12455756921

Committed 22 Dec 2024 04:07PM UTC coverage: 62.943% (-0.7%) from 63.692%

Build # 12455756921

Build Type

push

github

Committed by

AntiCompositeNumber

Commit Message

Check untagged error log directly instead of skip list

Run Details

191 of 346 branches covered (55.2%)

Branch coverage included in aggregate %.

7 of 9 new or added lines in 2 files covered. (77.78%)

66 existing lines in 1 file now uncovered.

643 of 979 relevant lines covered (65.68%)

0.66 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/src/inrcli.py

#!/usr/bin/env python3
# coding: utf-8
# SPDX-License-Identifier: GPL-3.0-or-later
# Copyright 2023 AntiCompositeNumber

import click
import pywikibot  # type: ignore
import pywikibot.bot  # type: ignore
import logging
import os
import sys
import difflib
import webbrowser
from typing import Sequence, Dict, Optional, Tuple

os.environ["LOG_FILE"] = "stderr"
os.environ["LOG_LEVEL"] = "WARNING"

import inrbot  # noqa: E402

inrbot.run_override = True
inrbot.summary_tag = f"(inrcli {inrbot.__version__})"
site = inrbot.site
logger = logging.getLogger("manual")
ids: Dict[pywikibot.Page, Optional[inrbot.iNaturalistID]] = {}
auto_open = False
last_ina_id = None

inrbot.config.update(
    {
        "fail_tag": "{{copyvio|License review not passed: "
        "iNaturalist author is using $review_license: $source_url}}\n",
        "fail_warn": "\n\n{{subst:Copyvionote |1=$filename "
        "|2=License review "
        "not passed: iNaturalist author is using $review_license: $source_url }} ~~~~",
        "review_summary": "Semi-automatic license review: "
        "$status $review_license $tag",
        "old_fail_warn": "\n\n{{subst:image permission|1=$filename}} "
        "License review not passed: iNaturalist author is "
        "using $review_license: $source_url. ~~~~",
        "use_wayback": False,
    }
)


class SkipFile(Exception):
    pass


def manual_compare(
    com_img: inrbot.CommonsImage, ina_img: inrbot.iNaturalistImage, **kwargs
):
    if ids.get(com_img.page, None) == ina_img.id and ina_img.id.type == "photos":
        return True
    else:
        return ask_compare(com_img, ina_img)


def ask_compare(com_img: inrbot.CommonsImage, ina_img: inrbot.iNaturalistImage):
    if click.confirm(f"Show {ina_img.id}?", default=False):
        com_img.image.show(title=com_img.page.title())
        ina_img.image.show(title=str(ina_img.id))
    res = click.confirm("Do these images match?", default=False)
    return res


inrbot.compare_methods.insert(0, ("manual", manual_compare))


class ManualCommonsPage(inrbot.CommonsPage):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

    def check_can_run(self) -> bool:
        """Determinies if the bot should run on this page and returns a bool."""
        page = self.page
        if (not page.has_permission("edit")) or (not page.botMayEdit()):
            return False
        else:
            return True

    def get_old_archive_(self):
        return super().get_old_archive()

    def get_old_archive(self):
        # Archives will have already been reviewed by the archive_status_hook
        if self.status != "fail":
            return super().get_old_archive()
        return ""

    def archive_status_hook(self) -> None:
        if self._status == "fail":
            super().get_old_archive()
            if self.archive:
                print(
                    f"This file would fail because of the {self.ina_license} license, "
                    f"but an archived copy is available at {self.archive}."
                )
                new_license = click.prompt(
                    "Archive license (leave blank for no change)",
                    default=self.ina_license,
                )
                if new_license:
                    self.ina_license = new_license
                    del self.status
                    self.status

