chore: update README

- ok i didnt expect github actions does not support regex wtf

.github/workflows/build-and-publish.yml

@@ -0,0 +1,103 @@
+name: Build, Release, Publish
+
+on:
+  workflow_dispatch:
+  push:
+    tags:
+      - "*.*.*"
+
+jobs:
+  build:
+    name: Build package
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Python environment
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.x"
+
+      - name: Build package
+        run: |
+          pip install build
+          python -m build
+
+      - name: Store the distribution files
+        uses: actions/upload-artifact@v4
+        with:
+          name: python-package-distributions
+          path: dist/
+
+  draft-release:
+    name: Draft a release
+    runs-on: ubuntu-latest
+    needs:
+      - build
+    permissions:
+      contents: write
+      discussions: write
+
+    steps:
+      - name: Download the distribution files
+        uses: actions/download-artifact@v4
+        with:
+          name: python-package-distributions
+          path: dist/
+
+      - name: Draft a release
+        uses: softprops/action-gh-release@v2
+        with:
+          discussion_category_name: New releases
+          draft: true
+          generate_release_notes: true
+          files: |
+            dist/*
+
+  publish-to-pypi:
+    name: Publish distribution to PyPI
+    runs-on: ubuntu-latest
+    if: startsWith(github.ref, 'refs/tags/') # only publish to PyPI on tag pushes
+    needs:
+      - build
+    environment:
+      name: pypi
+      url: https://pypi.org/p/arcaea-offline-ocr
+    permissions:
+      id-token: write
+
+    steps:
+      - name: Download the distribution files
+        uses: actions/download-artifact@v4
+        with:
+          name: python-package-distributions
+          path: dist/
+
+      - name: Publish distribution to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+
+  publish-to-testpypi:
+    name: Publish distribution to TestPyPI
+    runs-on: ubuntu-latest
+    needs:
+      - build
+    environment:
+      name: testpypi
+      url: https://test.pypi.org/p/arcaea-offline-ocr
+    permissions:
+      id-token: write
+
+    steps:
+      - name: Download the distribution files
+        uses: actions/download-artifact@v4
+        with:
+          name: python-package-distributions
+          path: dist/
+
+      - name: Publish distribution to TestPyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          repository-url: https://test.pypi.org/legacy/

.gitignore

@@ -1,6 +1,8 @@
 __debug*
 .vscode/
 
+*venv*/
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]

.pre-commit-config.yaml

@@ -4,13 +4,10 @@ repos:
     hooks:
       - id: end-of-file-fixer
       - id: trailing-whitespace
-  - repo: https://github.com/psf/black
+
-    rev: 23.1.0
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.11.13
     hooks:
-      - id: black
+      - id: ruff
-        exclude: _builtin_templates.py
+        args: ["--fix"]
-  - repo: https://github.com/PyCQA/isort
+      - id: ruff-format
-    rev: 5.12.0
-    hooks:
-      - id: isort
-        exclude: _builtin_templates.py

LICENSE

@@ -0,0 +1,675 @@
+# GNU GENERAL PUBLIC LICENSE
+
+Version 3, 29 June 2007
+
+Copyright (C) 2007 Free Software Foundation, Inc.
+<https://fsf.org/>
+
+Everyone is permitted to copy and distribute verbatim copies of this
+license document, but changing it is not allowed.
+
+## Preamble
+
+The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works. By contrast,
+the GNU General Public License is intended to guarantee your freedom
+to share and change all versions of a program--to make sure it remains
+free software for all its users. We, the Free Software Foundation, use
+the GNU General Public License for most of our software; it applies
+also to any other work released this way by its authors. You can apply
+it to your programs, too.
+
+When we speak of free software, we are referring to freedom, not
+price. Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights. Therefore, you
+have certain responsibilities if you distribute copies of the
+software, or if you modify it: responsibilities to respect the freedom
+of others.
+
+For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received. You must make sure that they, too, receive
+or can get the source code. And you must show them these terms so they
+know their rights.
+
+Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software. For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the
+manufacturer can do so. This is fundamentally incompatible with the
+aim of protecting users' freedom to change the software. The
+systematic pattern of such abuse occurs in the area of products for
+individuals to use, which is precisely where it is most unacceptable.
+Therefore, we have designed this version of the GPL to prohibit the
+practice for those products. If such problems arise substantially in
+other domains, we stand ready to extend this provision to those
+domains in future versions of the GPL, as needed to protect the
+freedom of users.
+
+Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish
+to avoid the special danger that patents applied to a free program
+could make it effectively proprietary. To prevent this, the GPL
+assures that patents cannot be used to render the program non-free.
+
+The precise terms and conditions for copying, distribution and
+modification follow.
+
+## TERMS AND CONDITIONS
+
+### 0. Definitions.
+
+"This License" refers to version 3 of the GNU General Public License.
+
+"Copyright" also means copyright-like laws that apply to other kinds
+of works, such as semiconductor masks.
+
+"The Program" refers to any copyrightable work licensed under this
+License. Each licensee is addressed as "you". "Licensees" and
+"recipients" may be individuals or organizations.
+
+To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of
+an exact copy. The resulting work is called a "modified version" of
+the earlier work or a work "based on" the earlier work.
+
+A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy. Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies. Mere interaction with a user
+through a computer network, with no transfer of a copy, is not
+conveying.
+
+An interactive user interface displays "Appropriate Legal Notices" to
+the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License. If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+### 1. Source Code.
+
+The "source code" for a work means the preferred form of the work for
+making modifications to it. "Object code" means any non-source form of
+a work.
+
+A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form. A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities. However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work. For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+The Corresponding Source need not include anything that users can
+regenerate automatically from other parts of the Corresponding Source.
+
+The Corresponding Source for a work in source code form is that same
+work.
+
+### 2. Basic Permissions.
+
+All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met. This License explicitly affirms your unlimited
+permission to run the unmodified Program. The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work. This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+You may make, run and propagate covered works that you do not convey,
+without conditions so long as your license otherwise remains in force.
+You may convey covered works to others for the sole purpose of having
+them make modifications exclusively for you, or provide you with
+facilities for running those works, provided that you comply with the
+terms of this License in conveying all material for which you do not
+control copyright. Those thus making or running the covered works for
+you must do so exclusively on your behalf, under your direction and
+control, on terms that prohibit them from making any copies of your
+copyrighted material outside their relationship with you.
+
+Conveying under any other circumstances is permitted solely under the
+conditions stated below. Sublicensing is not allowed; section 10 makes
+it unnecessary.
+
+### 3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such
+circumvention is effected by exercising rights under this License with
+respect to the covered work, and you disclaim any intention to limit
+operation or modification of the work as a means of enforcing, against
+the work's users, your or third parties' legal rights to forbid
+circumvention of technological measures.
+
+### 4. Conveying Verbatim Copies.
+
+You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+### 5. Conveying Modified Source Versions.
+
+You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these
+conditions:
+
+-   a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+-   b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under
+    section 7. This requirement modifies the requirement in section 4
+    to "keep intact all notices".
+-   c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy. This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged. This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+-   d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit. Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+### 6. Conveying Non-Source Forms.
+
+You may convey a covered work in object code form under the terms of
+sections 4 and 5, provided that you also convey the machine-readable
+Corresponding Source under the terms of this License, in one of these
+ways:
+
+-   a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+-   b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the Corresponding
+    Source from a network server at no charge.
+-   c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source. This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+-   d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge. You need not require recipients to copy the
+    Corresponding Source along with the object code. If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source. Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+-   e) Convey the object code using peer-to-peer transmission,
+    provided you inform other peers where the object code and
+    Corresponding Source of the work are being offered to the general
+    public at no charge under subsection 6d.
+
+A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal,
+family, or household purposes, or (2) anything designed or sold for
+incorporation into a dwelling. In determining whether a product is a
+consumer product, doubtful cases shall be resolved in favor of
+coverage. For a particular product received by a particular user,
+"normally used" refers to a typical or common use of that class of
+product, regardless of the status of the particular user or of the way
+in which the particular user actually uses, or expects or is expected
+to use, the product. A product is a consumer product regardless of
+whether the product has substantial commercial, industrial or
+non-consumer uses, unless such uses represent the only significant
+mode of use of the product.
+
+"Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to
+install and execute modified versions of a covered work in that User
+Product from a modified version of its Corresponding Source. The
+information must suffice to ensure that the continued functioning of
+the modified object code is in no case prevented or interfered with
+solely because modification has been made.
+
+If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information. But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or
+updates for a work that has been modified or installed by the
+recipient, or for the User Product in which it has been modified or
+installed. Access to a network may be denied when the modification
+itself materially and adversely affects the operation of the network
+or violates the rules and protocols for communication across the
+network.
+
+Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+### 7. Additional Terms.
+
+"Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law. If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it. (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.) You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders
+of that material) supplement the terms of this License with terms:
+
+-   a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+-   b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+-   c) Prohibiting misrepresentation of the origin of that material,
+    or requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+-   d) Limiting the use for publicity purposes of names of licensors
+    or authors of the material; or
+-   e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+-   f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions
+    of it) with contractual assumptions of liability to the recipient,
+    for any liability that these contractual assumptions directly
+    impose on those licensors and authors.
+
+All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10. If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term. If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions; the
+above requirements apply either way.
+
+### 8. Termination.
+
+You may not propagate or modify a covered work except as expressly
+provided under this License. Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+However, if you cease all violation of this License, then your license
+from a particular copyright holder is reinstated (a) provisionally,
+unless and until the copyright holder explicitly and finally
+terminates your license, and (b) permanently, if the copyright holder
+fails to notify you of the violation by some reasonable means prior to
+60 days after the cessation.
+
+Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License. If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+### 9. Acceptance Not Required for Having Copies.
+
+You are not required to accept this License in order to receive or run
+a copy of the Program. Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance. However,
+nothing other than this License grants you permission to propagate or
+modify any covered work. These actions infringe copyright if you do
+not accept this License. Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+### 10. Automatic Licensing of Downstream Recipients.
+
+Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License. You are not responsible
+for enforcing compliance by third parties with this License.
+
+An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations. If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License. For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+### 11. Patents.
+
+A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based. The
+work thus licensed is called the contributor's "contributor version".
+
+A contributor's "essential patent claims" are all patent claims owned
+or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version. For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement). To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients. "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+A patent license is "discriminatory" if it does not include within the
+scope of its coverage, prohibits the exercise of, or is conditioned on
+the non-exercise of one or more of the rights that are specifically
+granted under this License. You may not convey a covered work if you
+are a party to an arrangement with a third party that is in the
+business of distributing software, under which you make payment to the
+third party based on the extent of your activity of conveying the
+work, and under which the third party grants, to any of the parties
+who would receive the covered work from you, a discriminatory patent
+license (a) in connection with copies of the covered work conveyed by
+you (or copies made from those copies), or (b) primarily for and in
+connection with specific products or compilations that contain the
+covered work, unless you entered into that arrangement, or that patent
+license was granted, prior to 28 March 2007.
+
+Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+### 12. No Surrender of Others' Freedom.
+
+If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License. If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under
+this License and any other pertinent obligations, then as a
+consequence you may not convey it at all. For example, if you agree to
+terms that obligate you to collect a royalty for further conveying
+from those to whom you convey the Program, the only way you could
+satisfy both those terms and this License would be to refrain entirely
+from conveying the Program.
+
+### 13. Use with the GNU Affero General Public License.
+
+Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work. The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+### 14. Revised Versions of this License.
+
+The Free Software Foundation may publish revised and/or new versions
+of the GNU General Public License from time to time. Such new versions
+will be similar in spirit to the present version, but may differ in
+detail to address new problems or concerns.
+
+Each version is given a distinguishing version number. If the Program
+specifies that a certain numbered version of the GNU General Public
+License "or any later version" applies to it, you have the option of
+following the terms and conditions either of that numbered version or
+of any later version published by the Free Software Foundation. If the
+Program does not specify a version number of the GNU General Public
+License, you may choose any version ever published by the Free
+Software Foundation.
+
+If the Program specifies that a proxy can decide which future versions
+of the GNU General Public License can be used, that proxy's public
+statement of acceptance of a version permanently authorizes you to
+choose that version for the Program.
+
+Later license versions may give you additional or different
+permissions. However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+### 15. Disclaimer of Warranty.
+
+THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT
+WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND
+PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE
+DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR
+CORRECTION.
+
+### 16. Limitation of Liability.
+
+IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR
+CONVEYS THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
+INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES
+ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT
+NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR
+LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM
+TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER
+PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
+
+### 17. Interpretation of Sections 15 and 16.
+
+If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+END OF TERMS AND CONDITIONS
+
+## How to Apply These Terms to Your New Programs
+
+If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these
+terms.
+
+To do so, attach the following notices to the program. It is safest to
+attach them to the start of each source file to most effectively state
+the exclusion of warranty; and each file should have at least the
+"copyright" line and a pointer to where the full notice is found.
+
+        <one line to give the program's name and a brief idea of what it does.>
+        Copyright (C) <year>  <name of author>
+
+        This program is free software: you can redistribute it and/or modify
+        it under the terms of the GNU General Public License as published by
+        the Free Software Foundation, either version 3 of the License, or
+        (at your option) any later version.
+
+        This program is distributed in the hope that it will be useful,
+        but WITHOUT ANY WARRANTY; without even the implied warranty of
+        MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+        GNU General Public License for more details.
+
+        You should have received a copy of the GNU General Public License
+        along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper
+mail.
+
+If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+        <program>  Copyright (C) <year>  <name of author>
+        This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+        This is free software, and you are welcome to redistribute it
+        under certain conditions; type `show c' for details.
+
+The hypothetical commands \`show w' and \`show c' should show the
+appropriate parts of the General Public License. Of course, your
+program's commands might be different; for a GUI interface, you would
+use an "about box".
+
+You should also get your employer (if you work as a programmer) or
+school, if any, to sign a "copyright disclaimer" for the program, if
+necessary. For more information on this, and how to apply and follow
+the GNU GPL, see <https://www.gnu.org/licenses/>.
+
+The GNU General Public License does not permit incorporating your
+program into proprietary programs. If your program is a subroutine
+library, you may consider it more useful to permit linking proprietary
+applications with the library. If this is what you want to do, use the
+GNU Lesser General Public License instead of this License. But first,
+please read <https://www.gnu.org/licenses/why-not-lgpl.html>.

