def test_second_equal(self):
test_string_argument_second_equal = "BILD=Der Todesgang des armenischen Volkes.pdf{{!}}page=276"
test_string_second_equal = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_second_equal + "}}"
test_dict_second_equal = {"key": "BILD", "value": "Der Todesgang des armenischen Volkes.pdf{{!}}page=276"}
test_list_second_equal = [test_dict_argument_1, test_dict_second_equal]
handler = TemplateHandler(test_string_second_equal)
self.assertEqual(test_list_second_equal, handler.get_parameterlist())
def from_text(cls, article_text):
"""
main parser function for initiating a ReArticle from a given piece of text.
:param article_text: text that represent a valid ReArticle
:rtype: Article
"""
finder = TemplateFinder(article_text)
find_re_daten = finder.get_positions(RE_DATEN)
find_re_abschnitt = finder.get_positions(RE_ABSCHNITT)
# only one start template can be present
if len(find_re_daten) + len(find_re_abschnitt) != 1:
raise ReDatenException(
"Article has the wrong structure. There must one start template"
)
if find_re_daten:
find_re_start = find_re_daten
else:
find_re_start = find_re_abschnitt
find_re_author = finder.get_positions(RE_AUTHOR)
# only one end template can be present
if len(find_re_author) != 1:
raise ReDatenException(
"Article has the wrong structure. There must one stop template"
)
# the templates must have the right order
if find_re_start[0]["pos"][0] > find_re_author[0]["pos"][0]:
raise ReDatenException(
"Article has the wrong structure. Wrong order of templates.")
# it can only exists text between the start and the end template.
if find_re_start[0]["pos"][0] != 0:
raise ReDatenException(
"Article has the wrong structure. There is text in front of the article."
)
if find_re_author[0]["pos"][1] != len(article_text):
raise ReDatenException(
"Article has the wrong structure. There is text after the article."
)
try:
re_start = TemplateHandler(find_re_start[0]["text"])
except TemplateHandlerException as error:
raise ReDatenException(
"Start-Template has the wrong structure.") from error
try:
re_author = TemplateHandler(find_re_author[0]["text"])
except TemplateHandlerException as error:
raise ReDatenException(
"Author-Template has the wrong structure.") from error
properties_dict = cls._extract_properties(re_start.parameters)
author_name = re_author.parameters[0]["value"]
try:
author_issue = re_author.parameters[1]["value"]
except IndexError:
author_issue = ""
return Article(article_type=re_start.title,
re_daten_properties=properties_dict,
text=article_text[find_re_start[0]["pos"][1]:
find_re_author[0]["pos"][0]].strip(),
author=(author_name, author_issue))
def test_set_title(self):
test_string_12_test_title = "{{" + test_title_test + "|" \
+ test_string_argument_1 + "|" \
+ test_string_argument_2 + "}}"
handler = TemplateHandler(test_string_12_simple)
handler.set_title(test_title_test)
self.assertEqual(test_string_12_test_title,
handler.get_str(str_complex=False))
def test_update_parameters(self):
test_string_345_simple = "{{" + test_title + "|" + test_string_argument_3 + "|" + test_string_argument_4 + "|" + test_string_argument_5 + "}}"
test_list_345 = [test_dict_argument_3, test_dict_argument_4, test_dict_argument_5]
handler = TemplateHandler(test_string_12_simple)
self.assertEqual(test_dict_argument_1, handler.get_parameter("1"))
self.assertEqual(test_dict_argument_2, handler.get_parameter("2"))
handler.update_parameters(test_list_345)
self.assertEqual(test_string_345_simple, handler.get_str(str_complex=False))
def convert_re_nachtrag(template: str):
template_nachtrag = TemplateHandler(template)
template_daten = TemplateHandler()
new_list = template_nachtrag.get_parameterlist()
new_list.append({"key": "NACHTRAG", "value": "ON"})
template_daten.update_parameters(new_list)
template_daten.set_title("REDaten")
return template_daten.get_str()
def test_without_key(self):
test_string_12_no_key = "{{" + test_title + "|" \
+ test_string_argument_1_no_key + "|" \
+ test_string_argument_2 + "}}"