    @staticmethod
    def prompt_photo_url(
        default: Optional[str] = None,
    ) -> Tuple[Optional[inrbot.iNaturalistID], str]:
        global last_ina_id
        while True:
            # If default is not None, click will repeatedly prompt until it gets
            # an answer. That, however, doesn't mean we got an answer we like.
            url = click.prompt("iNaturalist Photos URL", default=default)
            ina_id = inrbot.parse_ina_url(url)
            if ina_id == last_ina_id:
                if click.confirm(
                    "That's the same URL you gave last time. Are you sure?"
                ):
                    break
                else:
                    continue
            break
        if ina_id is not None:
            last_ina_id = ina_id
            return ina_id, ""
        return ina_id, url

    def pre_save(self, new_text, summary, **kwargs):
        print(
            f"{self.page.title(as_link=True)} reviewed with status {self.status} "
            f"and license {self.ina_license}"
        )
        if self.status == "error":
            raise RuntimeError

        diff = difflib.unified_diff(
            self.page.get().split("\n"), new_text.split("\n"), lineterm=""
        )
        print("\n".join(diff))
        choice = click.confirm("Save the page?", default=True)
        if choice:
            return new_text, summary
        else:
            raise RuntimeError

    def id_hook(
        self,
        observations: Sequence[inrbot.iNaturalistID] = [],
        photos: Sequence[inrbot.iNaturalistID] = [],
        **kwargs,
    ):
        try:
            return ids[self.page]
        except KeyError:
            return self.ask_url(observations=observations, photos=photos)
        return None

    def ask_url(
        self,
        observations: Sequence[inrbot.iNaturalistID] = [],
        photos: Sequence[inrbot.iNaturalistID] = [],
    ):
        print(f"Commons page: {self.page.full_url()}")
        if observations:
            print(f"Observation ID found: {str(observations[0])}")
        if photos:
            print(f"Photo ID found: {str(photos[0])}")
        if auto_open:
            webbrowser.open(self.page.full_url())
            if observations and not photos:
                webbrowser.open(str(observations[0]))
            elif photos:
                webbrowser.open(str(photos[0]))

        res = click.prompt(
            "Is this ID correct? [Y/n/r/s/q]",
            default="Y",
            type=click.Choice("ynrsq", case_sensitive=False),
            show_default=False,
            show_choices=False,
        ).lower()
        if res == "y":
            correct_id = True
        elif res == "n":
            correct_id = False
        elif res == "r":
            self.remove_untagged_log()
            raise SkipFile
        elif res == "s":
            raise SkipFile
        elif res == "q":
            sys.exit()

        if not correct_id:
            ina_id, url = self.prompt_photo_url()
            if observations:
                self.page.text = self.page.text.replace(str(observations[0]), url)
            if photos:
                self.page.text = self.page.text.replace(str(photos[0]), url)
        elif observations and not photos:
            ina_id, url = self.prompt_photo_url(default="")
        elif photos:
            ina_id = photos[0]
        else:
            ina_id = None
        ids[self.page] = ina_id
        return ina_id

    def log_untagged_error(self) -> None:
        # Errors while running in CLI do not need to be logged on-wiki
        return


inrbot.id_hooks.append(ManualCommonsPage.id_hook)
inrbot.lock_hooks.append(ManualCommonsPage.archive_status_hook)
inrbot.pre_save_hooks.append(ManualCommonsPage.pre_save)


@click.command()
@click.argument("target")
@click.option("--url")
@click.option("--simulate/--no-simulate")
@click.option("--reverse", is_flag=True, default=False)
@click.option("-o", "--auto-open", "auto_open_", is_flag=True, default=False)
def main(target, url="", simulate=False, reverse=False, auto_open_=False):
    inrbot.simulate = simulate
    global auto_open
    auto_open = auto_open_
    if target == "auto":
        cat = pywikibot.Category(
            site, "Category:iNaturalist images needing human review"
        )
        dtt = pywikibot.Page(site, "Template:Deletion template tag")
        for page in cat.articles(namespaces=6, reverse=reverse):
            if dtt in set(page.itertemplates()):
                continue
            try:
                mcp = ManualCommonsPage(pywikibot.FilePage(page))
                mcp.review_file()
            except SkipFile:
                continue
    elif target == "errors":
        log_page = pywikibot.Page(site, inrbot.config["untagged_log_page"])
        dtt = pywikibot.Page(site, "Template:Deletion template tag")
        for page in log_page.linkedPages(namespaces=6, follow_redirects=True):
            mcp = ManualCommonsPage(pywikibot.FilePage(page))
            if (
                not page.exists()
                or dtt in set(page.itertemplates())
                or mcp.check_has_template()
            ):
                mcp.remove_untagged_log()
                continue
            try:
                mcp.review_file()
            except SkipFile:
                continue
    elif target == "ask":
        while True:
            new_target = click.prompt("Target", default="")
            if not new_target:
                break
            ManualCommonsPage(pywikibot.FilePage(site, new_target)).review_file()
    else:
        page = pywikibot.FilePage(site, target)
        if url:
            # TODO: Add validation
            ids[page] = inrbot.parse_ina_url(url)

