Add publisher and imprint handling

Imprint handling has been added to utils and uses a subclassed dict to return a tuple for imprint matching, this may not be the best idea but it works for now. Add settings option auto_imprint Add cli flag -a, --auto-import
2021-08-07 21:50:45 -07:00 · 2021-08-07 21:50:45 -07:00 · aefe778b36
commit aefe778b36
parent 444e67100c
9 changed files with 242 additions and 3 deletions
--- a/comicapi/genericmetadata.py
+++ b/comicapi/genericmetadata.py
@ -319,3 +319,21 @@ class GenericMetadata:
            outstr += fmt_str.format(i[0] + ":", i[1])

        return outstr
+
+    def fixPublisher(self):
+        if self.publisher is None:
+            return
+        if self.imprint is None:
+            self.imprint = ""
+
+        imprint, publisher = utils.getPublisher(self.publisher)
+
+        self.publisher = publisher
+
+        if self.imprint.lower() in publisher.lower():
+            self.imprint = None
+
+        if self.imprint is None or self.imprint == "":
+            self.imprint = imprint
+        elif self.imprint.lower() in imprint.lower():
+            self.imprint = imprint
--- a/comicapi/utils.py
+++ b/comicapi/utils.py
@ -615,3 +615,170 @@ def getLanguageFromISO(iso):
        return None
    else:
        return lang_dict[iso]
+
+
+def getPublisher(publisher):
+    if publisher is None:
+        return ("", "")
+    imprint = ""
+
+    for pub in publishers:
+        imprint, publisher, ok = pub[publisher]
+        if ok:
+            break
+
+    return (imprint, publisher)
+
+
+class ImprintDict(dict):
+    '''
+    ImprintDict takes a publisher and a dict or mapping of lowercased
+    imprint names to the proper imprint name. Retreiving a value from an
+    ImprintDict returns a tuple of (imprint, publisher, keyExists).
+    if the key does not exist the key is returned as the publisher unchanged
+    '''
+    def __init__(self, publisher, mapping=(), **kwargs):
+        super().__init__(mapping, **kwargs)
+        self.publisher = publisher
+
+    def __missing__(self, key):
+        return None
+
+    def __getitem__(self, k):
+        item = super().__getitem__(k.lower())
+        if item is None:
+            return ("", k, False)
+        else:
+            return (item, self.publisher, True)
+
+Marvel = ImprintDict("Marvel", {
+    "marvel comics": "",
+    "marvel": "",
+    "aircel comics": "Aircel Comics",
+    "aircel": "Aircel Comics",
+    "atlas comics": "Atlas Comics",
+    "atlas": "Atlas Comics",
+    "crossgen comics": "CrossGen comics",
+    "crossgen": "CrossGen comics",
+    "curtis magazines": "Curtis Magazines",
+    "disney books group": "Disney Books Group",
+    "disney books": "Disney Books Group",
+    "disney kingdoms": "Disney Kingdoms",
+    "epic comics": "Epic Comics",
+    "epic": "Epic Comics",
+    "epic comics group": "Epic Comics",
+    "eternity comics": "Eternity Comics",
+    "humorama": "Humorama",
+    "icon comics": "Icon Comics",
+    "infinite comics": "Infinite Comics",
+    "malibu comics": "Malibu Comics",
+    "malibu": "Malibu Comics",
+    "marvel 2099": "Marvel 2099",
+    "marvel absurd": "Marvel Absurd",
+    "marvel adventures": "Marvel Adventures",
+    "marvel age": "Marvel Age",
+    "marvel books": "Marvel Books",
+    "marvel comics 2": "Marvel Comics 2",
+    "marvel edge": "Marvel Edge",
+    "marvel frontier": "Marvel Frontier",
+    "marvel illustrated": "Marvel Illustrated",
+    "marvel knights": "Marvel Knights",
+    "marvel digital comics unlimited": "Marvel Unlimited",
+    "marvel magazine group": "Marvel Magazine Group",
+    "marvel mangaverse": "Marvel Mangaverse",
+    "marvel monsters group": "Marvel Monsters Group",
+    "marvel music": "Marvel Music",
+    "marvel next": "Marvel Next",
+    "marvel noir": "Marvel Noir",
+    "marvel press": "Marvel Press",
+    "marvel uk": "Marvel UK",
+    "marvel unlimited": "Marvel Unlimited",
+    "max": "MAX",
