use bulletind, abandoned work

setup.cfg

@@ -34,6 +34,7 @@ install_requires =
     platformdirs
     requests
     python-dateutil
+    bulletind
 
 [options.entry_points]
 console_scripts =

src/wwvb/iersdata_dist.py

@@ -23,16 +23,15 @@ DUT1_OFFSETS = str( # 19720601
     +i*66+h*52+g*67+f+p*103+o*56+n*68+m*69+l*107+k*82+j*72+i*67  # 19870518
     +h*63+g*113+f*63+e*51+o*11+n*60+m*59+l*121+k*71+j*71+i*67    # 19890531
     +h*57+g*93+f*61+e*48+d*12+n*41+m*44+l*46+k*61+j*66+i*47+h*45 # 19901231
-    +g*15+q*32+p*44+o*41+n*48+m*74+l*49+k*45+j*44+i*40+h*37+g*38 # 19920625
-    +f*50+e*5+o*60+n*49+m*40+l*40+k*38+j*38+i*36+h*39+g*25+q*31  # 19930919
-    +p*50+o*41+n*41+m*43+l*41+k*39+j*40+i*39+s*24+r*57+q*43+p*41 # 19950120
-    +o*39+n*38+m*35+l*37+k*43+j*69+i*44+h*42+g*37+q*4+p*51+o*45  # 19960523
-    +n*44+m*69+l*70+k*50+j*54+i*53+h*40+g*49+f*18+p*59+o*53+n*52 # 19980206
-    +m*57+l*48+k*53+j*127+i*70+h*30+r*62+q*79+p*152+o*82+n*106   # 20001026
-    +m*184+l*125+k*217+j*133+i*252+h*161+g*392+f*322+e*290+n*116 # 20060927
-    +m*154+l*85+k*83+j*91+i*168+h*105+g*147+f*105+e*42+o*70+n*91 # 20091111
-    +m*154+l*119+k*84+j*217+i*126+h*176+g*97+f*91+e*52+o*116     # 20130130
-    +n*98+m*70+l*133+k*91+j*91+i*77+h*140+g*91+f*84+e*70+d*34    # 20150910
-    +n*72+m*76+l*66+k*53+j*56+i*105+h*77+g*45+q*25+p*63+o*91     # 20171129
-    +n*154+m*105+l*190+k*118+j*105+i*807+j*376+k*351+l*86        # 20231007
+    +g*15+q*32+p*44+o*41+n*48+m*68+l*56+k*35+j*168+f*119+n*49    # 19921125
+    +m*35+l*49+k*112+h*56+q*98+o*42+n*42+m*77+k*35+j*71+s*237    # 19950315
+    +n*21+m*28+l*45+k*46+j*56+i*49+h*35+g*32+p*52+o*49+n*35+m*84 # 19961002
+    +l*56+k*63+j*63+i*42+h*49+g*49+f*5+p*79+o*42+n*49+m*63+l*35  # 19980506
+    +k*42+j*98+i*105+h*36+r*62+q*84+p*140+o*84+n*98+m*189+l*133  # 20011003
+    +k*217+j*133+i*252+h*161+g*392+f*322+e*290+n*116+m*154+l*85  # 20070314
+    +k*83+j*91+i*168+h*105+g*147+f*105+e*112+n*91+m*154+l*119    # 20100602
+    +k*84+j*217+i*126+h*176+g*97+f*91+e*52+o*116+n*98+m*70+l*133 # 20131120
+    +k*91+j*91+i*77+h*140+g*91+f*84+e*70+d*34+n*78+m*70+l*66     # 20160323
+    +k*53+j*56+i*105+h*77+g*45+q*25+p*63+o*91+n*154+m*105+l*190  # 20190116
+    +k*118+j*105+i*807+j*376+k*346+l*98                          # 20231014
 )

src/wwvb/updateiers.py

@@ -15,6 +15,7 @@ import pathlib
 from typing import Callable, List, Optional
 
 import bs4
+import bulletind
 import click
 import platformdirs
 import requests
@@ -33,7 +34,6 @@ try:
 except (ImportError, NameError) as e:
     pass
 IERS_URL = "https://datacenter.iers.org/data/csv/finals2000A.all.csv"
-NIST_URL = "https://www.nist.gov/pml/time-and-frequency-division/atomic-standards/leap-second-and-ut1-utc-information"
 
