From 6681f7dfd5a105757df2b859444c1462a70ffccb Mon Sep 17 00:00:00 2001
From: Daniel Tsvetkov <danieltcv@gmail.com>
Date: Mon, 11 May 2020 20:29:41 +0200
Subject: [PATCH] add query types resource parsing

---
 tww/data/.cache.csv      | 30 ++++++++++++++++++++++++++++++
 tww/lib.py               | 24 ++++++++++++++----------
 tww/templates/dt_tr.html |  8 ++++----
 tww/tokenizer.py         | 15 +++++++++++----
 tww/web.py               |  2 +-
 5 files changed, 60 insertions(+), 19 deletions(-)

diff --git a/tww/data/.cache.csv b/tww/data/.cache.csv
index 4486de0..1ddfc55 100644
--- a/tww/data/.cache.csv
+++ b/tww/data/.cache.csv
@@ -46,3 +46,33 @@ in,6.9833333,171.6999969
 ind,20.0,77.0
 indi,20.0,77.0
 local,37.6666667,-1.7
++02:,49.453285449999996,3.606899003594057
+it,33.6366667,42.8224983
+G,32.3293809,-83.1137366
+D,10.0,-67.166667
+dub,-32.25,148.6166687
+D,10.0,-67.166667
+dubli,53.3330556,-6.248889
+dublin,53.3330556,-6.248889
+Du,25.0750095,55.18876088183319
+dublin ,53.3330556,-6.248889
+S,35.7724185,127.79654346305617
+sof,42.6975135,23.3241463
+S,35.7724185,127.79654346305617
+sofia ,42.6975135,23.3241463
+S,35.7724185,127.79654346305617
+S,35.7724185,127.79654346305617
+du,-32.25,148.6166687
+D,10.0,-67.166667
+S,35.7724185,127.79654346305617
+S,35.7724185,127.79654346305617
+S,35.7724185,127.79654346305617
+S,35.7724185,127.79654346305617
+D,10.0,-67.166667
+D,10.0,-67.166667
+new y,6.9833333,171.6999969
+S,35.7724185,127.79654346305617
+S,35.7724185,127.79654346305617
+D,10.0,-67.166667
+dubl,53.3330556,-6.248889
+D,10.0,-67.166667
diff --git a/tww/lib.py b/tww/lib.py
index 6c2cc2e..cbde44c 100644
--- a/tww/lib.py
+++ b/tww/lib.py
@@ -33,12 +33,14 @@ from word2number import w2n
 from tww.common import logger
 
 FUZZ_THRESHOLD = 70
+ISO_Z_FORMAT = '%Y-%m-%dT%H:%M:%SZ'
 ISO_FORMAT = '%Y-%m-%dT%H:%M:%S%z'
+ISO_FORMAT_NO_TZ = '%Y-%m-%dT%H:%M:%S'
 DEFAULT_FORMAT = '%Y-%m-%d %H:%M:%S%z'
 
 basepath = os.path.dirname(os.path.abspath(__file__))
 
-SKIP_DATE_TOKENS = ['on', 'to', 'set', 'be']
+SKIP_DATE_TOKENS = ['on', 'to', 'set', 'be', 'an', "me"]
 
 
 def parse_to_iso(dt, dt_hint=None):
@@ -52,7 +54,7 @@ def parse_to_iso(dt, dt_hint=None):
 
 def find_dates_in(text):
     rv = []
-    dateparser_dates = search_dates(text, settings={"SKIP_TOKENS": SKIP_DATE_TOKENS})
+    dateparser_dates = search_dates(text)
     if dateparser_dates:
         rv += list(filter(_dateparser_search_filter, dateparser_dates))
     try:
@@ -604,15 +606,17 @@ def get_local_now(tzaware: bool = True):
     return datetime.utcnow()
 