+    "mc2": "Marvel Comics 2",
+    "new universe": "New Universe",
+    "non-pareil publishing corp.": "Non-Pareil Publishing Corp.",
+    "paramount comics": "Paramount Comics",
+    "power comics": "Power Comics",
+    "razorline": "Razorline",
+    "star comics": "Star Comics",
+    "timely comics": "Timely Comics",
+    "timely": "Timely Comics",
+    "tsunami": "Tsunami",
+    "ultimate comics": "Ultimate Comics",
+    "ultimate marvel": "Ultimate Marvel",
+    "vital publications, inc.": "Vital Publications, Inc."
+})
+
+DC_Comics = ImprintDict("DC Comics", {
+    "dc comics": "",
+    "dc_comics": "",
+    "dc": "",
+    "tangent comics": "Tangent Comics",
+    "dccomics": "",
+    "all star dc": "All-Star",
+    "all star": "All-Star",
+    "all-star dc": "All-Star",
+    "all-star": "All-Star",
+    "america's best comics": "America's Best Comics",
+    "black label": "DC Black Label",
+    "cliffhanger": "Cliffhanger",
+    "cmx manga": "CMX Manga",
+    "dc black label": "DC Black Label",
+    "dc focus": "DC Focus",
+    "dc ink": "DC Ink",
+    "dc zoom": "DC Zoom",
+    "earth m": "Earth M",
+    "earth one": "Earth One",
+    "earth-m": "Earth M",
+    "elseworlds": "Elseworlds",
+    "eo": "Earth One",
+    "first wave": "First Wave",
+    "focus": "DC Focus",
+    "helix": "Helix",
+    "homage comics": "Homage Comics",
+    "impact comics": "Impact Comics",
+    "impact! comics": "Impact Comics",
+    "!mpact comics": "Impact Comics",
+    "johnny dc": "Johnny DC",
+    "mad": "Mad",
+    "minx": "Minx",
+    "paradox press": "Paradox Press",
+    "piranha press": "Piranha Press",
+    "sandman universe": "Sandman Universe",
+    "tsr": "TSR",
+    "vertigo": "Vertigo",
+    "wildstorm productions": "WildStorm Productions",
+    "wildstorm signature": "WildStorm Productions",
+    "wildstorm": "WildStorm Productions",
+    "wonder comics": "Wonder Comics",
+    "young animal": "Young Animal",
+    "zuda comics": "Zuda Comics",
+    "zuda": "Zuda Comics",
+})
+
+Dark_Horse_Comics = ImprintDict("Dark Horse Comics", {
+    "legend": "Legend",
+    "comics' greatest world": "Dark Horse Heroes",
+    "dark horse heroes": "Dark Horse Heroes",
+    "dark horse manga": "Dark Horse Manga",
+    "maverick": "Maverick",
+    "dh press": "DH Press",
+    "m press": "M Press",
+    "dark horse digital": "Dark Horse Digital",
+    "dh deluxe": "DH Deluxe",
+    "kitchen sink books": "Kitchen Sink Books",
+    "berger books": "Berger Books",
+})
+
+Archie_Comics = ImprintDict("Archie Comics", {
+    "Archie Action": "Archie Action",
+    "Archie Horror": "Archie Horror",
+    "Dark Circle Comics": "Dark Circle Comics",
+    "Dark Circle": "Dark Circle Comics",
+    "Red Circle Comics": "Dark Circle Comics",
+    "Red Circle": "Dark Circle Comics",
+    "Archie Adventure Series": "Archie Adventure Series",
+    "Radio Comics": "Mighty Comics Group",
+    "Mighty Comics Group": "Mighty Comics Group",
+})
+publishers = [Marvel, DC_Comics, Dark_Horse_Comics, Archie_Comics]
--- a/comictaggerlib/autotagstartwindow.py
+++ b/comictaggerlib/autotagstartwindow.py
@ -45,6 +45,7 @@ class AutoTagStartWindow(QtWidgets.QDialog):
            QtCore.Qt.Unchecked)
        self.cbxRemoveAfterSuccess.setCheckState(QtCore.Qt.Unchecked)
        self.cbxSpecifySearchString.setCheckState(QtCore.Qt.Unchecked)
+        self.cbxAutoImprint.setCheckState(QtCore.Qt.Unchecked)
        self.leNameLengthMatchTolerance.setText(
            str(self.settings.id_length_delta_thresh))
        self.leSearchString.setEnabled(False)
@ -62,6 +63,9 @@ class AutoTagStartWindow(QtWidgets.QDialog):
            self.cbxRemoveAfterSuccess.setCheckState(QtCore.Qt.Checked)
        if self.settings.wait_and_retry_on_rate_limit:
            self.cbxWaitForRateLimit.setCheckState(QtCore.Qt.Checked)
+        if self.settings.auto_imprint:
+            self.cbxAutoImprint.setCheckState(QtCore.Qt.Checked)
+

        nlmtTip = (
            """ <html>The <b>Name Length Match Tolerance</b> is for eliminating automatic
--- a/comictaggerlib/cli.py
+++ b/comictaggerlib/cli.py
@ -132,6 +132,10 @@ def display_match_set_for_choice(label, match_set, opts, settings):
            cv_md = actual_issue_data_fetch(
                match_set.matches[int(i)], settings, opts)
            md.overlay(cv_md)
+
+            if settings.auto_imprint:
+                md.fixPublisher()
+
            actual_metadata_save(ca, opts, md)


@ -219,6 +223,8 @@ def process_file_cli(filename, opts, settings, match_results):

    batch_mode = len(opts.file_list) > 1

+    settings.auto_imprint = opts.auto_imprint
+
    ca = ComicArchive(
        filename,
        settings.rar_exe_path,
@ -473,6 +479,9 @@ def process_file_cli(filename, opts, settings, match_results):

            md.overlay(cv_md)

+            if settings.auto_imprint:
+                md.fixPublisher()
+
        # ok, done building our metadata. time to save
        if not actual_metadata_save(ca, opts, md):
            match_results.writeFailures.append(filename)
--- a/comictaggerlib/options.py
+++ b/comictaggerlib/options.py
@ -113,6 +113,7 @@ For more help visit the wiki at: https://github.com/comictagger/comictagger/wiki
        self.filename = None
        self.verbose = False
        self.terse = False
+        self.auto_imprint = False
        self.metadata = None
        self.print_tags = False
        self.copy_tags = False
@ -291,6 +292,8 @@ For more help visit the wiki at: https://github.com/comictagger/comictagger/wiki
                self.delete_tags = True
            if o in ("-i", "--interactive"):
                self.interactive = True
+            if o in ("-a", "--auto-imprint"):
+                self.auto_imprint = True
            if o in ("-c", "--copy"):
                self.copy_tags = True
                if a.lower() == "cr":
--- a/comictaggerlib/settings.py
+++ b/comictaggerlib/settings.py
@ -94,6 +94,7 @@ class ComicTaggerSettings:
        self.clear_form_before_populating_from_cv = False
        self.remove_html_tables = False
        self.cv_api_key = ""
+        self.auto_imprint = False

        # CBL Tranform settings

@ -323,6 +324,8 @@ class ComicTaggerSettings:
        if self.config.has_option('autotag', 'wait_and_retry_on_rate_limit'):
            self.wait_and_retry_on_rate_limit = self.config.getboolean(
                'autotag', 'wait_and_retry_on_rate_limit')
+        if self.config.has_option('autotag', 'auto_imprint'):
+            self.auto_imprint = self.config.getboolean('autotag', 'auto_imprint')

    def save(self):

@ -483,6 +486,7 @@ class ComicTaggerSettings:
            'autotag',
            'wait_and_retry_on_rate_limit',
            self.wait_and_retry_on_rate_limit)
+        self.config.set('autotag', 'auto_imprint', self.auto_imprint)

        with codecs.open(self.settings_file, 'wb', 'utf8') as configfile:
            self.config.write(configfile)
--- a/comictaggerlib/taggerwindow.py
+++ b/comictaggerlib/taggerwindow.py
@ -248,6 +248,8 @@ class TaggerWindow(QtWidgets.QMainWindow):
        self.fileSelectionList.addAppAction(self.actionRemoveAuto)
        self.fileSelectionList.addAppAction(self.actionRepackage)

+        self.btnAutoImprint.clicked.connect(self.autoImprint)
+
        if len(file_list) != 0:
            self.fileSelectionList.addPathList(file_list)

@ -384,6 +386,8 @@ class TaggerWindow(QtWidgets.QMainWindow):
        self.actionSearchOnline.setStatusTip('Search online for tags')
        self.actionSearchOnline.triggered.connect(self.queryOnline)

+        self.actionAutoImprint.triggered.connect(self.autoImprint)
+
        self.actionAutoIdentify.setShortcut('Ctrl+I')
        self.actionAutoIdentify.triggered.connect(self.autoIdentifySearch)