 
 def update_iersdata(  # pylint: disable=too-many-locals, too-many-branches, too-many-statements
@@ -51,6 +51,7 @@ def update_iersdata(  # pylint: disable=too-many-locals, too-many-branches, too-
             offs = int(round(float(offs_str) * 10))
             if not offsets:
                 table_start = datetime.date(1858, 11, 17) + datetime.timedelta(jd)
+                # If the table starts after June 1, 1972 (it does), add the 1972 leap second
                 if table_start > datetime.date(1972, 6, 1):
                     when = datetime.date(1972, 6, 1)
                     while when < datetime.date(1972, 7, 1):
@@ -62,18 +63,6 @@ def update_iersdata(  # pylint: disable=too-many-locals, too-many-branches, too-
                     table_start = datetime.date(1972, 6, 1)
             offsets.append(offs)
 
-    wwvb_text = requests.get(NIST_URL).text
-    wwvb_data = bs4.BeautifulSoup(wwvb_text, features="html.parser")
-    wwvb_dut1_table = wwvb_data.findAll("table")[2]
-    assert wwvb_dut1_table
-    meta = wwvb_data.find("meta", property="article:modified_time")
-    assert isinstance(meta, bs4.Tag)
-    wwvb_data_stamp = (
-        datetime.datetime.fromisoformat(meta.attrs["content"])
-        .replace(tzinfo=None)
-        .date()
-    )
-
     def patch(patch_start: datetime.date, patch_end: datetime.date, val: int) -> None:
         off_start = (patch_start - table_start).days
         off_end = (patch_end - table_start).days
@@ -81,29 +70,26 @@ def update_iersdata(  # pylint: disable=too-many-locals, too-many-branches, too-
 
     wwvb_dut1: Optional[int] = None
     wwvb_start: Optional[datetime.date] = None
-    for row in wwvb_dut1_table.findAll("tr")[1:][::-1]:
-        cells = row.findAll("td")
-        when = datetime.datetime.strptime(cells[0].text, "%Y-%m-%d").date()
-        dut1 = cells[2].text.replace("s", "").replace(" ", "")
-        dut1 = int(round(float(dut1) * 10))
+    for data in bulletind.get_cached_bulletin_d_data():
+        when = data.start_date
+        dut1 = round(10 * data.dut1)
         if wwvb_dut1 is not None:
             assert wwvb_start is not None
             patch(wwvb_start, when, wwvb_dut1)
         wwvb_dut1 = dut1
         wwvb_start = when
 
-    # As of 2021-06-14, NIST website incorrectly indicates the offset of -600ms
-    # persisted through 2009-03-12, causing an incorrect leap second inference.
-    # Assume instead that NIST started broadcasting +400ms on January 1, 2009,
-    # causing the leap second to occur on 2008-12-31.
-    patch(datetime.date(2009, 1, 1), datetime.date(2009, 3, 12), 4)
-
-    # this is the final (most recent) wwvb DUT1 value broadcast.  We want to
-    # extend it some distance into the future, but how far?  We will use the
-    # modified timestamp of the NIST data.
-    assert wwvb_dut1 is not None
-    assert wwvb_start is not None
-    patch(wwvb_start, wwvb_data_stamp + datetime.timedelta(days=1), wwvb_dut1)
+    # this is the final (most recent) Bulletin D DUT1 value specified.  We
+    # might want to extend it some distance into the future, but how far?  The
+    # textual Bulletin D says something like "Bulletin D 136 should be issued
+    # in August 2017" or "Bulletin D 143 should be issued in 2023" but this is
+    # not in the xml format and thus not in the data from the bulletind
+    # package. Historically, bulletins have even been published AFTER their
+    # effective date. As such, transition immediately back to the IERS Bulletin
+    # A data immediately after the last Bulletin D data and do _NOT_ patch any
+    # additonal days. This is in contrast to previous behavior, where the NIST
+    # data was propagated until the reported modification date of the scraped
+    # web page.
 
     with open(target_file, "wt", encoding="utf-8") as output: