex),
+            ("%d-%m-%Y", dmY_date_regex),
+            ("%d/%m/%Y", dmY_date_regex),
+            ("%d.%m.%Y", dmY_dot_date_regex),
+            ("%d-%m-%y", dmy_date_regex),
+            ("%d/%m/%y", dmy_date_regex),
+            # Natural dates
+            ("%d %B %Y", dBY_regex),
+            ("%d %B %y", dBy_regex),
+            ("%B %d %Y", BdY_regex),
+            ("%B %d %y", Bdy_regex),
+            ("%d %b %Y", dbY_regex),
+            ("%d %b %y", dby_regex),
+            ("%b %d %Y", bdY_regex),
+            ("%b %d %y", bdy_regex),
+            # Partial natural dates
+            ("%B %Y", BY_regex),
+            ("%B %y", By_regex),
+            ("%b %Y", bY_regex),
+            ("%b %y", by_regex),
+        ]
+        return dtparser_regexes
+
     def extract_dates(self, content):
-        "Extract all natural and structured dates across formats and locales from content"
+        "Extract natural and structured dates from content"
-        excluded_parsers = ["relative-time"]
+        valid_dates = set()
-        dtparser_settings = merge_dicts(
+        for date_format, date_regex in self.dtparser_regexes:
-            {
+            matched_dates = date_regex.findall(content)
-                # Exclude relative dates for date extraction from content as very ambiguous
+            for date_str in matched_dates:
-                "PARSERS": [parser for parser in default_parsers if parser not in excluded_parsers],
+                # Remove ordinal suffixes to parse date
-                "RETURN_AS_TIMEZONE_AWARE": False,
+                date_str = self.dtparser_ordinal_suffixes.sub("", date_str)
-            },
+                try:
-            self.dtparser_settings,
+                    valid_dates.add(datetime.strptime(date_str, date_format))
-        )
+                except ValueError:
-        try:
+                    continue
-            valid_dates = [
-                dt_item[1] for dt_item in search_dates(content, settings=dtparser_settings, languages=["en"]) or []
-            ]
-            return valid_dates
-        except Exception as e:
-            logger.warning(
-                f"Failed to extract natural dates from content with error: {e}. Fallback to regex based extraction."
-            )
 
-        # Fallback to extracting YYYY-MM-DD format dates from content
+        return list(valid_dates)
-        pattern_matched_dates = re.findall(self.raw_date_regex, content)
-        valid_dates = []
-        for date_str in pattern_matched_dates:
-            try:
-                valid_dates.append(datetime.strptime(date_str, "%Y-%m-%d"))
-            except ValueError:
-                continue
-
-        return valid_dates
 
     def get_filter_terms(self, query: str) -> List[str]:
         "Get all filter terms in query"

tests/test_date_filter.py

@@ -147,16 +147,70 @@ def test_date_extraction():
     assert extracted_dates == [], "Expected relative date to be ignored"
 
     extracted_dates = DateFilter().extract_dates("head CREATED: today SCHEDULED: 1984-04-01 tail")
-    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected only YMD structured date to be extracted"
+    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected only Y-m-d structured date to be extracted"
 
-    extracted_dates = DateFilter().extract_dates("head CREATED: today SCHEDULED: 01-04-1984 tail")
+    extracted_dates = DateFilter().extract_dates("head SCHEDULED: 01-04-1984 tail")
-    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected DMY structured date to be extracted"
+    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected d-m-Y structured date to be extracted"
 
-    extracted_dates = DateFilter().extract_dates("head Updates from April 1984 tail")
+    extracted_dates = DateFilter().extract_dates("head CREATED: today SCHEDULED: 1984/04/01 tail")
-    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected natural date to be extracted"
+    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected only Y/m/d structured date to be extracted"
 
-    extracted_dates = DateFilter().extract_dates("CLOCK: [1984-04-01 mer 09:50]--[1984-04-01 mer 10:10] => 24:20")
+    extracted_dates = DateFilter().extract_dates("head SCHEDULED: 01/04/1984 tail")
-    expected_dates = [datetime(1984, 4, 1, 9, 50, 0), datetime(1984, 4, 1, 10, 10, 0)]
+    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected d/m/Y structured date to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head DEADLINE: 01.04.1984 tail")
+    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected d.m.Y structured date to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("CLOCK: [1984-04-01 Sun 09:50]--[1984-04-01 Sun 10:10] => 24:20")
+    assert extracted_dates == [
+        datetime(1984, 4, 1, 0, 0, 0)
+    ], "Expected single deduplicated date extracted from logbook entry"
+
+    extracted_dates = DateFilter().extract_dates("CLOCK: [1984/03/31 mer 09:50]--[1984/04/01 mer 10:10] => 24:20")
+    expected_dates = [datetime(1984, 4, 1, 0, 0, 0), datetime(1984, 3, 31, 0, 0, 0)]
     assert all(
         [dt in extracted_dates for dt in expected_dates]
-    ), "Expected multiple non-english dates extracted from logbook entry"
+    ), "Expected multiple different dates extracted from logbook entry"
+
+
+def test_natual_date_extraction():
+    extracted_dates = DateFilter().extract_dates("head 1 April 1984 tail")
+    assert datetime(1984, 4, 1, 0, 0, 0) in extracted_dates, "Expected natural date to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head 1st April 1984 tail")
+    assert datetime(1984, 4, 1, 0, 0, 0) in extracted_dates, "Expected natural date to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head 2nd Apr 1984 tail")
+    assert datetime(1984, 4, 2, 0, 0, 0) in extracted_dates, "Expected natural date with short month to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head 4th Apr 1984 tail")
+    assert datetime(1984, 4, 4, 0, 0, 0) in extracted_dates, "Expected natural date to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head 11th april 1984 tail")
+    assert (
+        datetime(1984, 4, 11, 0, 0, 0) in extracted_dates
+    ), "Expected natural date with lowercase month to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head 23rd april 84 tail")
+    assert datetime(1984, 4, 23, 0, 0, 0) in extracted_dates, "Expected natural date with 2-digit year to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head 31st march 84 tail")
+    assert datetime(1984, 3, 31, 0, 0, 0) in extracted_dates, "Expected natural date with 2-digit year to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head April 1984 tail")
+    assert extracted_dates == [datetime(1984, 4, 1, 0, 0, 0)], "Expected partial natural date to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head Apr 1984 tail")
+    assert extracted_dates == [
+        datetime(1984, 4, 1, 0, 0, 0)
+    ], "Expected partial natural date with short month to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head apr 1984 tail")
+    assert extracted_dates == [
+        datetime(1984, 4, 1, 0, 0, 0)
+    ], "Expected partial natural date with lowercase month to be extracted"
+
+    extracted_dates = DateFilter().extract_dates("head apr 84 tail")
+    assert extracted_dates == [
+        datetime(1984, 4, 1, 0, 0, 0)
+    ], "Expected partial natural date with 2-digit year to be extracted"