README.md

@@ -1,28 +1,171 @@
 # Arcaea Offline OCR
 
+[![Ruff](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/v2.json)](https://github.com/astral-sh/ruff)
+
 ## Example
 
+> Results from `arcaea_offline_ocr 0.1.0a2`
+
+### Build an image hash database (ihdb)
+
 ```py
-from arcaea_offline_ocr.device.v2.rois import DeviceV2AutoRois
+import sqlite3
-from arcaea_offline_ocr.device.v2.ocr import DeviceV2Ocr
+from pathlib import Path
-from arcaea_offline_ocr.sift_db import SIFTDatabase
+
-from arcaea_offline_ocr.utils import imread_unicode
 import cv2
 
-knn_model = cv2.ml.KNearest_load(r'/path/to/knn/model')
+from arcaea_offline_ocr.builders.ihdb import (
-sift_db = SIFTDatabase(r'/path/to/sift/database.db')
+    ImageHashDatabaseBuildTask,
+    ImageHashesDatabaseBuilder,
+)
+from arcaea_offline_ocr.providers import ImageCategory, ImageHashDatabaseIdProvider
+from arcaea_offline_ocr.scenarios.device import DeviceScenario
 
-rois = DeviceV2AutoRois(imread_unicode(r'/path/to/your/screenshot.jpg'))  # any format that opencv-python supports
+def build():
-ocr = DeviceV2Ocr(knn_model, sift_db)
+    def _read_partner_icon(image_path: str):
-result = ocr.ocr(rois)
+        return DeviceScenario.preprocess_char_icon(
-print(result)
+            cv2.cvtColor(cv2.imread(image_path, cv2.IMREAD_COLOR), cv2.COLOR_BGR2GRAY),
+        )
+
+    builder = ImageHashesDatabaseBuilder()
+    tasks = [
+        ImageHashDatabaseBuildTask(
+            image_path=str(file),
+            image_id=file.stem,
+            category=ImageCategory.JACKET,
+        )
+        for file in Path("/path/to/some/jackets").glob("*.jpg")
+    ]
+
+    tasks.extend(
+        [
+            ImageHashDatabaseBuildTask(
+                image_path=str(file),
+                image_id=file.stem,
+                category=ImageCategory.PARTNER_ICON,
+                imread_function=_read_partner_icon,
+            )
+            for file in Path("/path/to/some/partner_icons").glob("*.png")
+        ],
+    )
+
+    with sqlite3.connect("/path/to/ihdb-X.Y.Z.db") as conn:
+        builder.build(conn, tasks)
 ```
 
