iamraw 4.91.1__py3-none-any.whl

iamraw/__init__.py

@@ -0,0 +1,353 @@
+# =============================================================================
+# C O P Y R I G H T
+# -----------------------------------------------------------------------------
+# Copyright (c) 2019-2023 by Helmut Konrad Schewe. All rights reserved.
+# This file is property of Helmut Konrad Schewe. Any unauthorized copy,
+# use or distribution is an offensive act against international law and may
+# be prosecuted under federal law. Its content is company confidential.
+# =============================================================================
+
+import os
+
+# path
+from iamraw import path
+# mixin
+from iamraw.__mixin__ import extracted
+from iamraw.__mixin__ import hasstrategy
+# abbreviation
+from iamraw.abbreviation import Abbreviation
+from iamraw.abbreviation import AbbreviationList
+from iamraw.abbreviation import AbbreviationListLookup
+from iamraw.abbreviation import AbbreviationLists
+from iamraw.abbreviation import AbbreviationPosition
+from iamraw.abbreviation import AbbreviationResult
+from iamraw.abbreviation import Abbreviations
+from iamraw.abbreviation import ExtractedTextAbbreviation
+from iamraw.abbreviation import ExtractedTextAbbreviations
+# annotation
+from iamraw.annotation import Annotation
+from iamraw.annotation import HyperLink
+from iamraw.annotation import Link
+from iamraw.annotation import PageAnnotation
+from iamraw.annotation import PageAnnotations
+from iamraw.annotation import PageLink
+from iamraw.annotation import hyperlink_annotations
+from iamraw.annotation import pagelink_annotations
+# bib
+from iamraw.bibliography import BibliographyReference
+from iamraw.bibliography import BibliographyReferences
+from iamraw.bibliography import BibliographyTable
+# border
+from iamraw.border import Border
+from iamraw.border import Borders
+# bounding
+from iamraw.bounding import BoundingBox
+from iamraw.bounding import BoundingBoxes
+from iamraw.bounding import PageBoundings
+from iamraw.bounding import PageBoundingsList
+from iamraw.bounding import between
+from iamraw.bounding import split_x
+from iamraw.bounding import split_y
+# boxes
+from iamraw.boxes import Box
+from iamraw.boxes import HorizontalLine
+from iamraw.boxes import PageContentBoxes
+from iamraw.boxes import PageContentHorizontals
+from iamraw.boxes import PagesWithBoxList
+from iamraw.boxes import PagesWithHorizontalList
+# captions
+from iamraw.caption import Caption
+from iamraw.caption import Captions
+from iamraw.caption import CaptionType
+from iamraw.caption import PageContentCaption
+from iamraw.caption import PageContentCaptions
+# code
+from iamraw.code import PageContentCode
+from iamraw.code import PageContentCodes
+from iamraw.code import PeaceOfCode
+from iamraw.code import PeaceOfCodes
+# content
+from iamraw.content import ContentBoundingBox
+from iamraw.content import ContentBoundingBoxes
+# distance
+from iamraw.distance import AreaDistance
+from iamraw.distance import AreaDistances
+from iamraw.distance import PageContentAreaDistance
+from iamraw.distance import PageContentAreaDistances
+# docinfo
+from iamraw.docinfo import DocContentType
+from iamraw.docinfo import DocInfo
+from iamraw.docinfo import DocumentType
+from iamraw.docinfo import Generator
