extern/pdf-to-markdown
1
0
Fork 0
mirror of https://github.com/jzillmann/pdf-to-markdown.git synced 2024-11-25 01:03:59 +01:00
Code Issues Packages Projects Releases Wiki Activity

Fix loading when PDF has non-resolvable fonts

Browse Source 
* Sometimes pdf.js gave fontId like Helvetica & Times instead of g_d0_f1, etc…, but those never got resolved through the callback. Now we simply ignore those.
* Also fixed that no fonts could be parsed
This commit is contained in:
Johannes Zillmann 2017-03-29 08:11:14 +02:00
parent c4c23ac6ee
commit 96d4f72889
1 changed files with 25 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
src/javascript/components/LoadingView.jsx
View File

@ -23,7 +23,7 @@ export default class LoadingView extends React.Component {
stages: [ stages: [
new ProgressStage('Parsing Metadata', 2), new ProgressStage('Parsing Metadata', 2),
new ProgressStage('Parsing Pages'), new ProgressStage('Parsing Pages'),
new ProgressStage('Parsing Fonts') new ProgressStage('Parsing Fonts', 0)
] ]
}); });
Progress.prototype.metadataStage = () => { Progress.prototype.metadataStage = () => {
@ -78,18 +78,6 @@ export default class LoadingView extends React.Component {
pageParsed(index, textItems) { pageParsed(index, textItems) {
const pageStage = this.state.progress.pageStage(); const pageStage = this.state.progress.pageStage();
const fontStage = this.state.progress.fontStage();
const self = this;
textItems.forEach(item => {
const fontId = item.font;
if (!this.state.fontIds.has(fontId)) {
this.state.document.transport.commonObjs.get(fontId, function(font) {
self.fontParsed(fontId, font);
});
this.state.fontIds.add(fontId);
fontStage.steps = this.state.fontIds.size;
}
});
pageStage.stepsDone = pageStage.stepsDone + 1; pageStage.stepsDone = pageStage.stepsDone + 1;
this.state.pages[index].items = textItems; // eslint-disable-line react/no-direct-mutation-state this.state.pages[index].items = textItems; // eslint-disable-line react/no-direct-mutation-state
@ -111,6 +99,8 @@ export default class LoadingView extends React.Component {
componentWillMount() { componentWillMount() {
const self = this; const self = this;
const fontStage = this.state.progress.fontStage();
PDFJS.getDocument(this.props.fileBuffer).then(function(pdfDocument) { // eslint-disable-line no-undef PDFJS.getDocument(this.props.fileBuffer).then(function(pdfDocument) { // eslint-disable-line no-undef
// console.debug(pdfDocument); // console.debug(pdfDocument);
pdfDocument.getMetadata().then(function(metadata) { pdfDocument.getMetadata().then(function(metadata) {
@ -127,6 +117,16 @@ export default class LoadingView extends React.Component {
page.getTextContent().then(function(textContent) { page.getTextContent().then(function(textContent) {
// console.debug(textContent); // console.debug(textContent);
const textItems = textContent.items.map(function(item) { const textItems = textContent.items.map(function(item) {
//trigger resolving of fonts
const fontId = item.fontName;
if (!self.state.fontIds.has(fontId) && fontId.startsWith('g_d0')) {
self.state.document.transport.commonObjs.get(fontId, function(font) {
self.fontParsed(fontId, font);
});
self.state.fontIds.add(fontId);
fontStage.steps = self.state.fontIds.size;
}
const tx = PDFJS.Util.transform( // eslint-disable-line no-undef const tx = PDFJS.Util.transform( // eslint-disable-line no-undef
viewport.transform, viewport.transform,
item.transform item.transform
@ -179,6 +179,12 @@ export default class LoadingView extends React.Component {
<div> <div>
{ stageItems } { stageItems }
</div> </div>
<br/>
<br/>
<br/>
<br/>
<br/>
<br/>
</div>); </div>);
} }
} }
@ -231,9 +237,14 @@ class ProgressStage {
} }
percentDone() { percentDone() {
if (!this.steps) { if (typeof this.steps === 'undefined') {
// if (!this.steps) {
return 0; return 0;
} }
if (this.steps == 0) {
return 100;
}
return this.stepsDone / this.steps * 100; return this.stepsDone / this.steps * 100;
} }
} }