-```sh
+### Device OCR
-$ python example.py
+
-DeviceOcrResult(rating_class=2, pure=1371, far=62, lost=34, score=9558078, max_recall=330, song_id='abstrusedilemma', title=None, clear_type=None)
+```py
+import json
+import sqlite3
+from dataclasses import asdict
+
+import cv2
+
+from arcaea_offline_ocr.providers import (
+    ImageHashDatabaseIdProvider,
+    OcrKNearestTextProvider,
+)
+from arcaea_offline_ocr.scenarios.device import (
+    DeviceRoisAutoT2,
+    DeviceRoisExtractor,
+    DeviceRoisMaskerAutoT2,
+    DeviceScenario,
+)
+
+
+with sqlite3.connect("/path/to/ihdb-X.Y.Z.db") as conn:
+    img = cv2.imread("/path/to/your/screenshot.jpg")
+    h, w = img.shape[:2]
+
+    r = DeviceRoisAutoT2(w, h)
+    m = DeviceRoisMaskerAutoT2()
+    e = DeviceRoisExtractor(img, r)
+
+    scenario = DeviceScenario(
+        extractor=e,
+        masker=m,
+        knn_provider=OcrKNearestTextProvider(
+            cv2.ml.KNearest.load("/path/to/knn_model.dat"),
+        ),
+        image_id_provider=ImageHashDatabaseIdProvider(conn),
+    )
+    result = scenario.result()
+
+    with open("result.jsonc", "w", encoding="utf-8") as jf:
+        json.dump(asdict(result), jf, indent=2, ensure_ascii=False)
 ```
 
+```jsonc
+// result.json
+{
+  "song_id": "vector",
+  "rating_class": 1,
+  "score": 9990996,
+  "song_id_results": [
+    {
+      "image_id": "vector",
+      "category": 0,
+      "confidence": 1.0,
+      "image_hash_type": 0
+    },
+    {
+      "image_id": "clotho",
+      "category": 0,
+      "confidence": 0.71875,
+      "image_hash_type": 0
+    }
+    // 28 more results omitted…
+  ],
+  "partner_id_results": [
+    {
+      "image_id": "23",
+      "category": 1,
+      "confidence": 0.90625,
+      "image_hash_type": 0
+    },
+    {
+      "image_id": "45",
+      "category": 1,
+      "confidence": 0.8828125,
+      "image_hash_type": 0
+    }
+    // 28 more results omitted…
+  ],
+  "pure": 1000,
+  "pure_inaccurate": null,
+  "pure_early": null,
+  "pure_late": null,
+  "far": 2,
+  "far_inaccurate": null,
+  "far_early": null,
+  "far_late": null,
+  "lost": 0,
+  "played_at": null,
+  "max_recall": 1002,
+  "clear_status": 2,
+  "clear_type": null,
+  "modifier": null
+}
+```
+
+## License
+
+This file is part of arcaea-offline-ocr, as called "This program" below.
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
 ## Credits
 
-[283375/image-sift-database](https://github.com/283375/image-sift-database)
+- [JohannesBuchner/imagehash](https://github.com/JohannesBuchner/imagehash): `arcaea_offline_ocr.core.hashers` implementations reference

pyproject.toml

@@ -1,27 +1,33 @@
 [build-system]
-requires = ["setuptools>=61.0"]
+requires = ["setuptools>=64", "setuptools-scm>=8"]
 build-backend = "setuptools.build_meta"
 
 [project]
+dynamic = ["version"]
+
 name = "arcaea-offline-ocr"
-version = "0.0.95"
 authors = [{ name = "283375", email = "log_283375@163.com" }]
 description = "Extract your Arcaea play result from screenshot."
 readme = "README.md"
 requires-python = ">=3.8"
-dependencies = ["attrs==23.1.0", "numpy==1.26.1", "opencv-python==4.8.1.78"]
+dependencies = ["numpy~=2.3", "opencv-python~=4.11"]
 classifiers = [
   "Development Status :: 3 - Alpha",
   "Programming Language :: Python :: 3",
 ]
 
-[project.urls]
+[project.optional-dependencies]
-"Homepage" = "https://github.com/283375/arcaea-offline-ocr"
+dev = ["ruff", "pre-commit"]
-"Bug Tracker" = "https://github.com/283375/arcaea-offline-ocr/issues"
 
-[tool.isort]
+[project.urls]
-profile = "black"
+"Homepage" = "https://github.com/ArcaeaOffline/core-ocr"
-src_paths = ["src/arcaea_offline_ocr"]
+"Bug Tracker" = "https://github.com/ArcaeaOffline/core-ocr/issues"
+
+[tool.setuptools_scm]
 
 [tool.pyright]
 ignore = ["**/__debug*.*"]
+
+[tool.ruff.lint]
+select = ["ALL"]
+ignore = ["ANN", "D", "ERA", "PLR"]

requirements.dev.txt

@@ -1,3 +1,2 @@
-black==23.7.0
+ruff
-isort==5.12.0
+pre-commit
-pre-commit==3.3.3

requirements.txt

@@ -1,3 +1,2 @@
-attrs==23.1.0
+numpy~=2.3
-numpy==1.26.1
+opencv-python~=4.11
-opencv-python==4.8.1.78

src/arcaea_offline_ocr/__init__.py

@@ -1,4 +0,0 @@
-from .crop import *
-from .device import *
-from .ocr import *
-from .utils import *

src/arcaea_offline_ocr/b30/shared.py

@@ -1,16 +0,0 @@
-from datetime import datetime
-from typing import Optional, Union
-
-import attrs
-
-
-@attrs.define
-class B30OcrResultItem:
-    rating_class: int
-    score: int
-    pure: Optional[int] = None
-    far: Optional[int] = None
-    lost: Optional[int] = None
-    date: Optional[datetime] = None
-    title: Optional[str] = None
-    song_id: Optional[str] = None

src/arcaea_offline_ocr/builders/__init__.py

@@ -0,0 +1,6 @@
+from .ihdb import ImageHashDatabaseBuildTask, ImageHashesDatabaseBuilder
+
+__all__ = [
+    "ImageHashDatabaseBuildTask",
+    "ImageHashesDatabaseBuilder",
+]

src/arcaea_offline_ocr/builders/ihdb.py

@@ -0,0 +1,115 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import TYPE_CHECKING, Callable
+
+import cv2
+
+from arcaea_offline_ocr.core import hashers
+from arcaea_offline_ocr.providers.ihdb import (
+    PROP_KEY_BUILT_AT,
+    PROP_KEY_HASH_SIZE,
+    PROP_KEY_HIGH_FREQ_FACTOR,
+    ImageHashDatabaseIdProvider,
+    ImageHashType,
+)
+
+if TYPE_CHECKING:
+    from sqlite3 import Connection
+
+    from arcaea_offline_ocr.providers import ImageCategory
+    from arcaea_offline_ocr.types import Mat
+
+
+def _default_imread_gray(image_path: str):
+    return cv2.cvtColor(cv2.imread(image_path, cv2.IMREAD_COLOR), cv2.COLOR_BGR2GRAY)
+
+
+@dataclass
+class ImageHashDatabaseBuildTask:
+    image_path: str
+    image_id: str
+    category: ImageCategory
+    imread_function: Callable[[str], Mat] = _default_imread_gray
+
+
+@dataclass
+class _ImageHash:
+    image_id: str
+    category: ImageCategory
+    image_hash_type: ImageHashType
+    hash: bytes
+
+
+class ImageHashesDatabaseBuilder:
+    @staticmethod
+    def __insert_property(conn: Connection, key: str, value: str):
+        return conn.execute(
+            "INSERT INTO properties (key, value) VALUES (?, ?)",
+            (key, value),
+        )
+
+    @classmethod
+    def build(
+        cls,
+        conn: Connection,
+        tasks: list[ImageHashDatabaseBuildTask],
+        *,
+        hash_size: int = 16,
+        high_freq_factor: int = 4,
+    ):
+        hashes: list[_ImageHash] = []
+
+        for task in tasks:
+            img_gray = task.imread_function(task.image_path)
+
+            for hash_type, hash_mat in [
+                (
+                    ImageHashType.AVERAGE,
+                    hashers.average(img_gray, hash_size),
+                ),
+                (
+                    ImageHashType.DCT,
+                    hashers.dct(img_gray, hash_size, high_freq_factor),
+                ),
+                (
+                    ImageHashType.DIFFERENCE,
+                    hashers.difference(img_gray, hash_size),
+                ),
+            ]:
+                hashes.append(
+                    _ImageHash(
+                        image_id=task.image_id,
+                        image_hash_type=hash_type,
+                        category=task.category,
+                        hash=ImageHashDatabaseIdProvider.hash_mat_to_bytes(hash_mat),
+                    ),
+                )
+
+        conn.execute("CREATE TABLE properties (`key` VARCHAR, `value` VARCHAR)")
+        conn.execute(
+            """CREATE TABLE hashes (
+`id` VARCHAR,
+`category` INTEGER,
+`hash_type` INTEGER,
+`hash` BLOB
+)""",
+        )
+
+        now = datetime.now(tz=timezone.utc)
+        timestamp = int(now.timestamp() * 1000)
+
+        cls.__insert_property(conn, PROP_KEY_HASH_SIZE, str(hash_size))
+        cls.__insert_property(conn, PROP_KEY_HIGH_FREQ_FACTOR, str(high_freq_factor))
+        cls.__insert_property(conn, PROP_KEY_BUILT_AT, str(timestamp))
+
+        conn.executemany(
+            """INSERT INTO hashes (`id`, `category`, `hash_type`, `hash`)
+                            VALUES (?, ?, ?, ?)""",
+            [
+                (it.image_id, it.category.value, it.image_hash_type.value, it.hash)
+                for it in hashes
+            ],
+        )
+        conn.commit()

src/arcaea_offline_ocr/core/hashers/__init__.py

@@ -0,0 +1,3 @@
+from .index import average, dct, difference
+
+__all__ = ["average", "dct", "difference"]

src/arcaea_offline_ocr/core/hashers/_common.py

@@ -0,0 +1,7 @@
+import cv2
+
+from arcaea_offline_ocr.types import Mat
+
+
+def _resize_image(src: Mat, dsize: ...) -> Mat:
+    return cv2.resize(src, dsize, fx=0, fy=0, interpolation=cv2.INTER_AREA)

src/arcaea_offline_ocr/core/hashers/index.py

@@ -0,0 +1,35 @@
+import cv2
+import numpy as np
+
+from arcaea_offline_ocr.types import Mat
+
+from ._common import _resize_image
+
+
+def average(img_gray: Mat, hash_size: int) -> Mat:
+    img_resized = _resize_image(img_gray, (hash_size, hash_size))
+    diff = img_resized > img_resized.mean()
+    return diff.flatten()
+
+
+def difference(img_gray: Mat, hash_size: int) -> Mat:
+    img_size = (hash_size + 1, hash_size)
+    img_resized = _resize_image(img_gray, img_size)
+
+    previous = img_resized[:, :-1]
+    current = img_resized[:, 1:]
+    diff = previous > current
+    return diff.flatten()
+
+
+def dct(img_gray: Mat, hash_size: int = 16, high_freq_factor: int = 4) -> Mat:
+    # TODO: consistency?  # noqa: FIX002, TD002, TD003
+    img_size_base = hash_size * high_freq_factor
+    img_size = (img_size_base, img_size_base)
+
+    img_resized = _resize_image(img_gray, img_size)
+    img_resized = img_resized.astype(np.float32)
+    dct_mat = cv2.dct(img_resized)
+
+    hash_mat = dct_mat[:hash_size, :hash_size]
+    return hash_mat > hash_mat.mean()

src/arcaea_offline_ocr/crop.py

@@ -1,29 +1,32 @@
+from __future__ import annotations
+
 import math
-from typing import Tuple
+from typing import TYPE_CHECKING
 
 import cv2
 import numpy as np
 
+if TYPE_CHECKING:
     from .types import Mat
 
-__all__ = ["crop_xywh", "CropBlackEdges"]
+__all__ = ["CropBlackEdges", "crop_xywh"]
 
 
-def crop_xywh(mat: Mat, rect: Tuple[int, int, int, int]):
+def crop_xywh(mat: Mat, rect: tuple[int, int, int, int]):
     x, y, w, h = rect
     return mat[y : y + h, x : x + w]
 
 
 class CropBlackEdges:
     @staticmethod
-    def is_black_edge(__img_gray_slice: Mat, black_pixel: int, ratio: float = 0.6):
+    def is_black_edge(img_gray_slice: Mat, black_pixel: int, ratio: float = 0.6):
-        pixels_compared = __img_gray_slice < black_pixel
+        pixels_compared = img_gray_slice < black_pixel
         return np.count_nonzero(pixels_compared) > math.floor(
-            __img_gray_slice.size * ratio
+            img_gray_slice.size * ratio,
         )
 
     @classmethod
-    def get_crop_rect(cls, img_gray: Mat, black_threshold: int = 25):
+    def get_crop_rect(cls, img_gray: Mat, black_threshold: int = 25):  # noqa: C901
         height, width = img_gray.shape[:2]
         left = 0
         right = width
@@ -54,13 +57,22 @@ class CropBlackEdges:
                 break
             bottom -= 1
 
-        assert right > left, "cropped width < 0"
+        if right <= left:
-        assert bottom > top, "cropped height < 0"
+            msg = "cropped width < 0"
+            raise ValueError(msg)
+
+        if bottom <= top:
+            msg = "cropped height < 0"
+            raise ValueError(msg)
+
         return (left, top, right - left, bottom - top)
 
     @classmethod
     def crop(
-        cls, img: Mat, convert_flag: cv2.COLOR_BGR2GRAY, black_threshold: int = 25
+        cls,
+        img: Mat,
+        convert_flag: cv2.COLOR_BGR2GRAY,
+        black_threshold: int = 25,
     ) -> Mat:
         rect = cls.get_crop_rect(cv2.cvtColor(img, convert_flag), black_threshold)
         return crop_xywh(img, rect)

src/arcaea_offline_ocr/device/__init__.py

@@ -1,2 +0,0 @@
-from .common import DeviceOcrResult
-from .ocr import DeviceOcr

src/arcaea_offline_ocr/device/common.py

@@ -1,18 +0,0 @@
-from typing import Optional
-
-import attrs
-
-
-@attrs.define
-class DeviceOcrResult:
-    rating_class: int
-    pure: int
-    far: int
-    lost: int
-    score: int
-    max_recall: Optional[int] = None
-    song_id: Optional[str] = None
-    song_id_possibility: Optional[float] = None
-    clear_status: Optional[int] = None
-    partner_id: Optional[str] = None
-    partner_id_possibility: Optional[float] = None

src/arcaea_offline_ocr/device/rois/__init__.py

@@ -1,3 +0,0 @@
-from .definition import *
-from .extractor import *
-from .masker import *

src/arcaea_offline_ocr/device/rois/definition/__init__.py

@@ -1,2 +0,0 @@
-from .auto import *
-from .common import DeviceRois

src/arcaea_offline_ocr/device/rois/definition/common.py

@@ -1,15 +0,0 @@
-from typing import Tuple
-
-Rect = Tuple[int, int, int, int]
-
-
-class DeviceRois:
-    pure: Rect
-    far: Rect
-    lost: Rect
-    score: Rect
-    rating_class: Rect
-    max_recall: Rect
-    jacket: Rect
-    clear_status: Rect
-    partner_icon: Rect

src/arcaea_offline_ocr/device/rois/extractor/__init__.py

@@ -1 +0,0 @@
-from .common import DeviceRoisExtractor

src/arcaea_offline_ocr/device/rois/extractor/common.py

@@ -1,48 +0,0 @@
-from ....crop import crop_xywh
-from ....types import Mat
-from ..definition.common import DeviceRois
-
-
-class DeviceRoisExtractor:
-    def __init__(self, img: Mat, rois: DeviceRois):
-        self.img = img
-        self.sizes = rois
-
-    def __construct_int_rect(self, rect):
-        return tuple(round(r) for r in rect)
-
-    @property
-    def pure(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.pure))
-
-    @property
-    def far(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.far))
-
-    @property
-    def lost(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.lost))
-
-    @property
-    def score(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.score))
-
-    @property
-    def jacket(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.jacket))