test_list_12_no_key = [
test_dict_argument_1_no_key, test_dict_argument_2
]
handler = TemplateHandler(test_string_12_no_key)
self.assertEqual(test_list_12_no_key, handler.get_parameterlist())
def convert_re_platzhalter(self, template: str):
template_platzhalter = TemplateHandler(template)
template_daten = TemplateHandler()
new_list = template_platzhalter.get_parameterlist()
new_list = self._gemeinfrei_todesjahr(new_list)
template_daten.update_parameters(new_list)
template_daten.set_title("REDaten")
return template_daten.get_str()
def _get_pre_text(self):
template_handler = TemplateHandler()
template_handler.title = RE_DATEN
list_of_properties = []
for re_property in self._properties:
list_of_properties.append({"key": re_property.name,
"value": re_property.value_to_string()})
template_handler.update_parameters(list_of_properties)
return template_handler.get_str(str_complex=True)
def convert_re_nachtrag_platzhalter(self, template: str):
template_platzhalter = TemplateHandler(template)
template_daten = TemplateHandler()
new_list = template_platzhalter.get_parameterlist()
new_list.append({"key": "NACHTRAG", "value": "ON"})
new_list = self._gemeinfrei_todesjahr(new_list)
template_daten.update_parameters(new_list)
template_daten.set_title("REDaten")
return template_daten.get_str()
def test_second_equal(self):
test_string_argument_second_equal = "BILD=Der Todesgang des armenischen Volkes.pdf{{!}}page=276"
test_string_second_equal = "{{" + test_title_test + "|" + \
test_string_argument_1 + "|" + \
test_string_argument_second_equal + "}}"
test_dict_second_equal = {
"key": "BILD",
"value": "Der Todesgang des armenischen Volkes.pdf{{!}}page=276"
}
test_list_second_equal = [test_dict_argument_1, test_dict_second_equal]
handler = TemplateHandler(test_string_second_equal)
self.assertEqual(test_list_second_equal, handler.get_parameterlist())
def substitute_sperrsatz(template):
handler = TemplateHandler(template.group(0))
handler.set_title('SperrSchrift')
parameters = handler.get_parameterlist()
parameters.append({'key': 'satz', 'value': '1'})
handler.update_parameters(parameters)
return handler.get_str(str_complex=False)
def test_get_str(self):
handler = TemplateHandler()
handler.set_title(test_title)
handler.update_parameters(test_list_12)
self.assertEqual(test_string_12_simple,
handler.get_str(str_complex=False))
self.assertEqual(test_string_12_complex,
handler.get_str(str_complex=True))
def test_link_with_text(self):
test_string_argument_2_link = "2 = [[link|text for link]] more"
test_string_12_link = "{{" + test_title + "|" + test_string_argument_1_no_key + "|" + test_string_argument_2_link + "}}"
test_dict_link = {"key": "2", "value": "[[link|text for link]] more"}
test_list_link = [test_dict_argument_1_no_key, test_dict_link]
handler = TemplateHandler(test_string_12_link)
self.assertEqual(test_list_link, handler.get_parameterlist())
del handler
test_string_argument_link = "[[link|text for link]] more"
test_string_12_link_no_key = "{{" + test_title + "|" + test_string_argument_1_no_key + "|" + test_string_argument_link + "}}"
test_dict_link_no_key = {"key": None, "value": "[[link|text for link]] more"}
test_list_link_no_key = [test_dict_argument_1_no_key, test_dict_link_no_key]
handler = TemplateHandler(test_string_12_link_no_key)
self.assertEqual(test_list_link_no_key, handler.get_parameterlist())
def convert_re_platzhalter(self, template: str):
template_platzhalter = TemplateHandler(template)
template_daten = TemplateHandler()
new_list = template_platzhalter.get_parameterlist()
new_list = self._gemeinfrei_todesjahr(new_list)
template_daten.update_parameters(new_list)
template_daten.set_title("REDaten")
return template_daten.get_str()
def convert_re_nachtrag(template: str):
template_nachtrag = TemplateHandler(template)
template_daten = TemplateHandler()
new_list = template_nachtrag.get_parameterlist()
new_list.append({"key": "NACHTRAG", "value": "ON"})
template_daten.update_parameters(new_list)
template_daten.set_title("REDaten")
return template_daten.get_str()
def convert_re_nachtrag_platzhalter(self, template: str):
template_platzhalter = TemplateHandler(template)
template_daten = TemplateHandler()
new_list = template_platzhalter.get_parameterlist()
new_list.append({"key": "NACHTRAG", "value": "ON"})
new_list = self._gemeinfrei_todesjahr(new_list)
template_daten.update_parameters(new_list)
template_daten.set_title("REDaten")
return template_daten.get_str()
def test_bug_authorlist(self):
test_string_argument_bug = "STERBEDATUM = 2. Januar < ref name = \"adp\" / > oder 31. Januar < ref > 49. Jahres - Bericht d.Schles.Ges.für vaterländische Cultur, S. 317, Nekrolog {{GBS|hP1DAAAAIAAJ|PA317}} < / ref > 1871"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug = {"key": "STERBEDATUM", "value": "2. Januar < ref name = \"adp\" / > oder 31. Januar < ref > 49. Jahres - Bericht d.Schles.Ges.für vaterländische Cultur, S. 317, Nekrolog {{GBS|hP1DAAAAIAAJ|PA317}} < / ref > 1871"}
test_list_bug = [test_dict_argument_1, test_dict_bug]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "GEBURTSDATUM=1783 < ref name = \"EB\" > Encyclopaedia Britannica. 11. Auflage(1911), Bd. 1, S.[[:en:Page:EB1911 - Volume 01. djvu / 792 | 748]] {{an | englisch, im Artikel}} < / ref >"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug = {"key": "GEBURTSDATUM", "value": "1783 < ref name = \"EB\" > Encyclopaedia Britannica. 11. Auflage(1911), Bd. 1, S.[[:en:Page:EB1911 - Volume 01. djvu / 792 | 748]] {{an | englisch, im Artikel}} < / ref >"}
test_list_bug = [test_dict_argument_1, test_dict_bug]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "GEBURTSORT=Klein Flottbek (heute zu [[Hamburg]])|STERBEDATUM=28. Oktober 1929|STERBEORT=[[Rom]]"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug_1 = {"key": "GEBURTSORT", "value": "Klein Flottbek (heute zu [[Hamburg]])"}
test_dict_bug_2 = {"key": "STERBEDATUM", "value": "28. Oktober 1929"}
test_dict_bug_3 = {"key": "STERBEORT", "value": "[[Rom]]"}
test_list_bug = [test_dict_argument_1, test_dict_bug_1, test_dict_bug_2, test_dict_bug_3]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "ALTERNATIVNAMEN = Carl Biedermann; Friedrich Karl Biedermann; Karl Friedrich 4[Pseudonym]|SONSTIGES=[http://gso.gbv.de/DB=1.28/REL?PPN=004072189&RELTYPE=TT Martin Opitz im VD 17]"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug_1 = {"key": "ALTERNATIVNAMEN", "value": "Carl Biedermann; Friedrich Karl Biedermann; Karl Friedrich 4[Pseudonym]"}
test_dict_bug_2 = {"key": "SONSTIGES", "value": "[http://gso.gbv.de/DB=1.28/REL?PPN=004072189&RELTYPE=TT Martin Opitz im VD 17]"}
test_list_bug = [test_dict_argument_1, test_dict_bug_1, test_dict_bug_2]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "SONSTIGES=Pächter der [[w:Harste|Domäne Harste]], Vater von [[w:Karl Henrici|Karl Henrici]]<ref>""Zeitschrift des Vereins für Hamburgische Geschichte."" Band 42. 1953, S. 135 [http://books.google.de/books?id=1XISAAAAIAAJ&q=%2B%22henrici%22+%2B%221885%22+%2B%22harste%22&dq=%2B%22henrici%22+%2B%221885%22+%2B%22harste%22 Google]</ref>"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug = {"key": "SONSTIGES", "value": "Pächter der [[w:Harste|Domäne Harste]], Vater von [[w:Karl Henrici|Karl Henrici]]<ref>""Zeitschrift des Vereins für Hamburgische Geschichte."" Band 42. 1953, S. 135 [http://books.google.de/books?id=1XISAAAAIAAJ&q=%2B%22henrici%22+%2B%221885%22+%2B%22harste%22&dq=%2B%22henrici%22+%2B%221885%22+%2B%22harste%22 Google]</ref>"}
test_list_bug = [test_dict_argument_1, test_dict_bug]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
def test_link_with_text(self):
test_string_argument_2_link = "2 = [[link|text for link]] more"
test_string_12_link = "{{" + test_title + "|" \
+ test_string_argument_1_no_key + "|" \
+ test_string_argument_2_link + "}}"
test_dict_link = {"key": "2", "value": "[[link|text for link]] more"}
test_list_link = [test_dict_argument_1_no_key, test_dict_link]
handler = TemplateHandler(test_string_12_link)
self.assertEqual(test_list_link, handler.get_parameterlist())
del handler
test_string_argument_link = "[[link|text for link]] more"
test_string_12_link_no_key = "{{" + test_title + "|" \
+ test_string_argument_1_no_key + "|" \
+ test_string_argument_link + "}}"
test_dict_link_no_key = {
"key": None,
"value": "[[link|text for link]] more"
}
test_list_link_no_key = [
test_dict_argument_1_no_key, test_dict_link_no_key
]
handler = TemplateHandler(test_string_12_link_no_key)
self.assertEqual(test_list_link_no_key, handler.get_parameterlist())
def _get_pre_text(self):
template_handler = TemplateHandler()
template_handler.title = RE_DATEN
list_of_properties = []
for re_property in self._properties:
list_of_properties.append({
"key": re_property.name,
"value": re_property.value_to_string()
})
template_handler.update_parameters(list_of_properties)
return template_handler.get_str(str_complex=True)
def test_template_in_template(self):
test_string_argument_template = "{{otherTemplate|other_argument}}"
test_string_12_template = "{{" + test_title + "|" \
+ test_string_argument_template + "|" \
+ test_string_argument_2 + "}}"
test_dict_template_no_key = {
"key": None,
"value": "{{otherTemplate|other_argument}}"
}
test_list_template_no_key = [
test_dict_template_no_key, test_dict_argument_2
]
handler = TemplateHandler(test_string_12_template)
self.assertListEqual(test_list_template_no_key,
handler.get_parameterlist())
del handler
test_string_argument_template2 = "{{Kapitaelchen|Test}}"
test_string_template_2 = "{{" + test_title_sperr + "|" + test_string_argument_template2 + "}}"
test_dict_template_2 = {"key": None, "value": "{{Kapitaelchen|Test}}"}
test_list_template_2 = [test_dict_template_2]
handler = TemplateHandler(test_string_template_2)
self.assertListEqual(test_list_template_2, handler.get_parameterlist())
del handler
test_string_argument_1_template = "1={{otherTemplate|other_argument}}"
test_string_12_template_no_key = "{{" + test_title + "|" \
+ test_string_argument_1_template + "|" \
+ test_string_argument_2 + "}}"
test_dict_template = {
"key": "1",
"value": "{{otherTemplate|other_argument}}"
}
test_list_template = [test_dict_template, test_dict_argument_2]
handler = TemplateHandler(test_string_12_template_no_key)
self.assertListEqual(test_list_template, handler.get_parameterlist())
def test_update_parameters(self):
test_string_345_simple = "{{" + test_title + "|" \
+ test_string_argument_3 + "|" \
+ test_string_argument_4 + "|" \
+ test_string_argument_5 + "}}"
test_list_345 = [
test_dict_argument_3, test_dict_argument_4, test_dict_argument_5
]
handler = TemplateHandler(test_string_12_simple)
self.assertEqual(test_dict_argument_1, handler.get_parameter("1"))
self.assertEqual(test_dict_argument_2, handler.get_parameter("2"))
handler.update_parameters(test_list_345)
self.assertEqual(test_string_345_simple,
handler.get_str(str_complex=False))
def test_template_in_template(self):
test_string_argument_template = "{{otherTemplate|other_argument}}"
test_string_12_template = "{{" + test_title + "|" + test_string_argument_template + "|" + test_string_argument_2 + "}}"
test_dict_template_no_key = {"key": None, "value": "{{otherTemplate|other_argument}}"}
test_list_template_no_key = [test_dict_template_no_key, test_dict_argument_2]
handler = TemplateHandler(test_string_12_template)
self.assertListEqual(test_list_template_no_key, handler.get_parameterlist())
del handler
test_string_argument_template2 = "{{Kapitaelchen|Test}}"
test_string_template_2 = "{{" + test_title_sperr + "|" + test_string_argument_template2 + "}}"
test_dict_template_2 = {"key": None, "value": "{{Kapitaelchen|Test}}"}
test_list_template_2 = [test_dict_template_2]
handler = TemplateHandler(test_string_template_2)
self.assertListEqual(test_list_template_2, handler.get_parameterlist())
del handler
test_string_argument_1_template = "1={{otherTemplate|other_argument}}"
test_string_12_template_no_key = "{{" + test_title + "|" + test_string_argument_1_template + "|" + test_string_argument_2 + "}}"
test_dict_template = {"key": "1", "value": "{{otherTemplate|other_argument}}"}
test_list_template = [test_dict_template, test_dict_argument_2]
handler = TemplateHandler(test_string_12_template_no_key)
self.assertListEqual(test_list_template, handler.get_parameterlist())
def test_get_parameter(self):
handler = TemplateHandler(test_string_12_complex)
self.assertEqual(test_dict_argument_1, handler.get_parameter("1"))
self.assertEqual(test_dict_argument_2, handler.get_parameter("2"))
def test_get_parameter(self):
handler = TemplateHandler(test_string_12_complex)
self.assertEqual(test_dict_argument_1, handler.get_parameter("1"))
self.assertEqual(test_dict_argument_2, handler.get_parameter("2"))
def test_template_from_page(self):
handler = TemplateHandler(test_string_12_complex)
self.assertEqual(test_list_12, handler.get_parameterlist())
def test_bug_authorlist(self):
test_string_argument_bug = "STERBEDATUM = 2. Januar < ref name = \"adp\" / > oder 31. Januar " \
"< ref > 49. Jahres - Bericht d.Schles.Ges.für vaterländische Cultur, S. 317, " \
"Nekrolog {{GBS|hP1DAAAAIAAJ|PA317}} < / ref > 1871"
test_string_bug = "{{" + test_title_test + "|" \
+ test_string_argument_1 + "|" \
+ test_string_argument_bug + "}}"
test_dict_bug = {
"key":
"STERBEDATUM",
"value":
"2. Januar < ref name = \"adp\" / > oder 31. Januar "
"< ref > 49. Jahres - Bericht d.Schles.Ges.für vaterländische Cultur, S. 317, "
"Nekrolog {{GBS|hP1DAAAAIAAJ|PA317}} < / ref > 1871"
}
test_list_bug = [test_dict_argument_1, test_dict_bug]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "GEBURTSDATUM=1783 < ref name = \"EB\" > Encyclopaedia Britannica. " \
"11. Auflage(1911), Bd. 1, S.[[:en:Page:EB1911 - Volume 01. djvu / 792 | 748]] " \
"{{an | englisch, im Artikel}} < / ref >"
test_string_bug = "{{" + test_title_test + "|" \
+ test_string_argument_1 + "|" \
+ test_string_argument_bug + "}}"
test_dict_bug = {
"key":
"GEBURTSDATUM",
"value":
"1783 < ref name = \"EB\" > Encyclopaedia Britannica. "
"11. Auflage(1911), Bd. 1, S.[[:en:Page:EB1911 - Volume 01. djvu / 792 | 748]] "
"{{an | englisch, im Artikel}} < / ref >"
}
test_list_bug = [test_dict_argument_1, test_dict_bug]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "GEBURTSORT=Klein Flottbek (heute zu [[Hamburg]])" \
"|STERBEDATUM=28. Oktober 1929|STERBEORT=[[Rom]]"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug_1 = {
"key": "GEBURTSORT",
"value": "Klein Flottbek (heute zu [[Hamburg]])"
}
test_dict_bug_2 = {"key": "STERBEDATUM", "value": "28. Oktober 1929"}
test_dict_bug_3 = {"key": "STERBEORT", "value": "[[Rom]]"}
test_list_bug = [
test_dict_argument_1, test_dict_bug_1, test_dict_bug_2,
test_dict_bug_3
]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "ALTERNATIVNAMEN = Carl Biedermann; Friedrich Karl Biedermann; " \
"Karl Friedrich 4[Pseudonym]" \
"|SONSTIGES=[http://gso.gbv.de/DB=1.28/REL?PPN=004072189&RELTYPE=TT " \
"Martin Opitz im VD 17]"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug_1 = {
"key":
"ALTERNATIVNAMEN",
"value":
"Carl Biedermann; Friedrich Karl Biedermann; Karl Friedrich 4[Pseudonym]"
}
test_dict_bug_2 = {
"key":
"SONSTIGES",
"value":
"[http://gso.gbv.de/DB=1.28/REL?PPN=004072189&RELTYPE=TT Martin Opitz im VD 17]"
}
test_list_bug = [
test_dict_argument_1, test_dict_bug_1, test_dict_bug_2
]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
test_string_argument_bug = "SONSTIGES=Pächter der [[w:Harste|Domäne Harste]], " \
"Vater von [[w:Karl Henrici|Karl Henrici]]<ref>Zeitschrift des Vereins für " \
"Hamburgische Geschichte."" Band 42. 1953, S. 135 " \
"[http://books.google.de/books?id=1XISAAAAIAAJ Google]</ref>"
test_string_bug = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_bug + "}}"
test_dict_bug = {
"key":
"SONSTIGES",
"value":
"Pächter der [[w:Harste|Domäne Harste]], "
"Vater von [[w:Karl Henrici|Karl Henrici]]<ref>"
"Zeitschrift des Vereins für "
"Hamburgische Geschichte."
" Band 42. 1953, S. 135 "
"[http://books.google.de/books?id=1XISAAAAIAAJ Google]</ref>"
}
test_list_bug = [test_dict_argument_1, test_dict_bug]
handler = TemplateHandler(test_string_bug)
real_dict = handler.get_parameterlist()
self.assertEqual(test_list_bug, real_dict)
def test_template_from_page(self):
handler = TemplateHandler(test_string_12_complex)
self.assertEqual(test_list_12, handler.get_parameterlist())
def test_without_key(self):
test_string_12_no_key = "{{" + test_title + "|" + test_string_argument_1_no_key + "|" + test_string_argument_2 + "}}"
test_list_12_no_key = [test_dict_argument_1_no_key, test_dict_argument_2]
handler = TemplateHandler(test_string_12_no_key)
self.assertEqual(test_list_12_no_key, handler.get_parameterlist())
elif number < 1000:
for members in range(digits - 3):
number_str = "0" + number_str
return number_str
site = pywikibot.Site()
for idx, i in enumerate(range(6, 296, 2)):
#for idx, i in enumerate(range(120, 144, 2)):
if i < 206:
continue
page = pywikibot.Page(site,
'Orbis sensualium pictus/{}'.format(titles[idx]))
print(i, titles[idx])
handler = TemplateHandler()
first_page = proofreadpage.ProofreadPage(
site, "Seite:OrbisPictus {}.jpg".format(add_zeros(i, 3)))
second_page = proofreadpage.ProofreadPage(
site, "Seite:OrbisPictus {}.jpg".format(add_zeros(i + 1, 3)))
status_1 = first_page.status
status_2 = second_page.status
if status_1 == "Fertig" and status_2 == "Fertig":
status = "Fertig"
elif status_1 == "Unkorrigiert" or status_2 == "Unkorrigiert":
status = "Unkorrigiert"
else:
status = "Korrigiert"
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil',
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil/Achtes Buch',
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil/Eilftes Buch',
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil/Neuntes Buch',
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil/Siebentes Buch',
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil/Zehntes Buch',
'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Zweyter Theil/Zwölftes Buch'
]
wiki = pywikibot.Site()
lemma_dummy = 'Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung/Erster Theil/Erstes Buch'
page = pywikibot.Page(wiki, title='Benutzer:THEbotIT/Test2')
template_textdaten = re.search('\{\{Textdaten(?:\{\{.*?\}\}|.)*?\}\}',
page.text, re.DOTALL).group()
list_textdaten = TemplateHandler(template_textdaten).get_parameterlist()
template_navigation1 = TemplateHandler()
template_navigation1.set_title('Navigation2')
template_navigation2 = TemplateHandler()
template_navigation2.set_title('Navigation2')
list_navigation = []
list_navigation.append({'key': 'AUTOR', 'value': '[[Basilius von Ramdohr]]'})
list_navigation.append({
'key':
'ARTIKEL',
'value':
'[[Venus Urania. Ueber die Natur der Liebe, über ihre Veredelung und Verschönerung|Venus Urania]]'
})
list_navigation.append({'key': 'KAPITEL', 'value': list_textdaten[3]['value']})
def test_bug_no_arguments(self):
test_string = "{{just_this}}"
handler = TemplateHandler(test_string)
self.assertListEqual([], handler.get_parameterlist())
def _build_database(self, lemma_list):
# pylint: disable=too-many-statements
for idx, author in enumerate(lemma_list):
self.logger.debug(f"{idx + 1}/{len(lemma_list)} {author['title']}")
# delete preexisting data of this author
try:
del self.data[str(author["id"])]
except KeyError:
if self.last_run_successful:
self.logger.info(
f"Can't delete old entry of [[{author['title']}]]")
dict_author = {"title": author["title"]}
# extract the Personendaten-block form the wikisource page
page = Page(self.wiki, author["title"])
try:
try:
personendaten = re.search(
r"\{\{Personendaten(?:.|\n)*?\n\}\}\n",
page.text).group()
except AttributeError:
self.logger.error(
f"No valid block \"Personendaten\" was found for "
f"[[{author['title']}]].")
personendaten = None
if personendaten:
# personendaten = re.sub('<ref.*?>.*?<\/ref>|<ref.*?\/>', '', personendaten)
# personendaten = re.sub('\{\{CRef|.*?(?:\{\{.*?\}\})?}}', '', personendaten)
template_extractor = TemplateHandler(personendaten)
dict_author.update({
"name":
self._strip_spaces(
template_extractor.get_parameter("NACHNAME")
["value"])
})
dict_author.update({
"first_name":
self._strip_spaces(
template_extractor.get_parameter("VORNAMEN")
["value"])
})
try:
dict_author.update({
"birth":
self._strip_spaces(
template_extractor.get_parameter(
"GEBURTSDATUM")["value"])
})
except Exception:
dict_author.update({"birth": ""})
self.logger.warning(
f"Templatehandler couldn't find a birthdate for: "
f"[[{author['title']}]]")
try:
dict_author.update({
"death":
self._strip_spaces(
template_extractor.get_parameter("STERBEDATUM")
["value"])
})
except Exception:
dict_author.update({"death": ""})
self.logger.warning(
f"Templatehandler couldn't find a deathdate for: "
f"[[{author['title']}]]")
try:
dict_author.update({
"description":
template_extractor.get_parameter(
"KURZBESCHREIBUNG")["value"]
})
except Exception:
dict_author.update({"description": ""})
self.logger.warning(
f"Templatehandler couldn't find a description for: "
f"[[{author['title']}]]")
try:
dict_author.update({
"synonyms":
template_extractor.get_parameter("ALTERNATIVNAMEN")
["value"]
})
except Exception:
dict_author.update({"synonyms": ""})
self.logger.warning(
f"Templatehandler couldn't find synonyms for: "
f"[[{author['title']}]]")
try:
dict_author.update({
"sortkey":
template_extractor.get_parameter("SORTIERUNG")
["value"]
})
if dict_author["sortkey"] == "":
raise ValueError
except Exception:
self.logger.debug(
f"there is no sortkey for [[{author['title']}]].")
# make a dummy key
if not dict_author["name"]:
dict_author["sortkey"] = dict_author["first_name"]
self.logger.warning("Author has no last name.")
elif not dict_author["first_name"]:
dict_author["sortkey"] = dict_author["name"]
self.logger.warning(
"Author has no last first_name.")
else:
dict_author["sortkey"] = \
dict_author["name"] + ", " + dict_author["first_name"]
try:
dict_author.update({"wikidata": author["q"]})
except KeyError:
self.logger.warning(
f"The autor [[{author['title']}]] has no wikidata_item"
)
self.data.update({author["id"]: dict_author})
except Exception as exception:
self.logger.exception("Exception not catched: ",
exc_info=exception)
self.logger.error(f"author {author['title']} have a problem")
def test_set_title(self):
test_string_12_test_title = "{{" + test_title_test + "|" + test_string_argument_1 + "|" + test_string_argument_2 + "}}"
handler = TemplateHandler(test_string_12_simple)
handler.set_title(test_title_test)
self.assertEqual(test_string_12_test_title, handler.get_str(str_complex=False))
def test_bug_no_arguments(self):
test_string = "{{just_this}}"
handler = TemplateHandler(test_string)
self.assertListEqual([], handler.get_parameterlist())
def _build_database(self, lemma_list):
# pylint: disable=too-many-statements
for idx, author in enumerate(lemma_list):
self.logger.debug(f"{idx + 1}/{len(lemma_list)} {author['title']}")
# delete preexisting data of this author
try:
del self.data[str(author["id"])]
except KeyError:
if self.last_run_successful:
self.logger.info(f"Can't delete old entry of [[{author['title']}]]")
dict_author = {"title": author["title"]}
# extract the Personendaten-block form the wikisource page
page = Page(self.wiki, author["title"])
try:
try:
personendaten = re.search(r"\{\{Personendaten(?:.|\n)*?\n\}\}\n",
page.text).group()
except AttributeError:
self.logger.error(f"No valid block \"Personendaten\" was found for "
f"[[{author['title']}]].")
personendaten = None
if personendaten:
# personendaten = re.sub('<ref.*?>.*?<\/ref>|<ref.*?\/>', '', personendaten)
# personendaten = re.sub('\{\{CRef|.*?(?:\{\{.*?\}\})?}}', '', personendaten)
template_extractor = TemplateHandler(personendaten)
dict_author.update({"name": self._strip_spaces(
template_extractor.get_parameter("NACHNAME")["value"])})
dict_author.update({"first_name": self._strip_spaces(
template_extractor.get_parameter("VORNAMEN")["value"])})
try:
dict_author.update({"birth": self._strip_spaces(
template_extractor.get_parameter("GEBURTSDATUM")["value"])})
except Exception:
dict_author.update({"birth": ""})
self.logger.warning(f"Templatehandler couldn't find a birthdate for: "
f"[[{author['title']}]]")
try:
dict_author.update({"death": self._strip_spaces(
template_extractor.get_parameter("STERBEDATUM")["value"])})
except Exception:
dict_author.update({"death": ""})
self.logger.warning(f"Templatehandler couldn't find a deathdate for: "
f"[[{author['title']}]]")
try:
dict_author.update(
{"description":
template_extractor.get_parameter("KURZBESCHREIBUNG")["value"]})
except Exception:
dict_author.update({"description": ""})
self.logger.warning(
f"Templatehandler couldn't find a description for: "
f"[[{author['title']}]]")
try:
dict_author.update(
{"synonyms":
template_extractor.get_parameter("ALTERNATIVNAMEN")["value"]})
except Exception:
dict_author.update({"synonyms": ""})
self.logger.warning(f"Templatehandler couldn't find synonyms for: "
f"[[{author['title']}]]")
try:
dict_author.update(
{"sortkey": template_extractor.get_parameter("SORTIERUNG")["value"]})
if dict_author["sortkey"] == "":
raise ValueError
except Exception:
self.logger.debug(f"there is no sortkey for [[{author['title']}]].")
# make a dummy key
if not dict_author["name"]:
dict_author["sortkey"] = dict_author["first_name"]
self.logger.warning("Author has no last name.")
elif not dict_author["first_name"]:
dict_author["sortkey"] = dict_author["name"]
self.logger.warning("Author has no last first_name.")
else:
dict_author["sortkey"] = \
dict_author["name"] + ", " + dict_author["first_name"]
try:
dict_author.update({"wikidata": author["q"]})
except KeyError:
self.logger.warning(f"The autor [[{author['title']}]] has no wikidata_item")
self.data.update({author["id"]: dict_author})
except Exception as exception:
self.logger.exception("Exception not catched: ", exc_info=exception)
self.logger.error(f"author {author['title']} have a problem")
for members in range(digits-2):
number_str = "0" + number_str
elif number < 1000:
for members in range(digits-3):
number_str = "0" + number_str
return number_str
site = pywikibot.Site()
for idx, i in enumerate(range(6, 296, 2)):
#for idx, i in enumerate(range(120, 144, 2)):
if i < 206:
continue
page = pywikibot.Page(site, 'Orbis sensualium pictus/{}'.format(titles[idx]))
print(i, titles[idx])
handler = TemplateHandler()
first_page = proofreadpage.ProofreadPage(site, "Seite:OrbisPictus {}.jpg".format(add_zeros(i, 3)))
second_page = proofreadpage.ProofreadPage(site, "Seite:OrbisPictus {}.jpg".format(add_zeros(i + 1, 3)))
status_1 = first_page.status
status_2 = second_page.status
if status_1 == "Fertig" and status_2 == "Fertig":
status = "Fertig"
elif status_1 == "Unkorrigiert" or status_2 == "Unkorrigiert":
status = "Unkorrigiert"
else:
status = "Korrigiert"
print(status_1, status_2, status)
def test_get_str(self):
handler = TemplateHandler()
handler.set_title(test_title)
handler.update_parameters(test_list_12)
self.assertEqual(test_string_12_simple, handler.get_str(str_complex=False))
self.assertEqual(test_string_12_complex, handler.get_str(str_complex=True))