Namespace LMKit.Document.Parsing

ChartContent: A chart read as data: Text is the plotted series as a Markdown pipe table, series and category labels as header and first-column cells.

ChartContent.Panel: One panel of a figure: the title it prints and the table read from it.

DocumentParser: Parses documents into pages of positioned, categorized and typed elements (text, headings, tables, formulas, figures and the data of the charts they plot) in reading order, and renders them as Markdown, JSON, HTML and DocLang.

ElementContent: Typed content of a ParsedElement. The concrete type says how Text is encoded: Markdown for running text, HTML for a table, LaTeX for a formula, a data table for a chart.

FigureCaptionContent: The caption of a figure read as data, in as many printed lines as the page set it: a title with its subtitle, or the label lines set over the plot. The whole of it heads the figure's table as one heading; a line of it left as plain text would read as body text, and a consumer looking for the figure's title would find half of it.

FigureContent: A figure read as a picture: an illustration, a photo, a logo, a diagram, or a chart whose data was not read. Text is the figure's words when it prints words that are not data (a diagram's labels, a form or a certificate shown as a picture), one printed row per line, or else an optional description; empty when it has neither.

FigureLabelContent: A label line the legend of a chart took over: the series names or the axis title the page prints under the plot. The words head the chart's table as its legend (Legend), so the line renders nowhere in the page rendering; the element keeps its box and its text for a layout view, where the line is still a text block printed at that place.

FormulaContent: A mathematical formula: Text is LaTeX without delimiters; the page rendering fences it.

PageParsedEventArgs: Progress of a parse: raised by PageParsed each time a page has been read.

PageProgressEventArgs: Progress within a page: raised by PageProgress while a page is read, between the page's start and its PageParsed.

ParsedBlock: A layout block of a parsed page: elements the page sets as one group (the entries of a list, a label-value block, a caption over several lines, a column of contents, a label and its value on one row), with the box they share and the probability that they do form one group. A block is a level of the page's layout above its elements and never replaces them: its members stay on the page, each in its own place in the reading order, and the page's rendering does not read blocks.

ParsedDocument: The result of a DocumentParser parse: the parsed pages, in document order, and their renderings in the standard formats.

ParsedElement: One region of a parsed page: its role, where it sits, its place in the reading order, and its typed content.

ParsedFigureText: A run of text a figure prints: a label, a legend entry, a callout or an annotation the page's text layer sets inside a figure's box and no other element of the page stands for, its lines grouped into one block. It is a child of Figure in the page's layout and never an element of the page: the figure stays as it is, and the page's rendering prints it after a figure that renders nothing of its own when the page shows it.

ParsedPage: One parsed page: its geometry and its elements in reading order.

ParsedRegion: A grounding region of a parsed page: the box a reader points at for one unit of the page's content. Most regions hold one element; a region holds several when the page sets them as one unit that the reading split, such as a label and its value printed on one row, or a short caption over two elements stacked in one block. The regions partition the page's elements: every element belongs to exactly one region, and the page's rendering does not read them. A page grounded hierarchically then lists its grounding units (a list as a whole, a run of lines, a row of fields): each is a region of one layout-only member (RepeatedTextContent) that is not one of the page's elements, carrying the words it covers.

RepeatedTextContent: Words another element of the page already carries: a running head or foot spanning a band whose pieces stand as elements of their own. The element keeps its box and its words for a layout view, where the band is one block printed at that place, and renders nowhere in the page rendering, where the words would print twice. Its text holds only characters that name a glyph.

TextContent: Running text, headings, captions and footnotes: Text is Markdown.

ChartMarkdownStyle: How a chart reads in the Markdown of a parsed document (ToMarkdown(ChartMarkdownStyle)). A chart can be read two ways: as the data it plots, a table of values, or as the words and numbers it prints, as a reader sees them on the page. The parse holds both; the style picks what the Markdown writes.

PageParsingStage: What the parser is doing on a page, as PageProgress reports it.

ParsingEffort: How much work a DocumentParser puts into each page: the trade-off between the time a document takes to parse and the fidelity of the result.