-
-    @property
-    def rating_class(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.rating_class))
-
-    @property
-    def max_recall(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.max_recall))
-
-    @property
-    def clear_status(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.clear_status))
-
-    @property
-    def partner_icon(self):
-        return crop_xywh(self.img, self.__construct_int_rect(self.sizes.partner_icon))

src/arcaea_offline_ocr/device/rois/masker/__init__.py

@@ -1,2 +0,0 @@
-from .auto import *
-from .common import DeviceRoisMasker

src/arcaea_offline_ocr/device/rois/masker/common.py

@@ -1,55 +0,0 @@
-from ....types import Mat
-
-
-class DeviceRoisMasker:
-    @classmethod
-    def pure(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def far(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def lost(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def score(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def rating_class_pst(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def rating_class_prs(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def rating_class_ftr(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def rating_class_byd(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def max_recall(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def clear_status_track_lost(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def clear_status_track_complete(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def clear_status_full_recall(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()
-
-    @classmethod
-    def clear_status_pure_memory(cls, roi_bgr: Mat) -> Mat:
-        raise NotImplementedError()

src/arcaea_offline_ocr/phash_db.py

@@ -1,118 +0,0 @@
-import sqlite3
-from typing import List, Union
-
-import cv2
-import numpy as np
-
-from .types import Mat
-
-
-def phash_opencv(img_gray, hash_size=8, highfreq_factor=4):
-    # type: (Union[Mat, np.ndarray], int, int) -> np.ndarray
-    """
-    Perceptual Hash computation.
-
-    Implementation follows http://www.hackerfactor.com/blog/index.php?/archives/432-Looks-Like-It.html
-
-    Adapted from `imagehash.phash`, pure opencv implementation
-
-    The result is slightly different from `imagehash.phash`.
-    """
-    if hash_size < 2:
-        raise ValueError("Hash size must be greater than or equal to 2")
-
-    img_size = hash_size * highfreq_factor
-    image = cv2.resize(img_gray, (img_size, img_size), interpolation=cv2.INTER_LANCZOS4)
-    image = np.float32(image)
-    dct = cv2.dct(image)
-    dctlowfreq = dct[:hash_size, :hash_size]
-    med = np.median(dctlowfreq)
-    diff = dctlowfreq > med
-    return diff
-
-
-def hamming_distance_sql_function(user_input, db_entry) -> int:
-    return np.count_nonzero(
-        np.frombuffer(user_input, bool) ^ np.frombuffer(db_entry, bool)
-    )
-
-
-class ImagePhashDatabase:
-    def __init__(self, db_path: str):
-        with sqlite3.connect(db_path) as conn:
-            self.hash_size = int(
-                conn.execute(
-                    "SELECT value FROM properties WHERE key = 'hash_size'"
-                ).fetchone()[0]
-            )
-            self.highfreq_factor = int(
-                conn.execute(
-                    "SELECT value FROM properties WHERE key = 'highfreq_factor'"
-                ).fetchone()[0]
-            )
-            self.built_timestamp = int(
-                conn.execute(
-                    "SELECT value FROM properties WHERE key = 'built_timestamp'"
-                ).fetchone()[0]
-            )
-
-            self.ids: List[str] = [
-                i[0] for i in conn.execute("SELECT id FROM hashes").fetchall()
-            ]
-            self.hashes_byte = [
-                i[0] for i in conn.execute("SELECT hash FROM hashes").fetchall()
-            ]
-            self.hashes = [np.frombuffer(hb, bool) for hb in self.hashes_byte]
-
-            self.jacket_ids: List[str] = []
-            self.jacket_hashes = []
-            self.partner_icon_ids: List[str] = []
-            self.partner_icon_hashes = []
-
-            for id, hash in zip(self.ids, self.hashes):
-                id_splitted = id.split("||")
-                if len(id_splitted) > 1 and id_splitted[0] == "partner_icon":
-                    self.partner_icon_ids.append(id_splitted[1])
-                    self.partner_icon_hashes.append(hash)
-                else:
-                    self.jacket_ids.append(id)
-                    self.jacket_hashes.append(hash)
-
-    def calculate_phash(self, img_gray: Mat):
-        return phash_opencv(
-            img_gray, hash_size=self.hash_size, highfreq_factor=self.highfreq_factor
-        )
-
-    def lookup_hash(self, image_hash: np.ndarray, *, limit: int = 5):
-        image_hash = image_hash.flatten()
-        xor_results = [
-            (id, np.count_nonzero(image_hash ^ h))
-            for id, h in zip(self.ids, self.hashes)
-        ]
-        return sorted(xor_results, key=lambda r: r[1])[:limit]
-
-    def lookup_image(self, img_gray: Mat):
-        image_hash = self.calculate_phash(img_gray)
-        return self.lookup_hash(image_hash)[0]
-
-    def lookup_jackets(self, img_gray: Mat, *, limit: int = 5):
-        image_hash = self.calculate_phash(img_gray).flatten()
-        xor_results = [
-            (id, np.count_nonzero(image_hash ^ h))
-            for id, h in zip(self.jacket_ids, self.jacket_hashes)
-        ]
-        return sorted(xor_results, key=lambda r: r[1])[:limit]
-
-    def lookup_jacket(self, img_gray: Mat):
-        return self.lookup_jackets(img_gray)[0]
-
-    def lookup_partner_icons(self, img_gray: Mat, *, limit: int = 5):
-        image_hash = self.calculate_phash(img_gray).flatten()
-        xor_results = [
-            (id, np.count_nonzero(image_hash ^ h))
-            for id, h in zip(self.partner_icon_ids, self.partner_icon_hashes)
-        ]
-        return sorted(xor_results, key=lambda r: r[1])[:limit]
-
-    def lookup_partner_icon(self, img_gray: Mat):
-        return self.lookup_partner_icons(img_gray)[0]

src/arcaea_offline_ocr/providers/__init__.py

@@ -0,0 +1,12 @@
+from .base import ImageCategory, ImageIdProvider, ImageIdProviderResult, OcrTextProvider
+from .ihdb import ImageHashDatabaseIdProvider
+from .knn import OcrKNearestTextProvider
+
+__all__ = [
+    "ImageCategory",
+    "ImageHashDatabaseIdProvider",
+    "ImageIdProvider",
+    "ImageIdProviderResult",
+    "OcrKNearestTextProvider",
+    "OcrTextProvider",
+]

src/arcaea_offline_ocr/providers/base.py

@@ -0,0 +1,50 @@
+from __future__ import annotations
+
+from abc import ABC, abstractmethod
+from dataclasses import dataclass
+from enum import IntEnum
+from typing import TYPE_CHECKING, Any, Sequence
+
+if TYPE_CHECKING:
+    from arcaea_offline_ocr.types import Mat
+
+
+class OcrTextProvider(ABC):
+    @abstractmethod
+    def result_raw(self, img: Mat, /, *args, **kwargs) -> Any: ...
+    @abstractmethod
+    def result(self, img: Mat, /, *args, **kwargs) -> str | None: ...
+
+
+class ImageCategory(IntEnum):
+    JACKET = 0
+    PARTNER_ICON = 1
+
+
+@dataclass(kw_only=True)
+class ImageIdProviderResult:
+    image_id: str
+    category: ImageCategory
+    confidence: float
+
+
+class ImageIdProvider(ABC):
+    @abstractmethod
+    def result(
+        self,
+        img: Mat,
+        category: ImageCategory,
+        /,
+        *args,
+        **kwargs,
+    ) -> ImageIdProviderResult: ...
+
+    @abstractmethod
+    def results(
+        self,
+        img: Mat,
+        category: ImageCategory,
+        /,
+        *args,
+        **kwargs,
+    ) -> Sequence[ImageIdProviderResult]: ...

src/arcaea_offline_ocr/providers/ihdb.py

@@ -0,0 +1,203 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from enum import IntEnum
+from typing import TYPE_CHECKING, Any, Callable, TypeVar
+
+from arcaea_offline_ocr.core import hashers
+
+from .base import ImageCategory, ImageIdProvider, ImageIdProviderResult
+
+if TYPE_CHECKING:
+    import sqlite3
+
+    from arcaea_offline_ocr.types import Mat
+
+
+T = TypeVar("T")
+PROP_KEY_HASH_SIZE = "hash_size"
+PROP_KEY_HIGH_FREQ_FACTOR = "high_freq_factor"
+PROP_KEY_BUILT_AT = "built_at"
+
+
+def _sql_hamming_distance(hash1: bytes, hash2: bytes):
+    if len(hash1) != len(hash2):
+        msg = "hash size does not match!"
+        raise ValueError(msg)
+
+    return sum(1 for byte1, byte2 in zip(hash1, hash2) if byte1 != byte2)
+
+
+class ImageHashType(IntEnum):
+    AVERAGE = 0
+    DIFFERENCE = 1
+    DCT = 2
+
+
+@dataclass(kw_only=True)
+class ImageHashDatabaseIdProviderResult(ImageIdProviderResult):
+    image_hash_type: ImageHashType
+
+
+class MissingPropertiesError(Exception):
+    keys: list[str]
+
+    def __init__(self, keys, *args):
+        super().__init__(*args)
+        self.keys = keys
+
+
+class ImageHashDatabaseIdProvider(ImageIdProvider):
+    def __init__(self, conn: sqlite3.Connection):
+        self.conn = conn
+        self.conn.create_function("HAMMING_DISTANCE", 2, _sql_hamming_distance)
+
+        self.properties = {
+            PROP_KEY_HASH_SIZE: -1,
+            PROP_KEY_HIGH_FREQ_FACTOR: -1,
+            PROP_KEY_BUILT_AT: None,
+        }
+
+        self._hashes_count = {
+            ImageCategory.JACKET: 0,
+            ImageCategory.PARTNER_ICON: 0,
+        }
+
+        self._hash_length: int = -1
+
+        self._initialize()
+
+    @property
+    def hash_size(self) -> int:
+        return self.properties[PROP_KEY_HASH_SIZE]
+
+    @property
+    def high_freq_factor(self) -> int:
+        return self.properties[PROP_KEY_HIGH_FREQ_FACTOR]
+
+    @property
+    def built_at(self) -> datetime | None:
+        return self.properties.get(PROP_KEY_BUILT_AT)
+
+    @property
+    def hash_length(self):
+        return self._hash_length
+
+    def _initialize(self):
+        def get_property(key, converter: Callable[[Any], T]) -> T | None:
+            result = self.conn.execute(
+                "SELECT value FROM properties WHERE key = ?",
+                (key,),
+            ).fetchone()
+            return converter(result[0]) if result is not None else None
+
+        def set_hashes_count(category: ImageCategory):
+            self._hashes_count[category] = self.conn.execute(
+                "SELECT COUNT(DISTINCT `id`) FROM hashes WHERE category = ?",
+                (category.value,),
+            ).fetchone()[0]
+
+        properties_converter_map = {
+            PROP_KEY_HASH_SIZE: lambda x: int(x),
+            PROP_KEY_HIGH_FREQ_FACTOR: lambda x: int(x),
+            PROP_KEY_BUILT_AT: lambda ts: datetime.fromtimestamp(
+                int(ts) / 1000,
+                tz=timezone.utc,
+            ),
+        }
+        required_properties = [PROP_KEY_HASH_SIZE, PROP_KEY_HIGH_FREQ_FACTOR]
+
+        missing_properties = []
+        for property_key, converter in properties_converter_map.items():
+            value = get_property(property_key, converter)
+            if value is None:
+                if property_key in required_properties:
+                    missing_properties.append(property_key)
+
+                continue
+
+            self.properties[property_key] = value
+
+        if missing_properties:
+            raise MissingPropertiesError(keys=missing_properties)
+
+        set_hashes_count(ImageCategory.JACKET)
+        set_hashes_count(ImageCategory.PARTNER_ICON)
+
+        self._hash_length = self.hash_size**2
+
+    def lookup_hash(
+        self,
+        category: ImageCategory,
+        hash_type: ImageHashType,
+        hash_data: bytes,
+    ) -> list[ImageHashDatabaseIdProviderResult]:
+        cursor = self.conn.execute(
+            """
+SELECT
+    `id`,
+    HAMMING_DISTANCE(hash, ?) AS distance
+FROM hashes
+WHERE category = ? AND hash_type = ?
+ORDER BY distance ASC LIMIT 10""",
+            (hash_data, category.value, hash_type.value),
+        )
+
+        results = []
+        for id_, distance in cursor.fetchall():
+            results.append(
+                ImageHashDatabaseIdProviderResult(
+                    image_id=id_,
+                    category=category,
+                    confidence=(self.hash_length - distance) / self.hash_length,
+                    image_hash_type=hash_type,
+                ),
+            )
+
+        return results
+
+    @staticmethod
+    def hash_mat_to_bytes(hash_mat: Mat) -> bytes:
+        return bytes([255 if b else 0 for b in hash_mat.flatten()])
+
+    def results(self, img: Mat, category: ImageCategory, /):
+        results: list[ImageHashDatabaseIdProviderResult] = []
+
+        results.extend(
+            self.lookup_hash(
+                category,
+                ImageHashType.AVERAGE,
+                self.hash_mat_to_bytes(hashers.average(img, self.hash_size)),
+            ),
+        )
+        results.extend(
+            self.lookup_hash(
+                category,
+                ImageHashType.DIFFERENCE,
+                self.hash_mat_to_bytes(hashers.difference(img, self.hash_size)),
+            ),
+        )
+        results.extend(
+            self.lookup_hash(
+                category,
+                ImageHashType.DCT,
+                self.hash_mat_to_bytes(
+                    hashers.dct(img, self.hash_size, self.high_freq_factor),
+                ),
+            ),
+        )
+
+        return results
+
+    def result(
+        self,
+        img: Mat,
+        category: ImageCategory,
+        /,
+        *,
+        hash_type: ImageHashType = ImageHashType.DCT,
+    ):
+        return next(
+            it for it in self.results(img, category) if it.image_hash_type == hash_type
+        )

src/arcaea_offline_ocr/providers/knn.py

@@ -1,27 +1,31 @@
+from __future__ import annotations
+
+import logging
 import math
-from typing import Optional, Sequence, Tuple
+from typing import TYPE_CHECKING, Callable, Sequence
 
 import cv2
 import numpy as np
 
-from .crop import crop_xywh
+from arcaea_offline_ocr.crop import crop_xywh
-from .types import Mat
 
-__all__ = [
+from .base import OcrTextProvider
-    "FixRects",
+
-    "preprocess_hog",
+if TYPE_CHECKING:
-    "ocr_digits_by_contour_get_samples",
+    from cv2.ml import KNearest
-    "ocr_digits_by_contour_knn",
+
-]
+    from arcaea_offline_ocr.types import Mat
+
+logger = logging.getLogger(__name__)
 
 
 class FixRects:
     @staticmethod
     def connect_broken(
-        rects: Sequence[Tuple[int, int, int, int]],
+        rects: Sequence[tuple[int, int, int, int]],
         img_width: int,
         img_height: int,
-        tolerance: Optional[int] = None,
+        tolerance: int | None = None,
     ):
         # for a "broken" digit, please refer to
         # /assets/fix_rects/broken_masked.jpg
@@ -36,7 +40,7 @@ class FixRects:
             if rect in consumed_rects:
                 continue
 
-            x, y, w, h = rect
+            x, _, w, h = rect
             # grab those small rects
             if not img_height * 0.1 <= h <= img_height * 0.6:
                 continue
@@ -46,7 +50,7 @@ class FixRects:
             for other_rect in rects:
                 if rect == other_rect:
                     continue
-                ox, oy, ow, oh = other_rect
+                ox, _, ow, _ = other_rect
                 if abs(x - ox) < tolerance and abs((x + w) - (ox + ow)) < tolerance:
                     group.append(other_rect)
 
@@ -69,7 +73,7 @@ class FixRects:
     @staticmethod
     def split_connected(
         img_masked: Mat,
-        rects: Sequence[Tuple[int, int, int, int]],
+        rects: Sequence[tuple[int, int, int, int]],
         rect_wh_ratio: float = 1.05,
         width_range_ratio: float = 0.1,
     ):
@@ -110,7 +114,7 @@ class FixRects:
 
             # split the rect
             new_rects.extend(
-                [(rx, ry, x_mid - rx, rh), (x_mid, ry, rx + rw - x_mid, rh)]
+                [(rx, ry, x_mid - rx, rh), (x_mid, ry, rx + rw - x_mid, rh)],
             )
 
         return_rects = [r for r in rects if r not in connected_rects]
@@ -131,11 +135,21 @@ def resize_fill_square(img: Mat, target: int = 20):
     border_size = math.ceil((max(new_w, new_h) - min(new_w, new_h)) / 2)
     if new_w < new_h:
         resized = cv2.copyMakeBorder(
-            resized, 0, 0, border_size, border_size, cv2.BORDER_CONSTANT
+            resized,
+            0,
+            0,
+            border_size,
+            border_size,
+            cv2.BORDER_CONSTANT,
         )
     else:
         resized = cv2.copyMakeBorder(
-            resized, border_size, border_size, 0, 0, cv2.BORDER_CONSTANT
+            resized,
+            border_size,
+            border_size,
+            0,
+            0,
+            cv2.BORDER_CONSTANT,
         )
     return cv2.resize(resized, (target, target))
 
@@ -150,31 +164,94 @@ def preprocess_hog(digit_rois):
     return np.float32(samples)
 
 
-def ocr_digit_samples_knn(__samples, knn_model: cv2.ml.KNearest, k: int = 4):
+def ocr_digit_samples_knn(samples, knn_model: cv2.ml.KNearest, k: int = 4):
-    _, results, _, _ = knn_model.findNearest(__samples, k)
+    _, results, _, _ = knn_model.findNearest(samples, k)
-    result_list = [int(r) for r in results.ravel()]
+    return [int(r) for r in results.ravel()]
-    result_str = "".join(str(r) for r in result_list if r > -1)
-    return int(result_str) if result_str else 0
 
 
-def ocr_digits_by_contour_get_samples(__roi_gray: Mat, size: int):
+class OcrKNearestTextProvider(OcrTextProvider):
-    roi = __roi_gray.copy()
+    _ContourFilter = Callable[["Mat"], bool]
-    contours, _ = cv2.findContours(roi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
+    _RectsFilter = Callable[[Sequence[int]], bool]
-    rects = [cv2.boundingRect(c) for c in contours]
-    rects = FixRects.connect_broken(rects, roi.shape[1], roi.shape[0])
-    rects = FixRects.split_connected(roi, rects)
-    rects = sorted(rects, key=lambda r: r[0])
-    # digit_rois = [cv2.resize(crop_xywh(roi, rect), size) for rect in rects]
-    digit_rois = [resize_fill_square(crop_xywh(roi, rect), size) for rect in rects]
-    return preprocess_hog(digit_rois)
 
+    def __init__(self, model: KNearest):
+        self.model = model
 
-def ocr_digits_by_contour_knn(
+    def contours(
-    __roi_gray: Mat,
+        self,
-    knn_model: cv2.ml.KNearest,
+        img: Mat,
+        /,
         *,
-    k=4,
+        contours_filter: _ContourFilter | None = None,
-    size: int = 20,
+    ):
-) -> int:
+        cnts, _ = cv2.findContours(img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
-    samples = ocr_digits_by_contour_get_samples(__roi_gray, size)
+        if contours_filter:
-    return ocr_digit_samples_knn(samples, knn_model, k)
+            cnts = list(filter(contours_filter, cnts))
+
+        return cnts
+
+    def result_raw(
+        self,
+        img: Mat,
+        /,
+        *,
+        fix_rects: bool = True,
+        contours_filter: _ContourFilter | None = None,
+        rects_filter: _RectsFilter | None = None,
+    ):
+        """
+        :param img: grayscaled roi
+        """
+
+        try:
+            cnts, _ = cv2.findContours(img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+            if contours_filter:
+                cnts = list(filter(contours_filter, cnts))
+
+            rects = [cv2.boundingRect(cnt) for cnt in cnts]
+            if fix_rects and rects_filter:
+                rects = FixRects.connect_broken(rects, img.shape[1], img.shape[0])
+                rects = list(filter(rects_filter, rects))
+                rects = FixRects.split_connected(img, rects)
+            elif fix_rects:
+                rects = FixRects.connect_broken(rects, img.shape[1], img.shape[0])
+                rects = FixRects.split_connected(img, rects)
+            elif rects_filter:
+                rects = list(filter(rects_filter, rects))
+
+            rects = sorted(rects, key=lambda r: r[0])
+
+            digits = []
+            for rect in rects:
+                digit = crop_xywh(img, rect)
+                digit = resize_fill_square(digit, 20)
+                digits.append(digit)
+            samples = preprocess_hog(digits)
+            return ocr_digit_samples_knn(samples, self.model)
+        except Exception:
+            logger.exception("Error occurred during KNearest OCR")
+            return None
+
+    def result(
+        self,
+        img: Mat,
+        /,
+        *,
+        fix_rects: bool = True,
+        contours_filter: _ContourFilter | None = None,
+        rects_filter: _RectsFilter | None = None,
+    ):
+        """
+        :param img: grayscaled roi
+        """
+
+        raw = self.result_raw(
+            img,
+            fix_rects=fix_rects,
+            contours_filter=contours_filter,
+            rects_filter=rects_filter,
+        )
+        return (
+            "".join(["".join(str(r) for r in raw if r > -1)])
+            if raw is not None
+            else None
+        )

src/arcaea_offline_ocr/scenarios/b30/__init__.py

@@ -0,0 +1,3 @@
+from .chieri import ChieriBotV4Best30Scenario
+
+__all__ = ["ChieriBotV4Best30Scenario"]

src/arcaea_offline_ocr/scenarios/b30/base.py

@@ -0,0 +1,24 @@
+from __future__ import annotations
+
+from abc import abstractmethod
+from typing import TYPE_CHECKING
+
+from arcaea_offline_ocr.scenarios.base import OcrScenario, OcrScenarioResult
+
+if TYPE_CHECKING:
+    from arcaea_offline_ocr.types import Mat
+
+
+class Best30Scenario(OcrScenario):
+    @abstractmethod
+    def components(self, img: Mat, /) -> list[Mat]: ...
+
+    @abstractmethod
+    def result(self, component_img: Mat, /, *args, **kwargs) -> OcrScenarioResult: ...
+
+    @abstractmethod
+    def results(self, img: Mat, /, *args, **kwargs) -> list[OcrScenarioResult]:
+        """
+        Commonly a shorthand for `[self.result(comp) for comp in self.components(img)]`
+        """
+        ...

src/arcaea_offline_ocr/scenarios/b30/chieri/__init__.py

@@ -0,0 +1,3 @@
+from .v4 import ChieriBotV4Best30Scenario
+
+__all__ = ["ChieriBotV4Best30Scenario"]

src/arcaea_offline_ocr/scenarios/b30/chieri/v4/__init__.py

@@ -0,0 +1,3 @@
+from .impl import ChieriBotV4Best30Scenario
+
+__all__ = ["ChieriBotV4Best30Scenario"]

src/arcaea_offline_ocr/scenarios/b30/chieri/v4/colors.py

@@ -1,19 +1,19 @@
 import numpy as np
 
 __all__ = [
-    "FONT_THRESHOLD",
-    "PURE_BG_MIN_HSV",
-    "PURE_BG_MAX_HSV",
-    "FAR_BG_MIN_HSV",
-    "FAR_BG_MAX_HSV",
-    "LOST_BG_MIN_HSV",
-    "LOST_BG_MAX_HSV",
-    "BYD_MIN_HSV",
     "BYD_MAX_HSV",
-    "FTR_MIN_HSV",
+    "BYD_MIN_HSV",
+    "FAR_BG_MAX_HSV",
+    "FAR_BG_MIN_HSV",
+    "FONT_THRESHOLD",
     "FTR_MAX_HSV",
-    "PRS_MIN_HSV",
+    "FTR_MIN_HSV",
+    "LOST_BG_MAX_HSV",
+    "LOST_BG_MIN_HSV",
     "PRS_MAX_HSV",
+    "PRS_MIN_HSV",
+    "PURE_BG_MAX_HSV",
+    "PURE_BG_MIN_HSV",
 ]
 
 FONT_THRESHOLD = 160
@@ -27,11 +27,11 @@ FAR_BG_MAX_HSV = np.array([20, 255, 255], np.uint8)
 LOST_BG_MIN_HSV = np.array([115, 60, 150], np.uint8)
 LOST_BG_MAX_HSV = np.array([140, 255, 255], np.uint8)
 
-BYD_MIN_HSV = (158, 120, 0)
+BYD_MIN_HSV = np.array([158, 120, 0], np.uint8)
-BYD_MAX_HSV = (172, 255, 255)
+BYD_MAX_HSV = np.array([172, 255, 255], np.uint8)
 
-FTR_MIN_HSV = (145, 70, 0)
+FTR_MIN_HSV = np.array([145, 70, 0], np.uint8)
-FTR_MAX_HSV = (160, 255, 255)
+FTR_MAX_HSV = np.array([160, 255, 255], np.uint8)
 
-PRS_MIN_HSV = (45, 60, 0)
+PRS_MIN_HSV = np.array([45, 60, 0], np.uint8)
-PRS_MAX_HSV = (70, 255, 255)
+PRS_MAX_HSV = np.array([70, 255, 255], np.uint8)

src/arcaea_offline_ocr/scenarios/b30/chieri/v4/impl.py

@@ -1,60 +1,51 @@
-from math import floor
+from __future__ import annotations
-from typing import List, Optional, Tuple
+
+from typing import TYPE_CHECKING
 
 import cv2
 import numpy as np
 
-from ....crop import crop_xywh
+from arcaea_offline_ocr.crop import crop_xywh
-from ....ocr import (
+from arcaea_offline_ocr.providers import (
-    FixRects,
+    ImageCategory,
-    ocr_digits_by_contour_knn,
+    ImageIdProvider,
-    preprocess_hog,
+    OcrKNearestTextProvider,
-    resize_fill_square,
+)
+from arcaea_offline_ocr.scenarios.b30.base import Best30Scenario
+from arcaea_offline_ocr.scenarios.base import OcrScenarioResult
+
+if TYPE_CHECKING:
+    from arcaea_offline_ocr.types import Mat
+
+from .colors import (
+    BYD_MAX_HSV,
+    BYD_MIN_HSV,
+    FAR_BG_MAX_HSV,
+    FAR_BG_MIN_HSV,
+    FTR_MAX_HSV,
+    FTR_MIN_HSV,
+    LOST_BG_MAX_HSV,
+    LOST_BG_MIN_HSV,
+    PRS_MAX_HSV,
+    PRS_MIN_HSV,
+    PURE_BG_MAX_HSV,
+    PURE_BG_MIN_HSV,
 )
-from ....phash_db import ImagePhashDatabase
-from ....types import Mat
-from ....utils import construct_int_xywh_rect
-from ...shared import B30OcrResultItem
-from .colors import *
 from .rois import ChieriBotV4Rois
 
 
-class ChieriBotV4Ocr:
+class ChieriBotV4Best30Scenario(Best30Scenario):
     def __init__(
         self,
-        score_knn: cv2.ml.KNearest,
+        score_knn_provider: OcrKNearestTextProvider,
-        pfl_knn: cv2.ml.KNearest,
+        pfl_knn_provider: OcrKNearestTextProvider,
-        phash_db: ImagePhashDatabase,
+        image_id_provider: ImageIdProvider,
-        factor: Optional[float] = 1.0,
+        factor: float = 1.0,
     ):
-        self.__score_knn = score_knn
-        self.__pfl_knn = pfl_knn
-        self.__phash_db = phash_db
         self.__rois = ChieriBotV4Rois(factor)
-
+        self.pfl_knn_provider = pfl_knn_provider
-    @property
+        self.score_knn_provider = score_knn_provider
-    def score_knn(self):
+        self.image_id_provider = image_id_provider
-        return self.__score_knn
-
-    @score_knn.setter
-    def score_knn(self, knn_digits_model: cv2.ml.KNearest):
-        self.__score_knn = knn_digits_model
-
-    @property
-    def pfl_knn(self):
-        return self.__pfl_knn
-
-    @pfl_knn.setter
-    def pfl_knn(self, knn_digits_model: cv2.ml.KNearest):
-        self.__pfl_knn = knn_digits_model
-
-    @property
-    def phash_db(self):
-        return self.__phash_db
-
-    @phash_db.setter
-    def phash_db(self, phash_db: ImagePhashDatabase):
-        self.__phash_db = phash_db
 
     @property
     def rois(self):
@@ -72,9 +63,8 @@ class ChieriBotV4Ocr:
         self.factor = img.shape[0] / 4400
 
     def ocr_component_rating_class(self, component_bgr: Mat) -> int:
-        rating_class_rect = construct_int_xywh_rect(
+        rating_class_rect = self.rois.component_rois.rating_class_rect.rounded()
-            self.rois.component_rois.rating_class_rect
+
-        )
         rating_class_roi = crop_xywh(component_bgr, rating_class_rect)
         rating_class_roi = cv2.cvtColor(rating_class_roi, cv2.COLOR_BGR2HSV)
         rating_class_masks = [
@@ -85,41 +75,50 @@ class ChieriBotV4Ocr:
         rating_class_results = [np.count_nonzero(m) for m in rating_class_masks]
         if max(rating_class_results) < 70:
             return 0
-        else:
         return max(enumerate(rating_class_results), key=lambda i: i[1])[0] + 1
 
-    def ocr_component_song_id(self, component_bgr: Mat):
+    def ocr_component_song_id_results(self, component_bgr: Mat):
-        jacket_rect = construct_int_xywh_rect(
+        jacket_rect = self.rois.component_rois.jacket_rect.floored()
-            self.rois.component_rois.jacket_rect, floor
-        )
         jacket_roi = cv2.cvtColor(
-            crop_xywh(component_bgr, jacket_rect), cv2.COLOR_BGR2GRAY
+            crop_xywh(component_bgr, jacket_rect),
+            cv2.COLOR_BGR2GRAY,
         )
-        return self.phash_db.lookup_jacket(jacket_roi)[0]
+        return self.image_id_provider.results(jacket_roi, ImageCategory.JACKET)
 
     def ocr_component_score_knn(self, component_bgr: Mat) -> int:
         # sourcery skip: inline-immediately-returned-variable
-        score_rect = construct_int_xywh_rect(self.rois.component_rois.score_rect)
+        score_rect = self.rois.component_rois.score_rect.rounded()
