extern/pdf-to-markdown
1
0
Fork 0
mirror of https://github.com/jzillmann/pdf-to-markdown.git synced 2025-01-21 04:58:36 +01:00
Code Issues Packages Projects Releases Wiki Activity

[WIP] remove explicit Footnotes transformation

Browse Source
This commit is contained in:
Johannes Zillmann 2017-03-10 12:12:20 +01:00
parent 68e3fd7a9f
commit 15c5946073
2 changed files with 0 additions and 69 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/javascript/models/AppState.jsx
View File

@ -7,7 +7,6 @@ import VerticalToHorizontal from './transformations/VerticalToHorizontal.jsx';
import DetectTOC from './transformations/DetectTOC.jsx' import DetectTOC from './transformations/DetectTOC.jsx'
import GatherBlocks from './transformations/GatherBlocks.jsx' import GatherBlocks from './transformations/GatherBlocks.jsx'
import DetectFootnotes from './transformations/DetectFootnotes.jsx'
import DetectLists from './transformations/DetectLists.jsx' import DetectLists from './transformations/DetectLists.jsx'
import DetectCodeBlocks from './transformations/DetectCodeBlocks.jsx' import DetectCodeBlocks from './transformations/DetectCodeBlocks.jsx'
import DetectHeadlines from './transformations/DetectHeadlines.jsx' import DetectHeadlines from './transformations/DetectHeadlines.jsx'
@ -37,7 +36,6 @@ export default class AppState {
new DetectTOC(), new DetectTOC(),
new GatherBlocks(), new GatherBlocks(),
new DetectFootnotes(),
new DetectLists(), new DetectLists(),
new DetectCodeBlocks(), new DetectCodeBlocks(),
new DetectHeadlines(), new DetectHeadlines(),

67
src/javascript/models/transformations/DetectFootnotes.jsx
View File

@ -1,67 +0,0 @@
import ToTextItemBlockTransformation from './ToTextItemBlockTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItemBlock from '../TextItemBlock.jsx';
import TextItemCombiner from '../TextItemCombiner.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../Annotation.jsx';
import { FOOTNOTE_BLOCK } from '../MarkdownElements.jsx';
//Detect quotes, code etc.. which is transformed to markdown code syntax
export default class DetectFootnotes extends ToTextItemBlockTransformation {
constructor() {
super("Detect Footnotes");
}
transform(parseResult:ParseResult) {
const {mostUsedDistance} = parseResult.globals;
var foundFootnotes = [];
const textCombiner = new TextItemCombiner({
mostUsedDistance: mostUsedDistance,
});
parseResult.pages.forEach(page => {
const newBlocks = [];
var lastFootnote;
page.items.forEach(block => {
newBlocks.push(block);
if (!block.type && block.textItems[0].y < 200) {
const combineResult = textCombiner.combine(block.textItems);
if (combineResult.parsedElements.footnotes.length > 0) {
block.annotation = REMOVED_ANNOTATION;
foundFootnotes.push.apply(foundFootnotes, combineResult.parsedElements.footnotes);
lastFootnote = new TextItemBlock({
textItems: combineResult.textItems,
type: FOOTNOTE_BLOCK,
annotation: ADDED_ANNOTATION,
parsedElements: combineResult.parsedElements
});
newBlocks.push(lastFootnote);
} else if (lastFootnote) {
// likely to be the second line of aboves footnote
block.annotation = REMOVED_ANNOTATION;
lastFootnote.textItems = lastFootnote.textItems.concat(combineResult.textItems);
lastFootnote.parsedElements.add(combineResult.parsedElements);
newBlocks[newBlocks.length - 2] = block;
newBlocks[newBlocks.length - 1] = lastFootnote;
}
} else {
lastFootnote = null;
}
});
page.items = newBlocks;
});
return new ParseResult({
...parseResult,
messages: [
'Detected ' + foundFootnotes.length + ' footnotes:',
foundFootnotes.join(', ')
]
});
}
}