@ -425,6 +429,8 @@ class TaggerWindow(QtWidgets.QMainWindow):
            QtGui.QIcon(ComicTaggerSettings.getGraphic('auto.png')))
        self.actionAutoTag.setIcon(
            QtGui.QIcon(ComicTaggerSettings.getGraphic('autotag.png')))
+        self.actionAutoImprint.setIcon(
+            QtGui.QIcon(ComicTaggerSettings.getGraphic('autotag.png')))
        self.actionClearEntryForm.setIcon(
            QtGui.QIcon(ComicTaggerSettings.getGraphic('clear.png')))
        self.actionPageBrowser.setIcon(
@ -438,6 +444,7 @@ class TaggerWindow(QtWidgets.QMainWindow):
        self.toolBar.addAction(self.actionAutoTag)
        self.toolBar.addAction(self.actionClearEntryForm)
        self.toolBar.addAction(self.actionPageBrowser)
+        self.toolBar.addAction(self.actionAutoImprint)

    def repackageArchive(self):
        ca_list = self.fileSelectionList.getSelectedArchiveList()
@ -1816,6 +1823,9 @@ class TaggerWindow(QtWidgets.QMainWindow):
            if cv_md is not None:
                md.overlay(cv_md)

+                if self.settings.auto_imprint:
+                    md.fixPublisher()
+
                if not ca.writeMetadata(md, self.save_data_style):
                    match_results.writeFailures.append(ca.path)
                    self.autoTagLog("Save failed ;-(\n")
@ -2167,3 +2177,8 @@ class TaggerWindow(QtWidgets.QMainWindow):
            # self.show()
            self.setWindowFlags(flags)
            self.show()
+
+    def autoImprint(self):
+        self.formToMetadata()
+        self.metadata.fixPublisher()
+        self.metadataToForm()
--- a/comictaggerlib/ui/autotagstartwindow.ui
+++ b/comictaggerlib/ui/autotagstartwindow.ui
@ -44,7 +44,7 @@
   </item>
   <item row="1" column="0">
    <layout class="QGridLayout" name="gridLayout">
-     <item row="6" column="0">
+     <item row="7" column="0">
      <widget class="QCheckBox" name="cbxSpecifySearchString">
       <property name="sizePolicy">
        <sizepolicy hsizetype="Minimum" vsizetype="Fixed">
@ -129,6 +129,16 @@
       </property>
      </widget>
     </item>
+     <item row="6" column="0">
+      <widget class="QCheckBox" name="cbxAutoImprint">
+       <property name="toolTip">
+        <string>&lt;html&gt;&lt;head/&gt;&lt;body&gt;&lt;p&gt;Checks the publisher against a list of imprints.&lt;/p&gt;&lt;/body&gt;&lt;/html&gt;</string>
+       </property>
+       <property name="text">
+        <string>Auto Imprint</string>
+       </property>
+      </widget>
+     </item>
     <item row="10" column="0">
      <widget class="QLineEdit" name="leNameLengthMatchTolerance">
       <property name="sizePolicy">
@ -145,7 +155,7 @@
       </property>
      </widget>
     </item>
-     <item row="7" column="0">
+     <item row="8" column="0">
      <widget class="QLineEdit" name="leSearchString">
       <property name="sizePolicy">
        <sizepolicy hsizetype="Expanding" vsizetype="Fixed">
@ -155,7 +165,7 @@
       </property>
      </widget>
     </item>
-     <item row="8" column="0">
+     <item row="9" column="0">
      <widget class="QLabel" name="label_3">
       <property name="sizePolicy">
        <sizepolicy hsizetype="Preferred" vsizetype="Fixed">
--- a/comictaggerlib/ui/taggerwindow.ui
+++ b/comictaggerlib/ui/taggerwindow.ui
@ -1176,6 +1176,7 @@
    <addaction name="actionParse_Filename"/>
    <addaction name="actionSearchOnline"/>
    <addaction name="actionAutoIdentify"/>
+    <addaction name="actionAutoImprint"/>
    <addaction name="separator"/>
    <addaction name="actionApplyCBLTransform"/>
   </widget>
@ -1373,6 +1374,14 @@
    <string>Search online for tags,auto-identify best match, and save to archive</string>
   </property>
  </action>
+  <action name="actionAutoImprint">
+   <property name="text">
+    <string>Auto Imprint</string>
+   </property>
+   <property name="toolTip">
+    <string>Normalize the publisher and map imprints to their parent publisher (e.g. Vertigo is an imprint of DC Comics)</string>
+   </property>
+  </action>
 </widget>
 <layoutdefault spacing="6" margin="11"/>
 <resources/>