extern/pdf-to-markdown
1
0
Fork 0
mirror of https://github.com/jzillmann/pdf-to-markdown.git synced 2024-11-21 15:23:26 +01:00
Code Issues Packages Projects Releases Wiki Activity

[WIP] move different typed transformations to different folders

Browse Source
This commit is contained in:
Johannes Zillmann 2017-03-15 06:09:18 +01:00
parent 739d20d83b
commit 93f15a38b5
16 changed files with 78 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/javascript/components/DebugView.jsx
View File

@ -101,6 +101,18 @@ export default class DebugView extends React.Component {
</li>
});
const transformationMenuItems = [];
var lastItemType;
transformations.forEach((transformation, i) => {
if (lastItemType && transformation.itemType !== lastItemType) {
transformationMenuItems.push(<MenuItem key={ i + '-divider' } divider />);
}
transformationMenuItems.push(<MenuItem key={ i } eventKey={ i } onSelect={ this.selectTransformation.bind(this, i) }>
{ transformation.name }
</MenuItem>);
lastItemType = transformation.itemType;
});
return (
<div>
<table>
@ -149,9 +161,7 @@ export default class DebugView extends React.Component {
</ButtonGroup>
<ButtonGroup>
<DropdownButton title={ currentTransformationName } id="dropdown-size-medium">
{ transformations.map((transformation, i) => <MenuItem key={ i } eventKey={ i } onSelect={ this.selectTransformation.bind(this, i) }>
{ transformation.name }
</MenuItem>) }
{ transformationMenuItems }
</DropdownButton>
</ButtonGroup>
<ButtonGroup>

24
src/javascript/models/AppState.jsx
View File

@ -1,16 +1,17 @@
import { Enum } from 'enumify';
import CalculateGlobalStats from './transformations/CalculateGlobalStats.jsx';
import CompactLines from './transformations/CompactLines.jsx';
import RemoveRepetitiveElements from './transformations/RemoveRepetitiveElements.jsx'
import VerticalToHorizontal from './transformations/VerticalToHorizontal.jsx';
import DetectTOC from './transformations/DetectTOC.jsx'
import DetectListItems from './transformations/DetectListItems.jsx'
import CalculateGlobalStats from './transformations/textitem/CalculateGlobalStats.jsx';
import CompactLines from './transformations/textitem/CompactLines.jsx';
import RemoveRepetitiveElements from './transformations/textitem/RemoveRepetitiveElements.jsx'
import VerticalToHorizontal from './transformations/textitem/VerticalToHorizontal.jsx';
import DetectTOC from './transformations/textitem/DetectTOC.jsx'
import DetectListItems from './transformations/textitem/DetectListItems.jsx'
// import DetectHeaders from './transformations/textitem/DetectHeaders.jsx'
import GatherBlocks from './transformations/GatherBlocks.jsx'
import DetectCodeQuoteBlocks from './transformations/DetectCodeQuoteBlocks.jsx'
import DetectListLevels from './transformations/DetectListLevels.jsx'
import DetectHeadlines from './transformations/DetectHeadlines.jsx'
import GatherBlocks from './transformations/textitemblock/GatherBlocks.jsx'
import DetectCodeQuoteBlocks from './transformations/textitemblock/DetectCodeQuoteBlocks.jsx'
import DetectListLevels from './transformations/textitemblock/DetectListLevels.jsx'
// import DetectHeadlines from './transformations/textitemblock/DetectHeadlines.jsx'
// import DetectFormats from './transformations/DetectFormats.jsx'
// import RemoveWhitespaces from './transformations/RemoveWhitespaces.jsx'
// import DetectLinks from './transformations/DetectLinks.jsx'
@ -34,11 +35,12 @@ export default class AppState {
new VerticalToHorizontal(),
new DetectTOC(),
new DetectListItems(),
new DetectHeaders(),
new GatherBlocks(),
new DetectCodeQuoteBlocks(),
new DetectListLevels(),
new DetectHeadlines(),
// new DetectHeadlines(),
// new DetectFormats(),
// new RemoveWhitespaces(),

2
src/javascript/models/transformations/ToMarkdown.jsx
View File

@ -6,7 +6,7 @@ import ParseResult from '../ParseResult.jsx';
export default class ToMarkdown extends Transformation {
constructor() {
super("To Markdown");
super("To Markdown", "String");
}
createPageView(page, modificationsOnly) { // eslint-disable-line no-unused-vars

2
src/javascript/models/transformations/ToTextBlocks.jsx
View File

@ -7,7 +7,7 @@ import { blockToText } from '../ElementType.jsx';
export default class ToTextBlocks extends Transformation {
constructor() {
super("To Text Blocks");
super("To Text Blocks", "TextBlock");
}
createPageView(page, modificationsOnly) { // eslint-disable-line no-unused-vars

3
src/javascript/models/transformations/ToTextItemBlockTransformation.jsx
View File

@ -1,6 +1,7 @@
import React from 'react';
import Transformation from './Transformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItemBlock from '../TextItemBlock.jsx';
import TextItemBlockPageView from '../../components/debug/TextItemBlockPageView.jsx';
import { REMOVED_ANNOTATION } from '../Annotation.jsx';
@ -8,7 +9,7 @@ import { REMOVED_ANNOTATION } from '../Annotation.jsx';
export default class ToTextItemBlockTransformation extends Transformation {
constructor(name) {
super(name);
super(name, TextItemBlock.name);
if (this.constructor === ToTextItemBlockTransformation) {
throw new TypeError("Can not construct abstract class.");
}

3
src/javascript/models/transformations/ToTextItemTransformation.jsx
View File

@ -1,6 +1,7 @@
import React from 'react';
import Transformation from './Transformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItem from '../TextItem.jsx';
import TextItemPageView from '../../components/debug/TextItemPageView.jsx';
import { REMOVED_ANNOTATION } from '../Annotation.jsx';
@ -8,7 +9,7 @@ import { REMOVED_ANNOTATION } from '../Annotation.jsx';
export default class ToTextItemTransformation extends Transformation {
constructor(name) {
super(name);
super(name, TextItem.name);
if (this.constructor === ToTextItemTransformation) {
throw new TypeError("Can not construct abstract class.");
}

3
src/javascript/models/transformations/Transformation.jsx
View File

@ -3,7 +3,7 @@ import ParseResult from '../ParseResult.jsx';
// A transformation from an PdfPage to an PdfPage
export default class Transformation {
constructor(name) {
constructor(name, itemType) {
if (this.constructor === Transformation) {
throw new TypeError("Can not construct abstract class.");
}
@ -11,6 +11,7 @@ export default class Transformation {
throw new TypeError("Please implement abstract method 'transform()'.");
}
this.name = name;
this.itemType = itemType;
}
showPageSelection() {

4
src/javascript/models/transformations/CalculateGlobalStats.jsx → src/javascript/models/transformations/textitem/CalculateGlobalStats.jsx
View File

@ -1,5 +1,5 @@
import ToTextItemTransformation from './ToTextItemTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import ToTextItemTransformation from '../ToTextItemTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
export default class CalculateGlobalStats extends ToTextItemTransformation {

12
src/javascript/models/transformations/CompactLines.jsx → src/javascript/models/transformations/textitem/CompactLines.jsx
View File

@ -1,11 +1,11 @@
import React from 'react';
import ToTextItemTransformation from './ToTextItemTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItemLineGrouper from '../TextItemLineGrouper.jsx';
import TextItemLineCompactor from '../TextItemLineCompactor.jsx';
import ElementType from '../ElementType.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../Annotation.jsx';
import ToTextItemTransformation from '../ToTextItemTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import TextItemLineGrouper from '../../TextItemLineGrouper.jsx';
import TextItemLineCompactor from '../../TextItemLineCompactor.jsx';
import ElementType from '../../ElementType.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../../Annotation.jsx';
// gathers text items on the same y line to one text item

12
src/javascript/models/transformations/DetectListItems.jsx → src/javascript/models/transformations/textitem/DetectListItems.jsx
View File

@ -1,9 +1,9 @@
import ToTextItemTransformation from './ToTextItemTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItem from '../TextItem.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION, DETECTED_ANNOTATION } from '../Annotation.jsx';
import ElementType from '../ElementType.jsx';
import { isListItem, isNumberedListItem, removeLeadingWhitespaces } from '../../functions.jsx';
import ToTextItemTransformation from '../ToTextItemTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import TextItem from '../../TextItem.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION, DETECTED_ANNOTATION } from '../../Annotation.jsx';
import ElementType from '../../ElementType.jsx';
import { isListItem, isNumberedListItem, removeLeadingWhitespaces } from '../../../functions.jsx';
//Detect items starting with -, •, etc...
export default class DetectListItems extends ToTextItemTransformation {

16
src/javascript/models/transformations/DetectTOC.jsx → src/javascript/models/transformations/textitem/DetectTOC.jsx
View File

@ -1,11 +1,11 @@
import ToTextItemTransformation from './ToTextItemTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItem from '../TextItem.jsx';
import HeadlineFinder from '../HeadlineFinder.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../Annotation.jsx';
import ElementType from '../ElementType.jsx';
import { headlineByLevel } from '../ElementType.jsx';
import { isDigit } from '../../functions.jsx'
import ToTextItemTransformation from '../ToTextItemTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import TextItem from '../../TextItem.jsx';
import HeadlineFinder from '../../HeadlineFinder.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../../Annotation.jsx';
import ElementType from '../../ElementType.jsx';
import { headlineByLevel } from '../../ElementType.jsx';
import { isDigit } from '../../../functions.jsx'
//Detect table of contents pages
export default class DetectTOC extends ToTextItemTransformation {

8
src/javascript/models/transformations/RemoveRepetitiveElements.jsx → src/javascript/models/transformations/textitem/RemoveRepetitiveElements.jsx
View File

@ -1,8 +1,8 @@
import ToTextItemTransformation from './ToTextItemTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import { REMOVED_ANNOTATION } from '../Annotation.jsx';
import ToTextItemTransformation from '../ToTextItemTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import { REMOVED_ANNOTATION } from '../../Annotation.jsx';
import { isDigit } from '../../functions.jsx'
import { isDigit } from '../../../functions.jsx'
function hashCodeIgnoringSpacesAndNumbers(string) {

8
src/javascript/models/transformations/VerticalToHorizontal.jsx → src/javascript/models/transformations/textitem/VerticalToHorizontal.jsx
View File

@ -1,7 +1,7 @@
import ToTextItemTransformation from './ToTextItemTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItem from '../TextItem.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../Annotation.jsx';
import ToTextItemTransformation from '../ToTextItemTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import TextItem from '../../TextItem.jsx';
import { REMOVED_ANNOTATION, ADDED_ANNOTATION } from '../../Annotation.jsx';
// Converts vertical text to horizontal
export default class VerticalToHorizontal extends ToTextItemTransformation {

10
src/javascript/models/transformations/DetectCodeQuoteBlocks.jsx → src/javascript/models/transformations/textitemblock/DetectCodeQuoteBlocks.jsx
View File

@ -1,8 +1,8 @@
import ToTextItemBlockTransformation from './ToTextItemBlockTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import { DETECTED_ANNOTATION } from '../Annotation.jsx';
import ElementType from '../ElementType.jsx';
import { minXFromBlocks } from '../../textItemFunctions.jsx';
import ToTextItemBlockTransformation from '..//ToTextItemBlockTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import { DETECTED_ANNOTATION } from '../../Annotation.jsx';
import ElementType from '../../ElementType.jsx';
import { minXFromBlocks } from '../../../textItemFunctions.jsx';
//Detect items which are code/quote blocks
export default class DetectCodeQuoteBlocks extends ToTextItemBlockTransformation {

8
src/javascript/models/transformations/DetectListLevels.jsx → src/javascript/models/transformations/textitemblock/DetectListLevels.jsx
View File

@ -1,7 +1,7 @@
import ToTextItemBlockTransformation from './ToTextItemBlockTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import { MODIFIED_ANNOTATION, UNCHANGED_ANNOTATION } from '../Annotation.jsx';
import ElementType from '../ElementType.jsx';
import ToTextItemBlockTransformation from '..//ToTextItemBlockTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import { MODIFIED_ANNOTATION, UNCHANGED_ANNOTATION } from '../../Annotation.jsx';
import ElementType from '../../ElementType.jsx';
// Cares for proper sub-item spacing/leveling
export default class DetectListLevels extends ToTextItemBlockTransformation {

10
src/javascript/models/transformations/GatherBlocks.jsx → src/javascript/models/transformations/textitemblock/GatherBlocks.jsx
View File

@ -1,8 +1,8 @@
import ToTextItemBlockTransformation from './ToTextItemBlockTransformation.jsx';
import ParseResult from '../ParseResult.jsx';
import TextItemBlock from '../TextItemBlock.jsx';
import { DETECTED_ANNOTATION } from '../Annotation.jsx';
import { minXFromTextItems } from '../../textItemFunctions.jsx';
import ToTextItemBlockTransformation from '../ToTextItemBlockTransformation.jsx';
import ParseResult from '../../ParseResult.jsx';
import TextItemBlock from '../../TextItemBlock.jsx';
import { DETECTED_ANNOTATION } from '../../Annotation.jsx';
import { minXFromTextItems } from '../../../textItemFunctions.jsx';
// Gathers lines to blocks
export default class GatherBlocks extends ToTextItemBlockTransformation {