 
+def split_offset(offset):
+    if ':' in offset:
+        to_shh, to_mm = offset.split(':')
+    else:
+        to_shh, to_mm = offset[:-2], offset[-2:]
+    return int(to_shh), int(to_mm)
+
+
 def dt_tz_translation(dt: datetime, to_tz_offset: str, from_tz_offset: str = "+00:00") -> datetime:
-    if ':' in to_tz_offset:
-        to_shh, to_mm = to_tz_offset.split(':')
-    else:
-        to_shh, to_mm = to_tz_offset[:-2], to_tz_offset[-2:]
-    if ':' in from_tz_offset:
-        from_shh, from_mm = from_tz_offset.split(':')
-    else:
-        from_shh, from_mm = from_tz_offset[:-2], to_tz_offset[-2:]
+    to_shh, to_mm = split_offset(to_tz_offset)
+    from_shh, from_mm = split_offset(from_tz_offset)
     tzinfo = tzinfo_from_offset(to_tz_offset)[0]
     if dt.tzinfo:
         return dt.astimezone(tzinfo)
diff --git a/tww/templates/dt_tr.html b/tww/templates/dt_tr.html
index af2275b..1ee2cd8 100644
--- a/tww/templates/dt_tr.html
+++ b/tww/templates/dt_tr.html
@@ -1,12 +1,12 @@
 <input type="text" class="datetime" name="datetime" disabled
-       value="{{ dst_dt.iso8601_full }}" style="width: 320px;"/>
+       value="{{ dt.iso8601_full }}" style="width: 320px;"/>
 </br></br>
 <input type="text" class="datetime__date_txt" name="datetime__date_txt" disabled
-       value="{{ dst_dt.iso8601_date }}"/>
+       value="{{ dt.iso8601_date }}"/>
 <input type="text" class="datetime__time_txt" name="datetime__time_txt" disabled
-       value="{{ dst_dt.iso8601_time }}"/>
+       value="{{ dt.iso8601_time }}"/>
 <input type="text" class="datetime__tz_offset" name="datetime__tz_offset" disabled
-       value="{{ dst_dt.tz_offset }}" size="8"/>
+       value="{{ dt.tz_offset }}" size="8"/>
 <select class="datetime__tz_name" name="datetime__tz">
     <option disabled selected value> -- select timezone --</option>
     {% for tz in all_tz %}
diff --git a/tww/tokenizer.py b/tww/tokenizer.py
index 6cb2041..0e81cf6 100644
--- a/tww/tokenizer.py
+++ b/tww/tokenizer.py
@@ -9,7 +9,8 @@ from pygments import highlight, lexers, formatters
 from scalpl import Cut
 
 from tww.localization import setlocale, resolve_locale
-from tww.lib import ISO_FORMAT, time_to_emoji, time_ago, workday_diff, workhours_diff, td_remainders, td_totals, td_iso8601
+from tww.lib import ISO_FORMAT, time_to_emoji, time_ago, workday_diff, workhours_diff, td_remainders, td_totals, \
+    td_iso8601, ISO_Z_FORMAT
 from tww.lib import resolve_timezone, dateparser_parse_dt, get_utcnow, get_s_since_epoch, get_ms_since_epoch, \
     dt_tz_translation, get_local_now, query_to_format_result
 from tww.common import logger
@@ -129,7 +130,7 @@ h_unix_s = 'dt->unix_s'
 h_unix_ms = 'dt->unix_ms'
 h_tz_offset = 'tz->tz_offset'
 h_time_in = 'dt->hh:mm'
-h_translation = 'dst_dt->iso8601_full'
+h_translation = 'dt->iso8601_full'
 h_default_dt = 'dt->iso8601_full'
 h_default_td = 'timedelta->diff->duration_iso8601'
 
@@ -207,6 +208,8 @@ def dt_pretty(dt):
     rv = {}
     global custom_locale
     rv["iso8601_full"] = dt.strftime(ISO_FORMAT)
+    rv["iso8601_utc"] = dt_tz_translation(dt, to_tz_offset='+00:00',
+                                          from_tz_offset=dt.strftime('%z')).strftime(ISO_Z_FORMAT)
     rv["iso8601_date"] = dt.strftime('%Y-%m-%d')
     rv["iso8601_time"] = dt.strftime('%H:%M:%S')
     rv["locale_dt"] = dt.strftime("%c")
@@ -247,12 +250,14 @@ def resolve_query_type(query):
     return solutions
 
 
-def resolve_query(query):
+def resolve_query(query, allowed_queries=None):
     rv = {
         "query": query,
         "solutions": [],
     }
     solutions = resolve_query_type(query)
+    if not allowed_queries:
+        allowed_queries = [QUERY_TYPE_DT, QUERY_TYPE_DT_TR, QUERY_TYPE_TD, QUERY_TYPE_TZ]
     for sol_id, solution in enumerate(solutions):
         element = {}
         handler, results, query_type, hi = solution
@@ -261,11 +266,13 @@ def resolve_query(query):
         element["highlight"] = hi
         element["solution_id"] = sol_id
         try:
+            if query_type not in allowed_queries:
+                continue
             if query_type == QUERY_TYPE_DT:
                 element["dt"] = dt_pretty(results)
             elif query_type == QUERY_TYPE_DT_TR:
                 element["src_dt"] = dt_pretty(results[0])
-                element["dst_dt"] = dt_pretty(results[1])
+                element["dt"] = dt_pretty(results[1])
                 element["tz"] = results[2]
             elif query_type == QUERY_TYPE_TZ:
                 element["tz"] = results
diff --git a/tww/web.py b/tww/web.py
index 76b29bc..8d18145 100644
--- a/tww/web.py
+++ b/tww/web.py
@@ -6,7 +6,7 @@ from tokenizer import resolve_query, QUERY_TYPE_DT, QUERY_TYPE_DT_TR, QUERY_TYPE
 app = Flask(__name__)
 
 import dateparser
-from tww import resolve_timezone
+from tww.lib import resolve_timezone
 
 IN_KW = " in "
 TO_KW = " to "