         score_roi = cv2.cvtColor(
-            crop_xywh(component_bgr, score_rect), cv2.COLOR_BGR2GRAY
+            crop_xywh(component_bgr, score_rect),
+            cv2.COLOR_BGR2GRAY,
         )
         _, score_roi = cv2.threshold(
-            score_roi, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU
+            score_roi,
+            0,
+            255,
+            cv2.THRESH_BINARY + cv2.THRESH_OTSU,
         )
         if score_roi[1][1] == 255:
             score_roi = 255 - score_roi
 
         contours, _ = cv2.findContours(
-            score_roi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE
+            score_roi,
+            cv2.RETR_EXTERNAL,
+            cv2.CHAIN_APPROX_SIMPLE,
         )
         for contour in contours:
             rect = cv2.boundingRect(contour)
             if rect[3] > score_roi.shape[0] * 0.5:
                 continue
             score_roi = cv2.fillPoly(score_roi, [contour], 0)
-        return ocr_digits_by_contour_knn(score_roi, self.score_knn)
 
-    def find_pfl_rects(self, component_pfl_processed: Mat) -> List[List[int]]:
+        ocr_result = self.score_knn_provider.result(score_roi)
+        return int(ocr_result) if ocr_result else 0
+
+    def find_pfl_rects(
+        self,
+        component_pfl_processed: Mat,
+    ) -> list[tuple[int, int, int, int]]:
         # sourcery skip: inline-immediately-returned-variable
         pfl_roi_find = cv2.morphologyEx(
             component_pfl_processed,
@@ -127,14 +126,16 @@ class ChieriBotV4Ocr:
             cv2.getStructuringElement(cv2.MORPH_RECT, [10, 1]),
         )
         pfl_contours, _ = cv2.findContours(
-            pfl_roi_find, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE
+            pfl_roi_find,
+            cv2.RETR_EXTERNAL,
+            cv2.CHAIN_APPROX_NONE,
         )
         pfl_rects = [cv2.boundingRect(c) for c in pfl_contours]
         pfl_rects = [
             r for r in pfl_rects if r[3] > component_pfl_processed.shape[0] * 0.1
         ]
         pfl_rects = sorted(pfl_rects, key=lambda r: r[1])
-        pfl_rects_adjusted = [
+        return [
             (
                 max(rect[0] - 2, 0),
                 rect[1],
@@ -143,10 +144,9 @@ class ChieriBotV4Ocr:
             )
             for rect in pfl_rects
         ]
-        return pfl_rects_adjusted
 
     def preprocess_component_pfl(self, component_bgr: Mat) -> Mat:
-        pfl_rect = construct_int_xywh_rect(self.rois.component_rois.pfl_rect)
+        pfl_rect = self.rois.component_rois.pfl_rect.rounded()
         pfl_roi = crop_xywh(component_bgr, pfl_rect)
         pfl_roi_hsv = cv2.cvtColor(pfl_roi, cv2.COLOR_BGR2HSV)
 
@@ -166,11 +166,17 @@ class ChieriBotV4Ocr:
         pfl_roi_blurred = cv2.GaussianBlur(pfl_roi, (5, 5), 0)
         # pfl_roi_blurred = cv2.medianBlur(pfl_roi, 3)
         _, pfl_roi_blurred_threshold = cv2.threshold(
-            pfl_roi_blurred, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU
+            pfl_roi_blurred,
+            0,
+            255,
+            cv2.THRESH_BINARY + cv2.THRESH_OTSU,
         )
         # and a threshold of the original roi
         _, pfl_roi_threshold = cv2.threshold(
-            pfl_roi, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU
+            pfl_roi,
+            0,
+            255,
+            cv2.THRESH_BINARY + cv2.THRESH_OTSU,
         )
         # turn thresholds into black background
         if pfl_roi_blurred_threshold[2][2] == 255:
@@ -180,64 +186,58 @@ class ChieriBotV4Ocr:
         # return a bitwise_and result
         result = cv2.bitwise_and(pfl_roi_blurred_threshold, pfl_roi_threshold)
         result_eroded = cv2.erode(
-            result, cv2.getStructuringElement(cv2.MORPH_CROSS, (2, 2))
+            result,
+            cv2.getStructuringElement(cv2.MORPH_CROSS, (2, 2)),
         )
         return result_eroded if len(self.find_pfl_rects(result_eroded)) == 3 else result
 
     def ocr_component_pfl(
-        self, component_bgr: Mat
+        self,
-    ) -> Tuple[Optional[int], Optional[int], Optional[int]]:
+        component_bgr: Mat,
+    ) -> tuple[int | None, int | None, int | None]:
         try:
             pfl_roi = self.preprocess_component_pfl(component_bgr)
             pfl_rects = self.find_pfl_rects(pfl_roi)
             pure_far_lost = []
             for pfl_roi_rect in pfl_rects:
                 roi = crop_xywh(pfl_roi, pfl_roi_rect)
-                digit_contours, _ = cv2.findContours(
+                result = self.pfl_knn_provider.result(roi)
-                    roi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE
+                pure_far_lost.append(int(result) if result else None)
-                )
-                digit_rects = [cv2.boundingRect(c) for c in digit_contours]
-                digit_rects = FixRects.connect_broken(
-                    digit_rects, roi.shape[1], roi.shape[0]
-                )
-                digit_rects = FixRects.split_connected(roi, digit_rects)
-                digit_rects = sorted(digit_rects, key=lambda r: r[0])
-                digits = []
-                for digit_rect in digit_rects:
-                    digit = crop_xywh(roi, digit_rect)
-                    digit = resize_fill_square(digit, 20)
-                    digits.append(digit)
-                samples = preprocess_hog(digits)
 
-                _, results, _, _ = self.pfl_knn.findNearest(samples, 4)
-                results = [str(int(i)) for i in results.ravel()]
-                pure_far_lost.append(int("".join(results)))
             return tuple(pure_far_lost)
-        except Exception:
+        except Exception:  # noqa: BLE001
             return (None, None, None)
 
-    def ocr_component(self, component_bgr: Mat) -> B30OcrResultItem:
+    def ocr_component(self, component_bgr: Mat) -> OcrScenarioResult:
         component_blur = cv2.GaussianBlur(component_bgr, (5, 5), 0)
         rating_class = self.ocr_component_rating_class(component_blur)
-        song_id = self.ocr_component_song_id(component_bgr)
+        song_id_results = self.ocr_component_song_id_results(component_bgr)
-        # title = self.ocr_component_title(component_blur)
         # score = self.ocr_component_score(component_blur)
         score = self.ocr_component_score_knn(component_bgr)
         pure, far, lost = self.ocr_component_pfl(component_bgr)
-        return B30OcrResultItem(
+        return OcrScenarioResult(
-            song_id=song_id,
+            song_id=song_id_results[0].image_id,
+            song_id_results=song_id_results,
             rating_class=rating_class,
-            # title=title,
             score=score,
             pure=pure,
             far=far,
             lost=lost,
-            date=None,
+            played_at=None,
         )
 
-    def ocr(self, img_bgr: Mat) -> List[B30OcrResultItem]:
+    def components(self, img: Mat, /):
-        self.set_factor(img_bgr)
+        """
-        return [
+        :param img: BGR format image
-            self.ocr_component(component_bgr)
+        """
-            for component_bgr in self.rois.components(img_bgr)
+        self.set_factor(img)
-        ]
+        return self.rois.components(img)
+
+    def result(self, component_img: Mat, /):
+        return self.ocr_component(component_img)
+
+    def results(self, img: Mat, /) -> list[OcrScenarioResult]:
+        """
+        :param img: BGR format image
+        """
+        return [self.ocr_component(component) for component in self.components(img)]

src/arcaea_offline_ocr/scenarios/b30/chieri/v4/rois.py

@@ -1,12 +1,11 @@
-from typing import List, Optional
+from __future__ import annotations
 
-from ....crop import crop_xywh
+from arcaea_offline_ocr.crop import crop_xywh
-from ....types import Mat, XYWHRect
+from arcaea_offline_ocr.types import Mat, XYWHRect
-from ....utils import apply_factor, construct_int_xywh_rect
 
 
 class ChieriBotV4ComponentRois:
-    def __init__(self, factor: Optional[float] = 1.0):
+    def __init__(self, factor: float = 1.0):
         self.__factor = factor
 
     @property
@@ -19,43 +18,43 @@ class ChieriBotV4ComponentRois:
 
     @property
     def top_font_color_detect(self):
-        return apply_factor((35, 10, 120, 100), self.factor)
+        return XYWHRect(35, 10, 120, 100), self.factor
 
     @property
     def bottom_font_color_detect(self):
-        return apply_factor((30, 125, 175, 110), self.factor)
+        return XYWHRect(30, 125, 175, 110) * self.factor
 
     @property
     def bg_point(self):
-        return apply_factor((75, 10), self.factor)
+        return (75 * self.factor, 10 * self.factor)
 
     @property
     def rating_class_rect(self):
-        return apply_factor((21, 40, 7, 20), self.factor)
+        return XYWHRect(21, 40, 7, 20) * self.factor
 
     @property
     def title_rect(self):
-        return apply_factor((35, 10, 430, 50), self.factor)
+        return XYWHRect(35, 10, 430, 50) * self.factor
 
     @property
     def jacket_rect(self):
-        return apply_factor((263, 0, 239, 239), self.factor)
+        return XYWHRect(263, 0, 239, 239) * self.factor
 
     @property
     def score_rect(self):
-        return apply_factor((30, 60, 270, 55), self.factor)
+        return XYWHRect(30, 60, 270, 55) * self.factor
 
     @property
     def pfl_rect(self):
-        return apply_factor((50, 125, 80, 100), self.factor)
+        return XYWHRect(50, 125, 80, 100) * self.factor
 
     @property
     def date_rect(self):
-        return apply_factor((205, 200, 225, 25), self.factor)
+        return XYWHRect(205, 200, 225, 25) * self.factor
 
 
 class ChieriBotV4Rois:
-    def __init__(self, factor: Optional[float] = 1.0):
+    def __init__(self, factor: float = 1.0):
         self.__factor = factor
         self.__component_rois = ChieriBotV4ComponentRois(factor)
 
@@ -74,54 +73,53 @@ class ChieriBotV4Rois:
 
     @property
     def top(self):
-        return apply_factor(823, self.factor)
+        return 823 * self.factor
 
     @property
     def left(self):
-        return apply_factor(107, self.factor)
+        return 107 * self.factor
 
     @property
     def width(self):
-        return apply_factor(502, self.factor)
+        return 502 * self.factor
 
     @property
     def height(self):
-        return apply_factor(240, self.factor)
+        return 240 * self.factor
 
     @property
     def vertical_gap(self):
-        return apply_factor(74, self.factor)
+        return 74 * self.factor
 
     @property
     def horizontal_gap(self):
-        return apply_factor(40, self.factor)
+        return 40 * self.factor
 
     @property
     def horizontal_items(self):
         return 3
 
-    @property
+    vertical_items = 10
-    def vertical_items(self):
-        return 10
 
     @property
     def b33_vertical_gap(self):
-        return apply_factor(121, self.factor)
+        return 121 * self.factor
 
-    def components(self, img_bgr: Mat) -> List[Mat]:
+    def components(self, img_bgr: Mat) -> list[Mat]:
         first_rect = XYWHRect(x=self.left, y=self.top, w=self.width, h=self.height)
         results = []
 
+        last_rect = first_rect
         for vi in range(self.vertical_items):
             rect = XYWHRect(*first_rect)
             rect += (0, (self.vertical_gap + self.height) * vi, 0, 0)
             for hi in range(self.horizontal_items):
                 if hi > 0:
                     rect += ((self.width + self.horizontal_gap), 0, 0, 0)
-                int_rect = construct_int_xywh_rect(rect)
+                results.append(crop_xywh(img_bgr, rect.rounded()))
-                results.append(crop_xywh(img_bgr, int_rect))
+            last_rect = rect
 
-        rect += (
+        last_rect += (
             -(self.width + self.horizontal_gap) * 2,
             self.height + self.b33_vertical_gap,
             0,
@@ -129,8 +127,7 @@ class ChieriBotV4Rois:
         )
         for hi in range(self.horizontal_items):
             if hi > 0:
-                rect += ((self.width + self.horizontal_gap), 0, 0, 0)
+                last_rect += ((self.width + self.horizontal_gap), 0, 0, 0)
-            int_rect = construct_int_xywh_rect(rect)
+            results.append(crop_xywh(img_bgr, last_rect.rounded()))
-            results.append(crop_xywh(img_bgr, int_rect))
 
         return results

src/arcaea_offline_ocr/scenarios/base.py

@@ -0,0 +1,42 @@
+from __future__ import annotations
+
+from abc import ABC
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, Sequence
+
+if TYPE_CHECKING:
+    from datetime import datetime
+
+    from arcaea_offline_ocr.providers import ImageIdProviderResult
+
+
+@dataclass(kw_only=True)
+class OcrScenarioResult:
+    song_id: str
+    rating_class: int
+    score: int
+
+    song_id_results: Sequence[ImageIdProviderResult] = field(default_factory=list)
+    partner_id_results: Sequence[ImageIdProviderResult] = field(
+        default_factory=list,
+    )
+
+    pure: int | None = None
+    pure_inaccurate: int | None = None
+    pure_early: int | None = None
+    pure_late: int | None = None
+    far: int | None = None
+    far_inaccurate: int | None = None
+    far_early: int | None = None
+    far_late: int | None = None
+    lost: int | None = None
+
+    played_at: datetime | None = None
+    max_recall: int | None = None
+    clear_status: int | None = None
+    clear_type: int | None = None
+    modifier: int | None = None
+
+
+class OcrScenario(ABC):  # noqa: B024
+    pass

src/arcaea_offline_ocr/scenarios/device/__init__.py

@@ -0,0 +1,13 @@
+from .extractor import DeviceRoisExtractor
+from .impl import DeviceScenario
+from .masker import DeviceRoisMaskerAutoT1, DeviceRoisMaskerAutoT2
+from .rois import DeviceRoisAutoT1, DeviceRoisAutoT2
+
+__all__ = [
+    "DeviceRoisAutoT1",
+    "DeviceRoisAutoT2",
+    "DeviceRoisExtractor",
+    "DeviceRoisMaskerAutoT1",
+    "DeviceRoisMaskerAutoT2",
+    "DeviceScenario",
+]

src/arcaea_offline_ocr/scenarios/device/base.py

@@ -0,0 +1,8 @@
+from abc import abstractmethod
+
+from arcaea_offline_ocr.scenarios.base import OcrScenario, OcrScenarioResult
+
+
+class DeviceScenarioBase(OcrScenario):
+    @abstractmethod
+    def result(self) -> OcrScenarioResult: ...

src/arcaea_offline_ocr/scenarios/device/extractor/__init__.py

@@ -0,0 +1,3 @@
+from .base import DeviceRoisExtractor
+
+__all__ = ["DeviceRoisExtractor"]

src/arcaea_offline_ocr/scenarios/device/extractor/base.py

@@ -0,0 +1,45 @@
+from arcaea_offline_ocr.crop import crop_xywh
+from arcaea_offline_ocr.scenarios.device.rois import DeviceRois
+from arcaea_offline_ocr.types import Mat
+
+
+class DeviceRoisExtractor:
+    def __init__(self, img: Mat, rois: DeviceRois):
+        self.img = img
+        self.sizes = rois
+
+    @property
+    def pure(self):
+        return crop_xywh(self.img, self.sizes.pure.rounded())
+
+    @property
+    def far(self):
+        return crop_xywh(self.img, self.sizes.far.rounded())
+
+    @property
+    def lost(self):
+        return crop_xywh(self.img, self.sizes.lost.rounded())
+
+    @property
+    def score(self):
+        return crop_xywh(self.img, self.sizes.score.rounded())
+
+    @property
+    def jacket(self):
+        return crop_xywh(self.img, self.sizes.jacket.rounded())
+
+    @property
+    def rating_class(self):
+        return crop_xywh(self.img, self.sizes.rating_class.rounded())
+
+    @property
+    def max_recall(self):
+        return crop_xywh(self.img, self.sizes.max_recall.rounded())
+
+    @property
+    def clear_status(self):
+        return crop_xywh(self.img, self.sizes.clear_status.rounded())
+
+    @property
+    def partner_icon(self):
+        return crop_xywh(self.img, self.sizes.partner_icon.rounded())

src/arcaea_offline_ocr/scenarios/device/impl.py

@@ -1,58 +1,56 @@
 import cv2
 import numpy as np
 
-from ..crop import crop_xywh
+from arcaea_offline_ocr.providers import (
-from ..ocr import (
+    ImageCategory,
-    FixRects,
+    ImageIdProvider,
-    ocr_digit_samples_knn,
+    OcrKNearestTextProvider,
-    ocr_digits_by_contour_knn,
-    preprocess_hog,
-    resize_fill_square,
 )
-from ..phash_db import ImagePhashDatabase
+from arcaea_offline_ocr.scenarios.base import OcrScenarioResult
-from ..types import Mat
+from arcaea_offline_ocr.types import Mat
-from .common import DeviceOcrResult
+
-from .rois.extractor import DeviceRoisExtractor
+from .base import DeviceScenarioBase
-from .rois.masker import DeviceRoisMasker
+from .extractor import DeviceRoisExtractor
+from .masker import DeviceRoisMasker
 
 
-class DeviceOcr:
+class DeviceScenario(DeviceScenarioBase):
     def __init__(
         self,
         extractor: DeviceRoisExtractor,
         masker: DeviceRoisMasker,
-        knn_model: cv2.ml.KNearest,
+        knn_provider: OcrKNearestTextProvider,
-        phash_db: ImagePhashDatabase,
+        image_id_provider: ImageIdProvider,
     ):
         self.extractor = extractor
         self.masker = masker
-        self.knn_model = knn_model
+        self.knn_provider = knn_provider
-        self.phash_db = phash_db
+        self.image_id_provider = image_id_provider
 
     def pfl(self, roi_gray: Mat, factor: float = 1.25):
-        contours, _ = cv2.findContours(
+        def contour_filter(cnt):
-            roi_gray, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE
+            return cv2.contourArea(cnt) >= 5 * factor
-        )
-        filtered_contours = [c for c in contours if cv2.contourArea(c) >= 5 * factor]
-        rects = [cv2.boundingRect(c) for c in filtered_contours]
-        rects = FixRects.connect_broken(rects, roi_gray.shape[1], roi_gray.shape[0])
 
-        filtered_rects = [r for r in rects if r[2] >= 5 * factor and r[3] >= 6 * factor]
+        contours = self.knn_provider.contours(roi_gray)
-        filtered_rects = FixRects.split_connected(roi_gray, filtered_rects)
+        contours_filtered = self.knn_provider.contours(
-        filtered_rects = sorted(filtered_rects, key=lambda r: r[0])
+            roi_gray,
+            contours_filter=contour_filter,
+        )
 
         roi_ocr = roi_gray.copy()
-        filtered_contours_flattened = {tuple(c.flatten()) for c in filtered_contours}
+        contours_filtered_flattened = {tuple(c.flatten()) for c in contours_filtered}
         for contour in contours:
-            if tuple(contour.flatten()) in filtered_contours_flattened:
+            if tuple(contour.flatten()) in contours_filtered_flattened:
                 continue
             roi_ocr = cv2.fillPoly(roi_ocr, [contour], [0])
-        digit_rois = [
-            resize_fill_square(crop_xywh(roi_ocr, r), 20) for r in filtered_rects
-        ]
 
-        samples = preprocess_hog(digit_rois)
+        ocr_result = self.knn_provider.result(
-        return ocr_digit_samples_knn(samples, self.knn_model)
+            roi_ocr,
+            contours_filter=lambda cnt: cv2.contourArea(cnt) >= 5 * factor,
+            rects_filter=lambda rect: rect[2] >= 5 * factor and rect[3] >= 6 * factor,
+        )
+
+        return int(ocr_result) if ocr_result else 0
 
     def pure(self):
         return self.pfl(self.masker.pure(self.extractor.pure))
@@ -65,12 +63,14 @@ class DeviceOcr:
 
     def score(self):
         roi = self.masker.score(self.extractor.score)
-        contours, _ = cv2.findContours(roi, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
+        contours = self.knn_provider.contours(roi)
         for contour in contours:
-            x, y, w, h = cv2.boundingRect(contour)
+            if (
-            if h < roi.shape[0] * 0.6:
+                cv2.boundingRect(contour)[3] < roi.shape[0] * 0.6
+            ):  # h < score_component_h * 0.6
                 roi = cv2.fillPoly(roi, [contour], [0])
-        return ocr_digits_by_contour_knn(roi, self.knn_model)
+        ocr_result = self.knn_provider.result(roi)
+        return int(ocr_result) if ocr_result else 0
 
     def rating_class(self):
         roi = self.extractor.rating_class
@@ -79,13 +79,15 @@ class DeviceOcr:
             self.masker.rating_class_prs(roi),
             self.masker.rating_class_ftr(roi),
             self.masker.rating_class_byd(roi),
+            self.masker.rating_class_etr(roi),
         ]
         return max(enumerate(results), key=lambda i: np.count_nonzero(i[1]))[0]
 
     def max_recall(self):
-        return ocr_digits_by_contour_knn(
+        ocr_result = self.knn_provider.result(
-            self.masker.max_recall(self.extractor.max_recall), self.knn_model
+            self.masker.max_recall(self.extractor.max_recall),
         )
+        return int(ocr_result) if ocr_result else None
 
     def clear_status(self):
         roi = self.extractor.clear_status
@@ -97,20 +99,18 @@ class DeviceOcr:
         ]
         return max(enumerate(results), key=lambda i: np.count_nonzero(i[1]))[0]
 
-    def lookup_song_id(self):
+    def song_id_results(self):
-        return self.phash_db.lookup_jacket(
+        return self.image_id_provider.results(
-            cv2.cvtColor(self.extractor.jacket, cv2.COLOR_BGR2GRAY)
+            cv2.cvtColor(self.extractor.jacket, cv2.COLOR_BGR2GRAY),
+            ImageCategory.JACKET,
         )
 
-    def song_id(self):
-        return self.lookup_song_id()[0]
-
     @staticmethod
     def preprocess_char_icon(img_gray: Mat):
         h, w = img_gray.shape[:2]
-        img = cv2.copyMakeBorder(img_gray, w - h, 0, 0, 0, cv2.BORDER_REPLICATE)
+        img = cv2.copyMakeBorder(img_gray, max(w - h, 0), 0, 0, 0, cv2.BORDER_REPLICATE)
         h, w = img.shape[:2]
-        img = cv2.fillPoly(
+        return cv2.fillPoly(
             img,
             [
                 np.array([[0, 0], [round(w / 2), 0], [0, round(h / 2)]], np.int32),
@@ -118,21 +118,18 @@ class DeviceOcr:
                 np.array([[0, h], [round(w / 2), h], [0, round(h / 2)]], np.int32),
                 np.array([[w, h], [round(w / 2), h], [w, round(h / 2)]], np.int32),
             ],
-            (128),
+            (128,),
         )
-        return img
 
-    def lookup_partner_id(self):
+    def partner_id_results(self):
-        return self.phash_db.lookup_partner_icon(
+        return self.image_id_provider.results(
             self.preprocess_char_icon(
-                cv2.cvtColor(self.extractor.partner_icon, cv2.COLOR_BGR2GRAY)
+                cv2.cvtColor(self.extractor.partner_icon, cv2.COLOR_BGR2GRAY),
-            )
+            ),
+            ImageCategory.PARTNER_ICON,
         )
 
-    def partner_id(self):
+    def result(self):
-        return self.lookup_partner_id()[0]
-
-    def ocr(self) -> DeviceOcrResult:
         rating_class = self.rating_class()
         pure = self.pure()
         far = self.far()
@@ -141,20 +138,18 @@ class DeviceOcr:
         max_recall = self.max_recall()
         clear_status = self.clear_status()
 
-        hash_len = self.phash_db.hash_size**2
+        song_id_results = self.song_id_results()
-        song_id, song_id_distance = self.lookup_song_id()
+        partner_id_results = self.partner_id_results()
-        partner_id, partner_id_distance = self.lookup_partner_id()
 
-        return DeviceOcrResult(
+        return OcrScenarioResult(
+            song_id=song_id_results[0].image_id,
+            song_id_results=song_id_results,
             rating_class=rating_class,
             pure=pure,
             far=far,
             lost=lost,
             score=score,
             max_recall=max_recall,
-            song_id=song_id,
+            partner_id_results=partner_id_results,
-            song_id_possibility=1 - song_id_distance / hash_len,
             clear_status=clear_status,
-            partner_id=partner_id,
-            partner_id_possibility=1 - partner_id_distance / hash_len,
         )

src/arcaea_offline_ocr/scenarios/device/masker/__init__.py

@@ -0,0 +1,9 @@
+from .auto import DeviceRoisMaskerAuto, DeviceRoisMaskerAutoT1, DeviceRoisMaskerAutoT2
+from .base import DeviceRoisMasker
+
+__all__ = [
+    "DeviceRoisMasker",
+    "DeviceRoisMaskerAuto",
+    "DeviceRoisMaskerAutoT1",
+    "DeviceRoisMaskerAutoT2",
+]

src/arcaea_offline_ocr/scenarios/device/masker/auto.py

@@ -1,20 +1,27 @@
 import cv2
 import numpy as np
 
-from ....types import Mat
+from arcaea_offline_ocr.types import Mat
-from .common import DeviceRoisMasker
+
+from .base import DeviceRoisMasker
 
 
 class DeviceRoisMaskerAuto(DeviceRoisMasker):
-    ...
+    @staticmethod
+    def mask_bgr_in_hsv(roi_bgr: Mat, hsv_lower: Mat, hsv_upper: Mat):
+        return cv2.inRange(
+            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            hsv_lower,
+            hsv_upper,
+        )
 
 
 class DeviceRoisMaskerAutoT1(DeviceRoisMaskerAuto):
     GRAY_BGR_MIN = np.array([50] * 3, np.uint8)
     GRAY_BGR_MAX = np.array([160] * 3, np.uint8)
 
-    WHITE_HSV_MIN = np.array([0, 0, 240], np.uint8)
+    SCORE_HSV_MIN = np.array([0, 0, 180], np.uint8)
-    WHITE_HSV_MAX = np.array([179, 10, 255], np.uint8)
+    SCORE_HSV_MAX = np.array([179, 255, 255], np.uint8)
 
     PST_HSV_MIN = np.array([100, 50, 80], np.uint8)
     PST_HSV_MAX = np.array([100, 255, 255], np.uint8)
@@ -28,6 +35,9 @@ class DeviceRoisMaskerAutoT1(DeviceRoisMaskerAuto):
     BYD_HSV_MIN = np.array([170, 50, 50], np.uint8)
     BYD_HSV_MAX = np.array([179, 210, 198], np.uint8)
 
+    ETR_HSV_MIN = np.array([130, 60, 80], np.uint8)
+    ETR_HSV_MAX = np.array([140, 145, 180], np.uint8)
+
     TRACK_LOST_HSV_MIN = np.array([170, 75, 90], np.uint8)
     TRACK_LOST_HSV_MAX = np.array([175, 170, 160], np.uint8)
 
@@ -63,35 +73,27 @@ class DeviceRoisMaskerAutoT1(DeviceRoisMaskerAuto):
 
     @classmethod
     def score(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.SCORE_HSV_MIN, cls.SCORE_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
-            cls.WHITE_HSV_MIN,
-            cls.WHITE_HSV_MAX,
-        )
 
     @classmethod
     def rating_class_pst(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.PST_HSV_MIN, cls.PST_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.PST_HSV_MIN, cls.PST_HSV_MAX
-        )
 
     @classmethod
     def rating_class_prs(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.PRS_HSV_MIN, cls.PRS_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.PRS_HSV_MIN, cls.PRS_HSV_MAX
-        )
 
     @classmethod
     def rating_class_ftr(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.FTR_HSV_MIN, cls.FTR_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.FTR_HSV_MIN, cls.FTR_HSV_MAX
-        )
 
     @classmethod
     def rating_class_byd(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.BYD_HSV_MIN, cls.BYD_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.BYD_HSV_MIN, cls.BYD_HSV_MAX
+
-        )
+    @classmethod
+    def rating_class_etr(cls, roi_bgr: Mat) -> Mat:
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.ETR_HSV_MIN, cls.ETR_HSV_MAX)
 
     @classmethod
     def max_recall(cls, roi_bgr: Mat) -> Mat:
@@ -99,32 +101,32 @@ class DeviceRoisMaskerAutoT1(DeviceRoisMaskerAuto):
 
     @classmethod
     def clear_status_track_lost(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.TRACK_LOST_HSV_MIN,
             cls.TRACK_LOST_HSV_MAX,
         )
 
     @classmethod
     def clear_status_track_complete(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.TRACK_COMPLETE_HSV_MIN,
             cls.TRACK_COMPLETE_HSV_MAX,
         )
 
     @classmethod
     def clear_status_full_recall(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.FULL_RECALL_HSV_MIN,
             cls.FULL_RECALL_HSV_MAX,
         )
 
     @classmethod
     def clear_status_pure_memory(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.PURE_MEMORY_HSV_MIN,
             cls.PURE_MEMORY_HSV_MAX,
         )
@@ -132,10 +134,10 @@ class DeviceRoisMaskerAutoT1(DeviceRoisMaskerAuto):
 
 class DeviceRoisMaskerAutoT2(DeviceRoisMaskerAuto):
     PFL_HSV_MIN = np.array([0, 0, 248], np.uint8)
-    PFL_HSV_MAX = np.array([179, 10, 255], np.uint8)