        ManualCommonsPage(page).review_file()


if __name__ == "__main__":
    main()

1	#!/usr/bin/env python3
2	# coding: utf-8
3	# SPDX-License-Identifier: GPL-3.0-or-later
4	# Copyright 2023 AntiCompositeNumber
5
6	import click	×
7	import pywikibot # type: ignore	×
8	import pywikibot.bot # type: ignore	×
9	import logging	×
10	import os	×
11	import sys	×
12	import difflib	×
13	import webbrowser	×
14	from typing import Sequence, Dict, Optional, Tuple	×
15
16	os.environ["LOG_FILE"] = "stderr"	×
17	os.environ["LOG_LEVEL"] = "WARNING"	×
18
19	import inrbot # noqa: E402	×
20
21	inrbot.run_override = True	×
22	inrbot.summary_tag = f"(inrcli {inrbot.__version__})"	×
23	site = inrbot.site	×
24	logger = logging.getLogger("manual")	×
25	ids: Dict[pywikibot.Page, Optional[inrbot.iNaturalistID]] = {}	×
26	auto_open = False	×
UNCOV 27	last_ina_id = None	×
28
UNCOV 29	inrbot.config.update(	×
30	{
31	"fail_tag": "{{copyvio\|License review not passed: "
32	"iNaturalist author is using $review_license: $source_url}}\n",
33	"fail_warn": "\n\n{{subst:Copyvionote \|1=$filename "
34	"\|2=License review "
35	"not passed: iNaturalist author is using $review_license: $source_url }} ~~~~",
36	"review_summary": "Semi-automatic license review: "
37	"$status $review_license $tag",
38	"old_fail_warn": "\n\n{{subst:image permission\|1=$filename}} "
39	"License review not passed: iNaturalist author is "
40	"using $review_license: $source_url. ~~~~",
41	"use_wayback": False,
42	}
43	)
44
45
46	class SkipFile(Exception):	×
UNCOV 47	pass	×
48
49
UNCOV 50	def manual_compare(	×
51	com_img: inrbot.CommonsImage, ina_img: inrbot.iNaturalistImage, **kwargs
52	):
53	if ids.get(com_img.page, None) == ina_img.id and ina_img.id.type == "photos":	×
UNCOV 54	return True	×
55	else:
UNCOV 56	return ask_compare(com_img, ina_img)	×
57
58
59	def ask_compare(com_img: inrbot.CommonsImage, ina_img: inrbot.iNaturalistImage):	×
60	if click.confirm(f"Show {ina_img.id}?", default=False):	×
61	com_img.image.show(title=com_img.page.title())	×
62	ina_img.image.show(title=str(ina_img.id))	×
63	res = click.confirm("Do these images match?", default=False)	×
UNCOV 64	return res	×
65
66
UNCOV 67	inrbot.compare_methods.insert(0, ("manual", manual_compare))	×
68
69
70	class ManualCommonsPage(inrbot.CommonsPage):	×
71	def __init__(self, args, *kwargs):	×
UNCOV 72	super().__init__(args, *kwargs)	×
73
UNCOV 74	def check_can_run(self) -> bool:	×
75	"""Determinies if the bot should run on this page and returns a bool."""
76	page = self.page	×
NEW 77	if (not page.has_permission("edit")) or (not page.botMayEdit()):	×
UNCOV 78	return False	×
79	else:
UNCOV 80	return True	×
81
82	def get_old_archive_(self):	×
UNCOV 83	return super().get_old_archive()	×
84
UNCOV 85	def get_old_archive(self):	×
86	# Archives will have already been reviewed by the archive_status_hook
87	if self.status != "fail":	×
88	return super().get_old_archive()	×
UNCOV 89	return ""	×
90
91	def archive_status_hook(self) -> None:	×
92	if self._status == "fail":	×
93	super().get_old_archive()	×
