extern/pdf-to-markdown
1
0
Fork 0
mirror of https://github.com/jzillmann/pdf-to-markdown.git synced 2025-06-24 19:41:24 +02:00
Code Issues Packages Projects Releases Wiki Activity

Test for remote PDFS

Browse Source
This commit is contained in:
Johannes Zillmann 2021-03-22 09:03:26 +01:00
parent f5a180113d
commit c98145a63c
7 changed files with 194 additions and 49 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@ -1,3 +1,4 @@
.DS_Store
node_modules/
build/
npm-debug.log

2
core/src/Debugger.ts
View File

@ -33,7 +33,7 @@ export default class Debugger {
this.stageResultCache = [initialStage(inputSchema, inputItems)];
}
stageResults(stageIndex: number): StageResult {
stageResult(stageIndex: number): StageResult {
for (let idx = 0; idx < stageIndex + 1; idx++) {
if (!this.stageResultCache[idx]) {
const transformer = this.transformers[idx - 1];

32
core/test/Debugger.test.ts
View File

@ -39,14 +39,14 @@ describe('Transform Items', () => {
const debug = new Debugger(1, parsedSchema, parsedItems, { fontMap: new Map(), pageViewports: [] }, transformers);
expect(debug.stageNames).toEqual(['Parse Result', 'Trans1']);
expect(debug.stageResults(0).schema).toEqual(parsedSchema.map((column) => ({ name: column })));
expect(debug.stageResults(1).schema).toEqual([
expect(debug.stageResult(0).schema).toEqual(parsedSchema.map((column) => ({ name: column })));
expect(debug.stageResult(1).schema).toEqual([
...parsedSchema.map((column) => ({ name: column, annotation: ColumnAnnotation.REMOVED })),
{ name: 'C', annotation: ColumnAnnotation.ADDED },
]);
expect(debug.stageResults(0).itemsUnpacked()).toEqual(parsedItems);
expect(debug.stageResults(1).itemsUnpacked()).toEqual(trans1Items);
expect(debug.stageResult(0).itemsUnpacked()).toEqual(parsedItems);
expect(debug.stageResult(1).itemsUnpacked()).toEqual(trans1Items);
});
test('Line Merge', async () => {
@ -65,17 +65,17 @@ describe('Transform Items', () => {
const debug = new Debugger(1, parsedSchema, parsedItems, { fontMap: new Map(), pageViewports: [] }, transformers);
expect(debug.stageNames).toEqual(['Parse Result', 'Trans1']);
expect(debug.stageResults(0).schema).toEqual([{ name: 'id' }, { name: 'y' }]);
expect(debug.stageResults(1).schema).toEqual([
expect(debug.stageResult(0).schema).toEqual([{ name: 'id' }, { name: 'y' }]);
expect(debug.stageResult(1).schema).toEqual([
{ name: 'id' },
{ name: 'y', annotation: ColumnAnnotation.REMOVED },
{ name: 'line', annotation: ColumnAnnotation.ADDED },
]);
expect(debug.stageResults(0).itemsUnpacked()).toEqual(parsedItems);
expect(debug.stageResults(1).itemsUnpacked()).toEqual(trans1Items);
expect(debug.stageResult(0).itemsUnpacked()).toEqual(parsedItems);
expect(debug.stageResult(1).itemsUnpacked()).toEqual(trans1Items);
const lineMergingStage = debug.stageResults(1);
const lineMergingStage = debug.stageResult(1);
const { changes, pages } = lineMergingStage;
//verify item groups
@ -103,12 +103,12 @@ test('Change inside of Line', async () => {
const debug = new Debugger(1, parsedSchema, parsedItems, { fontMap: new Map(), pageViewports: [] }, transformers);
expect(debug.stageNames).toEqual(['Parse Result', 'Trans1']);
expect(debug.stageResults(0).schema).toEqual([{ name: 'id' }, { name: 'line' }]);
expect(debug.stageResults(1).schema).toEqual([{ name: 'id' }, { name: 'line' }]);
expect(debug.stageResults(0).itemsUnpacked()).toEqual(parsedItems);
expect(debug.stageResults(1).itemsUnpacked()).toEqual(trans1Items);
expect(debug.stageResult(0).schema).toEqual([{ name: 'id' }, { name: 'line' }]);
expect(debug.stageResult(1).schema).toEqual([{ name: 'id' }, { name: 'line' }]);
expect(debug.stageResult(0).itemsUnpacked()).toEqual(parsedItems);
expect(debug.stageResult(1).itemsUnpacked()).toEqual(trans1Items);
const { changes, pages } = debug.stageResults(1);
const { changes, pages } = debug.stageResult(1);
//verify item groups
expect(pages[0].itemGroups.map((itemGroup) => changes.hasChanged(itemGroup.top))).toEqual([true, false]);
@ -116,7 +116,7 @@ test('Change inside of Line', async () => {
//verify unpacked items
expect(
debug
.stageResults(1)
.stageResult(1)
.itemsUnpacked()
.map((item) => changes.hasChanged(item)),
).toEqual([true, true, false, false]);
@ -135,7 +135,7 @@ describe('build schemas', () => {
function calculateSchema(inputSchema: string[], outputSchema: string[]): AnnotatedColumn[] {
const transformers = [new TestTransformer('Trans1', {}, outputSchema, items)];
const debug = new Debugger(1, inputSchema, items, { fontMap: new Map(), pageViewports: [] }, transformers);
return debug.stageResults(1).schema;
return debug.stageResult(1).schema;
}
test('Add', async () => {

143
core/test/Files.test.ts
View File

@ -2,21 +2,34 @@ import { toMatchFile } from 'jest-file-snapshot';
import * as pdfjs from 'pdfjs-dist/es5/build/pdf';
import * as fs from 'fs';
import * as path from 'path';
import * as http from 'http';
import * as https from 'https';
import PdfParser from 'src/PdfParser';
import PdfPipeline from 'src/PdfPipeline';
import { transformers } from 'src/index';
import Debugger from 'src/Debugger';
import Item from 'src/Item';
import RemoveRepetitiveItems from 'src/transformer/RemoveRepetitiveItems';
import StageResult from 'src/debug/StageResult';
const parser = new PdfParser(pdfjs);
const pipeline = new PdfPipeline(parser, transformers);
const folder = '../examples';
const files = fs.readdirSync(folder).filter((file) => file.endsWith('.pdf'));
const urls = ['https://homepages.cwi.nl/~lex/files/dict.pdf'];
const downloadCache = 'node_modules/.cache/files';
expect.extend({ toMatchFile });
// Test is for debugging purpose
test.skip('Debug', async () => {
const data = fs.readFileSync(`${folder}/Adventures-Of-Sherlock-Holmes.pdf`);
await pipeline.execute(data, () => {});
});
describe.each(files)('Test %p', (file) => {
const data = fs.readFileSync(`${folder}/${file}`);
@ -26,15 +39,15 @@ describe.each(files)('Test %p', (file) => {
test.each(transformers.map((t) => t.name).filter((name) => name !== 'Does nothing'))(
'stage %p',
(transformerName) => {
const stageResults = debug.stageResults(debug.stageNames.indexOf(transformerName));
const stageResult = debug.stageResult(debug.stageNames.indexOf(transformerName));
const chunkedLines: string[][] = [[]];
let resultIndex = 0;
let collectedItems = 0;
stageResults.selectPages(true, true).forEach((page) => {
stageResult.selectPages(true, true).forEach((page) => {
page.itemGroups.forEach((itemGroup) => {
const change = stageResults.changes.change(itemGroup.top);
if (change || stageResults.descriptor.debug?.showAll) {
const change = stageResult.changes.change(itemGroup.top);
if (change || stageResult.descriptor.debug?.showAll) {
const item = itemGroup.top;
const changeType = change?.constructor.name || 'none';
chunkedLines[resultIndex].push(itemToString(debug.fontMap, item, changeType));
@ -50,36 +63,70 @@ describe.each(files)('Test %p', (file) => {
});
// Global characteristics
let groupedItemCount = stageResults
.selectPages(false, true)
.reduce((itemCount, page) => itemCount + page.itemGroups.length, 0);
chunkedLines[0].unshift(
JSON.stringify(
{
pages: stageResults.pages.length,
items: stageResults.itemsUnpacked().length,
groupedItems: groupedItemCount,
changes: stageResults.changes.changeCount(),
schema: stageResults.schema,
// messages: stageResults.messages,
},
null,
2,
),
);
chunkedLines[0].unshift(toHeader(stageResult));
chunkedLines.forEach((lines, idx) => {
const transformerResultAsString = lines.join('\n') || '{}';
const resultFolder = `${folder}/${file.substr(0, file.length - 4)}`;
const fileName = `${transformerName[0].toLowerCase() + transformerName.slice(1).replace(/\s/g, '')}`;
const fileIndex = chunkedLines.length > 1 ? `.${idx}` : '';
const resultFile = `${resultFolder}/${fileName}${fileIndex}.json`;
expect(transformerResultAsString).toMatchFile(resultFile);
expect(transformerResultAsString).toMatchFile(matchFilePath(file, transformerName, chunkedLines.length, idx));
});
},
);
});
function matchFilePath(pdfFileName: string, transformerName: string, chunkCount = 1, chunkIndex = 0): string {
const pdfFileNameWithoutExtension = pdfFileName.substr(0, pdfFileName.length - 4);
