Fix the parsing of the ISO 8601 Z UTC designator (#448)

* Fix parsing UTC designator

* Test the pure Python version first

Author: sdispater

.github/workflows/tests.yml

@@ -48,15 +48,16 @@ jobs:
       run: |
         source $HOME/.poetry/env
         poetry install
+    - name: Test Pure Python
+      run: |
+        source $HOME/.poetry/env
+        PENDULUM_EXTENSIONS=0 poetry run pytest -q tests
     - name: Test
       run: |
         source $HOME/.poetry/env
         poetry run pytest -q tests
         poetry install
-    - name: Test Pure Python
-      run: |
-        source $HOME/.poetry/env
-        PENDULUM_EXTENSIONS=0 poetry run pytest -q tests
+
   MacOS:
     needs: Linting
     runs-on: macos-latest
@@ -89,14 +90,14 @@ jobs:
       run: |
         source $HOME/.poetry/env
         poetry install
-    - name: Test
-      run: |
-        source $HOME/.poetry/env
-        poetry run pytest -q tests
     - name: Test Pure Python
       run: |
         source $HOME/.poetry/env
         PENDULUM_EXTENSIONS=0 poetry run pytest -q tests
+    - name: Test
+      run: |
+        source $HOME/.poetry/env
+        poetry run pytest -q tests
   Windows:
     needs: Linting
     runs-on: windows-latest
@@ -130,12 +131,12 @@ jobs:
       run: |
         $env:Path += ";$env:Userprofile\.poetry\bin"
         poetry install
-    - name: Test
+    - name: Test Pure Python
       run: |
         $env:Path += ";$env:Userprofile\.poetry\bin"
+        $env:PENDULUM_EXTENSIONS = "0"
         poetry run pytest -q tests
-    - name: Test Pure Python
+    - name: Test
       run: |
         $env:Path += ";$env:Userprofile\.poetry\bin"
-        $env:PENDULUM_EXTENSIONS = "0"
         poetry run pytest -q tests

pendulum/__init__.py

@@ -79,6 +79,8 @@ def _safe_timezone(obj):
         # pytz
         if hasattr(obj, "localize"):
             obj = obj.zone
+        elif obj.tzname(None) == "UTC":
+            return UTC
         else:
             offset = obj.utcoffset(None)
 

pendulum/parsing/_iso8601.c

@@ -178,6 +178,7 @@ int is_long_year(int year) {
 typedef struct {
     PyObject_HEAD
     int offset;
+    char *tzname;
 } FixedOffset;
 
 /*
@@ -186,10 +187,16 @@ typedef struct {
 */
 static int FixedOffset_init(FixedOffset *self, PyObject *args, PyObject *kwargs) {
     int offset;
-    if (!PyArg_ParseTuple(args, "i", &offset))
+    char *tzname = NULL;
+
+    static char *kwlist[] = {"offset", "tzname", NULL};
+
+    if (!PyArg_ParseTupleAndKeywords(args, kwargs, "i|s", kwlist, &offset, &tzname))
         return -1;
 
     self->offset = offset;
+    self->tzname = tzname;
+
     return 0;
 }
 
@@ -217,6 +224,10 @@ static PyObject *FixedOffset_dst(FixedOffset *self, PyObject *args) {
  *     return "%s%d:%d" % (sign, self.offset / 60, self.offset % 60)
  */
 static PyObject *FixedOffset_tzname(FixedOffset *self, PyObject *args) {
+    if (self->tzname != NULL) {
+        return PyUnicode_FromString(self->tzname);
+    }
+
     char tzname_[7] = {0};
     char sign = '+';
     int offset = self->offset;
@@ -292,16 +303,17 @@ static PyTypeObject FixedOffset_type = {
  * Skip overhead of calling PyObject_New and PyObject_Init.
  * Directly allocate object.
  */
-static PyObject *new_fixed_offset_ex(int offset, PyTypeObject *type) {
+static PyObject *new_fixed_offset_ex(int offset, char *name, PyTypeObject *type) {
     FixedOffset *self = (FixedOffset *) (type->tp_alloc(type, 0));
 
     if (self != NULL)
         self->offset = offset;
+        self->tzname = name;
 
     return (PyObject *) self;
 }
 
-#define new_fixed_offset(offset) new_fixed_offset_ex(offset, &FixedOffset_type)
+#define new_fixed_offset(offset, name) new_fixed_offset_ex(offset, name, &FixedOffset_type)
 
 
 /*
@@ -455,6 +467,7 @@ typedef struct {
     int microsecond;
     int offset;
     int has_offset;
+    char *tzname;
     int years;
     int months;
     int weeks;
@@ -487,6 +500,7 @@ Parsed* new_parsed() {
         parsed->microsecond = 0;
         parsed->offset = 0;
         parsed->has_offset = 0;
+        parsed->tzname = NULL;
 
         parsed->years = 0;
         parsed->months = 0;
@@ -585,7 +599,7 @@ Parsed* _parse_iso8601_datetime(char *str, Parsed *parsed) {
         }
 
         // Checks
-        if (week > 53 || week > 52 && !is_long_year(parsed->year)) {
+        if (week > 53 || (week > 52 && !is_long_year(parsed->year))) {
             parsed->error = PARSER_INVALID_WEEK_NUMBER;
 
             return NULL;
@@ -850,6 +864,7 @@ Parsed* _parse_iso8601_datetime(char *str, Parsed *parsed) {
         // Timezone
         if (*c == 'Z') {
             parsed->has_offset = 1;
+            parsed->tzname = "UTC";
             c++;
         } else if (*c == '+' || *c == '-') {
             tz_sign = 1;
@@ -1258,7 +1273,7 @@ PyObject* parse_iso8601(PyObject *self, PyObject *args) {
         if (!parsed->has_offset) {
             tzinfo = Py_BuildValue("");
         } else {
-            tzinfo = new_fixed_offset(parsed->offset);
+            tzinfo = new_fixed_offset(parsed->offset, parsed->tzname);
         }
 
         obj = PyDateTimeAPI->DateTime_FromDateAndTime(

pendulum/parsing/iso8601.py

@@ -12,6 +12,7 @@
 from ..helpers import is_leap
 from ..helpers import is_long_year
 from ..helpers import week_day
+from ..tz.timezone import UTC
 from ..tz.timezone import FixedTimezone
 from .exceptions import ParserError
 
@@ -230,7 +231,7 @@ def parse_iso8601(text):
         tz = m.group("tz")
         if tz:
             if tz == "Z":
-                offset = 0
+                tzinfo = UTC
             else:
                 negative = True if tz.startswith("-") else False
                 tz = tz[1:]
@@ -248,7 +249,7 @@ def parse_iso8601(text):
                 if negative:
                     offset = -1 * offset
 
-            tzinfo = FixedTimezone(offset)
+                tzinfo = FixedTimezone(offset)
 
         if is_time:
             return datetime.time(hour, minute, second, microsecond)

tests/test_parsing.py

@@ -131,3 +131,9 @@ def test_parse_now():
 
     with pendulum.test(mock_now):
         assert pendulum.parse("now") == mock_now
+
+
+def test_parse_with_utc_timezone():
+    dt = pendulum.parse("2020-02-05T20:05:37.364951Z")
+
+    assert "2020-02-05T20:05:37.364951Z" == dt.to_iso8601_string()