+from iamraw.docinfo import SectionLookup
+# docref
+from iamraw.docref import DocRef
+from iamraw.docref import DocRefs
+from iamraw.docref import TextAdvice
+from iamraw.docref import TextAdviceDelete
+from iamraw.docref import TextAdviceReplacement
+from iamraw.docref import TextAdvices
+# document
+from iamraw.document import Boxed
+from iamraw.document import Char
+from iamraw.document import Chars
+from iamraw.document import Document
+from iamraw.document import Line
+from iamraw.document import Lines
+from iamraw.document import Page
+from iamraw.document import PageObject
+from iamraw.document import Pages
+from iamraw.document import PageSize
+from iamraw.document import PageSizes
+from iamraw.document import TextContainer
+from iamraw.document import TextContainers
+from iamraw.document import UnicodeChar
+from iamraw.document import VerticalTextContainer
+from iamraw.document import VerticalTextContainers
+from iamraw.document import VirtualChar
+# figures
+from iamraw.figure import Figure
+from iamraw.figure import Figures
+# findings
+from iamraw.findings import BoundingLocation
+from iamraw.findings import Finding
+from iamraw.findings import FindingLevel
+from iamraw.findings import Findings
+from iamraw.findings import Location
+from iamraw.findings import PageFinding
+from iamraw.findings import PageFindings
+from iamraw.findings import RangedLocation
+from iamraw.findings import select_findings
+# fonts
+from iamraw.fonts import DEFAULT_STRETCH
+from iamraw.fonts import DEFAULT_STYLE
+from iamraw.fonts import DEFAULT_WEIGHT
+from iamraw.fonts import Font
+from iamraw.fonts import FontFlag
+from iamraw.fonts import FontFlags
+from iamraw.fonts import PageFontContent
+from iamraw.fonts import PageFontContents
+from iamraw.fonts import Stretch
+from iamraw.fonts import Style
+from iamraw.fonts import Weight
+from iamraw.fontstore import NO_FONT
+from iamraw.fontstore import FontChunk
+from iamraw.fontstore import FontChunks
+from iamraw.fontstore import FontContentStore
+from iamraw.fontstore import FontStore
+# footnotes
+from iamraw.footnotes import PageContentFootnote
+from iamraw.footnotes import PageContentFootnotes
+# formula
+from iamraw.formula import Formula
+from iamraw.formula import FormulaRaw
+from iamraw.formula import Formulas
+from iamraw.formula import FormulasRaw
+from iamraw.formula import MathChar
+from iamraw.formula import MathChars
+from iamraw.formula import PageContentFormula
+from iamraw.formula import PageContentFormulas
+from iamraw.formula import PageContentRawFormula
+from iamraw.formula import PageContentRawFormulas
+# headerfooter
+from iamraw.headerfooter import FixedFooterInfo
+from iamraw.headerfooter import FixedHeaderInfo
+from iamraw.headerfooter import FooterInfo
+from iamraw.headerfooter import FootJudgedNote
+from iamraw.headerfooter import FootNote
+from iamraw.headerfooter import FootNoteMerged
+from iamraw.headerfooter import FootNoteRaw
+from iamraw.headerfooter import FootNotes
+from iamraw.headerfooter import HeaderImages
+from iamraw.headerfooter import HeaderInfo
+from iamraw.headerfooter import HeaderTitle
+from iamraw.headerfooter import MovingFooterInfo
+from iamraw.headerfooter import PageContentFooterHeader
+from iamraw.headerfooter import PageContentFooterHeaders
+from iamraw.headerfooter import PageInformation