94	if self.archive:	×
UNCOV 95	print(	×
96	f"This file would fail because of the {self.ina_license} license, "
97	f"but an archived copy is available at {self.archive}."
98	)
UNCOV 99	new_license = click.prompt(	×
100	"Archive license (leave blank for no change)",
101	default=self.ina_license,
102	)
103	if new_license:	×
104	self.ina_license = new_license	×
105	del self.status	×
UNCOV 106	self.status	×
107
108	@staticmethod	×
UNCOV 109	def prompt_photo_url(	×
110	default: Optional[str] = None,
111	) -> Tuple[Optional[inrbot.iNaturalistID], str]:
112	global last_ina_id
113	while True:	×
114	# If default is not None, click will repeatedly prompt until it gets
115	# an answer. That, however, doesn't mean we got an answer we like.
UNCOV 116	url = click.prompt("iNaturalist Photos URL", default=default)	×
UNCOV 117	ina_id = inrbot.parse_ina_url(url)	×
118	if ina_id == last_ina_id:	×
119	if click.confirm(	×
120	"That's the same URL you gave last time. Are you sure?"
121	):
UNCOV 122	break	×
123	else:
UNCOV 124	continue	×
125	break	×
UNCOV 126	if ina_id is not None:	×
UNCOV 127	last_ina_id = ina_id	×
UNCOV 128	return ina_id, ""	×
UNCOV 129	return ina_id, url	×
130
131	def pre_save(self, new_text, summary, **kwargs):	×
132	print(	×
133	f"{self.page.title(as_link=True)} reviewed with status {self.status} "
134	f"and license {self.ina_license}"
135	)
UNCOV 136	if self.status == "error":	×
137	raise RuntimeError	×
138
UNCOV 139	diff = difflib.unified_diff(	×
140	self.page.get().split("\n"), new_text.split("\n"), lineterm=""
141	)
142	print("\n".join(diff))	×
143	choice = click.confirm("Save the page?", default=True)	×
144	if choice:	×
145	return new_text, summary	×
146	else:
147	raise RuntimeError	×
148
149	def id_hook(	×
150	self,
151	observations: Sequence[inrbot.iNaturalistID] = [],
152	photos: Sequence[inrbot.iNaturalistID] = [],
153	**kwargs,
154	):
UNCOV 155	try:	×
UNCOV 156	return ids[self.page]	×
UNCOV 157	except KeyError:	×
UNCOV 158	return self.ask_url(observations=observations, photos=photos)	×
159	return None
160
161	def ask_url(	×
162	self,
163	observations: Sequence[inrbot.iNaturalistID] = [],
164	photos: Sequence[inrbot.iNaturalistID] = [],
165	):
166	print(f"Commons page: {self.page.full_url()}")	×
167	if observations:	×
168	print(f"Observation ID found: {str(observations[0])}")	×
169	if photos:	×
170	print(f"Photo ID found: {str(photos[0])}")	×
171	if auto_open:	×
UNCOV 172	webbrowser.open(self.page.full_url())	×
173	if observations and not photos:	×
174	webbrowser.open(str(observations[0]))	×
175	elif photos:	×
176	webbrowser.open(str(photos[0]))	×
177
178	res = click.prompt(	×
179	"Is this ID correct? [Y/n/r/s/q]",
180	default="Y",
181	type=click.Choice("ynrsq", case_sensitive=False),
182	show_default=False,
183	show_choices=False,
184	).lower()
UNCOV 185	if res == "y":	×
186	correct_id = True	×
187	elif res == "n":	×
188	correct_id = False	×
UNCOV 189	elif res == "r":	×
190	self.remove_untagged_log()	×
UNCOV 191	raise SkipFile	×
192	elif res == "s":	×
