|
|
@ -12,7 +12,7 @@ from .common import Extractor, AsynchronousExtractor, Message
|
|
|
|
from .. import text
|
|
|
|
from .. import text
|
|
|
|
|
|
|
|
|
|
|
|
class SpectrumNexusMangaExtractor(Extractor):
|
|
|
|
class SpectrumNexusMangaExtractor(Extractor):
|
|
|
|
|
|
|
|
"""Extract all manga-chapters and -volumes from spectrumnexus"""
|
|
|
|
category = "spectrumnexus"
|
|
|
|
category = "spectrumnexus"
|
|
|
|
subcategory = "manga"
|
|
|
|
subcategory = "manga"
|
|
|
|
pattern = [r"(?:https?://)?view\.thespectrum\.net/series/([^\.]+)\.html$"]
|
|
|
|
pattern = [r"(?:https?://)?view\.thespectrum\.net/series/([^\.]+)\.html$"]
|
|
|
@ -35,24 +35,27 @@ class SpectrumNexusMangaExtractor(Extractor):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class SpectrumNexusChapterExtractor(AsynchronousExtractor):
|
|
|
|
class SpectrumNexusChapterExtractor(AsynchronousExtractor):
|
|
|
|
|
|
|
|
"""Extract a single manga-chapter or -volume from spectrumnexus"""
|
|
|
|
category = "spectrumnexus"
|
|
|
|
category = "spectrumnexus"
|
|
|
|
subcategory = "chapter"
|
|
|
|
subcategory = "chapter"
|
|
|
|
directory_fmt = ["{category}", "{manga}", "c{chapter:>03}"]
|
|
|
|
directory_fmt = ["{category}", "{manga}", "{identifier}"]
|
|
|
|
filename_fmt = "{manga}_c{chapter:>03}_{page:>03}.{extension}"
|
|
|
|
filename_fmt = "{manga} {identifier} {page:>03}.{extension}"
|
|
|
|
pattern = [
|
|
|
|
pattern = [
|
|
|
|
r"(?:https?://)?(view\.thespectrum\.net/series/[^\.]+\.html)\?ch=Chapter\+(\d+)",
|
|
|
|
(r"(?:https?://)?(view\.thespectrum\.net/series/[^\.]+\.html)"
|
|
|
|
r"(?:https?://)?(view\.thespectrum\.net/series/[^/]+-chapter-(\d+)\.html)",
|
|
|
|
r"\?ch=(Chapter\+(\d+)|Volume\+(\d+))"),
|
|
|
|
|
|
|
|
(r"(?:https?://)?(view\.thespectrum\.net/series/[^/]+-chapter-(\d+)\.html)"),
|
|
|
|
]
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
|
|
def __init__(self, match):
|
|
|
|
def __init__(self, match):
|
|
|
|
AsynchronousExtractor.__init__(self)
|
|
|
|
AsynchronousExtractor.__init__(self)
|
|
|
|
self.url = "http://" + match.group(1)
|
|
|
|
self.url = "http://" + match.group(1)
|
|
|
|
self.chapter = match.group(2)
|
|
|
|
self.identifier = match.group(2)
|
|
|
|
|
|
|
|
self.chapter = match.group(3)
|
|
|
|
|
|
|
|
self.volume = match.group(4)
|
|
|
|
|
|
|
|
|
|
|
|
def items(self):
|
|
|
|
def items(self):
|
|
|
|
params = {
|
|
|
|
params = {
|
|
|
|
"ch": "Chapter " + self.chapter,
|
|
|
|
"ch": self.identifier,
|
|
|
|
"page": 1,
|
|
|
|
"page": 1,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
page = self.request(self.url, params=params).text
|
|
|
|
page = self.request(self.url, params=params).text
|
|
|
@ -73,7 +76,9 @@ class SpectrumNexusChapterExtractor(AsynchronousExtractor):
|
|
|
|
"""Collect metadata for extractor-job"""
|
|
|
|
"""Collect metadata for extractor-job"""
|
|
|
|
data = {
|
|
|
|
data = {
|
|
|
|
"category": self.category,
|
|
|
|
"category": self.category,
|
|
|
|
"chapter": self.chapter,
|
|
|
|
"chapter": self.chapter or "",
|
|
|
|
|
|
|
|
"volume": self.volume or "",
|
|
|
|
|
|
|
|
"identifier": self.identifier.replace("+", " "),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return text.extract_all(page, (
|
|
|
|
return text.extract_all(page, (
|
|
|
|
('manga', '<title>', ' · SPECTRUM NEXUS </title>'),
|
|
|
|
('manga', '<title>', ' · SPECTRUM NEXUS </title>'),
|
|
|
|