+from iamraw.headerfooter import PagesFooterInfo
+from iamraw.headerfooter import RawText
+# headlines
+from iamraw.headlines import Headline
+from iamraw.headlines import HeadlineGroup
+from iamraw.headlines import HeadlineGroups
+from iamraw.headlines import HeadlineResult
+from iamraw.headlines import Headlines
+from iamraw.headlines import PagesHeadlineList
+from iamraw.headlines import headlines_totoc
+# hits
+from iamraw.hits import PageContentHit
+from iamraw.hits import PageContentHits
+# href
+from iamraw.href import ExtractedHyperLink
+from iamraw.href import ExtractedHyperLinks
+# images
+from iamraw.images import ImageInformation
+from iamraw.images import ImageInformations
+from iamraw.images import PageContentImageInfo
+from iamraw.images import PageContentImageInfos
+# index
+from iamraw.index import DocumentIndex
+from iamraw.index import DocumentIndexElement
+# lang
+from iamraw.lang import Language
+from iamraw.lang import simplelang
+# layout
+from iamraw.layout import Layout
+from iamraw.layout import Layouts
+# likelihood
+from iamraw.likelihood import Likelihood
+from iamraw.likelihood import PageContentLikelihood
+from iamraw.likelihood import PageContentLikelihoods
+# lines
+from iamraw.line import PageContentLine
+from iamraw.line import PageContentLines
+# list
+from iamraw.list import ListType
+from iamraw.list import PageContentList
+from iamraw.list import PageContentLists
+from iamraw.list import PageList
+# magic
+from iamraw.magic import PageContentContentType
+from iamraw.magic import PageContentContentTypes
+from iamraw.magic import PageContentType
+# pages
+from iamraw.page import PageSizeBorder
+from iamraw.page import PageSizeBorderList
+# content
+from iamraw.pagecontent import PageContent
+from iamraw.pagecontent import PageContents
+# pagenumber
+from iamraw.pagenumbers import PageNumber
+from iamraw.pagenumbers import PageNumberOrientation
+# pdf
+from iamraw.pdfinfo import InvalidPDF
+from iamraw.pdfinfo import PDFDate
+from iamraw.pdfinfo import PDFInfo
+from iamraw.pdfinfo import PDFVersion
+# person
+from iamraw.person import NoPerson
+from iamraw.person import Person
+from iamraw.person import Persons
+# quote
+from iamraw.quote import ExtractedQuotation
+from iamraw.quote import ExtractedQuotations
+from iamraw.quote import PageContentBlockQuotes
+from iamraw.quote import PageContentBlockQuotesList
+# sections
+from iamraw.sections import AbbreviationTable
+from iamraw.sections import Abstract
+from iamraw.sections import Acknowledgments
+from iamraw.sections import Appendix
+from iamraw.sections import AreaItem
+from iamraw.sections import AreaItems
+from iamraw.sections import Bibliography
+from iamraw.sections import Chapter
+from iamraw.sections import CiteContent
+from iamraw.sections import CitePart
+from iamraw.sections import CodeTable
+from iamraw.sections import DocumentSection
+from iamraw.sections import FigureTable
+from iamraw.sections import Glossary
+from iamraw.sections import Index
+from iamraw.sections import LegalInformation
+from iamraw.sections import MainPart
+from iamraw.sections import MultipleSection
+from iamraw.sections import NotImplementedItem
+from iamraw.sections import PartOfDocMixin
+from iamraw.sections import PartsOfDoc
+from iamraw.sections import Publication
+from iamraw.sections import SectionMixin
+from iamraw.sections import Sections
+from iamraw.sections import SectionsList
+from iamraw.sections import SymbolTable
+from iamraw.sections import TableOfContent
+from iamraw.sections import TableTable
+from iamraw.sections import TitlePageSection
+from iamraw.sections import Unknown
+# solution
+from iamraw.solution import Doctails
+from iamraw.solution import ProblemStatus
+from iamraw.solution import Solution
+from iamraw.solution import Solutions
+from iamraw.solution import Text
+from iamraw.solution import Web
+# spacestation
+from iamraw.spacestation import DocumentCharDist
+from iamraw.spacestation import DocumentWordDist
+# style
+from iamraw.style import DocTextStyle
+from iamraw.style import PageTextProperties
+from iamraw.style import PageTextPropertiesList
+from iamraw.style import TextProperties
+from iamraw.style import TextProperty
+# table
+from iamraw.table import PageContentTableBounding
+from iamraw.table import PageContentTableBoundings
+from iamraw.table import TableBounding
+from iamraw.table import TableBoundings
+# text
+from iamraw.text import ChapterText
+from iamraw.text import ChapterTextList
+from iamraw.text import ContentType
+from iamraw.text import DFormula
+from iamraw.text import DocumentContent
+from iamraw.text import HeadlineWithContent
+from iamraw.text import PageContentText
+from iamraw.text import PageContentTexts
+from iamraw.text import Paragraph
+from iamraw.text import Paragraphs
+from iamraw.text import TextSection
+from iamraw.text import TextSections
+from iamraw.text import Undefined
+# textposition
+from iamraw.textposition import PageContentTextPosition
+from iamraw.textposition import PageContentTextPositions
+from iamraw.textposition import TextPosition
+from iamraw.textposition import TextPositions
+# title
+from iamraw.title import PROF_DR
+from iamraw.title import AcademicTitle
+# title
+from iamraw.titlepage import THESIS
+from iamraw.titlepage import Institution
+from iamraw.titlepage import Matrikel
+from iamraw.titlepage import TitleDate
+from iamraw.titlepage import TitlePage
+from iamraw.titlepage import TitlePages
+from iamraw.titlepage import TitleThesisType
+# table of content
+from iamraw.toc import AppendixLevel
+from iamraw.toc import Level
+from iamraw.toc import RomanLevel
+from iamraw.toc import Section
+from iamraw.toc import SectionList
+from iamraw.toc import SectionRaw
+from iamraw.toc import StepLevel
+from iamraw.toc import Toc
+from iamraw.toc import TocLinkMixin
+from iamraw.toc import TocLinkMixins
+from iamraw.toc import TocStyle
+from iamraw.toc import create_toc
+from iamraw.toc import merge_toc
+from iamraw.toc import tosection
+from iamraw.toc import tosectionraw
+# webconfig
+from iamraw.webconfig import WebConfig
+# whitepage
+from iamraw.whitepage import PageContentWhitepage
+from iamraw.whitepage import PageContentWhitepages
+from iamraw.whitepage import WhitePage
+
+# TODO: REMOVE LATER
+FootRawNote = FootNoteRaw
+FixedFooterInformation = FixedFooterInfo
+FixedHeaderInformation = FixedHeaderInfo
+HeaderInformation = HeaderInfo
+FooterInformation = FooterInfo
+MovingFooterInformation = MovingFooterInfo
+PagesFooterInformation = PagesFooterInfo
+
+__version__ = '4.91.0'
+
+ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), '..'))