UNCOV 193	raise SkipFile	×
UNCOV 194	elif res == "q":	×
195	sys.exit()	×
196
197	if not correct_id:	×
UNCOV 198	ina_id, url = self.prompt_photo_url()	×
UNCOV 199	if observations:	×
200	self.page.text = self.page.text.replace(str(observations[0]), url)	×
201	if photos:	×
202	self.page.text = self.page.text.replace(str(photos[0]), url)	×
203	elif observations and not photos:	×
204	ina_id, url = self.prompt_photo_url(default="")	×
205	elif photos:	×
206	ina_id = photos[0]	×
207	else:
UNCOV 208	ina_id = None	×
209	ids[self.page] = ina_id	×
210	return ina_id	×
211
UNCOV 212	def log_untagged_error(self) -> None:	×
213	# Errors while running in CLI do not need to be logged on-wiki
214	return	×
215
216
217	inrbot.id_hooks.append(ManualCommonsPage.id_hook)	×
218	inrbot.lock_hooks.append(ManualCommonsPage.archive_status_hook)	×
219	inrbot.pre_save_hooks.append(ManualCommonsPage.pre_save)	×
220
221
222	@click.command()	×
223	@click.argument("target")	×
224	@click.option("--url")	×
225	@click.option("--simulate/--no-simulate")	×
226	@click.option("--reverse", is_flag=True, default=False)	×
227	@click.option("-o", "--auto-open", "auto_open_", is_flag=True, default=False)	×
228	def main(target, url="", simulate=False, reverse=False, auto_open_=False):	×
UNCOV 229	inrbot.simulate = simulate	×
230	global auto_open
UNCOV 231	auto_open = auto_open_	×
UNCOV 232	if target == "auto":	×
233	cat = pywikibot.Category(	×
234	site, "Category:iNaturalist images needing human review"
235	)
236	dtt = pywikibot.Page(site, "Template:Deletion template tag")	×
237	for page in cat.articles(namespaces=6, reverse=reverse):	×
238	if dtt in set(page.itertemplates()):	×
239	continue	×
240	try:	×
241	mcp = ManualCommonsPage(pywikibot.FilePage(page))	×
242	mcp.review_file()	×
243	except SkipFile:	×
244	continue	×
UNCOV 245	elif target == "errors":	×
246	log_page = pywikibot.Page(site, inrbot.config["untagged_log_page"])	×
247	dtt = pywikibot.Page(site, "Template:Deletion template tag")	×
UNCOV 248	for page in log_page.linkedPages(namespaces=6, follow_redirects=True):	×
249	mcp = ManualCommonsPage(pywikibot.FilePage(page))	×
UNCOV 250	if (	×
251	not page.exists()
252	or dtt in set(page.itertemplates())
253	or mcp.check_has_template()
254	):
255	mcp.remove_untagged_log()	×
UNCOV 256	continue	×
UNCOV 257	try:	×
UNCOV 258	mcp.review_file()	×
UNCOV 259	except SkipFile:	×
UNCOV 260	continue	×
UNCOV 261	elif target == "ask":	×
UNCOV 262	while True:	×
UNCOV 263	new_target = click.prompt("Target", default="")	×
UNCOV 264	if not new_target:	×
UNCOV 265	break	×
UNCOV 266	ManualCommonsPage(pywikibot.FilePage(site, new_target)).review_file()	×
267	else:
UNCOV 268	page = pywikibot.FilePage(site, target)	×
UNCOV 269	if url:	×
270	# TODO: Add validation
UNCOV 271	ids[page] = inrbot.parse_ina_url(url)	×
272
UNCOV 273	ManualCommonsPage(page).review_file()	×
274
275
UNCOV 276	if __name__ == "__main__":	×
UNCOV 277	main()	×

AntiCompositeNumber / iNaturalistReviewer / 12455756921

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous