extensive search: regex_parse on whole document (#67)

adbar · Nov 15, 2022 · 522db7d · 522db7d
1 parent dffa53b
commit 522db7d
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 17 deletions.
diff --git a/htmldate/core.py b/htmldate/core.py
@@ -26,7 +26,7 @@
  idiosyncrasies_search,
  img_search,
  json_search,
- # regex_parse,
+ regex_parse,
  timestamp_search,
  try_date_expr,
  DATE_EXPRESSIONS,
@@ -846,27 +846,28 @@ def search_page(
  if result is not None:
  return result
 
+ # try full-blown text regex on all HTML?
+ dateobject = regex_parse(htmlstring)
+ # todo: find all candidates and disambiguate?
+ if date_validator(
+ dateobject, outputformat, earliest=min_date, latest=max_date
+ ) is True and (
+ copyear == 0 or dateobject.year >= copyear # type: ignore[union-attr]
+ ):
+ try:
+ LOGGER.debug("regex result on HTML: %s", dateobject)
+ return dateobject.strftime(outputformat) # type: ignore
+ except ValueError as err:
+ LOGGER.error("value error during conversion: %s %s", dateobject, err)
+
  # catchall: copyright mention
  if copyear != 0:
  LOGGER.debug("using copyright year as default")
  return convert_date(
  "-".join([str(copyear), "01", "01"]), "%Y-%m-%d", outputformat
  )
 
- # try full-blown text regex on all HTML?
- # dateobject = regex_parse(htmlstring)
- # todo: find all candidates and disambiguate?
- # if (
- # date_validator(dateobject, outputformat, earliest=min_date, latest=max_date)
- # is True
- # ):
- # try:
- # LOGGER.debug("regex result on HTML: %s", dateobject)
- # return dateobject.strftime(outputformat) # type: ignore
- # except ValueError as err:
- # LOGGER.error("value error during conversion: %s %s", string, err)
-
- # 1 component, last try
+ # last resort: 1 component
  LOGGER.debug("switching to one component")
  bestmatch = search_pattern(
  htmlstring,

diff --git a/tests/unit_tests.py b/tests/unit_tests.py
@@ -559,8 +559,7 @@ def test_exact_date():
  '<html><body><time class="Feed-module--feed__item-meta-time--3t1fg" dateTime="November 29, 2020">November 2020</time></body></html>',
  outputformat="%Y-%m-%d %H:%m:%S",
  )
- == "2020-01-01 00:01:00" # solution so far
- # == "2020-11-29 00:11:00" # better
+ == "2020-11-29 00:11:00" # 00:11 unclear
  )
  assert (
  find_date(