iamraw/__mixin__.py

@@ -0,0 +1,33 @@
+# =============================================================================
+# C O P Y R I G H T
+# -----------------------------------------------------------------------------
+# Copyright (c) 2022-2023 by Helmut Konrad Schewe. All rights reserved.
+# This file is property of Helmut Konrad Schewe. Any unauthorized copy,
+# use or distribution is an offensive act against international law and may
+# be prosecuted under federal law. Its content is company confidential.
+# =============================================================================
+"""\
+>>> @extracted
+... class Helmut:
+...     pass
+>>> first = Helmut()
+>>> first.__strategy__ = 'hello'
+>>> first.__strategy__
+'hello'
+"""
+
+
+def extracted(item):
+    setattr(item, '__strategy__', None)
+    setattr(item, '__strategy_location__', None)
+
+    def raw(self) -> str:
+        """Element in document which is converted to current result."""
+        raise NotImplementedError
+
+    setattr(item, '__strategy_raw__', raw)
+    return item
+
+
+def hasstrategy(item) -> bool:
+    return hasattr(item, '__strategy__')

iamraw/abbreviation.py

@@ -0,0 +1,102 @@
+# =============================================================================
+# C O P Y R I G H T
+# -----------------------------------------------------------------------------
+# Copyright (c) 2020-2023 by Helmut Konrad Schewe. All rights reserved.
+# This file is property of Helmut Konrad Schewe. Any unauthorized copy,
+# use or distribution is an offensive act against international law and may
+# be prosecuted under federal law. Its content is company confidential.
+# =============================================================================
+
+import collections
+import dataclasses
+
+import utilo
+
+
+@dataclasses.dataclass
+class AbbreviationPosition:
+    page: int = None
+    sentence: int = None
+    word: int = None
+
+
+@dataclasses.dataclass
+class Abbreviation:
+    short: str = None
+    description: str = None
+    position: AbbreviationPosition = None
+
+    def __lt__(self, item):
+        if utilo.alphabetically(self.short) <= utilo.alphabetically(item.short):
+            return True
+        return False
+
+
+Abbreviations = list[Abbreviation]
+
+
+@dataclasses.dataclass
+class AbbreviationResult:
+
+    abbreviations: Abbreviations = dataclasses.field(default_factory=list)
+    pdfpages: list = dataclasses.field(default_factory=list)
+
+    def append(self, item):
+        self.abbreviations.append(item)  # pylint:disable=E1101
+
+    def __getitem__(self, index):
+        return self.abbreviations[index]  # pylint:disable=E1136
+
+    def __len__(self):
+        return len(self.abbreviations)
+
+    def short_inside(self, abbrev: str) -> bool:
+        """\
+        >>> AbbreviationResult().short_inside('')
+        False
+        """
+        return any((item.short.lower() == abbrev for item in self.abbreviations))  # yapf:disable
+
+
+ExtractedTextAbbreviation = collections.namedtuple(
+    'ExtractedTextAbbreviation',
+    'page, content',
+)
+ExtractedTextAbbreviations = list[ExtractedTextAbbreviation]
+
+
+@dataclasses.dataclass
+class AbbreviationList:
+    data: set = dataclasses.field(default_factory=set)
+
+    def append(self, item):
+        self.data.add(item)  # pylint:disable=E1101
+
+    def __contains__(self, item):
+        return item in self.data  # pylint:disable=unsupported-membership-test
+
+
+AbbreviationLists = list[AbbreviationList]
+
+
+@dataclasses.dataclass
+class AbbreviationListLookup:
+    table: AbbreviationList = dataclasses.field(default=AbbreviationList)
+    other: AbbreviationLists = dataclasses.field(default_factory=list)
+
+    def __contains__(self, item):
+        if item in self.table:  # pylint:disable=E1135
+            return True
+        if self.other:
+            for table in self.other:
+                if item in table:
+                    return True
+        return False
+
+    @classmethod
+    def fromparsed(cls, parsed=None, other=None):
+        assert parsed or other, 'empty input'
+        if parsed is None:
+            parsed = AbbreviationList()
+        lookup = cls(table=parsed, other=other)
+        return lookup

iamraw/annotation.py

@@ -0,0 +1,52 @@
+# =============================================================================
+# C O P Y R I G H T
+# -----------------------------------------------------------------------------
+# Copyright (c) 2019-2023 by Helmut Konrad Schewe. All rights reserved.
+# This file is property of Helmut Konrad Schewe. Any unauthorized copy,
+# use or distribution is an offensive act against international law and may
+# be prosecuted under federal law. Its content is company confidential.
+# =============================================================================
+
+import collections
+import dataclasses
+import enum
+
+import iamraw.bounding
+
+
+class Link(enum.Enum):
+    UNDEFINED = -1
+    INTERNAL = 0
+    HYPERLINK = 1
+
+
+@dataclasses.dataclass
+class Annotation:
+    goal: str
+    bounds: iamraw.bounding.BoundingBox
+    typ: Link = Link.UNDEFINED
+
+
+@dataclasses.dataclass
+class HyperLink(Annotation):
+    typ: Link = Link.HYPERLINK
+
+
+@dataclasses.dataclass
+class PageLink(Annotation):
+    typ: Link = Link.INTERNAL
+
+
+PageAnnotation = collections.namedtuple(
+    'PageAnnotation',
+    'pagelinks hyperlinks page',
+)
+PageAnnotations = list[PageAnnotation]
+
+
+def pagelink_annotations(annos: PageAnnotations) -> list[PageLink]:
+    return [item.pagelinks for item in annos]
+
+
+def hyperlink_annotations(annos: PageAnnotations) -> list[HyperLink]:
+    return [item.hyperlinks for item in annos]