+    PFL_HSV_MAX = np.array([179, 40, 255], np.uint8)
 
-    WHITE_HSV_MIN = np.array([0, 0, 240], np.uint8)
+    SCORE_HSV_MIN = np.array([0, 0, 180], np.uint8)
-    WHITE_HSV_MAX = np.array([179, 10, 255], np.uint8)
+    SCORE_HSV_MAX = np.array([179, 255, 255], np.uint8)
 
     PST_HSV_MIN = np.array([100, 50, 80], np.uint8)
     PST_HSV_MAX = np.array([100, 255, 255], np.uint8)
@@ -149,6 +151,9 @@ class DeviceRoisMaskerAutoT2(DeviceRoisMaskerAuto):
     BYD_HSV_MIN = np.array([170, 50, 50], np.uint8)
     BYD_HSV_MAX = np.array([179, 210, 198], np.uint8)
 
+    ETR_HSV_MIN = np.array([130, 60, 80], np.uint8)
+    ETR_HSV_MAX = np.array([140, 145, 180], np.uint8)
+
     MAX_RECALL_HSV_MIN = np.array([125, 0, 0], np.uint8)
     MAX_RECALL_HSV_MAX = np.array([145, 100, 150], np.uint8)
 
@@ -166,9 +171,7 @@ class DeviceRoisMaskerAutoT2(DeviceRoisMaskerAuto):
 
     @classmethod
     def pfl(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.PFL_HSV_MIN, cls.PFL_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.PFL_HSV_MIN, cls.PFL_HSV_MAX
-        )
 
     @classmethod
     def pure(cls, roi_bgr: Mat) -> Mat:
@@ -184,72 +187,64 @@ class DeviceRoisMaskerAutoT2(DeviceRoisMaskerAuto):
 
     @classmethod
     def score(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.SCORE_HSV_MIN, cls.SCORE_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
-            cls.WHITE_HSV_MIN,
-            cls.WHITE_HSV_MAX,
-        )
 
     @classmethod
     def rating_class_pst(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.PST_HSV_MIN, cls.PST_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.PST_HSV_MIN, cls.PST_HSV_MAX
-        )
 
     @classmethod
     def rating_class_prs(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.PRS_HSV_MIN, cls.PRS_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.PRS_HSV_MIN, cls.PRS_HSV_MAX
-        )
 
     @classmethod
     def rating_class_ftr(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.FTR_HSV_MIN, cls.FTR_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.FTR_HSV_MIN, cls.FTR_HSV_MAX
-        )
 
     @classmethod
     def rating_class_byd(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.BYD_HSV_MIN, cls.BYD_HSV_MAX)
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV), cls.BYD_HSV_MIN, cls.BYD_HSV_MAX
+
-        )
+    @classmethod
+    def rating_class_etr(cls, roi_bgr: Mat) -> Mat:
+        return cls.mask_bgr_in_hsv(roi_bgr, cls.ETR_HSV_MIN, cls.ETR_HSV_MAX)
 
     @classmethod
     def max_recall(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.MAX_RECALL_HSV_MIN,
             cls.MAX_RECALL_HSV_MAX,
         )
 
     @classmethod
     def clear_status_track_lost(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.TRACK_LOST_HSV_MIN,
             cls.TRACK_LOST_HSV_MAX,
         )
 
     @classmethod
     def clear_status_track_complete(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.TRACK_COMPLETE_HSV_MIN,
             cls.TRACK_COMPLETE_HSV_MAX,
         )
 
     @classmethod
     def clear_status_full_recall(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.FULL_RECALL_HSV_MIN,
             cls.FULL_RECALL_HSV_MAX,
         )
 
     @classmethod
     def clear_status_pure_memory(cls, roi_bgr: Mat) -> Mat:
-        return cv2.inRange(
+        return cls.mask_bgr_in_hsv(
-            cv2.cvtColor(roi_bgr, cv2.COLOR_BGR2HSV),
+            roi_bgr,
             cls.PURE_MEMORY_HSV_MIN,
             cls.PURE_MEMORY_HSV_MAX,
         )

src/arcaea_offline_ocr/scenarios/device/masker/base.py

@@ -0,0 +1,61 @@
+from abc import ABC, abstractmethod
+
+from arcaea_offline_ocr.types import Mat
+
+
+class DeviceRoisMasker(ABC):
+    @classmethod
+    @abstractmethod
+    def pure(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def far(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def lost(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def score(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def rating_class_pst(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def rating_class_prs(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def rating_class_ftr(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def rating_class_byd(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def rating_class_etr(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def max_recall(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def clear_status_track_lost(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def clear_status_track_complete(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def clear_status_full_recall(cls, roi_bgr: Mat) -> Mat: ...
+
+    @classmethod
+    @abstractmethod
+    def clear_status_pure_memory(cls, roi_bgr: Mat) -> Mat: ...

src/arcaea_offline_ocr/scenarios/device/rois/__init__.py

@@ -0,0 +1,9 @@
+from .auto import DeviceRoisAuto, DeviceRoisAutoT1, DeviceRoisAutoT2
+from .base import DeviceRois
+
+__all__ = [
+    "DeviceRois",
+    "DeviceRoisAuto",
+    "DeviceRoisAutoT1",
+    "DeviceRoisAutoT2",
+]

src/arcaea_offline_ocr/scenarios/device/rois/auto.py

@@ -1,6 +1,6 @@
-from .common import DeviceRois
+from arcaea_offline_ocr.types import XYWHRect
 
-__all__ = ["DeviceRoisAuto", "DeviceRoisAutoT1", "DeviceRoisAutoT2"]
+from .base import DeviceRois
 
 
 class DeviceRoisAuto(DeviceRois):
@@ -50,7 +50,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
 
     @property
     def pure(self):
-        return (
+        return XYWHRect(
             self.pfl_x,
             self.layout_area_h_mid + 110 * self.factor,
             self.pfl_w,
@@ -59,7 +59,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
 
     @property
     def far(self):
-        return (
+        return XYWHRect(
             self.pfl_x,
             self.pure[1] + self.pure[3] + 12 * self.factor,
             self.pfl_w,
@@ -68,7 +68,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
 
     @property
     def lost(self):
-        return (
+        return XYWHRect(
             self.pfl_x,
             self.far[1] + self.far[3] + 10 * self.factor,
             self.pfl_w,
@@ -79,7 +79,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
     def score(self):
         w = 280 * self.factor
         h = 45 * self.factor
-        return (
+        return XYWHRect(
             self.w_mid - w / 2,
             self.layout_area_h_mid - 75 * self.factor - h,
             w,
@@ -88,7 +88,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
 
     @property
     def rating_class(self):
-        return (
+        return XYWHRect(
             self.w_mid - 610 * self.factor,
             self.layout_area_h_mid - 180 * self.factor,
             265 * self.factor,
@@ -97,7 +97,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
 
     @property
     def max_recall(self):
-        return (
+        return XYWHRect(
             self.w_mid - 465 * self.factor,
             self.layout_area_h_mid - 215 * self.factor,
             150 * self.factor,
@@ -106,7 +106,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
 
     @property
     def jacket(self):
-        return (
+        return XYWHRect(
             self.w_mid - 610 * self.factor,
             self.layout_area_h_mid - 143 * self.factor,
             375 * self.factor,
@@ -117,10 +117,10 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
     def clear_status(self):
         w = 550 * self.factor
         h = 60 * self.factor
-        return (
+        return XYWHRect(
             self.w_mid - w / 2,
             self.layout_area_h_mid - 155 * self.factor - h,
-            w,
+            w * 0.4,
             h,
         )
 
@@ -128,7 +128,7 @@ class DeviceRoisAutoT1(DeviceRoisAuto):
     def partner_icon(self):
         w = 90 * self.factor
         h = 75 * self.factor
-        return (self.w_mid - w / 2, 0, w, h)
+        return XYWHRect(self.w_mid - w / 2, 0, w, h)
 
 
 class DeviceRoisAutoT2(DeviceRoisAuto):
@@ -174,7 +174,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
 
     @property
     def pure(self):
-        return (
+        return XYWHRect(
             self.pfl_x,
             self.layout_area_h_mid + 175 * self.factor,
             self.pfl_w,
@@ -183,7 +183,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
 
     @property
     def far(self):
-        return (
+        return XYWHRect(
             self.pfl_x,
             self.pure[1] + self.pure[3] + 30 * self.factor,
             self.pfl_w,
@@ -192,7 +192,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
 
     @property
     def lost(self):
-        return (
+        return XYWHRect(
             self.pfl_x,
             self.far[1] + self.far[3] + 35 * self.factor,
             self.pfl_w,
@@ -203,7 +203,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
     def score(self):
         w = 420 * self.factor
         h = 70 * self.factor
-        return (
+        return XYWHRect(
             self.w_mid - w / 2,
             self.layout_area_h_mid - 110 * self.factor - h,
             w,
@@ -212,7 +212,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
 
     @property
     def rating_class(self):
-        return (
+        return XYWHRect(
             max(0, self.w_mid - 965 * self.factor),
             self.layout_area_h_mid - 330 * self.factor,
             350 * self.factor,
@@ -221,7 +221,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
 
     @property
     def max_recall(self):
-        return (
+        return XYWHRect(
             self.w_mid - 625 * self.factor,
             self.layout_area_h_mid - 275 * self.factor,
             150 * self.factor,
@@ -230,7 +230,7 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
 
     @property
     def jacket(self):
-        return (
+        return XYWHRect(
             self.w_mid - 915 * self.factor,
             self.layout_area_h_mid - 215 * self.factor,
             565 * self.factor,
@@ -241,10 +241,10 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
     def clear_status(self):
         w = 825 * self.factor
         h = 90 * self.factor
-        return (
+        return XYWHRect(
             self.w_mid - w / 2,
             self.layout_area_h_mid - 235 * self.factor - h,
-            w,
+            w * 0.4,
             h,
         )
 
@@ -252,4 +252,4 @@ class DeviceRoisAutoT2(DeviceRoisAuto):
     def partner_icon(self):
         w = 135 * self.factor
         h = 110 * self.factor
-        return (self.w_mid - w / 2, 0, w, h)
+        return XYWHRect(self.w_mid - w / 2, 0, w, h)

src/arcaea_offline_ocr/scenarios/device/rois/base.py

@@ -0,0 +1,33 @@
+from abc import ABC, abstractmethod
+
+from arcaea_offline_ocr.types import XYWHRect
+
+
+class DeviceRois(ABC):
+    @property
+    @abstractmethod
+    def pure(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def far(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def lost(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def score(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def rating_class(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def max_recall(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def jacket(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def clear_status(self) -> XYWHRect: ...
+    @property
+    @abstractmethod
+    def partner_icon(self) -> XYWHRect: ...

src/arcaea_offline_ocr/types.py

@@ -1,25 +1,42 @@
-from collections.abc import Iterable
+from math import floor
-from typing import NamedTuple, Tuple, Union
+from typing import Callable, NamedTuple, Union
 
 import numpy as np
 
 Mat = np.ndarray
 
+_IntOrFloat = Union[int, float]
+
 
 class XYWHRect(NamedTuple):
-    x: int
+    x: _IntOrFloat
-    y: int
+    y: _IntOrFloat
-    w: int
+    w: _IntOrFloat
-    h: int
+    h: _IntOrFloat
 
-    def __add__(self, other: Union["XYWHRect", Tuple[int, int, int, int]]):
+    def _to_int(self, func: Callable[[_IntOrFloat], int]):
-        if not isinstance(other, Iterable) or len(other) != 4:
+        return (func(self.x), func(self.y), func(self.w), func(self.h))
-            raise ValueError()
+
+    def rounded(self):
+        return self._to_int(round)
+
+    def floored(self):
+        return self._to_int(floor)
+
+    def __add__(self, other):
+        if not isinstance(other, (list, tuple)) or len(other) != 4:
+            raise TypeError
 
         return self.__class__(*[a + b for a, b in zip(self, other)])
 
-    def __sub__(self, other: Union["XYWHRect", Tuple[int, int, int, int]]):
+    def __sub__(self, other):
-        if not isinstance(other, Iterable) or len(other) != 4:
+        if not isinstance(other, (list, tuple)) or len(other) != 4:
-            raise ValueError()
+            raise TypeError
 
         return self.__class__(*[a - b for a, b in zip(self, other)])
+
+    def __mul__(self, other):
+        if not isinstance(other, (int, float)):
+            raise TypeError
+
+        return self.__class__(*[v * other for v in self])

src/arcaea_offline_ocr/utils.py

@@ -1,11 +1,6 @@
-from collections.abc import Iterable
-from typing import Callable, TypeVar, Union, overload
-
 import cv2
 import numpy as np
 
-from .types import XYWHRect
-
 __all__ = ["imread_unicode"]
 
 
@@ -13,34 +8,3 @@ def imread_unicode(filepath: str, flags: int = cv2.IMREAD_UNCHANGED):
     # https://stackoverflow.com/a/57872297/16484891
     # CC BY-SA 4.0
     return cv2.imdecode(np.fromfile(filepath, dtype=np.uint8), flags)
-
-
-def construct_int_xywh_rect(
-    rect: XYWHRect, func: Callable[[Union[int, float]], int] = round
-):
-    return XYWHRect(*[func(num) for num in rect])
-
-
-@overload
-def apply_factor(item: int, factor: float) -> float:
-    ...
-
-
-@overload
-def apply_factor(item: float, factor: float) -> float:
-    ...
-
-
-T = TypeVar("T", bound=Iterable)
-
-
-@overload
-def apply_factor(item: T, factor: float) -> T:
-    ...
-
-
-def apply_factor(item, factor: float):
-    if isinstance(item, (int, float)):
-        return item * factor
-    elif isinstance(item, Iterable):
-        return item.__class__([i * factor for i in item])