const resultFileName = `${transformerName[0].toLowerCase() + transformerName.slice(1).replace(/\s/g, '')}`;
const fileIndex = chunkCount > 1 ? `.${chunkIndex}` : '';
return `${folder}/${pdfFileNameWithoutExtension}/${resultFileName}${fileIndex}.json`;
}
describe('Remove repetitive items from online resources', () => {
const transformerName = new RemoveRepetitiveItems().name;
test.each(urls)('URL %p', async (url) => {
console.log(url);
const { fileName, data } = download(url);
const debug = await pipeline.debug(data, () => {});
const stageResult = debug.stageResult(debug.stageNames.indexOf(transformerName));
const pages = stageResult.selectPages(true, true);
const lines: string[] = [];
lines.push(toHeader(stageResult));
pages.forEach((page) =>
page.itemGroups.forEach((itemGroup) => {
const change = stageResult.changes.change(itemGroup.top);
if (change) {
const item = itemGroup.top;
const changeType = change?.constructor.name || 'none';
lines.push(itemToString(debug.fontMap, item, changeType));
}
}),
);
console.log(lines);
const transformerResultAsString = lines.join('\n') || '{}';
expect(transformerResultAsString).toMatchFile(matchFilePath(fileName, transformerName));
});
});
function toHeader(stageResult: StageResult): string {
let groupedItemCount = stageResult
.selectPages(false, true)
.reduce((itemCount, page) => itemCount + page.itemGroups.length, 0);
return JSON.stringify(
{
pages: stageResult.pages.length,
items: stageResult.itemsUnpacked().length,
groupedItems: groupedItemCount,
changes: stageResult.changes.changeCount(),
schema: stageResult.schema,
// messages: stageResults.messages,
},
null,
2,
);
}
function itemToString(fontMap: Map<string, object>, item: Item, changeType: string): string {
const fontName: string | Array<string> = item.data['fontName'];
let newFontName: string | Array<string> | undefined = undefined;
@ -91,7 +138,7 @@ function itemToString(fontMap: Map<string, object>, item: Item, changeType: stri
}
}
const transform: undefined | number[] = item.data['transform'];
let newTransform;
let newTransform: undefined | string[];
if (transform) {
newTransform = transform.map((num) => num.toFixed(2));
}
@ -105,3 +152,45 @@ function itemToString(fontMap: Map<string, object>, item: Item, changeType: stri
transform: newTransform,
});
}
function download(url: string): { fileName: string; data: Buffer } {
const fileName = path.basename(new URL(url).pathname);
const localFilePath = `${downloadCache}/${fileName}`;
console.log(localFilePath);
if (!fs.existsSync(localFilePath)) {
fs.mkdirSync(downloadCache, { recursive: true });
downloadToFile(url, localFilePath);
}
return {
fileName,
data: fs.readFileSync(localFilePath),
};
}
function downloadToFile(url: string, dest: string): Promise<void> {
const uri = new URL(url);
const pkg = url.toLowerCase().startsWith('https:') ? https : http;
return new Promise((resolve, reject) => {
pkg.get(uri.href).on('response', (res) => {
if (res.statusCode === 200) {
const file = fs.createWriteStream(dest, { flags: 'wx' });
res
.on('end', () => {
file.end();
resolve();
})
.on('error', (err) => {
file.destroy();
fs.unlink(dest, () => reject(err));
})
.pipe(file);
} else if (res.statusCode === 302 || res.statusCode === 301) {
// Recursively follow redirects, only a 200 will resolve.
downloadToFile(res.headers.location as string, dest).then(() => resolve());
} else {
reject(new Error(`Download request failed, response status: ${res.statusCode} ${res.statusMessage}`));
}
});
});
}

29
examples/README.md
View File

@ -1,10 +1,21 @@
These PDFs are used in the parsers's test suite.
# Test PDFs
Self-generated files are:
This folder contains PDFs for testing purposes and the parse results of the PDFs. Generally there are 3 types of PDFs test setups:
1. Self generated PDFs
2. PDFs which entered `public domain` or have a otherwise permissive license like `Creative Commons SA`
3. PDFs where the license is unclear
For (1) and (2) we track the end-result and all transformation steps.
For (3) we only track the resulst of some transfomation stages (those who doesn't leak too much of the content)
## Self-generated PDFs
- [ExamplePdf](ExamplePdf.pdf)
All other PDFs are either entered `public domain` or have a otherwise permissive license like `Creative Commons SA`.
## Included Public PDFs
_(PDFs which entered `public domain` or have a otherwise permissive license like `Creative Commons SA`)_
| File | Source | Author | License Information |
| ------------------------------------------------------------------ | ------------------------------------- | ------------------------------------------------ | ------------------------- |
@ -21,3 +32,15 @@ All other PDFs are either entered `public domain` or have a otherwise permissive
| [The-War-of-the-Worlds](The-War-of-the-Worlds.pdf) | http://www.planetpdf.com/ | H.G Wells | Public Domain |
| [Tragedy-Of-The-Commons](Tragedy-Of-The-Commons.pdf) | https://science.sciencemag.org | Garrett Hardin | Public Domain |
| [WoodUp](WoodUp.pdf) | https://bupress.unibz.it/ | Freie Universität Bozen-Bolzano / Giustino Tonon | Creative Commons BY 4.0 |
## PDFs not stored but paritally tested
- https://homepages.cwi.nl/~lex/files/dict.pdf
# Known transformatino problems
_Tracks known problems with parsing and transforming certain PDFs ._
- `Remove Repetitive Elements`
- https://homepages.cwi.nl/~lex/files/dict.pdf
- Nothing gets detected cause the page-number line contains the current chapter

32
examples/dict/removeRepetitiveItems.json Normal file
View File

@ -0,0 +1,32 @@
{
"pages": 221,
"items": 51638,
"groupedItems": 8465,
"changes": 0,
"schema": [
{
"name": "line"
},
{
"name": "x"
},
{
"name": "y"
},
{
"name": "str"
},
{
"name": "fontName"
},
{
"name": "dir"
},
{
"name": "width"
},
{
"name": "height"
}
]
}

2
ui/src/debug/DebugView.svelte
View File

@ -18,7 +18,7 @@
let groupingEnabled = true;
let onlyRelevantItems = true;
$: stageResult = debug.stageResults($debugStage);
$: stageResult = debug.stageResult($debugStage);
$: supportsGrouping = !!stageResult.descriptor?.debug?.itemMerger;
$: supportsRelevanceFiltering = !stageResult.descriptor?.debug?.showAll;
$: visiblePages = pageControl.selectPages(stageResult, onlyRelevantItems, groupingEnabled, $pinnedPageIndex);