iamraw/bibliography.py

@@ -0,0 +1,79 @@
+# =============================================================================
+# C O P Y R I G H T
+# -----------------------------------------------------------------------------
+# Copyright (c) 2020-2023 by Helmut Konrad Schewe. All rights reserved.
+# This file is property of Helmut Konrad Schewe. Any unauthorized copy,
+# use or distribution is an offensive act against international law and may
+# be prosecuted under federal law. Its content is company confidential.
+# =============================================================================
+
+import contextlib
+import dataclasses
+
+import iamraw
+
+
+@dataclasses.dataclass(unsafe_hash=True)
+class BibliographyReference:  # pylint:disable=R0902
+
+    title: str = None
+    reference: str = None
+
+    data: str = None
+
+    page: int = None
+    pageend: int = None
+
+    year: int = None
+    yearend: int = None
+
+    hyperlink: str = None
+    accessed: str = None
+
+    # a,b,c... to differentiate item in the same year
+    number: str = None
+    authors: list[str] = dataclasses.field(default_factory=list)
+
+    publisher: str = None
+
+    raw: str = dataclasses.field(default=None, compare=False)
+    raw_pdfpage: int = None
+
+    @classmethod
+    def create(cls, author: str, title: str = '', year: int = 2000):
+        author = author.split(' ', maxsplit=1)
+        author = iamraw.Person(name=author[0], firstname=author[1])
+        with contextlib.suppress(TypeError):
+            year = int(year)
+        return cls(authors=[author], title=title, year=year)
+
+    @property
+    def author(self) -> str:
+        """Return family of first author."""
+        with contextlib.suppress(IndexError, AttributeError):
+            # IndexError: No author parsed
+            # AttributeError: NoPerson parsed
+            return self.authors[0].name  # pylint:disable=E1136
+        return None
+
+    def __post_init__(self):
+        # TODO: MAY REMOVE NO YEAR LATER
+        assert any((self.year is None, isinstance(self.year, int), self.year
+                    == 'no year'),), str(self)
+
+
+BibliographyReferences = list[BibliographyReference]
+
+
+@iamraw.extracted
+@dataclasses.dataclass
+class BibliographyTable:
+    headline: str = None
+    references: BibliographyReferences = dataclasses.field(default_factory=list)
+    pdfpages: tuple = None
+
+    def __getitem__(self, index):
+        return self.references[index]
+
+    def __len__(self):
+        return len(self.references)

iamraw/border.py

@@ -0,0 +1,36 @@
+# =============================================================================
+# C O P Y R I G H T
+# -----------------------------------------------------------------------------
+# Copyright (c) 2019-2023 by Helmut Konrad Schewe. All rights reserved.
+# This file is property of Helmut Konrad Schewe. Any unauthorized copy,
+# use or distribution is an offensive act against international law and may
+# be prosecuted under federal law. Its content is company confidential.
+# =============================================================================
+
+import collections
+
+import utilo
+
+Border = collections.namedtuple('Border', 'left right top bottom')
+Borders = list[Border]
+
+
+def validate(items) -> bool:
+    """Iterate throw elements and check if some element contains a negative
+    element.
+
+    Args:
+        items(List[Border/PageSize]): list or single item is supported
+    Returns:
+        True if all elements are positive, False if at least one is not
+    """
+    valid = True
+    if not isinstance(items, list):
+        items = [items]
+    for index, item in enumerate(items):
+        for itemindex, check in enumerate(item):
+            if check is not None and check < 0:
+                msg = 'invalid field(%d, %d): %r' % (index, itemindex, check)
+                utilo.error(msg)
+                valid = False
+    return valid